Merge branch 'main' of https://github.com/embassy-rs/embassy into fix_main

139 files changed, 3168 insertions, 1559 deletions

diff --git a/docs/pages/faq.adoc b/docs/pages/faq.adoc
index bb088dfb6..a2f56a539 100644
--- a/docs/pages/faq.adoc
+++ b/docs/pages/faq.adoc
@@ -299,7 +299,7 @@ You can add a section to your `Cargo.toml` file like this, you'll need to patch
 
 Using `patch` will replace all direct AND indirect dependencies.
 
-See the link:https://embassy.dev/book/dev/new_project.html#_cargo_toml[new project docs] for more details on this approach.
+See the link:https://embassy.dev/book/#_starting_a_new_project[new project docs] for more details on this approach.
 
 [source,toml]
 ----
diff --git a/docs/pages/getting_started.adoc b/docs/pages/getting_started.adoc
index 465059922..77275a857 100644
--- a/docs/pages/getting_started.adoc
+++ b/docs/pages/getting_started.adoc
@@ -131,7 +131,7 @@ If you’re using a raspberry pi pico-w, make sure you’re running `+cargo run
 
 If you’re using an rp2040 debug probe (e.g. the pico probe) and are having issues after running `probe-rs info`, unplug and reconnect the probe, letting it power cycle. Running `probe-rs info` is link:https://github.com/probe-rs/probe-rs/issues/1849[known to put the pico probe into an unusable state].
 
-If you’re still having problems, check the link:https://embassy.dev/book/dev/faq.html[FAQ], or ask for help in the link:https://matrix.to/#/#embassy-rs:matrix.org[Embassy Chat Room].
+If you’re still having problems, check the link:https://embassy.dev/book/#_frequently_asked_questions[FAQ], or ask for help in the link:https://matrix.to/#/#embassy-rs:matrix.org[Embassy Chat Room].
 
 == What's next?
 
diff --git a/embassy-boot-nrf/Cargo.toml b/embassy-boot-nrf/Cargo.toml
index e6bb27043..86bfc21f4 100644
--- a/embassy-boot-nrf/Cargo.toml
+++ b/embassy-boot-nrf/Cargo.toml
@@ -22,6 +22,7 @@ target = "thumbv7em-none-eabi"
 
 [dependencies]
 defmt = { version = "0.3", optional = true }
+log = { version = "0.4.17", optional = true }
 
 embassy-sync = { version = "0.6.0", path = "../embassy-sync" }
 embassy-nrf = { version = "0.1.0", path = "../embassy-nrf", default-features = false }
diff --git a/embassy-boot-rp/Cargo.toml b/embassy-boot-rp/Cargo.toml
index 0bd8abf4c..de2ae3780 100644
--- a/embassy-boot-rp/Cargo.toml
+++ b/embassy-boot-rp/Cargo.toml
@@ -21,7 +21,6 @@ target = "thumbv6m-none-eabi"
 
 [dependencies]
 defmt = { version = "0.3", optional = true }
-defmt-rtt = { version = "0.4", optional = true }
 log = { version = "0.4", optional = true }
 
 embassy-sync = { version = "0.6.0", path = "../embassy-sync" }
@@ -46,7 +45,6 @@ log = [
     "embassy-boot/log",
     "embassy-rp/log",
 ]
-debug = ["defmt-rtt"]
 
 [profile.dev]
 debug = 2
diff --git a/embassy-boot-rp/build.rs b/embassy-boot-rp/build.rs
index 2cbc7ef5e..bfaee3503 100644
--- a/embassy-boot-rp/build.rs
+++ b/embassy-boot-rp/build.rs
@@ -5,4 +5,5 @@ fn main() {
     if target.starts_with("thumbv6m-") {
         println!("cargo:rustc-cfg=armv6m");
     }
+    println!("cargo:rustc-check-cfg=cfg(armv6m)");
 }
diff --git a/embassy-boot-stm32/Cargo.toml b/embassy-boot-stm32/Cargo.toml
index a3661f6cb..d4525a6de 100644
--- a/embassy-boot-stm32/Cargo.toml
+++ b/embassy-boot-stm32/Cargo.toml
@@ -22,7 +22,6 @@ target = "thumbv7em-none-eabi"
 
 [dependencies]
 defmt = { version = "0.3", optional = true }
-defmt-rtt = { version = "0.4", optional = true }
 log = { version = "0.4", optional = true }
 
 embassy-sync = { version = "0.6.0", path = "../embassy-sync" }
@@ -37,7 +36,6 @@ cfg-if = "1.0.0"
 [features]
 defmt = ["dep:defmt", "embassy-boot/defmt", "embassy-stm32/defmt"]
 log = ["dep:log", "embassy-boot/log", "embassy-stm32/log"]
-debug = ["defmt-rtt"]
 
 [profile.dev]
 debug = 2
diff --git a/embassy-boot-stm32/build.rs b/embassy-boot-stm32/build.rs
index 2cbc7ef5e..bfaee3503 100644
--- a/embassy-boot-stm32/build.rs
+++ b/embassy-boot-stm32/build.rs
@@ -5,4 +5,5 @@ fn main() {
     if target.starts_with("thumbv6m-") {
         println!("cargo:rustc-cfg=armv6m");
     }
+    println!("cargo:rustc-check-cfg=cfg(armv6m)");
 }
diff --git a/embassy-boot/README.md b/embassy-boot/README.md
index 812c43524..c893c83e1 100644
--- a/embassy-boot/README.md
+++ b/embassy-boot/README.md
@@ -24,7 +24,7 @@ For any partition, the following preconditions are required:
 
 The linker scripts for the application and bootloader look similar, but the FLASH region must point to the BOOTLOADER partition for the bootloader, and the ACTIVE partition for the application.
 
-For more details on the bootloader, see [the documentation](https://embassy.dev/book/dev/bootloader.html).
+For more details on the bootloader, see [the documentation](https://embassy.dev/book/#_bootloader).
 
 ## Hardware support
 
diff --git a/embassy-executor/build.rs b/embassy-executor/build.rs
index 07f31e3fb..8a41d7503 100644
--- a/embassy-executor/build.rs
+++ b/embassy-executor/build.rs
@@ -3,6 +3,9 @@ use std::fmt::Write;
 use std::path::PathBuf;
 use std::{env, fs};
 
+#[path = "./build_common.rs"]
+mod common;
+
 static CONFIGS: &[(&str, usize)] = &[
     // BEGIN AUTOGENERATED CONFIG FEATURES
     // Generated by gen_config.py. DO NOT EDIT.
@@ -91,30 +94,6 @@ fn main() {
     let out_file = out_dir.join("config.rs").to_string_lossy().to_string();
     fs::write(out_file, data).unwrap();
 
-    // cortex-m targets
-    let target = env::var("TARGET").unwrap();
-
-    if target.starts_with("thumbv6m-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv6m");
-    } else if target.starts_with("thumbv7m-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv7m");
-    } else if target.starts_with("thumbv7em-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv7m");
-        println!("cargo:rustc-cfg=armv7em"); // (not currently used)
-    } else if target.starts_with("thumbv8m.base") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv8m");
-        println!("cargo:rustc-cfg=armv8m_base");
-    } else if target.starts_with("thumbv8m.main") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv8m");
-        println!("cargo:rustc-cfg=armv8m_main");
-    }
-
-    if target.ends_with("-eabihf") {
-        println!("cargo:rustc-cfg=has_fpu");
-    }
+    let mut rustc_cfgs = common::CfgSet::new();
+    common::set_target_cfgs(&mut rustc_cfgs);
 }
diff --git a/embassy-executor/build_common.rs b/embassy-executor/build_common.rs
new file mode 100644
index 000000000..2c65f8529
--- /dev/null
+++ b/embassy-executor/build_common.rs
@@ -0,0 +1,109 @@
+// NOTE: this file is copy-pasted between several Embassy crates, because there is no
+// straightforward way to share this code:
+// - it cannot be placed into the root of the repo and linked from each build.rs using `#[path =
+// "../build_common.rs"]`, because `cargo publish` requires that all files published with a crate
+// reside in the crate's directory,
+// - it cannot be symlinked from `embassy-xxx/build_common.rs` to `../build_common.rs`, because
+// symlinks don't work on Windows.
+
+use std::collections::HashSet;
+use std::env;
+use std::ffi::OsString;
+use std::process::Command;
+
+/// Helper for emitting cargo instruction for enabling configs (`cargo:rustc-cfg=X`) and declaring
+/// them (`cargo:rust-check-cfg=cfg(X)`).
+#[derive(Debug)]
+pub struct CfgSet {
+    enabled: HashSet<String>,
+    declared: HashSet<String>,
+    emit_declared: bool,
+}
+
+impl CfgSet {
+    pub fn new() -> Self {
+        Self {
+            enabled: HashSet::new(),
+            declared: HashSet::new(),
+            emit_declared: is_rustc_nightly(),
+        }
+    }
+
+    /// Enable a config, which can then be used in `#[cfg(...)]` for conditional compilation.
+    ///
+    /// All configs that can potentially be enabled should be unconditionally declared using
+    /// [`Self::declare()`].
+    pub fn enable(&mut self, cfg: impl AsRef<str>) {
+        if self.enabled.insert(cfg.as_ref().to_owned()) {
+            println!("cargo:rustc-cfg={}", cfg.as_ref());
+        }
+    }
+
+    pub fn enable_all(&mut self, cfgs: &[impl AsRef<str>]) {
+        for cfg in cfgs.iter() {
+            self.enable(cfg.as_ref());
+        }
+    }
+
+    /// Declare a valid config for conditional compilation, without enabling it.
+    ///
+    /// This enables rustc to check that the configs in `#[cfg(...)]` attributes are valid.
+    pub fn declare(&mut self, cfg: impl AsRef<str>) {
+        if self.declared.insert(cfg.as_ref().to_owned()) && self.emit_declared {
+            println!("cargo:rustc-check-cfg=cfg({})", cfg.as_ref());
+        }
+    }
+
+    pub fn declare_all(&mut self, cfgs: &[impl AsRef<str>]) {
+        for cfg in cfgs.iter() {
+            self.declare(cfg.as_ref());
+        }
+    }
+
+    pub fn set(&mut self, cfg: impl Into<String>, enable: bool) {
+        let cfg = cfg.into();
+        if enable {
+            self.enable(cfg.clone());
+        }
+        self.declare(cfg);
+    }
+}
+
+fn is_rustc_nightly() -> bool {
+    let rustc = env::var_os("RUSTC").unwrap_or_else(|| OsString::from("rustc"));
+
+    let output = Command::new(rustc)
+        .arg("--version")
+        .output()
+        .expect("failed to run `rustc --version`");
+
+    String::from_utf8_lossy(&output.stdout).contains("nightly")
+}
+
+/// Sets configs that describe the target platform.
+pub fn set_target_cfgs(cfgs: &mut CfgSet) {
+    let target = env::var("TARGET").unwrap();
+
+    if target.starts_with("thumbv6m-") {
+        cfgs.enable_all(&["cortex_m", "armv6m"]);
+    } else if target.starts_with("thumbv7m-") {
+        cfgs.enable_all(&["cortex_m", "armv7m"]);
+    } else if target.starts_with("thumbv7em-") {
+        cfgs.enable_all(&["cortex_m", "armv7m", "armv7em"]);
+    } else if target.starts_with("thumbv8m.base") {
+        cfgs.enable_all(&["cortex_m", "armv8m", "armv8m_base"]);
+    } else if target.starts_with("thumbv8m.main") {
+        cfgs.enable_all(&["cortex_m", "armv8m", "armv8m_main"]);
+    }
+    cfgs.declare_all(&[
+        "cortex_m",
+        "armv6m",
+        "armv7m",
+        "armv7em",
+        "armv8m",
+        "armv8m_base",
+        "armv8m_main",
+    ]);
+
+    cfgs.set("has_fpu", target.ends_with("-eabihf"));
+}
diff --git a/embassy-hal-internal/build.rs b/embassy-hal-internal/build.rs
index 6fe82b44f..ecd2c0c9f 100644
--- a/embassy-hal-internal/build.rs
+++ b/embassy-hal-internal/build.rs
@@ -1,29 +1,7 @@
-use std::env;
+#[path = "./build_common.rs"]
+mod common;
 
 fn main() {
-    let target = env::var("TARGET").unwrap();
-
-    if target.starts_with("thumbv6m-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv6m");
-    } else if target.starts_with("thumbv7m-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv7m");
-    } else if target.starts_with("thumbv7em-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv7m");
-        println!("cargo:rustc-cfg=armv7em"); // (not currently used)
-    } else if target.starts_with("thumbv8m.base") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv8m");
-        println!("cargo:rustc-cfg=armv8m_base");
-    } else if target.starts_with("thumbv8m.main") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv8m");
-        println!("cargo:rustc-cfg=armv8m_main");
-    }
-
-    if target.ends_with("-eabihf") {
-        println!("cargo:rustc-cfg=has_fpu");
-    }
+    let mut cfgs = common::CfgSet::new();
+    common::set_target_cfgs(&mut cfgs);
 }
diff --git a/embassy-hal-internal/build_common.rs b/embassy-hal-internal/build_common.rs
new file mode 100644
index 000000000..2c65f8529
--- /dev/null
+++ b/embassy-hal-internal/build_common.rs
@@ -0,0 +1,109 @@
+// NOTE: this file is copy-pasted between several Embassy crates, because there is no
+// straightforward way to share this code:
+// - it cannot be placed into the root of the repo and linked from each build.rs using `#[path =
+// "../build_common.rs"]`, because `cargo publish` requires that all files published with a crate
+// reside in the crate's directory,
+// - it cannot be symlinked from `embassy-xxx/build_common.rs` to `../build_common.rs`, because
+// symlinks don't work on Windows.
+
+use std::collections::HashSet;
+use std::env;
+use std::ffi::OsString;
+use std::process::Command;
+
+/// Helper for emitting cargo instruction for enabling configs (`cargo:rustc-cfg=X`) and declaring
+/// them (`cargo:rust-check-cfg=cfg(X)`).
+#[derive(Debug)]
+pub struct CfgSet {
+    enabled: HashSet<String>,
+    declared: HashSet<String>,
+    emit_declared: bool,
+}
+
+impl CfgSet {
+    pub fn new() -> Self {
+        Self {
+            enabled: HashSet::new(),
+            declared: HashSet::new(),
+            emit_declared: is_rustc_nightly(),
+        }
+    }
+
+    /// Enable a config, which can then be used in `#[cfg(...)]` for conditional compilation.
+    ///
+    /// All configs that can potentially be enabled should be unconditionally declared using
+    /// [`Self::declare()`].
+    pub fn enable(&mut self, cfg: impl AsRef<str>) {
+        if self.enabled.insert(cfg.as_ref().to_owned()) {
+            println!("cargo:rustc-cfg={}", cfg.as_ref());
+        }
+    }
+
+    pub fn enable_all(&mut self, cfgs: &[impl AsRef<str>]) {
+        for cfg in cfgs.iter() {
+            self.enable(cfg.as_ref());
+        }
+    }
+
+    /// Declare a valid config for conditional compilation, without enabling it.
+    ///
+    /// This enables rustc to check that the configs in `#[cfg(...)]` attributes are valid.
+    pub fn declare(&mut self, cfg: impl AsRef<str>) {
+        if self.declared.insert(cfg.as_ref().to_owned()) && self.emit_declared {
+            println!("cargo:rustc-check-cfg=cfg({})", cfg.as_ref());
+        }
+    }
+
+    pub fn declare_all(&mut self, cfgs: &[impl AsRef<str>]) {
+        for cfg in cfgs.iter() {
+            self.declare(cfg.as_ref());
+        }
+    }
+
+    pub fn set(&mut self, cfg: impl Into<String>, enable: bool) {
+        let cfg = cfg.into();
+        if enable {
+            self.enable(cfg.clone());
+        }
+        self.declare(cfg);
+    }
+}
+
+fn is_rustc_nightly() -> bool {
+    let rustc = env::var_os("RUSTC").unwrap_or_else(|| OsString::from("rustc"));
+
+    let output = Command::new(rustc)
+        .arg("--version")
+        .output()
+        .expect("failed to run `rustc --version`");
+
+    String::from_utf8_lossy(&output.stdout).contains("nightly")
+}
+
+/// Sets configs that describe the target platform.
+pub fn set_target_cfgs(cfgs: &mut CfgSet) {
+    let target = env::var("TARGET").unwrap();
+
+    if target.starts_with("thumbv6m-") {
+        cfgs.enable_all(&["cortex_m", "armv6m"]);
+    } else if target.starts_with("thumbv7m-") {
+        cfgs.enable_all(&["cortex_m", "armv7m"]);
+    } else if target.starts_with("thumbv7em-") {
+        cfgs.enable_all(&["cortex_m", "armv7m", "armv7em"]);
+    } else if target.starts_with("thumbv8m.base") {
+        cfgs.enable_all(&["cortex_m", "armv8m", "armv8m_base"]);
+    } else if target.starts_with("thumbv8m.main") {
+        cfgs.enable_all(&["cortex_m", "armv8m", "armv8m_main"]);
+    }
+    cfgs.declare_all(&[
+        "cortex_m",
+        "armv6m",
+        "armv7m",
+        "armv7em",
+        "armv8m",
+        "armv8m_base",
+        "armv8m_main",
+    ]);
+
+    cfgs.set("has_fpu", target.ends_with("-eabihf"));
+}
diff --git a/embassy-net-adin1110/Cargo.toml b/embassy-net-adin1110/Cargo.toml
index 97579a467..4a856fac3 100644
--- a/embassy-net-adin1110/Cargo.toml
+++ b/embassy-net-adin1110/Cargo.toml
@@ -29,7 +29,6 @@ critical-section = { version = "1.1.2", features = ["std"] }
 futures-test = "0.3.28"
 
 [features]
-default = [ ]
 defmt = [ "dep:defmt", "embedded-hal-1/defmt-03" ]
 log = ["dep:log"]
 
diff --git a/embassy-net-adin1110/src/lib.rs b/embassy-net-adin1110/src/lib.rs
index d98e98422..7f1c772e2 100644
--- a/embassy-net-adin1110/src/lib.rs
+++ b/embassy-net-adin1110/src/lib.rs
@@ -1,6 +1,6 @@
+#![cfg_attr(not(test), no_std)]
 #![deny(clippy::pedantic)]
 #![allow(async_fn_in_trait)]
-#![cfg_attr(not(any(test, feature = "std")), no_std)]
 #![allow(clippy::module_name_repetitions)]
 #![allow(clippy::missing_errors_doc)]
 #![allow(clippy::missing_panics_doc)]
diff --git a/embassy-nrf/src/qspi.rs b/embassy-nrf/src/qspi.rs
index 060fe72cd..d40096edc 100755
--- a/embassy-nrf/src/qspi.rs
+++ b/embassy-nrf/src/qspi.rs
@@ -166,7 +166,7 @@ impl<'d, T: Instance> Qspi<'d, T> {
                 $pin.conf().write(|w| {
                     w.dir().output();
                     w.drive().h0h1();
-                    #[cfg(feature = "_nrf5340-s")]
+                    #[cfg(all(feature = "_nrf5340", feature = "_s"))]
                     w.mcusel().peripheral();
                     w
                 });
diff --git a/embassy-stm32-wpan/Cargo.toml b/embassy-stm32-wpan/Cargo.toml
index 11a0adab2..9c81a71f7 100644
--- a/embassy-stm32-wpan/Cargo.toml
+++ b/embassy-stm32-wpan/Cargo.toml
@@ -28,6 +28,8 @@ embassy-embedded-hal = { version = "0.1.0", path = "../embassy-embedded-hal" }
 embassy-net-driver = { version = "0.2.0", path = "../embassy-net-driver", optional=true }
 
 defmt = { version = "0.3", optional = true }
+log = { version = "0.4.17", optional = true }
+
 cortex-m = "0.7.6"
 heapless = "0.8"
 aligned = "0.4.1"
diff --git a/embassy-stm32/Cargo.toml b/embassy-stm32/Cargo.toml
index fde5b4786..6a1dfec26 100644
--- a/embassy-stm32/Cargo.toml
+++ b/embassy-stm32/Cargo.toml
@@ -72,7 +72,7 @@ rand_core = "0.6.3"
 sdio-host = "0.5.0"
 critical-section = "1.1"
 #stm32-metapac = { version = "15" }
-stm32-metapac = { git = "https://github.com/embassy-rs/stm32-data-generated", tag = "stm32-data-34c0188a682b32c32ff147d377e0629b1ebe8318" }
+stm32-metapac = { git = "https://github.com/embassy-rs/stm32-data-generated", tag = "stm32-data-59b1f65bd109c3ef35782e6c44062208d0ef3d0e" }
 
 vcell = "0.1.3"
 nb = "1.0.0"
@@ -97,7 +97,7 @@ proc-macro2 = "1.0.36"
 quote = "1.0.15"
 
 #stm32-metapac = { version = "15", default-features = false, features = ["metadata"]}
-stm32-metapac = { git = "https://github.com/embassy-rs/stm32-data-generated", tag = "stm32-data-34c0188a682b32c32ff147d377e0629b1ebe8318", default-features = false, features = ["metadata"]}
+stm32-metapac = { git = "https://github.com/embassy-rs/stm32-data-generated", tag = "stm32-data-59b1f65bd109c3ef35782e6c44062208d0ef3d0e", default-features = false, features = ["metadata"] }
 
 [features]
 default = ["rt"]
diff --git a/embassy-stm32/build.rs b/embassy-stm32/build.rs
index 7bf6ffba2..6aedcc228 100644
--- a/embassy-stm32/build.rs
+++ b/embassy-stm32/build.rs
@@ -9,35 +9,16 @@ use proc_macro2::{Ident, TokenStream};
 use quote::{format_ident, quote};
 use stm32_metapac::metadata::ir::BitOffset;
 use stm32_metapac::metadata::{
-    MemoryRegionKind, PeripheralRccKernelClock, PeripheralRccRegister, PeripheralRegisters, StopMode, METADATA,
+    MemoryRegionKind, PeripheralRccKernelClock, PeripheralRccRegister, PeripheralRegisters, StopMode, ALL_CHIPS,
+    ALL_PERIPHERAL_VERSIONS, METADATA,
 };
 
-fn main() {
-    let target = env::var("TARGET").unwrap();
-
-    if target.starts_with("thumbv6m-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv6m");
-    } else if target.starts_with("thumbv7m-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv7m");
-    } else if target.starts_with("thumbv7em-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv7m");
-        println!("cargo:rustc-cfg=armv7em"); // (not currently used)
-    } else if target.starts_with("thumbv8m.base") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv8m");
-        println!("cargo:rustc-cfg=armv8m_base");
-    } else if target.starts_with("thumbv8m.main") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv8m");
-        println!("cargo:rustc-cfg=armv8m_main");
-    }
+#[path = "./build_common.rs"]
+mod common;
 
-    if target.ends_with("-eabihf") {
-        println!("cargo:rustc-cfg=has_fpu");
-    }
+fn main() {
+    let mut cfgs = common::CfgSet::new();
+    common::set_target_cfgs(&mut cfgs);
 
     let chip_name = match env::vars()
         .map(|(a, _)| a)
@@ -56,8 +37,15 @@ fn main() {
 
     for p in METADATA.peripherals {
         if let Some(r) = &p.registers {
-            println!("cargo:rustc-cfg={}", r.kind);
-            println!("cargo:rustc-cfg={}_{}", r.kind, r.version);
+            cfgs.enable(r.kind);
+            cfgs.enable(format!("{}_{}", r.kind, r.version));
+        }
+    }
+
+    for &(kind, versions) in ALL_PERIPHERAL_VERSIONS.iter() {
+        cfgs.declare(kind);
+        for &version in versions.iter() {
+            cfgs.declare(format!("{}_{}", kind, version));
         }
     }
 
@@ -67,7 +55,13 @@ fn main() {
     let mut singletons: Vec<String> = Vec::new();
     for p in METADATA.peripherals {
         if let Some(r) = &p.registers {
-            println!("cargo:rustc-cfg=peri_{}", p.name.to_ascii_lowercase());
+            if r.kind == "adccommon" || r.kind == "sai" || r.kind == "ucpd" {
+                // TODO: should we emit this for all peripherals? if so, we will need a list of all
+                // possible peripherals across all chips, so that we can declare the configs
+                // (replacing the hard-coded list of `peri_*` cfgs below)
+                cfgs.enable(format!("peri_{}", p.name.to_ascii_lowercase()));
+            }
+
             match r.kind {
                 // Generate singletons per pin, not per port
                 "gpio" => {
@@ -87,7 +81,7 @@ fn main() {
                         if pin.signal.starts_with("MCO") {
                             let name = pin.signal.replace('_', "").to_string();
                             if !singletons.contains(&name) {
-                                println!("cargo:rustc-cfg={}", name.to_ascii_lowercase());
+                                cfgs.enable(name.to_ascii_lowercase());
                                 singletons.push(name);
                             }
                         }
@@ -106,6 +100,20 @@ fn main() {
         }
     }
 
+    cfgs.declare_all(&[
+        "peri_adc1_common",
+        "peri_adc3_common",
+        "peri_adc12_common",
+        "peri_adc34_common",
+        "peri_sai1",
+        "peri_sai2",
+        "peri_sai3",
+        "peri_sai4",
+        "peri_ucpd1",
+        "peri_ucpd2",
+    ]);
+    cfgs.declare_all(&["mco", "mco1", "mco2"]);
+
     // One singleton per EXTI line
     for pin_num in 0..16 {
         singletons.push(format!("EXTI{}", pin_num));
@@ -221,7 +229,13 @@ fn main() {
     };
 
     if !time_driver_singleton.is_empty() {
-        println!("cargo:rustc-cfg=time_driver_{}", time_driver_singleton.to_lowercase());
+        cfgs.enable(format!("time_driver_{}", time_driver_singleton.to_lowercase()));
+    }
+    for tim in [
+        "tim1", "tim2", "tim3", "tim4", "tim5", "tim8", "tim9", "tim12", "tim15", "tim20", "tim21", "tim22", "tim23",
+        "tim24",
+    ] {
+        cfgs.declare(format!("time_driver_{}", tim));
     }
 
     // ========
@@ -367,9 +381,6 @@ fn main() {
         .filter_map(|p| p.registers.as_ref())
         .find(|r| r.kind == "rcc")
         .unwrap();
-    for b in rcc_registers.ir.blocks {
-        eprintln!("{}", b.name);
-    }
     let rcc_block = rcc_registers.ir.blocks.iter().find(|b| b.name == "Rcc").unwrap();
 
     // ========
@@ -388,7 +399,6 @@ fn main() {
         rcc_registers: &'a PeripheralRegisters,
         chained_muxes: HashMap<&'a str, &'a PeripheralRccRegister>,
 
-        refcount_statics: BTreeSet<Ident>,
         clock_names: BTreeSet<String>,
         muxes: BTreeSet<(Ident, Ident, Ident)>,
     }
@@ -397,7 +407,6 @@ fn main() {
         rcc_registers,
         chained_muxes: HashMap::new(),
 
-        refcount_statics: BTreeSet::new(),
         clock_names: BTreeSet::new(),
         muxes: BTreeSet::new(),
     };
@@ -516,80 +525,64 @@ fn main() {
         }
     }
 
+    let mut refcount_idxs = HashMap::new();
+
     for p in METADATA.peripherals {
         if !singletons.contains(&p.name.to_string()) {
             continue;
         }
 
         if let Some(rcc) = &p.rcc {
-            let en = rcc.enable.as_ref().unwrap();
-
-            let (start_rst, end_rst) = match &rcc.reset {
-                Some(rst) => {
-                    let rst_reg = format_ident!("{}", rst.register.to_ascii_lowercase());
-                    let set_rst_field = format_ident!("set_{}", rst.field.to_ascii_lowercase());
-                    (
-                        quote! {
-                            crate::pac::RCC.#rst_reg().modify(|w| w.#set_rst_field(true));
-                        },
-                        quote! {
-                            crate::pac::RCC.#rst_reg().modify(|w| w.#set_rst_field(false));
-                        },
-                    )
-                }
-                None => (TokenStream::new(), TokenStream::new()),
-            };
-
+            let rst_reg = rcc.reset.as_ref();
+            let en_reg = rcc.enable.as_ref().unwrap();
             let pname = format_ident!("{}", p.name);
-            let en_reg = format_ident!("{}", en.register.to_ascii_lowercase());
-            let set_en_field = format_ident!("set_{}", en.field.to_ascii_lowercase());
-            let en_reg_offs = rcc_block
-                .items
-                .iter()
-                .find(|i| i.name.eq_ignore_ascii_case(en.register))
-                .unwrap()
-                .byte_offset;
-            let en_reg_offs: u8 = (en_reg_offs / 4).try_into().unwrap();
-
-            let en_bit_offs = &rcc_registers
-                .ir
-                .fieldsets
-                .iter()
-                .find(|i| i.name.eq_ignore_ascii_case(en.register))
-                .unwrap()
-                .fields
-                .iter()
-                .find(|i| i.name.eq_ignore_ascii_case(en.field))
-                .unwrap()
-                .bit_offset;
-            let BitOffset::Regular(en_bit_offs) = en_bit_offs else {
-                panic!("cursed bit offset")
-            };
-            let en_bit_offs: u8 = en_bit_offs.offset.try_into().unwrap();
 
-            let refcount = *rcc_field_count.get(&(en.register, en.field)).unwrap() > 1;
-            let (before_enable, before_disable) = if refcount {
-                let refcount_static =
-                    format_ident!("{}_{}", en.register.to_ascii_uppercase(), en.field.to_ascii_uppercase());
+            let get_offset_and_bit = |reg: &PeripheralRccRegister| -> TokenStream {
+                let reg_offset = rcc_block
+                    .items
+                    .iter()
+                    .find(|i| i.name.eq_ignore_ascii_case(reg.register))
+                    .unwrap()
+                    .byte_offset;
+                let reg_offset: u8 = (reg_offset / 4).try_into().unwrap();
+
+                let bit_offset = &rcc_registers
+                    .ir
+                    .fieldsets
+                    .iter()
+                    .find(|i| i.name.eq_ignore_ascii_case(reg.register))
+                    .unwrap()
+                    .fields
+                    .iter()
+                    .find(|i| i.name.eq_ignore_ascii_case(reg.field))
+                    .unwrap()
+                    .bit_offset;
+                let BitOffset::Regular(bit_offset) = bit_offset else {
+                    panic!("cursed bit offset")
+                };
+                let bit_offset: u8 = bit_offset.offset.try_into().unwrap();
 
-                clock_gen.refcount_statics.insert(refcount_static.clone());
+                quote! { (#reg_offset, #bit_offset) }
+            };
 
-                (
-                    quote! {
-                        unsafe { refcount_statics::#refcount_static += 1 };
-                        if unsafe { refcount_statics::#refcount_static } > 1 {
-                            return;
-                        }
-                    },
-                    quote! {
-                        unsafe { refcount_statics::#refcount_static -= 1 };
-                        if unsafe { refcount_statics::#refcount_static } > 0  {
-                            return;
-                        }
-                    },
-                )
+            let reset_offset_and_bit = match rst_reg {
+                Some(rst_reg) => {
+                    let reset_offset_and_bit = get_offset_and_bit(rst_reg);
+                    quote! { Some(#reset_offset_and_bit) }
+                }
+                None => quote! { None },
+            };
+            let enable_offset_and_bit = get_offset_and_bit(en_reg);
+
+            let needs_refcount = *rcc_field_count.get(&(en_reg.register, en_reg.field)).unwrap() > 1;
+            let refcount_idx = if needs_refcount {
+                let next_refcount_idx = refcount_idxs.len() as u8;
+                let refcount_idx = *refcount_idxs
+                    .entry((en_reg.register, en_reg.field))
+                    .or_insert(next_refcount_idx);
+                quote! { Some(#refcount_idx) }
             } else {
-                (TokenStream::new(), TokenStream::new())
+                quote! { None }
             };
 
             let clock_frequency = match &rcc.kernel_clock {
@@ -599,24 +592,10 @@ fn main() {
 
             // A refcount leak can result if the same field is shared by peripherals with different stop modes
             // This condition should be checked in stm32-data
-            let stop_refcount = match rcc.stop_mode {
-                StopMode::Standby => None,
-                StopMode::Stop2 => Some(quote! { REFCOUNT_STOP2 }),
-                StopMode::Stop1 => Some(quote! { REFCOUNT_STOP1 }),
-            };
-
-            let (incr_stop_refcount, decr_stop_refcount) = match stop_refcount {
-                Some(stop_refcount) => (
-                    quote! {
-                        #[cfg(feature = "low-power")]
-                        unsafe { crate::rcc::#stop_refcount += 1 };
-                    },
-                    quote! {
-                        #[cfg(feature = "low-power")]
-                        unsafe { crate::rcc::#stop_refcount -= 1 };
-                    },
-                ),
-                None => (TokenStream::new(), TokenStream::new()),
+            let stop_mode = match rcc.stop_mode {
+                StopMode::Standby => quote! { crate::rcc::StopMode::Standby },
+                StopMode::Stop2 => quote! { crate::rcc::StopMode::Stop2 },
+                StopMode::Stop1 => quote! { crate::rcc::StopMode::Stop1 },
             };
 
             g.extend(quote! {
@@ -624,34 +603,16 @@ fn main() {
                     fn frequency() -> crate::time::Hertz {
                         #clock_frequency
                     }
-                    fn enable_and_reset_with_cs(_cs: critical_section::CriticalSection) {
-                        #before_enable
-                        #incr_stop_refcount
-
-                        #start_rst
-
-                        crate::pac::RCC.#en_reg().modify(|w| w.#set_en_field(true));
-
-                        // we must wait two peripheral clock cycles before the clock is active
-                        // this seems to work, but might be incorrect
-                        // see http://efton.sk/STM32/gotcha/g183.html
-
-                        // dummy read (like in the ST HALs)
-                        let _ = crate::pac::RCC.#en_reg().read();
-
-                        // DSB for good measure
-                        cortex_m::asm::dsb();
-
-                        #end_rst
-                    }
-                    fn disable_with_cs(_cs: critical_section::CriticalSection) {
-                        #before_disable
-                        crate::pac::RCC.#en_reg().modify(|w| w.#set_en_field(false));
-                        #decr_stop_refcount
-                    }
 
-                    const ENABLE_BIT: crate::rcc::ClockEnableBit =
-                        unsafe { crate::rcc::ClockEnableBit::new(#en_reg_offs, #en_bit_offs) };
+                    const RCC_INFO: crate::rcc::RccInfo = unsafe {
+                        crate::rcc::RccInfo::new(
+                            #reset_offset_and_bit,
+                            #enable_offset_and_bit,
+                            #refcount_idx,
+                            #[cfg(feature = "low-power")]
+                            #stop_mode,
+                        )
+                    };
                 }
 
                 impl crate::rcc::RccPeripheral for peripherals::#pname {}
@@ -659,6 +620,14 @@ fn main() {
         }
     }
 
+    g.extend({
+        let refcounts_len = refcount_idxs.len();
+        let refcount_zeros: TokenStream = refcount_idxs.iter().map(|_| quote! { 0u8, }).collect();
+        quote! {
+            pub(crate) static mut REFCOUNTS: [u8; #refcounts_len] = [#refcount_zeros];
+        }
+    });
+
     let struct_fields: Vec<_> = clock_gen
         .muxes
         .iter()
@@ -762,22 +731,6 @@ fn main() {
         }
     );
 
-    let refcount_mod: TokenStream = clock_gen
-        .refcount_statics
-        .iter()
-        .map(|refcount_static| {
-            quote! {
-                pub(crate) static mut #refcount_static: u8 = 0;
-            }
-        })
-        .collect();
-
-    g.extend(quote! {
-        mod refcount_statics {
-            #refcount_mod
-        }
-    });
-
     // ========
     // Generate fns to enable GPIO, DMA in RCC
 
@@ -1654,54 +1607,65 @@ fn main() {
     rustfmt(&out_file);
 
     // ========
-    // Multicore
-
-    let mut s = chip_name.split('_');
-    let mut chip_name: String = s.next().unwrap().to_string();
-    let core_name = if let Some(c) = s.next() {
-        if !c.starts_with("CM") {
-            chip_name.push('_');
-            chip_name.push_str(c);
-            None
-        } else {
-            Some(c)
-        }
-    } else {
-        None
-    };
+    // Configs for multicore and for targeting groups of chips
 
-    if let Some(core) = core_name {
-        println!("cargo:rustc-cfg={}_{}", &chip_name[..chip_name.len() - 2], core);
-    }
+    fn get_chip_cfgs(chip_name: &str) -> Vec<String> {
+        let mut cfgs = Vec::new();
 
-    // =======
-    // Features for targeting groups of chips
+        // Multicore
 
-    if &chip_name[..8] == "stm32wba" {
-        println!("cargo:rustc-cfg={}", &chip_name[..8]); // stm32wba
-        println!("cargo:rustc-cfg={}", &chip_name[..10]); // stm32wba52
-        println!("cargo:rustc-cfg=package_{}", &chip_name[10..11]);
-        println!("cargo:rustc-cfg=flashsize_{}", &chip_name[11..12]);
-    } else {
-        if &chip_name[..8] == "stm32h7r" || &chip_name[..8] == "stm32h7s" {
-            println!("cargo:rustc-cfg=stm32h7rs");
+        let mut s = chip_name.split('_');
+        let mut chip_name: String = s.next().unwrap().to_string();
+        let core_name = if let Some(c) = s.next() {
+            if !c.starts_with("CM") {
+                chip_name.push('_');
+                chip_name.push_str(c);
+                None
+            } else {
+                Some(c)
+            }
         } else {
-            println!("cargo:rustc-cfg={}", &chip_name[..7]); // stm32f4
+            None
+        };
+
+        if let Some(core) = core_name {
+            cfgs.push(format!("{}_{}", &chip_name[..chip_name.len() - 2], core));
         }
-        println!("cargo:rustc-cfg={}", &chip_name[..9]); // stm32f429
-        println!("cargo:rustc-cfg={}x", &chip_name[..8]); // stm32f42x
-        println!("cargo:rustc-cfg={}x{}", &chip_name[..7], &chip_name[8..9]); // stm32f4x9
-        println!("cargo:rustc-cfg=package_{}", &chip_name[9..10]);
-        println!("cargo:rustc-cfg=flashsize_{}", &chip_name[10..11]);
-    }
 
-    // Mark the L4+ chips as they have many differences to regular L4.
-    if &chip_name[..7] == "stm32l4" {
-        if "pqrs".contains(&chip_name[7..8]) {
-            println!("cargo:rustc-cfg=stm32l4_plus");
+        // Configs for targeting groups of chips
+        if &chip_name[..8] == "stm32wba" {
+            cfgs.push(chip_name[..8].to_owned()); // stm32wba
+            cfgs.push(chip_name[..10].to_owned()); // stm32wba52
+            cfgs.push(format!("package_{}", &chip_name[10..11]));
+            cfgs.push(format!("flashsize_{}", &chip_name[11..12]));
         } else {
-            println!("cargo:rustc-cfg=stm32l4_nonplus");
+            if &chip_name[..8] == "stm32h7r" || &chip_name[..8] == "stm32h7s" {
+                cfgs.push("stm32h7rs".to_owned());
+            } else {
+                cfgs.push(chip_name[..7].to_owned()); // stm32f4
+            }
+            cfgs.push(chip_name[..9].to_owned()); // stm32f429
+            cfgs.push(format!("{}x", &chip_name[..8])); // stm32f42x
+            cfgs.push(format!("{}x{}", &chip_name[..7], &chip_name[8..9])); // stm32f4x9
+            cfgs.push(format!("package_{}", &chip_name[9..10]));
+            cfgs.push(format!("flashsize_{}", &chip_name[10..11]));
+        }
+
+        // Mark the L4+ chips as they have many differences to regular L4.
+        if &chip_name[..7] == "stm32l4" {
+            if "pqrs".contains(&chip_name[7..8]) {
+                cfgs.push("stm32l4_plus".to_owned());
+            } else {
+                cfgs.push("stm32l4_nonplus".to_owned());
+            }
         }
+
+        cfgs
+    }
+
+    cfgs.enable_all(&get_chip_cfgs(&chip_name));
+    for &chip_name in ALL_CHIPS.iter() {
+        cfgs.declare_all(&get_chip_cfgs(&chip_name.to_ascii_lowercase()));
     }
 
     println!("cargo:rerun-if-changed=build.rs");
diff --git a/embassy-stm32/build_common.rs b/embassy-stm32/build_common.rs
new file mode 100644
index 000000000..2c65f8529
--- /dev/null
+++ b/embassy-stm32/build_common.rs
@@ -0,0 +1,109 @@
+// NOTE: this file is copy-pasted between several Embassy crates, because there is no
+// straightforward way to share this code:
+// - it cannot be placed into the root of the repo and linked from each build.rs using `#[path =
+// "../build_common.rs"]`, because `cargo publish` requires that all files published with a crate
+// reside in the crate's directory,
+// - it cannot be symlinked from `embassy-xxx/build_common.rs` to `../build_common.rs`, because
+// symlinks don't work on Windows.
+
+use std::collections::HashSet;
+use std::env;
+use std::ffi::OsString;
+use std::process::Command;
+
+/// Helper for emitting cargo instruction for enabling configs (`cargo:rustc-cfg=X`) and declaring
+/// them (`cargo:rust-check-cfg=cfg(X)`).
+#[derive(Debug)]
+pub struct CfgSet {
+    enabled: HashSet<String>,
+    declared: HashSet<String>,
+    emit_declared: bool,
+}
+
+impl CfgSet {
+    pub fn new() -> Self {
+        Self {
+            enabled: HashSet::new(),
+            declared: HashSet::new(),
+            emit_declared: is_rustc_nightly(),
+        }
+    }
+
+    /// Enable a config, which can then be used in `#[cfg(...)]` for conditional compilation.
+    ///
+    /// All configs that can potentially be enabled should be unconditionally declared using
+    /// [`Self::declare()`].
+    pub fn enable(&mut self, cfg: impl AsRef<str>) {
+        if self.enabled.insert(cfg.as_ref().to_owned()) {
+            println!("cargo:rustc-cfg={}", cfg.as_ref());
+        }
+    }
+
+    pub fn enable_all(&mut self, cfgs: &[impl AsRef<str>]) {
+        for cfg in cfgs.iter() {
+            self.enable(cfg.as_ref());
+        }
+    }
+
+    /// Declare a valid config for conditional compilation, without enabling it.
+    ///
+    /// This enables rustc to check that the configs in `#[cfg(...)]` attributes are valid.
+    pub fn declare(&mut self, cfg: impl AsRef<str>) {
+        if self.declared.insert(cfg.as_ref().to_owned()) && self.emit_declared {
+            println!("cargo:rustc-check-cfg=cfg({})", cfg.as_ref());
+        }
+    }
+
+    pub fn declare_all(&mut self, cfgs: &[impl AsRef<str>]) {
+        for cfg in cfgs.iter() {
+            self.declare(cfg.as_ref());
+        }
+    }
+
+    pub fn set(&mut self, cfg: impl Into<String>, enable: bool) {
+        let cfg = cfg.into();
+        if enable {
+            self.enable(cfg.clone());
+        }
+        self.declare(cfg);
+    }
+}
+
+fn is_rustc_nightly() -> bool {
+    let rustc = env::var_os("RUSTC").unwrap_or_else(|| OsString::from("rustc"));
+
+    let output = Command::new(rustc)
+        .arg("--version")
+        .output()
+        .expect("failed to run `rustc --version`");
+
+    String::from_utf8_lossy(&output.stdout).contains("nightly")
+}
+
+/// Sets configs that describe the target platform.
+pub fn set_target_cfgs(cfgs: &mut CfgSet) {
+    let target = env::var("TARGET").unwrap();
+
+    if target.starts_with("thumbv6m-") {
+        cfgs.enable_all(&["cortex_m", "armv6m"]);
+    } else if target.starts_with("thumbv7m-") {
+        cfgs.enable_all(&["cortex_m", "armv7m"]);
+    } else if target.starts_with("thumbv7em-") {
+        cfgs.enable_all(&["cortex_m", "armv7m", "armv7em"]);
+    } else if target.starts_with("thumbv8m.base") {
+        cfgs.enable_all(&["cortex_m", "armv8m", "armv8m_base"]);
+    } else if target.starts_with("thumbv8m.main") {
+        cfgs.enable_all(&["cortex_m", "armv8m", "armv8m_main"]);
+    }
+    cfgs.declare_all(&[
+        "cortex_m",
+        "armv6m",
+        "armv7m",
+        "armv7em",
+        "armv8m",
+        "armv8m_base",
+        "armv8m_main",
+    ]);
+
+    cfgs.set("has_fpu", target.ends_with("-eabihf"));
+}
diff --git a/embassy-stm32/src/adc/f1.rs b/embassy-stm32/src/adc/f1.rs
index 3822d5032..b37ec260f 100644
--- a/embassy-stm32/src/adc/f1.rs
+++ b/embassy-stm32/src/adc/f1.rs
@@ -7,7 +7,7 @@ use embassy_hal_internal::into_ref;
 use super::blocking_delay_us;
 use crate::adc::{Adc, AdcChannel, Instance, SampleTime};
 use crate::time::Hertz;
-use crate::{interrupt, Peripheral};
+use crate::{interrupt, rcc, Peripheral};
 
 pub const VDDA_CALIB_MV: u32 = 3300;
 pub const ADC_MAX: u32 = (1 << 12) - 1;
@@ -50,7 +50,7 @@ impl<T: Instance> super::SealedAdcChannel<T> for Temperature {
 impl<'d, T: Instance> Adc<'d, T> {
     pub fn new(adc: impl Peripheral<P = T> + 'd) -> Self {
         into_ref!(adc);
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
         T::regs().cr2().modify(|reg| reg.set_adon(true));
 
         // 11.4: Before starting a calibration, the ADC must have been in power-on state (ADON bit = ‘1’)
@@ -169,6 +169,6 @@ impl<'d, T: Instance> Drop for Adc<'d, T> {
     fn drop(&mut self) {
         T::regs().cr2().modify(|reg| reg.set_adon(false));
 
-        T::disable();
+        rcc::disable::<T>();
     }
 }
diff --git a/embassy-stm32/src/adc/f3.rs b/embassy-stm32/src/adc/f3.rs
index 3f076d64b..ac88c9742 100644
--- a/embassy-stm32/src/adc/f3.rs
+++ b/embassy-stm32/src/adc/f3.rs
@@ -8,7 +8,7 @@ use super::blocking_delay_us;
 use crate::adc::{Adc, AdcChannel, Instance, SampleTime};
 use crate::interrupt::typelevel::Interrupt;
 use crate::time::Hertz;
-use crate::{interrupt, Peripheral};
+use crate::{interrupt, rcc, Peripheral};
 
 pub const VDDA_CALIB_MV: u32 = 3300;
 pub const ADC_MAX: u32 = (1 << 12) - 1;
@@ -63,7 +63,7 @@ impl<'d, T: Instance> Adc<'d, T> {
 
         into_ref!(adc);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         // Enable the adc regulator
         T::regs().cr().modify(|w| w.set_advregen(vals::Advregen::INTERMEDIATE));
@@ -188,6 +188,6 @@ impl<'d, T: Instance> Drop for Adc<'d, T> {
         T::regs().cr().modify(|w| w.set_advregen(vals::Advregen::INTERMEDIATE));
         T::regs().cr().modify(|w| w.set_advregen(vals::Advregen::DISABLED));
 
-        T::disable();
+        rcc::disable::<T>();
     }
 }
diff --git a/embassy-stm32/src/adc/f3_v1_1.rs b/embassy-stm32/src/adc/f3_v1_1.rs
index 106956989..689c2871d 100644
--- a/embassy-stm32/src/adc/f3_v1_1.rs
+++ b/embassy-stm32/src/adc/f3_v1_1.rs
@@ -10,7 +10,7 @@ use super::Resolution;
 use crate::adc::{Adc, AdcChannel, Instance, SampleTime};
 use crate::interrupt::typelevel::Interrupt;
 use crate::time::Hertz;
-use crate::{interrupt, Peripheral};
+use crate::{interrupt, rcc, Peripheral};
 
 const ADC_FREQ: Hertz = crate::rcc::HSI_FREQ;
 
@@ -143,7 +143,7 @@ impl<'d, T: Instance> Adc<'d, T> {
     ) -> Self {
         into_ref!(adc);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         //let r = T::regs();
         //r.cr2().write(|w| w.set_align(true));
@@ -403,6 +403,6 @@ impl<'d, T: Instance> Drop for Adc<'d, T> {
 
         T::regs().cr2().modify(|w| w.set_adon(false));
 
-        T::disable();
+        rcc::disable::<T>();
     }
 }
diff --git a/embassy-stm32/src/adc/g4.rs b/embassy-stm32/src/adc/g4.rs
index ce7f5db70..6569361fe 100644
--- a/embassy-stm32/src/adc/g4.rs
+++ b/embassy-stm32/src/adc/g4.rs
@@ -4,7 +4,7 @@ use pac::adccommon::vals::Presc;
 
 use super::{blocking_delay_us, Adc, AdcChannel, Instance, Resolution, SampleTime};
 use crate::time::Hertz;
-use crate::{pac, Peripheral};
+use crate::{pac, rcc, Peripheral};
 
 /// Default VREF voltage used for sample conversion to millivolts.
 pub const VREF_DEFAULT_MV: u32 = 3300;
@@ -130,7 +130,7 @@ impl<'d, T: Instance> Adc<'d, T> {
     /// Create a new ADC driver.
     pub fn new(adc: impl Peripheral<P = T> + 'd) -> Self {
         embassy_hal_internal::into_ref!(adc);
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         let prescaler = Prescaler::from_ker_ck(T::frequency());
 
diff --git a/embassy-stm32/src/adc/mod.rs b/embassy-stm32/src/adc/mod.rs
index 040ee9c53..0c22a7dae 100644
--- a/embassy-stm32/src/adc/mod.rs
+++ b/embassy-stm32/src/adc/mod.rs
@@ -71,9 +71,9 @@ pub(crate) trait SealedAdcChannel<T> {
 /// Performs a busy-wait delay for a specified number of microseconds.
 #[allow(unused)]
 pub(crate) fn blocking_delay_us(us: u32) {
-    #[cfg(time)]
-    embassy_time::block_for(embassy_time::Duration::from_micros(us));
-    #[cfg(not(time))]
+    #[cfg(feature = "time")]
+    embassy_time::block_for(embassy_time::Duration::from_micros(us as u64));
+    #[cfg(not(feature = "time"))]
     {
         let freq = unsafe { crate::rcc::get_freqs() }.sys.unwrap().0 as u64;
         let us = us as u64;
diff --git a/embassy-stm32/src/adc/v1.rs b/embassy-stm32/src/adc/v1.rs
index 090790c39..9bec2e13b 100644
--- a/embassy-stm32/src/adc/v1.rs
+++ b/embassy-stm32/src/adc/v1.rs
@@ -10,7 +10,7 @@ use super::blocking_delay_us;
 use crate::adc::{Adc, AdcChannel, Instance, Resolution, SampleTime};
 use crate::interrupt::typelevel::Interrupt;
 use crate::peripherals::ADC1;
-use crate::{interrupt, Peripheral};
+use crate::{interrupt, rcc, Peripheral};
 
 pub const VDDA_CALIB_MV: u32 = 3300;
 pub const VREF_INT: u32 = 1230;
@@ -67,7 +67,7 @@ impl<'d, T: Instance> Adc<'d, T> {
         _irq: impl interrupt::typelevel::Binding<T::Interrupt, InterruptHandler<T>> + 'd,
     ) -> Self {
         into_ref!(adc);
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         // Delay 1μs when using HSI14 as the ADC clock.
         //
@@ -199,6 +199,6 @@ impl<'d, T: Instance> Drop for Adc<'d, T> {
         T::regs().cr().modify(|reg| reg.set_addis(true));
         while T::regs().cr().read().aden() {}
 
-        T::disable();
+        rcc::disable::<T>();
     }
 }
diff --git a/embassy-stm32/src/adc/v2.rs b/embassy-stm32/src/adc/v2.rs
index 033108195..e3175dff5 100644
--- a/embassy-stm32/src/adc/v2.rs
+++ b/embassy-stm32/src/adc/v2.rs
@@ -4,7 +4,7 @@ use super::blocking_delay_us;
 use crate::adc::{Adc, AdcChannel, Instance, Resolution, SampleTime};
 use crate::peripherals::ADC1;
 use crate::time::Hertz;
-use crate::Peripheral;
+use crate::{rcc, Peripheral};
 
 /// Default VREF voltage used for sample conversion to millivolts.
 pub const VREF_DEFAULT_MV: u32 = 3300;
@@ -31,7 +31,7 @@ impl AdcChannel<ADC1> for Temperature {}
 impl super::SealedAdcChannel<ADC1> for Temperature {
     fn channel(&self) -> u8 {
         cfg_if::cfg_if! {
-            if #[cfg(any(stm32f2, stm32f40, stm32f41))] {
+            if #[cfg(any(stm32f2, stm32f40x, stm32f41x))] {
                 16
             } else {
                 18
@@ -96,7 +96,7 @@ where
 {
     pub fn new(adc: impl Peripheral<P = T> + 'd) -> Self {
         into_ref!(adc);
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         let presc = Prescaler::from_pclk2(T::frequency());
         T::common_regs().ccr().modify(|w| w.set_adcpre(presc.adcpre()));
@@ -206,6 +206,6 @@ impl<'d, T: Instance> Drop for Adc<'d, T> {
             reg.set_adon(false);
         });
 
-        T::disable();
+        rcc::disable::<T>();
     }
 }
diff --git a/embassy-stm32/src/adc/v3.rs b/embassy-stm32/src/adc/v3.rs
index be857f4dd..398c57a92 100644
--- a/embassy-stm32/src/adc/v3.rs
+++ b/embassy-stm32/src/adc/v3.rs
@@ -3,7 +3,7 @@ use embassy_hal_internal::into_ref;
 
 use super::blocking_delay_us;
 use crate::adc::{Adc, AdcChannel, Instance, Resolution, SampleTime};
-use crate::Peripheral;
+use crate::{rcc, Peripheral};
 
 /// Default VREF voltage used for sample conversion to millivolts.
 pub const VREF_DEFAULT_MV: u32 = 3300;
@@ -94,7 +94,7 @@ cfg_if! {
 impl<'d, T: Instance> Adc<'d, T> {
     pub fn new(adc: impl Peripheral<P = T> + 'd) -> Self {
         into_ref!(adc);
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
         T::regs().cr().modify(|reg| {
             #[cfg(not(any(adc_g0, adc_u0)))]
             reg.set_deeppwd(false);
diff --git a/embassy-stm32/src/adc/v4.rs b/embassy-stm32/src/adc/v4.rs
index f564114c2..50db646fe 100644
--- a/embassy-stm32/src/adc/v4.rs
+++ b/embassy-stm32/src/adc/v4.rs
@@ -4,7 +4,7 @@ use pac::adccommon::vals::Presc;
 
 use super::{blocking_delay_us, Adc, AdcChannel, Instance, Resolution, SampleTime};
 use crate::time::Hertz;
-use crate::{pac, Peripheral};
+use crate::{pac, rcc, Peripheral};
 
 /// Default VREF voltage used for sample conversion to millivolts.
 pub const VREF_DEFAULT_MV: u32 = 3300;
@@ -130,7 +130,7 @@ impl<'d, T: Instance> Adc<'d, T> {
     /// Create a new ADC driver.
     pub fn new(adc: impl Peripheral<P = T> + 'd) -> Self {
         embassy_hal_internal::into_ref!(adc);
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         let prescaler = Prescaler::from_ker_ck(T::frequency());
 
diff --git a/embassy-stm32/src/can/bxcan/filter.rs b/embassy-stm32/src/can/bxcan/filter.rs
index 9940c7f50..167c6c572 100644
--- a/embassy-stm32/src/can/bxcan/filter.rs
+++ b/embassy-stm32/src/can/bxcan/filter.rs
@@ -2,7 +2,7 @@
 
 use core::marker::PhantomData;
 
-use super::{ExtendedId, Fifo, FilterOwner, Id, Instance, MasterInstance, StandardId};
+use super::{ExtendedId, Fifo, Id, StandardId};
 
 const F32_RTR: u32 = 0b010; // set the RTR bit to match remote frames
 const F32_IDE: u32 = 0b100; // set the IDE bit to match extended identifiers
@@ -210,24 +210,24 @@ impl From<Mask32> for BankConfig {
 }
 
 /// Interface to the filter banks of a CAN peripheral.
-pub struct MasterFilters<'a, I: FilterOwner> {
+pub struct MasterFilters<'a> {
     /// Number of assigned filter banks.
     ///
     /// On chips with splittable filter banks, this value can be dynamic.
     bank_count: u8,
-    _can: PhantomData<&'a mut I>,
-    canregs: crate::pac::can::Can,
+    _phantom: PhantomData<&'a ()>,
+    info: &'static crate::can::Info,
 }
 
 // NOTE: This type mutably borrows the CAN instance and has unique access to the registers while it
 // exists.
-impl<I: FilterOwner> MasterFilters<'_, I> {
-    pub(crate) unsafe fn new(canregs: crate::pac::can::Can) -> Self {
+impl MasterFilters<'_> {
+    pub(crate) unsafe fn new(info: &'static crate::can::Info) -> Self {
         // Enable initialization mode.
-        canregs.fmr().modify(|reg| reg.set_finit(true));
+        info.regs.0.fmr().modify(|reg| reg.set_finit(true));
 
         // Read the filter split value.
-        let bank_count = canregs.fmr().read().can2sb();
+        let bank_count = info.regs.0.fmr().read().can2sb();
 
         // (Reset value of CAN2SB is 0x0E, 14, which, in devices with 14 filter banks, assigns all
         // of them to the master peripheral, and in devices with 28, assigns them 50/50 to
@@ -235,8 +235,8 @@ impl<I: FilterOwner> MasterFilters<'_, I> {
 
         Self {
             bank_count,
-            _can: PhantomData,
-            canregs,
+            _phantom: PhantomData,
+            info,
         }
     }
 
@@ -244,7 +244,7 @@ impl<I: FilterOwner> MasterFilters<'_, I> {
         FilterBanks {
             start_idx: 0,
             bank_count: self.bank_count,
-            canregs: self.canregs,
+            info: self.info,
         }
     }
 
@@ -291,49 +291,49 @@ impl<I: FilterOwner> MasterFilters<'_, I> {
     }
 }
 
-impl<I: MasterInstance> MasterFilters<'_, I> {
+impl MasterFilters<'_> {
     /// Sets the index at which the filter banks owned by the slave peripheral start.
     pub fn set_split(&mut self, split_index: u8) -> &mut Self {
-        assert!(split_index <= I::NUM_FILTER_BANKS);
-        self.canregs.fmr().modify(|reg| reg.set_can2sb(split_index));
+        assert!(split_index <= self.info.num_filter_banks);
+        self.info.regs.0.fmr().modify(|reg| reg.set_can2sb(split_index));
         self.bank_count = split_index;
         self
     }
 
     /// Accesses the filters assigned to the slave peripheral.
-    pub fn slave_filters(&mut self) -> SlaveFilters<'_, I> {
+    pub fn slave_filters(&mut self) -> SlaveFilters<'_> {
         // NB: This mutably borrows `self`, so it has full access to the filter bank registers.
         SlaveFilters {
             start_idx: self.bank_count,
-            bank_count: I::NUM_FILTER_BANKS - self.bank_count,
-            _can: PhantomData,
-            canregs: self.canregs,
+            bank_count: self.info.num_filter_banks - self.bank_count,
+            _phantom: PhantomData,
+            info: self.info,
         }
     }
 }
 
-impl<I: FilterOwner> Drop for MasterFilters<'_, I> {
+impl Drop for MasterFilters<'_> {
     #[inline]
     fn drop(&mut self) {
         // Leave initialization mode.
-        self.canregs.fmr().modify(|regs| regs.set_finit(false));
+        self.info.regs.0.fmr().modify(|regs| regs.set_finit(false));
     }
 }
 
 /// Interface to the filter banks assigned to a slave peripheral.
-pub struct SlaveFilters<'a, I: Instance> {
+pub struct SlaveFilters<'a> {
     start_idx: u8,
     bank_count: u8,
-    _can: PhantomData<&'a mut I>,
-    canregs: crate::pac::can::Can,
+    _phantom: PhantomData<&'a ()>,
+    info: &'static crate::can::Info,
 }
 
-impl<I: Instance> SlaveFilters<'_, I> {
+impl SlaveFilters<'_> {
     fn banks_imm(&self) -> FilterBanks {
         FilterBanks {
             start_idx: self.start_idx,
             bank_count: self.bank_count,
-            canregs: self.canregs,
+            info: self.info,
         }
     }
 
@@ -377,14 +377,14 @@ impl<I: Instance> SlaveFilters<'_, I> {
 struct FilterBanks {
     start_idx: u8,
     bank_count: u8,
-    canregs: crate::pac::can::Can,
+    info: &'static crate::can::Info,
 }
 
 impl FilterBanks {
     fn clear(&mut self) {
         let mask = filter_bitmask(self.start_idx, self.bank_count);
 
-        self.canregs.fa1r().modify(|reg| {
+        self.info.regs.0.fa1r().modify(|reg| {
             for i in 0..28usize {
                 if (0x01u32 << i) & mask != 0 {
                     reg.set_fact(i, false);
@@ -399,7 +399,11 @@ impl FilterBanks {
 
     fn disable(&mut self, index: u8) {
         self.assert_bank_index(index);
-        self.canregs.fa1r().modify(|reg| reg.set_fact(index as usize, false))
+        self.info
+            .regs
+            .0
+            .fa1r()
+            .modify(|reg| reg.set_fact(index as usize, false))
     }
 
     fn enable(&mut self, index: u8, fifo: Fifo, config: BankConfig) {
@@ -407,11 +411,11 @@ impl FilterBanks {
 
         // Configure mode.
         let mode = matches!(config, BankConfig::List16(_) | BankConfig::List32(_));
-        self.canregs.fm1r().modify(|reg| reg.set_fbm(index as usize, mode));
+        self.info.regs.0.fm1r().modify(|reg| reg.set_fbm(index as usize, mode));
 
         // Configure scale.
         let scale = matches!(config, BankConfig::List32(_) | BankConfig::Mask32(_));
-        self.canregs.fs1r().modify(|reg| reg.set_fsc(index as usize, scale));
+        self.info.regs.0.fs1r().modify(|reg| reg.set_fsc(index as usize, scale));
 
         // Configure filter register.
         let (fxr1, fxr2);
@@ -433,12 +437,12 @@ impl FilterBanks {
                 fxr2 = a.mask;
             }
         };
-        let bank = self.canregs.fb(index as usize);
+        let bank = self.info.regs.0.fb(index as usize);
         bank.fr1().write(|w| w.0 = fxr1);
         bank.fr2().write(|w| w.0 = fxr2);
 
         // Assign to the right FIFO
-        self.canregs.ffa1r().modify(|reg| {
+        self.info.regs.0.ffa1r().modify(|reg| {
             reg.set_ffa(
                 index as usize,
                 match fifo {
@@ -449,7 +453,7 @@ impl FilterBanks {
         });
 
         // Set active.
-        self.canregs.fa1r().modify(|reg| reg.set_fact(index as usize, true))
+        self.info.regs.0.fa1r().modify(|reg| reg.set_fact(index as usize, true))
     }
 }
 
diff --git a/embassy-stm32/src/can/bxcan/mod.rs b/embassy-stm32/src/can/bxcan/mod.rs
index 0ac4cdab6..53b94b9e2 100644
--- a/embassy-stm32/src/can/bxcan/mod.rs
+++ b/embassy-stm32/src/can/bxcan/mod.rs
@@ -5,7 +5,8 @@ use core::future::poll_fn;
 use core::marker::PhantomData;
 use core::task::Poll;
 
-use embassy_hal_internal::{into_ref, PeripheralRef};
+use embassy_hal_internal::interrupt::InterruptExt;
+use embassy_hal_internal::into_ref;
 use embassy_sync::blocking_mutex::raw::CriticalSectionRawMutex;
 use embassy_sync::channel::Channel;
 use embassy_sync::waitqueue::AtomicWaker;
@@ -19,7 +20,7 @@ use super::util;
 use crate::can::enums::{BusError, TryReadError};
 use crate::gpio::AFType;
 use crate::interrupt::typelevel::Interrupt;
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::{interrupt, peripherals, Peripheral};
 
 /// Interrupt handler.
@@ -90,11 +91,13 @@ impl<T: Instance> interrupt::typelevel::Handler<T::SCEInterrupt> for SceInterrup
 }
 
 /// Configuration proxy returned by [`Can::modify_config`].
-pub struct CanConfig<'a, T: Instance> {
-    can: PhantomData<&'a mut T>,
+pub struct CanConfig<'a> {
+    phantom: PhantomData<&'a ()>,
+    info: &'static Info,
+    periph_clock: crate::time::Hertz,
 }
 
-impl<T: Instance> CanConfig<'_, T> {
+impl CanConfig<'_> {
     /// Configures the bit timings.
     ///
     /// You can use <http://www.bittiming.can-wiki.info/> to calculate the `btr` parameter. Enter
@@ -108,7 +111,7 @@ impl<T: Instance> CanConfig<'_, T> {
     /// Then copy the `CAN_BUS_TIME` register value from the table and pass it as the `btr`
     /// parameter to this method.
     pub fn set_bit_timing(self, bt: crate::can::util::NominalBitTiming) -> Self {
-        Registers(T::regs()).set_bit_timing(bt);
+        self.info.regs.set_bit_timing(bt);
         self
     }
 
@@ -116,20 +119,20 @@ impl<T: Instance> CanConfig<'_, T> {
     ///
     /// This is a helper that internally calls `set_bit_timing()`[Self::set_bit_timing].
     pub fn set_bitrate(self, bitrate: u32) -> Self {
-        let bit_timing = util::calc_can_timings(T::frequency(), bitrate).unwrap();
+        let bit_timing = util::calc_can_timings(self.periph_clock, bitrate).unwrap();
         self.set_bit_timing(bit_timing)
     }
 
     /// Enables or disables loopback mode: Internally connects the TX and RX
     /// signals together.
     pub fn set_loopback(self, enabled: bool) -> Self {
-        Registers(T::regs()).set_loopback(enabled);
+        self.info.regs.set_loopback(enabled);
         self
     }
 
     /// Enables or disables silent mode: Disconnects the TX signal from the pin.
     pub fn set_silent(self, enabled: bool) -> Self {
-        Registers(T::regs()).set_silent(enabled);
+        self.info.regs.set_silent(enabled);
         self
     }
 
@@ -140,21 +143,24 @@ impl<T: Instance> CanConfig<'_, T> {
     ///
     /// Automatic retransmission is enabled by default.
     pub fn set_automatic_retransmit(self, enabled: bool) -> Self {
-        Registers(T::regs()).set_automatic_retransmit(enabled);
+        self.info.regs.set_automatic_retransmit(enabled);
         self
     }
 }
 
-impl<T: Instance> Drop for CanConfig<'_, T> {
+impl Drop for CanConfig<'_> {
     #[inline]
     fn drop(&mut self) {
-        Registers(T::regs()).leave_init_mode();
+        self.info.regs.leave_init_mode();
     }
 }
 
 /// CAN driver
-pub struct Can<'d, T: Instance> {
-    peri: PeripheralRef<'d, T>,
+pub struct Can<'d> {
+    phantom: PhantomData<&'d ()>,
+    info: &'static Info,
+    state: &'static State,
+    periph_clock: crate::time::Hertz,
 }
 
 /// Error returned by `try_write`
@@ -165,11 +171,11 @@ pub enum TryWriteError {
     Full,
 }
 
-impl<'d, T: Instance> Can<'d, T> {
+impl<'d> Can<'d> {
     /// Creates a new Bxcan instance, keeping the peripheral in sleep mode.
     /// You must call [Can::enable_non_blocking] to use the peripheral.
-    pub fn new(
-        peri: impl Peripheral<P = T> + 'd,
+    pub fn new<T: Instance>(
+        _peri: impl Peripheral<P = T> + 'd,
         rx: impl Peripheral<P = impl RxPin<T>> + 'd,
         tx: impl Peripheral<P = impl TxPin<T>> + 'd,
         _irqs: impl interrupt::typelevel::Binding<T::TXInterrupt, TxInterruptHandler<T>>
@@ -178,15 +184,17 @@ impl<'d, T: Instance> Can<'d, T> {
             + interrupt::typelevel::Binding<T::SCEInterrupt, SceInterruptHandler<T>>
             + 'd,
     ) -> Self {
-        into_ref!(peri, rx, tx);
+        into_ref!(_peri, rx, tx);
+        let info = T::info();
+        let regs = &T::info().regs;
 
         rx.set_as_af(rx.af_num(), AFType::Input);
         tx.set_as_af(tx.af_num(), AFType::OutputPushPull);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         {
-            T::regs().ier().write(|w| {
+            regs.0.ier().write(|w| {
                 w.set_errie(true);
                 w.set_fmpie(0, true);
                 w.set_fmpie(1, true);
@@ -197,7 +205,7 @@ impl<'d, T: Instance> Can<'d, T> {
                 w.set_lecie(true);
             });
 
-            T::regs().mcr().write(|w| {
+            regs.0.mcr().write(|w| {
                 // Enable timestamps on rx messages
 
                 w.set_ttcm(true);
@@ -205,17 +213,14 @@ impl<'d, T: Instance> Can<'d, T> {
         }
 
         unsafe {
-            T::TXInterrupt::unpend();
-            T::TXInterrupt::enable();
-
-            T::RX0Interrupt::unpend();
-            T::RX0Interrupt::enable();
-
-            T::RX1Interrupt::unpend();
-            T::RX1Interrupt::enable();
-
-            T::SCEInterrupt::unpend();
-            T::SCEInterrupt::enable();
+            info.tx_interrupt.unpend();
+            info.tx_interrupt.enable();
+            info.rx0_interrupt.unpend();
+            info.rx0_interrupt.enable();
+            info.rx1_interrupt.unpend();
+            info.rx1_interrupt.enable();
+            info.sce_interrupt.unpend();
+            info.sce_interrupt.enable();
         }
 
         rx.set_as_af(rx.af_num(), AFType::Input);
@@ -223,12 +228,17 @@ impl<'d, T: Instance> Can<'d, T> {
 
         Registers(T::regs()).leave_init_mode();
 
-        Self { peri }
+        Self {
+            phantom: PhantomData,
+            info: T::info(),
+            state: T::state(),
+            periph_clock: T::frequency(),
+        }
     }
 
     /// Set CAN bit rate.
     pub fn set_bitrate(&mut self, bitrate: u32) {
-        let bit_timing = util::calc_can_timings(T::frequency(), bitrate).unwrap();
+        let bit_timing = util::calc_can_timings(self.periph_clock, bitrate).unwrap();
         self.modify_config().set_bit_timing(bit_timing);
     }
 
@@ -236,10 +246,14 @@ impl<'d, T: Instance> Can<'d, T> {
     ///
     /// Calling this method will enter initialization mode. You must enable the peripheral
     /// again afterwards with [`enable`](Self::enable).
-    pub fn modify_config(&mut self) -> CanConfig<'_, T> {
-        Registers(T::regs()).enter_init_mode();
+    pub fn modify_config(&mut self) -> CanConfig<'_> {
+        self.info.regs.enter_init_mode();
 
-        CanConfig { can: PhantomData }
+        CanConfig {
+            phantom: self.phantom,
+            info: self.info,
+            periph_clock: self.periph_clock,
+        }
     }
 
     /// Enables the peripheral and synchronizes with the bus.
@@ -247,7 +261,7 @@ impl<'d, T: Instance> Can<'d, T> {
     /// This will wait for 11 consecutive recessive bits (bus idle state).
     /// Contrary to enable method from bxcan library, this will not freeze the executor while waiting.
     pub async fn enable(&mut self) {
-        while Registers(T::regs()).enable_non_blocking().is_err() {
+        while self.info.regs.enable_non_blocking().is_err() {
             // SCE interrupt is only generated for entering sleep mode, but not leaving.
             // Yield to allow other tasks to execute while can bus is initializing.
             embassy_futures::yield_now().await;
@@ -257,7 +271,7 @@ impl<'d, T: Instance> Can<'d, T> {
     /// Enables or disables the peripheral from automatically wakeup when a SOF is detected on the bus
     /// while the peripheral is in sleep mode
     pub fn set_automatic_wakeup(&mut self, enabled: bool) {
-        Registers(T::regs()).set_automatic_wakeup(enabled);
+        self.info.regs.set_automatic_wakeup(enabled);
     }
 
     /// Manually wake the peripheral from sleep mode.
@@ -265,12 +279,12 @@ impl<'d, T: Instance> Can<'d, T> {
     /// Waking the peripheral manually does not trigger a wake-up interrupt.
     /// This will wait until the peripheral has acknowledged it has awoken from sleep mode
     pub fn wakeup(&mut self) {
-        Registers(T::regs()).wakeup()
+        self.info.regs.wakeup()
     }
 
     /// Check if the peripheral is currently in sleep mode
     pub fn is_sleeping(&self) -> bool {
-        T::regs().msr().read().slak()
+        self.info.regs.0.msr().read().slak()
     }
 
     /// Put the peripheral in sleep mode
@@ -282,11 +296,11 @@ impl<'d, T: Instance> Can<'d, T> {
     /// If the peripheral has automatic wakeup enabled, when a Start-of-Frame is detected
     /// the peripheral will automatically wake and receive the incoming message.
     pub async fn sleep(&mut self) {
-        T::regs().ier().modify(|i| i.set_slkie(true));
-        T::regs().mcr().modify(|m| m.set_sleep(true));
+        self.info.regs.0.ier().modify(|i| i.set_slkie(true));
+        self.info.regs.0.mcr().modify(|m| m.set_sleep(true));
 
         poll_fn(|cx| {
-            T::state().err_waker.register(cx.waker());
+            self.state.err_waker.register(cx.waker());
             if self.is_sleeping() {
                 Poll::Ready(())
             } else {
@@ -295,7 +309,7 @@ impl<'d, T: Instance> Can<'d, T> {
         })
         .await;
 
-        T::regs().ier().modify(|i| i.set_slkie(false));
+        self.info.regs.0.ier().modify(|i| i.set_slkie(false));
     }
 
     /// Enable FIFO scheduling of outgoing frames.
@@ -307,12 +321,12 @@ impl<'d, T: Instance> Can<'d, T> {
     ///
     /// FIFO scheduling is disabled by default.
     pub fn set_tx_fifo_scheduling(&mut self, enabled: bool) {
-        Registers(T::regs()).set_tx_fifo_scheduling(enabled)
+        self.info.regs.set_tx_fifo_scheduling(enabled)
     }
 
     /// Checks if FIFO scheduling of outgoing frames is enabled.
     pub fn tx_fifo_scheduling_enabled(&self) -> bool {
-        Registers(T::regs()).tx_fifo_scheduling_enabled()
+        self.info.regs.tx_fifo_scheduling_enabled()
     }
 
     /// Queues the message to be sent.
@@ -337,7 +351,13 @@ impl<'d, T: Instance> Can<'d, T> {
 
     /// Waits for a specific transmit mailbox to become empty
     pub async fn flush(&self, mb: Mailbox) {
-        CanTx::<T>::flush_inner(mb).await
+        CanTx {
+            _phantom: PhantomData,
+            info: self.info,
+            state: self.state,
+        }
+        .flush_inner(mb)
+        .await;
     }
 
     /// Waits until any of the transmit mailboxes become empty
@@ -347,12 +367,24 @@ impl<'d, T: Instance> Can<'d, T> {
     /// This will happen if FIFO scheduling of outgoing frames is not enabled,
     /// and a frame with equal priority is already queued for transmission.
     pub async fn flush_any(&self) {
-        CanTx::<T>::flush_any_inner().await
+        CanTx {
+            _phantom: PhantomData,
+            info: self.info,
+            state: self.state,
+        }
+        .flush_any_inner()
+        .await
     }
 
     /// Waits until all of the transmit mailboxes become empty
     pub async fn flush_all(&self) {
-        CanTx::<T>::flush_all_inner().await
+        CanTx {
+            _phantom: PhantomData,
+            info: self.info,
+            state: self.state,
+        }
+        .flush_all_inner()
+        .await
     }
 
     /// Attempts to abort the sending of a frame that is pending in a mailbox.
@@ -363,12 +395,12 @@ impl<'d, T: Instance> Can<'d, T> {
     /// If there is a frame in the provided mailbox, and it is canceled successfully, this function
     /// returns `true`.
     pub fn abort(&mut self, mailbox: Mailbox) -> bool {
-        Registers(T::regs()).abort(mailbox)
+        self.info.regs.abort(mailbox)
     }
 
     /// Returns `true` if no frame is pending for transmission.
     pub fn is_transmitter_idle(&self) -> bool {
-        Registers(T::regs()).is_idle()
+        self.info.regs.is_idle()
     }
 
     /// Read a CAN frame.
@@ -377,31 +409,35 @@ impl<'d, T: Instance> Can<'d, T> {
     ///
     /// Returns a tuple of the time the message was received and the message frame
     pub async fn read(&mut self) -> Result<Envelope, BusError> {
-        T::state().rx_mode.read::<T>().await
+        self.state.rx_mode.read(self.info, self.state).await
     }
 
     /// Attempts to read a CAN frame without blocking.
     ///
     /// Returns [Err(TryReadError::Empty)] if there are no frames in the rx queue.
     pub fn try_read(&mut self) -> Result<Envelope, TryReadError> {
-        T::state().rx_mode.try_read::<T>()
+        self.state.rx_mode.try_read(self.info)
     }
 
     /// Waits while receive queue is empty.
     pub async fn wait_not_empty(&mut self) {
-        T::state().rx_mode.wait_not_empty::<T>().await
+        self.state.rx_mode.wait_not_empty(self.info, self.state).await
     }
 
     /// Split the CAN driver into transmit and receive halves.
     ///
     /// Useful for doing separate transmit/receive tasks.
-    pub fn split<'c>(&'c mut self) -> (CanTx<'d, T>, CanRx<'d, T>) {
+    pub fn split<'c>(&'c mut self) -> (CanTx<'d>, CanRx<'d>) {
         (
             CanTx {
-                _peri: unsafe { self.peri.clone_unchecked() },
+                _phantom: PhantomData,
+                info: self.info,
+                state: self.state,
             },
             CanRx {
-                peri: unsafe { self.peri.clone_unchecked() },
+                _phantom: PhantomData,
+                info: self.info,
+                state: self.state,
             },
         )
     }
@@ -411,7 +447,7 @@ impl<'d, T: Instance> Can<'d, T> {
         &'c mut self,
         txb: &'static mut TxBuf<TX_BUF_SIZE>,
         rxb: &'static mut RxBuf<RX_BUF_SIZE>,
-    ) -> BufferedCan<'d, T, TX_BUF_SIZE, RX_BUF_SIZE> {
+    ) -> BufferedCan<'d, TX_BUF_SIZE, RX_BUF_SIZE> {
         let (tx, rx) = self.split();
         BufferedCan {
             tx: tx.buffered(txb),
@@ -420,23 +456,23 @@ impl<'d, T: Instance> Can<'d, T> {
     }
 }
 
-impl<'d, T: FilterOwner> Can<'d, T> {
+impl<'d> Can<'d> {
     /// Accesses the filter banks owned by this CAN peripheral.
     ///
     /// To modify filters of a slave peripheral, `modify_filters` has to be called on the master
     /// peripheral instead.
-    pub fn modify_filters(&mut self) -> MasterFilters<'_, T> {
-        unsafe { MasterFilters::new(T::regs()) }
+    pub fn modify_filters(&mut self) -> MasterFilters<'_> {
+        unsafe { MasterFilters::new(self.info) }
     }
 }
 
 /// Buffered CAN driver.
-pub struct BufferedCan<'d, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> {
-    tx: BufferedCanTx<'d, T, TX_BUF_SIZE>,
-    rx: BufferedCanRx<'d, T, RX_BUF_SIZE>,
+pub struct BufferedCan<'d, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> {
+    tx: BufferedCanTx<'d, TX_BUF_SIZE>,
+    rx: BufferedCanRx<'d, RX_BUF_SIZE>,
 }
 
-impl<'d, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> BufferedCan<'d, T, TX_BUF_SIZE, RX_BUF_SIZE> {
+impl<'d, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> BufferedCan<'d, TX_BUF_SIZE, RX_BUF_SIZE> {
     /// Async write frame to TX buffer.
     pub async fn write(&mut self, frame: &Frame) {
         self.tx.write(frame).await
@@ -471,18 +507,20 @@ impl<'d, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> Buffer
 }
 
 /// CAN driver, transmit half.
-pub struct CanTx<'d, T: Instance> {
-    _peri: PeripheralRef<'d, T>,
+pub struct CanTx<'d> {
+    _phantom: PhantomData<&'d ()>,
+    info: &'static Info,
+    state: &'static State,
 }
 
-impl<'d, T: Instance> CanTx<'d, T> {
+impl<'d> CanTx<'d> {
     /// Queues the message to be sent.
     ///
     /// If the TX queue is full, this will wait until there is space, therefore exerting backpressure.
     pub async fn write(&mut self, frame: &Frame) -> TransmitStatus {
         poll_fn(|cx| {
-            T::state().tx_mode.register(cx.waker());
-            if let Ok(status) = Registers(T::regs()).transmit(frame) {
+            self.state.tx_mode.register(cx.waker());
+            if let Ok(status) = self.info.regs.transmit(frame) {
                 return Poll::Ready(status);
             }
 
@@ -501,13 +539,13 @@ impl<'d, T: Instance> CanTx<'d, T> {
     /// This is done to work around a hardware limitation that could lead to out-of-order delivery
     /// of frames with the same priority.
     pub fn try_write(&mut self, frame: &Frame) -> Result<TransmitStatus, TryWriteError> {
-        Registers(T::regs()).transmit(frame).map_err(|_| TryWriteError::Full)
+        self.info.regs.transmit(frame).map_err(|_| TryWriteError::Full)
     }
 
-    async fn flush_inner(mb: Mailbox) {
+    async fn flush_inner(&self, mb: Mailbox) {
         poll_fn(|cx| {
-            T::state().tx_mode.register(cx.waker());
-            if T::regs().tsr().read().tme(mb.index()) {
+            self.state.tx_mode.register(cx.waker());
+            if self.info.regs.0.tsr().read().tme(mb.index()) {
                 return Poll::Ready(());
             }
 
@@ -518,14 +556,14 @@ impl<'d, T: Instance> CanTx<'d, T> {
 
     /// Waits for a specific transmit mailbox to become empty
     pub async fn flush(&self, mb: Mailbox) {
-        Self::flush_inner(mb).await
+        self.flush_inner(mb).await
     }
 
-    async fn flush_any_inner() {
+    async fn flush_any_inner(&self) {
         poll_fn(|cx| {
-            T::state().tx_mode.register(cx.waker());
+            self.state.tx_mode.register(cx.waker());
 
-            let tsr = T::regs().tsr().read();
+            let tsr = self.info.regs.0.tsr().read();
             if tsr.tme(Mailbox::Mailbox0.index())
                 || tsr.tme(Mailbox::Mailbox1.index())
                 || tsr.tme(Mailbox::Mailbox2.index())
@@ -545,14 +583,14 @@ impl<'d, T: Instance> CanTx<'d, T> {
     /// This will happen if FIFO scheduling of outgoing frames is not enabled,
     /// and a frame with equal priority is already queued for transmission.
     pub async fn flush_any(&self) {
-        Self::flush_any_inner().await
+        self.flush_any_inner().await
     }
 
-    async fn flush_all_inner() {
+    async fn flush_all_inner(&self) {
         poll_fn(|cx| {
-            T::state().tx_mode.register(cx.waker());
+            self.state.tx_mode.register(cx.waker());
 
-            let tsr = T::regs().tsr().read();
+            let tsr = self.info.regs.0.tsr().read();
             if tsr.tme(Mailbox::Mailbox0.index())
                 && tsr.tme(Mailbox::Mailbox1.index())
                 && tsr.tme(Mailbox::Mailbox2.index())
@@ -567,7 +605,7 @@ impl<'d, T: Instance> CanTx<'d, T> {
 
     /// Waits until all of the transmit mailboxes become empty
     pub async fn flush_all(&self) {
-        Self::flush_all_inner().await
+        self.flush_all_inner().await
     }
 
     /// Attempts to abort the sending of a frame that is pending in a mailbox.
@@ -578,20 +616,20 @@ impl<'d, T: Instance> CanTx<'d, T> {
     /// If there is a frame in the provided mailbox, and it is canceled successfully, this function
     /// returns `true`.
     pub fn abort(&mut self, mailbox: Mailbox) -> bool {
-        Registers(T::regs()).abort(mailbox)
+        self.info.regs.abort(mailbox)
     }
 
     /// Returns `true` if no frame is pending for transmission.
     pub fn is_idle(&self) -> bool {
-        Registers(T::regs()).is_idle()
+        self.info.regs.is_idle()
     }
 
     /// Return a buffered instance of driver. User must supply Buffers
     pub fn buffered<const TX_BUF_SIZE: usize>(
         self,
         txb: &'static mut TxBuf<TX_BUF_SIZE>,
-    ) -> BufferedCanTx<'d, T, TX_BUF_SIZE> {
-        BufferedCanTx::new(self, txb)
+    ) -> BufferedCanTx<'d, TX_BUF_SIZE> {
+        BufferedCanTx::new(self.info, self.state, self, txb)
     }
 }
 
@@ -599,23 +637,35 @@ impl<'d, T: Instance> CanTx<'d, T> {
 pub type TxBuf<const BUF_SIZE: usize> = Channel<CriticalSectionRawMutex, Frame, BUF_SIZE>;
 
 /// Buffered CAN driver, transmit half.
-pub struct BufferedCanTx<'d, T: Instance, const TX_BUF_SIZE: usize> {
-    _tx: CanTx<'d, T>,
+pub struct BufferedCanTx<'d, const TX_BUF_SIZE: usize> {
+    info: &'static Info,
+    state: &'static State,
+    _tx: CanTx<'d>,
     tx_buf: &'static TxBuf<TX_BUF_SIZE>,
 }
 
-impl<'d, T: Instance, const TX_BUF_SIZE: usize> BufferedCanTx<'d, T, TX_BUF_SIZE> {
-    fn new(_tx: CanTx<'d, T>, tx_buf: &'static TxBuf<TX_BUF_SIZE>) -> Self {
-        Self { _tx, tx_buf }.setup()
+impl<'d, const TX_BUF_SIZE: usize> BufferedCanTx<'d, TX_BUF_SIZE> {
+    fn new(info: &'static Info, state: &'static State, _tx: CanTx<'d>, tx_buf: &'static TxBuf<TX_BUF_SIZE>) -> Self {
+        Self {
+            info,
+            state,
+            _tx,
+            tx_buf,
+        }
+        .setup()
     }
 
     fn setup(self) -> Self {
         // We don't want interrupts being processed while we change modes.
-        critical_section::with(|_| unsafe {
+        critical_section::with(|_| {
             let tx_inner = super::common::ClassicBufferedTxInner {
                 tx_receiver: self.tx_buf.receiver().into(),
             };
-            T::mut_state().tx_mode = TxMode::Buffered(tx_inner);
+            let state = self.state as *const State;
+            unsafe {
+                let mut_state = state as *mut State;
+                (*mut_state).tx_mode = TxMode::Buffered(tx_inner);
+            }
         });
         self
     }
@@ -623,60 +673,67 @@ impl<'d, T: Instance, const TX_BUF_SIZE: usize> BufferedCanTx<'d, T, TX_BUF_SIZE
     /// Async write frame to TX buffer.
     pub async fn write(&mut self, frame: &Frame) {
         self.tx_buf.send(*frame).await;
-        T::TXInterrupt::pend(); // Wake for Tx
+        let waker = self.info.tx_waker;
+        waker(); // Wake for Tx
     }
 
     /// Returns a sender that can be used for sending CAN frames.
     pub fn writer(&self) -> BufferedCanSender {
         BufferedCanSender {
             tx_buf: self.tx_buf.sender().into(),
-            waker: T::TXInterrupt::pend,
+            waker: self.info.tx_waker,
         }
     }
 }
 
-impl<'d, T: Instance, const TX_BUF_SIZE: usize> Drop for BufferedCanTx<'d, T, TX_BUF_SIZE> {
+impl<'d, const TX_BUF_SIZE: usize> Drop for BufferedCanTx<'d, TX_BUF_SIZE> {
     fn drop(&mut self) {
-        critical_section::with(|_| unsafe {
-            T::mut_state().tx_mode = TxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+        critical_section::with(|_| {
+            let state = self.state as *const State;
+            unsafe {
+                let mut_state = state as *mut State;
+                (*mut_state).tx_mode = TxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+            }
         });
     }
 }
 
 /// CAN driver, receive half.
 #[allow(dead_code)]
-pub struct CanRx<'d, T: Instance> {
-    peri: PeripheralRef<'d, T>,
+pub struct CanRx<'d> {
+    _phantom: PhantomData<&'d ()>,
+    info: &'static Info,
+    state: &'static State,
 }
 
-impl<'d, T: Instance> CanRx<'d, T> {
+impl<'d> CanRx<'d> {
     /// Read a CAN frame.
     ///
     /// If no CAN frame is in the RX buffer, this will wait until there is one.
     ///
     /// Returns a tuple of the time the message was received and the message frame
     pub async fn read(&mut self) -> Result<Envelope, BusError> {
-        T::state().rx_mode.read::<T>().await
+        self.state.rx_mode.read(self.info, self.state).await
     }
 
     /// Attempts to read a CAN frame without blocking.
     ///
     /// Returns [Err(TryReadError::Empty)] if there are no frames in the rx queue.
     pub fn try_read(&mut self) -> Result<Envelope, TryReadError> {
-        T::state().rx_mode.try_read::<T>()
+        self.state.rx_mode.try_read(self.info)
     }
 
     /// Waits while receive queue is empty.
     pub async fn wait_not_empty(&mut self) {
-        T::state().rx_mode.wait_not_empty::<T>().await
+        self.state.rx_mode.wait_not_empty(self.info, self.state).await
     }
 
     /// Return a buffered instance of driver. User must supply Buffers
     pub fn buffered<const RX_BUF_SIZE: usize>(
         self,
         rxb: &'static mut RxBuf<RX_BUF_SIZE>,
-    ) -> BufferedCanRx<'d, T, RX_BUF_SIZE> {
-        BufferedCanRx::new(self, rxb)
+    ) -> BufferedCanRx<'d, RX_BUF_SIZE> {
+        BufferedCanRx::new(self.info, self.state, self, rxb)
     }
 }
 
@@ -684,23 +741,35 @@ impl<'d, T: Instance> CanRx<'d, T> {
 pub type RxBuf<const BUF_SIZE: usize> = Channel<CriticalSectionRawMutex, Result<Envelope, BusError>, BUF_SIZE>;
 
 /// CAN driver, receive half in Buffered mode.
-pub struct BufferedCanRx<'d, T: Instance, const RX_BUF_SIZE: usize> {
-    _rx: CanRx<'d, T>,
+pub struct BufferedCanRx<'d, const RX_BUF_SIZE: usize> {
+    info: &'static Info,
+    state: &'static State,
+    _rx: CanRx<'d>,
     rx_buf: &'static RxBuf<RX_BUF_SIZE>,
 }
 
-impl<'d, T: Instance, const RX_BUF_SIZE: usize> BufferedCanRx<'d, T, RX_BUF_SIZE> {
-    fn new(_rx: CanRx<'d, T>, rx_buf: &'static RxBuf<RX_BUF_SIZE>) -> Self {
-        BufferedCanRx { _rx, rx_buf }.setup()
+impl<'d, const RX_BUF_SIZE: usize> BufferedCanRx<'d, RX_BUF_SIZE> {
+    fn new(info: &'static Info, state: &'static State, _rx: CanRx<'d>, rx_buf: &'static RxBuf<RX_BUF_SIZE>) -> Self {
+        BufferedCanRx {
+            info,
+            state,
+            _rx,
+            rx_buf,
+        }
+        .setup()
     }
 
     fn setup(self) -> Self {
         // We don't want interrupts being processed while we change modes.
-        critical_section::with(|_| unsafe {
+        critical_section::with(|_| {
             let rx_inner = super::common::ClassicBufferedRxInner {
                 rx_sender: self.rx_buf.sender().into(),
             };
-            T::mut_state().rx_mode = RxMode::Buffered(rx_inner);
+            let state = self.state as *const State;
+            unsafe {
+                let mut_state = state as *mut State;
+                (*mut_state).rx_mode = RxMode::Buffered(rx_inner);
+            }
         });
         self
     }
@@ -714,7 +783,7 @@ impl<'d, T: Instance, const RX_BUF_SIZE: usize> BufferedCanRx<'d, T, RX_BUF_SIZE
     ///
     /// Returns [Err(TryReadError::Empty)] if there are no frames in the rx queue.
     pub fn try_read(&mut self) -> Result<Envelope, TryReadError> {
-        match &T::state().rx_mode {
+        match &self.state.rx_mode {
             RxMode::Buffered(_) => {
                 if let Ok(result) = self.rx_buf.try_receive() {
                     match result {
@@ -722,7 +791,7 @@ impl<'d, T: Instance, const RX_BUF_SIZE: usize> BufferedCanRx<'d, T, RX_BUF_SIZE
                         Err(e) => Err(TryReadError::BusError(e)),
                     }
                 } else {
-                    if let Some(err) = Registers(T::regs()).curr_error() {
+                    if let Some(err) = self.info.regs.curr_error() {
                         return Err(TryReadError::BusError(err));
                     } else {
                         Err(TryReadError::Empty)
@@ -746,20 +815,26 @@ impl<'d, T: Instance, const RX_BUF_SIZE: usize> BufferedCanRx<'d, T, RX_BUF_SIZE
     }
 }
 
-impl<'d, T: Instance, const RX_BUF_SIZE: usize> Drop for BufferedCanRx<'d, T, RX_BUF_SIZE> {
+impl<'d, const RX_BUF_SIZE: usize> Drop for BufferedCanRx<'d, RX_BUF_SIZE> {
     fn drop(&mut self) {
-        critical_section::with(|_| unsafe {
-            T::mut_state().rx_mode = RxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+        critical_section::with(|_| {
+            let state = self.state as *const State;
+            unsafe {
+                let mut_state = state as *mut State;
+                (*mut_state).rx_mode = RxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+            }
         });
     }
 }
 
-impl<'d, T: Instance> Drop for Can<'d, T> {
+impl Drop for Can<'_> {
     fn drop(&mut self) {
         // Cannot call `free()` because it moves the instance.
         // Manually reset the peripheral.
-        T::regs().mcr().write(|w| w.set_reset(true));
-        T::disable();
+        self.info.regs.0.mcr().write(|w| w.set_reset(true));
+        self.info.regs.enter_init_mode();
+        self.info.regs.leave_init_mode();
+        //rcc::disable::<T>();
     }
 }
 
@@ -839,13 +914,13 @@ impl RxMode {
         }
     }
 
-    pub async fn read<T: Instance>(&self) -> Result<Envelope, BusError> {
+    pub(crate) async fn read(&self, info: &Info, state: &State) -> Result<Envelope, BusError> {
         match self {
             Self::NonBuffered(waker) => {
                 poll_fn(|cx| {
-                    T::state().err_waker.register(cx.waker());
+                    state.err_waker.register(cx.waker());
                     waker.register(cx.waker());
-                    match self.try_read::<T>() {
+                    match self.try_read(info) {
                         Ok(result) => Poll::Ready(Ok(result)),
                         Err(TryReadError::Empty) => Poll::Pending,
                         Err(TryReadError::BusError(be)) => Poll::Ready(Err(be)),
@@ -858,17 +933,17 @@ impl RxMode {
             }
         }
     }
-    pub fn try_read<T: Instance>(&self) -> Result<Envelope, TryReadError> {
+    pub(crate) fn try_read(&self, info: &Info) -> Result<Envelope, TryReadError> {
         match self {
             Self::NonBuffered(_) => {
-                let registers = Registers(T::regs());
+                let registers = &info.regs;
                 if let Some(msg) = registers.receive_fifo(RxFifo::Fifo0) {
-                    T::regs().ier().write(|w| {
+                    registers.0.ier().write(|w| {
                         w.set_fmpie(0, true);
                     });
                     Ok(msg)
                 } else if let Some(msg) = registers.receive_fifo(RxFifo::Fifo1) {
-                    T::regs().ier().write(|w| {
+                    registers.0.ier().write(|w| {
                         w.set_fmpie(1, true);
                     });
                     Ok(msg)
@@ -883,12 +958,12 @@ impl RxMode {
             }
         }
     }
-    pub async fn wait_not_empty<T: Instance>(&self) {
-        match &T::state().rx_mode {
+    pub(crate) async fn wait_not_empty(&self, info: &Info, state: &State) {
+        match &state.rx_mode {
             Self::NonBuffered(waker) => {
                 poll_fn(|cx| {
                     waker.register(cx.waker());
-                    if Registers(T::regs()).receive_frame_available() {
+                    if info.regs.receive_frame_available() {
                         Poll::Ready(())
                     } else {
                         Poll::Pending
@@ -903,7 +978,7 @@ impl RxMode {
     }
 }
 
-enum TxMode {
+pub(crate) enum TxMode {
     NonBuffered(AtomicWaker),
     Buffered(super::common::ClassicBufferedTxInner),
 }
@@ -943,7 +1018,7 @@ impl TxMode {
     }
 }
 
-struct State {
+pub(crate) struct State {
     pub(crate) rx_mode: RxMode,
     pub(crate) tx_mode: TxMode,
     pub err_waker: AtomicWaker,
@@ -959,7 +1034,22 @@ impl State {
     }
 }
 
+pub(crate) struct Info {
+    regs: Registers,
+    tx_interrupt: crate::interrupt::Interrupt,
+    rx0_interrupt: crate::interrupt::Interrupt,
+    rx1_interrupt: crate::interrupt::Interrupt,
+    sce_interrupt: crate::interrupt::Interrupt,
+    tx_waker: fn(),
+
+    /// The total number of filter banks available to the instance.
+    ///
+    /// This is usually either 14 or 28, and should be specified in the chip's reference manual or datasheet.
+    num_filter_banks: u8,
+}
+
 trait SealedInstance {
+    fn info() -> &'static Info;
     fn regs() -> crate::pac::can::Can;
     fn state() -> &'static State;
     unsafe fn mut_state() -> &'static mut State;
@@ -1012,6 +1102,18 @@ foreach_peripheral!(
     (can, $inst:ident) => {
         impl SealedInstance for peripherals::$inst {
 
+            fn info() -> &'static Info {
+                static INFO: Info = Info {
+                    regs: Registers(crate::pac::$inst),
+                    tx_interrupt: crate::_generated::peripheral_interrupts::$inst::TX::IRQ,
+                    rx0_interrupt: crate::_generated::peripheral_interrupts::$inst::RX0::IRQ,
+                    rx1_interrupt: crate::_generated::peripheral_interrupts::$inst::RX1::IRQ,
+                    sce_interrupt: crate::_generated::peripheral_interrupts::$inst::SCE::IRQ,
+                    tx_waker: crate::_generated::peripheral_interrupts::$inst::TX::pend,
+                    num_filter_banks: peripherals::$inst::NUM_FILTER_BANKS,
+                };
+                &INFO
+            }
             fn regs() -> crate::pac::can::Can {
                 crate::pac::$inst
             }
@@ -1046,8 +1148,8 @@ foreach_peripheral!(
     (can, CAN1) => {
         cfg_if::cfg_if! {
             if #[cfg(all(
-                any(stm32l4, stm32f72, stm32f73),
-                not(any(stm32l49, stm32l4a))
+                any(stm32l4, stm32f72x, stm32f73x),
+                not(any(stm32l49x, stm32l4ax))
             ))] {
                 // Most L4 devices and some F7 devices use the name "CAN1"
                 // even if there is no "CAN2" peripheral.
@@ -1062,6 +1164,11 @@ foreach_peripheral!(
             }
         }
     };
+    (can, CAN2) => {
+        unsafe impl FilterOwner for peripherals::CAN2 {
+            const NUM_FILTER_BANKS: u8 = 0;
+        }
+    };
     (can, CAN3) => {
         unsafe impl FilterOwner for peripherals::CAN3 {
             const NUM_FILTER_BANKS: u8 = 14;
diff --git a/embassy-stm32/src/can/bxcan/registers.rs b/embassy-stm32/src/can/bxcan/registers.rs
index ad27e0744..5f3d70e25 100644
--- a/embassy-stm32/src/can/bxcan/registers.rs
+++ b/embassy-stm32/src/can/bxcan/registers.rs
@@ -11,7 +11,7 @@ use crate::can::frame::{Envelope, Frame, Header};
 pub(crate) struct Registers(pub crate::pac::can::Can);
 
 impl Registers {
-    pub fn enter_init_mode(&mut self) {
+    pub fn enter_init_mode(&self) {
         self.0.mcr().modify(|reg| {
             reg.set_sleep(false);
             reg.set_inrq(true);
@@ -25,7 +25,7 @@ impl Registers {
     }
 
     // Leaves initialization mode, enters sleep mode.
-    pub fn leave_init_mode(&mut self) {
+    pub fn leave_init_mode(&self) {
         self.0.mcr().modify(|reg| {
             reg.set_sleep(true);
             reg.set_inrq(false);
@@ -38,7 +38,7 @@ impl Registers {
         }
     }
 
-    pub fn set_bit_timing(&mut self, bt: crate::can::util::NominalBitTiming) {
+    pub fn set_bit_timing(&self, bt: crate::can::util::NominalBitTiming) {
         let prescaler = u16::from(bt.prescaler) & 0x1FF;
         let seg1 = u8::from(bt.seg1);
         let seg2 = u8::from(bt.seg2) & 0x7F;
@@ -84,7 +84,7 @@ impl Registers {
     /// receive the frame. If enabled, [`Interrupt::Wakeup`] will also be triggered by the incoming
     /// frame.
     #[allow(dead_code)]
-    pub fn set_automatic_wakeup(&mut self, enabled: bool) {
+    pub fn set_automatic_wakeup(&self, enabled: bool) {
         self.0.mcr().modify(|reg| reg.set_awum(enabled));
     }
 
@@ -96,7 +96,7 @@ impl Registers {
     /// If this returns [`WouldBlock`][nb::Error::WouldBlock], the peripheral will enable itself
     /// in the background. The peripheral is enabled and ready to use when this method returns
     /// successfully.
-    pub fn enable_non_blocking(&mut self) -> nb::Result<(), Infallible> {
+    pub fn enable_non_blocking(&self) -> nb::Result<(), Infallible> {
         let msr = self.0.msr().read();
         if msr.slak() {
             self.0.mcr().modify(|reg| {
@@ -131,7 +131,7 @@ impl Registers {
     /// Note that this will not trigger [`Interrupt::Wakeup`], only reception of an incoming CAN
     /// frame will cause that interrupt.
     #[allow(dead_code)]
-    pub fn wakeup(&mut self) {
+    pub fn wakeup(&self) {
         self.0.mcr().modify(|reg| {
             reg.set_sleep(false);
             reg.set_inrq(false);
@@ -186,7 +186,7 @@ impl Registers {
     /// If this is enabled, mailboxes are scheduled based on the time when the transmit request bit of the mailbox was set.
     ///
     /// If this is disabled, mailboxes are scheduled based on the priority of the frame in the mailbox.
-    pub fn set_tx_fifo_scheduling(&mut self, enabled: bool) {
+    pub fn set_tx_fifo_scheduling(&self, enabled: bool) {
         self.0.mcr().modify(|w| w.set_txfp(enabled))
     }
 
@@ -216,7 +216,7 @@ impl Registers {
     /// If FIFO scheduling is enabled, frames are transmitted in the order that they are passed to this function.
     ///
     /// If all transmit mailboxes are full, this function returns [`nb::Error::WouldBlock`].
-    pub fn transmit(&mut self, frame: &Frame) -> nb::Result<TransmitStatus, Infallible> {
+    pub fn transmit(&self, frame: &Frame) -> nb::Result<TransmitStatus, Infallible> {
         // Check if FIFO scheduling is enabled.
         let fifo_scheduling = self.0.mcr().read().txfp();
 
@@ -292,7 +292,7 @@ impl Registers {
         Ok(())
     }
 
-    fn write_mailbox(&mut self, idx: usize, frame: &Frame) {
+    fn write_mailbox(&self, idx: usize, frame: &Frame) {
         debug_assert!(idx < 3);
 
         let mb = self.0.tx(idx);
@@ -309,7 +309,7 @@ impl Registers {
         });
     }
 
-    fn read_pending_mailbox(&mut self, idx: usize) -> Option<Frame> {
+    fn read_pending_mailbox(&self, idx: usize) -> Option<Frame> {
         if self.abort_by_index(idx) {
             debug_assert!(idx < 3);
 
@@ -332,7 +332,7 @@ impl Registers {
     }
 
     /// Tries to abort a pending frame. Returns `true` when aborted.
-    fn abort_by_index(&mut self, idx: usize) -> bool {
+    fn abort_by_index(&self, idx: usize) -> bool {
         self.0.tsr().write(|reg| reg.set_abrq(idx, true));
 
         // Wait for the abort request to be finished.
@@ -351,7 +351,7 @@ impl Registers {
     ///
     /// If there is a frame in the provided mailbox, and it is canceled successfully, this function
     /// returns `true`.
-    pub fn abort(&mut self, mailbox: Mailbox) -> bool {
+    pub fn abort(&self, mailbox: Mailbox) -> bool {
         // If the mailbox is empty, the value of TXOKx depends on what happened with the previous
         // frame in that mailbox. Only call abort_by_index() if the mailbox is not empty.
         let tsr = self.0.tsr().read();
diff --git a/embassy-stm32/src/can/fd/peripheral.rs b/embassy-stm32/src/can/fd/peripheral.rs
index 9cd5f0785..07e3dddad 100644
--- a/embassy-stm32/src/can/fd/peripheral.rs
+++ b/embassy-stm32/src/can/fd/peripheral.rs
@@ -20,8 +20,9 @@ enum LoopbackMode {
 }
 
 pub struct Registers {
-    pub regs: &'static crate::pac::can::Fdcan,
-    pub msgram: &'static crate::pac::fdcanram::Fdcanram,
+    pub regs: crate::pac::can::Fdcan,
+    pub msgram: crate::pac::fdcanram::Fdcanram,
+    #[allow(dead_code)]
     pub msg_ram_offset: usize,
 }
 
@@ -72,7 +73,6 @@ impl Registers {
 
     pub fn put_tx_frame(&self, bufidx: usize, header: &Header, buffer: &[u8]) {
         let mailbox = self.tx_buffer_element(bufidx);
-
         mailbox.reset();
         put_tx_header(mailbox, header);
         put_tx_data(mailbox, &buffer[..header.len() as usize]);
@@ -244,12 +244,12 @@ impl Registers {
     }
 
     #[inline]
-    fn reset_msg_ram(&mut self) {
+    fn reset_msg_ram(&self) {
         self.msg_ram_mut().reset();
     }
 
     #[inline]
-    fn enter_init_mode(&mut self) {
+    fn enter_init_mode(&self) {
         self.regs.cccr().modify(|w| w.set_init(true));
         while false == self.regs.cccr().read().init() {}
         self.regs.cccr().modify(|w| w.set_cce(true));
@@ -258,7 +258,7 @@ impl Registers {
     /// Enables or disables loopback mode: Internally connects the TX and RX
     /// signals together.
     #[inline]
-    fn set_loopback_mode(&mut self, mode: LoopbackMode) {
+    fn set_loopback_mode(&self, mode: LoopbackMode) {
         let (test, mon, lbck) = match mode {
             LoopbackMode::None => (false, false, false),
             LoopbackMode::Internal => (true, true, true),
@@ -273,34 +273,34 @@ impl Registers {
 
     /// Enables or disables silent mode: Disconnects the TX signal from the pin.
     #[inline]
-    fn set_bus_monitoring_mode(&mut self, enabled: bool) {
+    fn set_bus_monitoring_mode(&self, enabled: bool) {
         self.regs.cccr().modify(|w| w.set_mon(enabled));
     }
 
     #[inline]
-    fn set_restricted_operations(&mut self, enabled: bool) {
+    fn set_restricted_operations(&self, enabled: bool) {
         self.regs.cccr().modify(|w| w.set_asm(enabled));
     }
 
     #[inline]
-    fn set_normal_operations(&mut self, _enabled: bool) {
+    fn set_normal_operations(&self, _enabled: bool) {
         self.set_loopback_mode(LoopbackMode::None);
     }
 
     #[inline]
-    fn set_test_mode(&mut self, enabled: bool) {
+    fn set_test_mode(&self, enabled: bool) {
         self.regs.cccr().modify(|w| w.set_test(enabled));
     }
 
     #[inline]
-    fn set_power_down_mode(&mut self, enabled: bool) {
+    fn set_power_down_mode(&self, enabled: bool) {
         self.regs.cccr().modify(|w| w.set_csr(enabled));
         while self.regs.cccr().read().csa() != enabled {}
     }
 
     /// Moves out of PoweredDownMode and into ConfigMode
     #[inline]
-    pub fn into_config_mode(mut self, _config: FdCanConfig) {
+    pub fn into_config_mode(self, _config: FdCanConfig) {
         self.set_power_down_mode(false);
         self.enter_init_mode();
         self.reset_msg_ram();
@@ -327,7 +327,7 @@ impl Registers {
 
     /// Applies the settings of a new FdCanConfig See [`FdCanConfig`]
     #[inline]
-    pub fn apply_config(&mut self, config: FdCanConfig) {
+    pub fn apply_config(&self, config: FdCanConfig) {
         self.set_tx_buffer_mode(config.tx_buffer_mode);
 
         // set standard filters list size to 28
@@ -388,7 +388,7 @@ impl Registers {
     }
 
     #[inline]
-    fn leave_init_mode(&mut self, config: FdCanConfig) {
+    fn leave_init_mode(&self, config: FdCanConfig) {
         self.apply_config(config);
 
         self.regs.cccr().modify(|w| w.set_cce(false));
@@ -398,7 +398,7 @@ impl Registers {
 
     /// Moves out of ConfigMode and into specified mode
     #[inline]
-    pub fn into_mode(mut self, config: FdCanConfig, mode: crate::can::_version::OperatingMode) {
+    pub fn into_mode(&self, config: FdCanConfig, mode: crate::can::_version::OperatingMode) {
         match mode {
             crate::can::OperatingMode::InternalLoopbackMode => self.set_loopback_mode(LoopbackMode::Internal),
             crate::can::OperatingMode::ExternalLoopbackMode => self.set_loopback_mode(LoopbackMode::External),
@@ -422,7 +422,7 @@ impl Registers {
     /// Then copy the `CAN_BUS_TIME` register value from the table and pass it as the `btr`
     /// parameter to this method.
     #[inline]
-    pub fn set_nominal_bit_timing(&mut self, btr: NominalBitTiming) {
+    pub fn set_nominal_bit_timing(&self, btr: NominalBitTiming) {
         self.regs.nbtp().write(|w| {
             w.set_nbrp(btr.nbrp() - 1);
             w.set_ntseg1(btr.ntseg1() - 1);
@@ -434,7 +434,7 @@ impl Registers {
     /// Configures the data bit timings for the FdCan Variable Bitrates.
     /// This is not used when frame_transmit is set to anything other than AllowFdCanAndBRS.
     #[inline]
-    pub fn set_data_bit_timing(&mut self, btr: DataBitTiming) {
+    pub fn set_data_bit_timing(&self, btr: DataBitTiming) {
         self.regs.dbtp().write(|w| {
             w.set_dbrp(btr.dbrp() - 1);
             w.set_dtseg1(btr.dtseg1() - 1);
@@ -450,39 +450,39 @@ impl Registers {
     ///
     /// Automatic retransmission is enabled by default.
     #[inline]
-    pub fn set_automatic_retransmit(&mut self, enabled: bool) {
+    pub fn set_automatic_retransmit(&self, enabled: bool) {
         self.regs.cccr().modify(|w| w.set_dar(!enabled));
     }
 
     /// Configures the transmit pause feature. See
     /// [`FdCanConfig::set_transmit_pause`]
     #[inline]
-    pub fn set_transmit_pause(&mut self, enabled: bool) {
+    pub fn set_transmit_pause(&self, enabled: bool) {
         self.regs.cccr().modify(|w| w.set_txp(!enabled));
     }
 
     /// Configures non-iso mode. See [`FdCanConfig::set_non_iso_mode`]
     #[inline]
-    pub fn set_non_iso_mode(&mut self, enabled: bool) {
+    pub fn set_non_iso_mode(&self, enabled: bool) {
         self.regs.cccr().modify(|w| w.set_niso(enabled));
     }
 
     /// Configures edge filtering. See [`FdCanConfig::set_edge_filtering`]
     #[inline]
-    pub fn set_edge_filtering(&mut self, enabled: bool) {
+    pub fn set_edge_filtering(&self, enabled: bool) {
         self.regs.cccr().modify(|w| w.set_efbi(enabled));
     }
 
     /// Configures TX Buffer Mode
     #[inline]
-    pub fn set_tx_buffer_mode(&mut self, tbm: TxBufferMode) {
+    pub fn set_tx_buffer_mode(&self, tbm: TxBufferMode) {
         self.regs.txbc().write(|w| w.set_tfqm(tbm.into()));
     }
 
     /// Configures frame transmission mode. See
     /// [`FdCanConfig::set_frame_transmit`]
     #[inline]
-    pub fn set_frame_transmit(&mut self, fts: FrameTransmissionConfig) {
+    pub fn set_frame_transmit(&self, fts: FrameTransmissionConfig) {
         let (fdoe, brse) = match fts {
             FrameTransmissionConfig::ClassicCanOnly => (false, false),
             FrameTransmissionConfig::AllowFdCan => (true, false),
@@ -500,14 +500,14 @@ impl Registers {
 
     /// Sets the protocol exception handling on/off
     #[inline]
-    pub fn set_protocol_exception_handling(&mut self, enabled: bool) {
+    pub fn set_protocol_exception_handling(&self, enabled: bool) {
         self.regs.cccr().modify(|w| w.set_pxhd(!enabled));
     }
 
     /// Configures and resets the timestamp counter
     #[inline]
     #[allow(unused)]
-    pub fn set_timestamp_counter_source(&mut self, select: TimestampSource) {
+    pub fn set_timestamp_counter_source(&self, select: TimestampSource) {
         #[cfg(can_fdcan_h7)]
         let (tcp, tss) = match select {
             TimestampSource::None => (0, 0),
@@ -531,7 +531,7 @@ impl Registers {
     #[cfg(not(can_fdcan_h7))]
     /// Configures the global filter settings
     #[inline]
-    pub fn set_global_filter(&mut self, filter: GlobalFilter) {
+    pub fn set_global_filter(&self, filter: GlobalFilter) {
         let anfs = match filter.handle_standard_frames {
             crate::can::fd::config::NonMatchingFilter::IntoRxFifo0 => stm32_metapac::can::vals::Anfs::ACCEPT_FIFO_0,
             crate::can::fd::config::NonMatchingFilter::IntoRxFifo1 => stm32_metapac::can::vals::Anfs::ACCEPT_FIFO_1,
@@ -554,7 +554,7 @@ impl Registers {
     #[cfg(can_fdcan_h7)]
     /// Configures the global filter settings
     #[inline]
-    pub fn set_global_filter(&mut self, filter: GlobalFilter) {
+    pub fn set_global_filter(&self, filter: GlobalFilter) {
         let anfs = match filter.handle_standard_frames {
             crate::can::fd::config::NonMatchingFilter::IntoRxFifo0 => 0,
             crate::can::fd::config::NonMatchingFilter::IntoRxFifo1 => 1,
@@ -576,10 +576,10 @@ impl Registers {
     }
 
     #[cfg(not(can_fdcan_h7))]
-    fn configure_msg_ram(&mut self) {}
+    fn configure_msg_ram(&self) {}
 
     #[cfg(can_fdcan_h7)]
-    fn configure_msg_ram(&mut self) {
+    fn configure_msg_ram(&self) {
         let r = self.regs;
 
         use crate::can::fd::message_ram::*;
diff --git a/embassy-stm32/src/can/fdcan.rs b/embassy-stm32/src/can/fdcan.rs
index 81ceb06aa..f3ad718ae 100644
--- a/embassy-stm32/src/can/fdcan.rs
+++ b/embassy-stm32/src/can/fdcan.rs
@@ -3,6 +3,7 @@ use core::future::poll_fn;
 use core::marker::PhantomData;
 use core::task::Poll;
 
+use embassy_hal_internal::interrupt::InterruptExt;
 use embassy_hal_internal::{into_ref, PeripheralRef};
 use embassy_sync::blocking_mutex::raw::CriticalSectionRawMutex;
 use embassy_sync::channel::{Channel, DynamicReceiver, DynamicSender};
@@ -11,7 +12,7 @@ use embassy_sync::waitqueue::AtomicWaker;
 use crate::can::fd::peripheral::Registers;
 use crate::gpio::AFType;
 use crate::interrupt::typelevel::Interrupt;
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::{interrupt, peripherals, Peripheral};
 
 pub(crate) mod fd;
@@ -40,7 +41,7 @@ pub struct IT0InterruptHandler<T: Instance> {
 // We use IT0 for everything currently
 impl<T: Instance> interrupt::typelevel::Handler<T::IT0Interrupt> for IT0InterruptHandler<T> {
     unsafe fn on_interrupt() {
-        let regs = T::regs();
+        let regs = T::registers().regs;
 
         let ir = regs.ir().read();
 
@@ -140,22 +141,16 @@ pub enum OperatingMode {
     //TestMode,
 }
 
-/// FDCAN Configuration instance instance
-/// Create instance of this first
-pub struct CanConfigurator<'d, T: Instance> {
-    config: crate::can::fd::config::FdCanConfig,
-    /// Reference to internals.
-    instance: FdcanInstance<'d, T>,
-    properties: Properties<T>,
-}
-
-fn calc_ns_per_timer_tick<T: Instance>(mode: crate::can::fd::config::FrameTransmissionConfig) -> u64 {
+fn calc_ns_per_timer_tick(
+    info: &'static Info,
+    freq: crate::time::Hertz,
+    mode: crate::can::fd::config::FrameTransmissionConfig,
+) -> u64 {
     match mode {
         // Use timestamp from Rx FIFO to adjust timestamp reported to user
         crate::can::fd::config::FrameTransmissionConfig::ClassicCanOnly => {
-            let freq = T::frequency();
-            let prescale: u64 =
-                ({ T::regs().nbtp().read().nbrp() } + 1) as u64 * ({ T::regs().tscc().read().tcp() } + 1) as u64;
+            let prescale: u64 = ({ info.regs.regs.nbtp().read().nbrp() } + 1) as u64
+                * ({ info.regs.regs.tscc().read().tcp() } + 1) as u64;
             1_000_000_000 as u64 / (freq.0 as u64 * prescale)
         }
         // For VBR this is too hard because the FDCAN timer switches clock rate you need to configure to use
@@ -164,23 +159,35 @@ fn calc_ns_per_timer_tick<T: Instance>(mode: crate::can::fd::config::FrameTransm
     }
 }
 
-impl<'d, T: Instance> CanConfigurator<'d, T> {
+/// FDCAN Configuration instance instance
+/// Create instance of this first
+pub struct CanConfigurator<'d> {
+    _phantom: PhantomData<&'d ()>,
+    config: crate::can::fd::config::FdCanConfig,
+    info: &'static Info,
+    state: &'static State,
+    /// Reference to internals.
+    properties: Properties,
+    periph_clock: crate::time::Hertz,
+}
+
+impl<'d> CanConfigurator<'d> {
     /// Creates a new Fdcan instance, keeping the peripheral in sleep mode.
     /// You must call [Fdcan::enable_non_blocking] to use the peripheral.
-    pub fn new(
-        peri: impl Peripheral<P = T> + 'd,
+    pub fn new<T: Instance>(
+        _peri: impl Peripheral<P = T> + 'd,
         rx: impl Peripheral<P = impl RxPin<T>> + 'd,
         tx: impl Peripheral<P = impl TxPin<T>> + 'd,
         _irqs: impl interrupt::typelevel::Binding<T::IT0Interrupt, IT0InterruptHandler<T>>
             + interrupt::typelevel::Binding<T::IT1Interrupt, IT1InterruptHandler<T>>
             + 'd,
-    ) -> CanConfigurator<'d, T> {
-        into_ref!(peri, rx, tx);
+    ) -> CanConfigurator<'d> {
+        into_ref!(_peri, rx, tx);
 
         rx.set_as_af(rx.af_num(), AFType::Input);
         tx.set_as_af(tx.af_num(), AFType::OutputPushPull);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         let mut config = crate::can::fd::config::FdCanConfig::default();
         config.timestamp_source = TimestampSource::Prescaler(TimestampPrescaler::_1);
@@ -196,16 +203,18 @@ impl<'d, T: Instance> CanConfigurator<'d, T> {
             T::IT1Interrupt::unpend(); // Not unsafe
             T::IT1Interrupt::enable();
         }
-
         Self {
+            _phantom: PhantomData,
             config,
-            instance: FdcanInstance(peri),
-            properties: Properties::new(),
+            info: T::info(),
+            state: T::state(),
+            properties: Properties::new(T::info()),
+            periph_clock: T::frequency(),
         }
     }
 
     /// Get driver properties
-    pub fn properties(&self) -> &Properties<T> {
+    pub fn properties(&self) -> &Properties {
         &self.properties
     }
 
@@ -221,7 +230,7 @@ impl<'d, T: Instance> CanConfigurator<'d, T> {
 
     /// Configures the bit timings calculated from supplied bitrate.
     pub fn set_bitrate(&mut self, bitrate: u32) {
-        let bit_timing = util::calc_can_timings(T::frequency(), bitrate).unwrap();
+        let bit_timing = util::calc_can_timings(self.periph_clock, bitrate).unwrap();
 
         let nbtr = crate::can::fd::config::NominalBitTiming {
             sync_jump_width: bit_timing.sync_jump_width,
@@ -234,7 +243,7 @@ impl<'d, T: Instance> CanConfigurator<'d, T> {
 
     /// Configures the bit timings for VBR data calculated from supplied bitrate. This also sets confit to allow can FD and VBR
     pub fn set_fd_data_bitrate(&mut self, bitrate: u32, transceiver_delay_compensation: bool) {
-        let bit_timing = util::calc_can_timings(T::frequency(), bitrate).unwrap();
+        let bit_timing = util::calc_can_timings(self.periph_clock, bitrate).unwrap();
         // Note, used existing calcluation for normal(non-VBR) bitrate, appears to work for 250k/1M
         let nbtr = crate::can::fd::config::DataBitTiming {
             transceiver_delay_compensation,
@@ -248,62 +257,68 @@ impl<'d, T: Instance> CanConfigurator<'d, T> {
     }
 
     /// Start in mode.
-    pub fn start(self, mode: OperatingMode) -> Can<'d, T> {
-        let ns_per_timer_tick = calc_ns_per_timer_tick::<T>(self.config.frame_transmit);
-        critical_section::with(|_| unsafe {
-            T::mut_state().ns_per_timer_tick = ns_per_timer_tick;
+    pub fn start(self, mode: OperatingMode) -> Can<'d> {
+        let ns_per_timer_tick = calc_ns_per_timer_tick(self.info, self.periph_clock, self.config.frame_transmit);
+        critical_section::with(|_| {
+            let state = self.state as *const State;
+            unsafe {
+                let mut_state = state as *mut State;
+                (*mut_state).ns_per_timer_tick = ns_per_timer_tick;
+            }
         });
-        T::registers().into_mode(self.config, mode);
-        let ret = Can {
+        self.info.regs.into_mode(self.config, mode);
+        Can {
+            _phantom: PhantomData,
             config: self.config,
-            instance: self.instance,
+            info: self.info,
+            state: self.state,
             _mode: mode,
-            properties: self.properties,
-        };
-        ret
+            properties: Properties::new(self.info),
+        }
     }
 
     /// Start, entering mode. Does same as start(mode)
-    pub fn into_normal_mode(self) -> Can<'d, T> {
+    pub fn into_normal_mode(self) -> Can<'d> {
         self.start(OperatingMode::NormalOperationMode)
     }
 
     /// Start, entering mode. Does same as start(mode)
-    pub fn into_internal_loopback_mode(self) -> Can<'d, T> {
+    pub fn into_internal_loopback_mode(self) -> Can<'d> {
         self.start(OperatingMode::InternalLoopbackMode)
     }
 
     /// Start, entering mode. Does same as start(mode)
-    pub fn into_external_loopback_mode(self) -> Can<'d, T> {
+    pub fn into_external_loopback_mode(self) -> Can<'d> {
         self.start(OperatingMode::ExternalLoopbackMode)
     }
 }
 
 /// FDCAN Instance
-pub struct Can<'d, T: Instance> {
+pub struct Can<'d> {
+    _phantom: PhantomData<&'d ()>,
     config: crate::can::fd::config::FdCanConfig,
-    /// Reference to internals.
-    instance: FdcanInstance<'d, T>,
+    info: &'static Info,
+    state: &'static State,
     _mode: OperatingMode,
-    properties: Properties<T>,
+    properties: Properties,
 }
 
-impl<'d, T: Instance> Can<'d, T> {
+impl<'d> Can<'d> {
     /// Get driver properties
-    pub fn properties(&self) -> &Properties<T> {
+    pub fn properties(&self) -> &Properties {
         &self.properties
     }
 
     /// Flush one of the TX mailboxes.
     pub async fn flush(&self, idx: usize) {
         poll_fn(|cx| {
-            T::state().tx_mode.register(cx.waker());
+            self.state.tx_mode.register(cx.waker());
 
             if idx > 3 {
                 panic!("Bad mailbox");
             }
             let idx = 1 << idx;
-            if !T::regs().txbrp().read().trp(idx) {
+            if !self.info.regs.regs.txbrp().read().trp(idx) {
                 return Poll::Ready(());
             }
 
@@ -317,12 +332,12 @@ impl<'d, T: Instance> Can<'d, T> {
     /// can be replaced, this call asynchronously waits for a frame to be successfully
     /// transmitted, then tries again.
     pub async fn write(&mut self, frame: &Frame) -> Option<Frame> {
-        T::state().tx_mode.write::<T>(frame).await
+        self.state.tx_mode.write(self.info, frame).await
     }
 
     /// Returns the next received message frame
     pub async fn read(&mut self) -> Result<Envelope, BusError> {
-        T::state().rx_mode.read_classic::<T>().await
+        self.state.rx_mode.read_classic(self.info, self.state).await
     }
 
     /// Queues the message to be sent but exerts backpressure.  If a lower-priority
@@ -330,58 +345,61 @@ impl<'d, T: Instance> Can<'d, T> {
     /// can be replaced, this call asynchronously waits for a frame to be successfully
     /// transmitted, then tries again.
     pub async fn write_fd(&mut self, frame: &FdFrame) -> Option<FdFrame> {
-        T::state().tx_mode.write_fd::<T>(frame).await
+        self.state.tx_mode.write_fd(self.info, frame).await
     }
 
     /// Returns the next received message frame
     pub async fn read_fd(&mut self) -> Result<FdEnvelope, BusError> {
-        T::state().rx_mode.read_fd::<T>().await
+        self.state.rx_mode.read_fd(self.info, self.state).await
     }
 
     /// Split instance into separate portions: Tx(write), Rx(read), common properties
-    pub fn split(self) -> (CanTx<'d, T>, CanRx<'d, T>, Properties<T>) {
+    pub fn split(self) -> (CanTx<'d>, CanRx<'d>, Properties) {
         (
             CanTx {
+                _phantom: PhantomData,
+                info: self.info,
+                state: self.state,
                 config: self.config,
-                _instance: self.instance,
                 _mode: self._mode,
             },
             CanRx {
-                _instance1: PhantomData::<T>,
-                _instance2: T::regs(),
+                _phantom: PhantomData,
+                info: self.info,
+                state: self.state,
                 _mode: self._mode,
             },
             self.properties,
         )
     }
-
     /// Join split rx and tx portions back together
-    pub fn join(tx: CanTx<'d, T>, rx: CanRx<'d, T>) -> Self {
+    pub fn join(tx: CanTx<'d>, rx: CanRx<'d>) -> Self {
         Can {
+            _phantom: PhantomData,
             config: tx.config,
-            //_instance2: T::regs(),
-            instance: tx._instance,
+            info: tx.info,
+            state: tx.state,
             _mode: rx._mode,
-            properties: Properties::new(),
+            properties: Properties::new(tx.info),
         }
     }
 
     /// Return a buffered instance of driver without CAN FD support. User must supply Buffers
     pub fn buffered<const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>(
-        &self,
+        self,
         tx_buf: &'static mut TxBuf<TX_BUF_SIZE>,
         rxb: &'static mut RxBuf<RX_BUF_SIZE>,
-    ) -> BufferedCan<'d, T, TX_BUF_SIZE, RX_BUF_SIZE> {
-        BufferedCan::new(PhantomData::<T>, T::regs(), self._mode, tx_buf, rxb)
+    ) -> BufferedCan<'d, TX_BUF_SIZE, RX_BUF_SIZE> {
+        BufferedCan::new(self.info, self.state, self._mode, tx_buf, rxb)
     }
 
     /// Return a buffered instance of driver with CAN FD support. User must supply Buffers
     pub fn buffered_fd<const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>(
-        &self,
+        self,
         tx_buf: &'static mut TxFdBuf<TX_BUF_SIZE>,
         rxb: &'static mut RxFdBuf<RX_BUF_SIZE>,
-    ) -> BufferedCanFd<'d, T, TX_BUF_SIZE, RX_BUF_SIZE> {
-        BufferedCanFd::new(PhantomData::<T>, T::regs(), self._mode, tx_buf, rxb)
+    ) -> BufferedCanFd<'d, TX_BUF_SIZE, RX_BUF_SIZE> {
+        BufferedCanFd::new(self.info, self.state, self._mode, tx_buf, rxb)
     }
 }
 
@@ -392,52 +410,56 @@ pub type RxBuf<const BUF_SIZE: usize> = Channel<CriticalSectionRawMutex, Result<
 pub type TxBuf<const BUF_SIZE: usize> = Channel<CriticalSectionRawMutex, Frame, BUF_SIZE>;
 
 /// Buffered FDCAN Instance
-pub struct BufferedCan<'d, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> {
-    _instance1: PhantomData<T>,
-    _instance2: &'d crate::pac::can::Fdcan,
+pub struct BufferedCan<'d, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> {
+    _phantom: PhantomData<&'d ()>,
+    info: &'static Info,
+    state: &'static State,
     _mode: OperatingMode,
     tx_buf: &'static TxBuf<TX_BUF_SIZE>,
     rx_buf: &'static RxBuf<RX_BUF_SIZE>,
-    properties: Properties<T>,
+    properties: Properties,
 }
 
-impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>
-    BufferedCan<'d, T, TX_BUF_SIZE, RX_BUF_SIZE>
-{
+impl<'c, 'd, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> BufferedCan<'d, TX_BUF_SIZE, RX_BUF_SIZE> {
     fn new(
-        _instance1: PhantomData<T>,
-        _instance2: &'d crate::pac::can::Fdcan,
+        info: &'static Info,
+        state: &'static State,
         _mode: OperatingMode,
         tx_buf: &'static TxBuf<TX_BUF_SIZE>,
         rx_buf: &'static RxBuf<RX_BUF_SIZE>,
     ) -> Self {
         BufferedCan {
-            _instance1,
-            _instance2,
+            _phantom: PhantomData,
+            info,
+            state,
             _mode,
             tx_buf,
             rx_buf,
-            properties: Properties::new(),
+            properties: Properties::new(info),
         }
         .setup()
     }
 
     /// Get driver properties
-    pub fn properties(&self) -> &Properties<T> {
+    pub fn properties(&self) -> &Properties {
         &self.properties
     }
 
     fn setup(self) -> Self {
         // We don't want interrupts being processed while we change modes.
-        critical_section::with(|_| unsafe {
+        critical_section::with(|_| {
             let rx_inner = super::common::ClassicBufferedRxInner {
                 rx_sender: self.rx_buf.sender().into(),
             };
             let tx_inner = super::common::ClassicBufferedTxInner {
                 tx_receiver: self.tx_buf.receiver().into(),
             };
-            T::mut_state().rx_mode = RxMode::ClassicBuffered(rx_inner);
-            T::mut_state().tx_mode = TxMode::ClassicBuffered(tx_inner);
+            let state = self.state as *const State;
+            unsafe {
+                let mut_state = state as *mut State;
+                (*mut_state).rx_mode = RxMode::ClassicBuffered(rx_inner);
+                (*mut_state).tx_mode = TxMode::ClassicBuffered(tx_inner);
+            }
         });
         self
     }
@@ -445,7 +467,8 @@ impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>
     /// Async write frame to TX buffer.
     pub async fn write(&mut self, frame: Frame) {
         self.tx_buf.send(frame).await;
-        T::IT0Interrupt::pend(); // Wake for Tx
+        self.info.interrupt0.pend(); // Wake for Tx
+                                     //T::IT0Interrupt::pend(); // Wake for Tx
     }
 
     /// Async read frame from RX buffer.
@@ -457,7 +480,7 @@ impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>
     pub fn writer(&self) -> BufferedCanSender {
         BufferedCanSender {
             tx_buf: self.tx_buf.sender().into(),
-            waker: T::IT0Interrupt::pend,
+            waker: self.info.tx_waker,
         }
     }
 
@@ -467,13 +490,15 @@ impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>
     }
 }
 
-impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> Drop
-    for BufferedCan<'d, T, TX_BUF_SIZE, RX_BUF_SIZE>
-{
+impl<'c, 'd, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> Drop for BufferedCan<'d, TX_BUF_SIZE, RX_BUF_SIZE> {
     fn drop(&mut self) {
-        critical_section::with(|_| unsafe {
-            T::mut_state().rx_mode = RxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
-            T::mut_state().tx_mode = TxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+        critical_section::with(|_| {
+            let state = self.state as *const State;
+            unsafe {
+                let mut_state = state as *mut State;
+                (*mut_state).rx_mode = RxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+                (*mut_state).tx_mode = TxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+            }
         });
     }
 }
@@ -484,16 +509,6 @@ pub type RxFdBuf<const BUF_SIZE: usize> = Channel<CriticalSectionRawMutex, Resul
 /// User supplied buffer for TX buffering
 pub type TxFdBuf<const BUF_SIZE: usize> = Channel<CriticalSectionRawMutex, FdFrame, BUF_SIZE>;
 
-/// Buffered FDCAN Instance
-pub struct BufferedCanFd<'d, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> {
-    _instance1: PhantomData<T>,
-    _instance2: &'d crate::pac::can::Fdcan,
-    _mode: OperatingMode,
-    tx_buf: &'static TxFdBuf<TX_BUF_SIZE>,
-    rx_buf: &'static RxFdBuf<RX_BUF_SIZE>,
-    properties: Properties<T>,
-}
-
 /// Sender that can be used for sending CAN frames.
 #[derive(Copy, Clone)]
 pub struct BufferedFdCanSender {
@@ -524,43 +539,57 @@ impl BufferedFdCanSender {
 /// Receiver that can be used for receiving CAN frames. Note, each CAN frame will only be received by one receiver.
 pub type BufferedFdCanReceiver = DynamicReceiver<'static, Result<FdEnvelope, BusError>>;
 
-impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>
-    BufferedCanFd<'d, T, TX_BUF_SIZE, RX_BUF_SIZE>
-{
+/// Buffered FDCAN Instance
+pub struct BufferedCanFd<'d, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> {
+    _phantom: PhantomData<&'d ()>,
+    info: &'static Info,
+    state: &'static State,
+    _mode: OperatingMode,
+    tx_buf: &'static TxFdBuf<TX_BUF_SIZE>,
+    rx_buf: &'static RxFdBuf<RX_BUF_SIZE>,
+    properties: Properties,
+}
+
+impl<'c, 'd, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> BufferedCanFd<'d, TX_BUF_SIZE, RX_BUF_SIZE> {
     fn new(
-        _instance1: PhantomData<T>,
-        _instance2: &'d crate::pac::can::Fdcan,
+        info: &'static Info,
+        state: &'static State,
         _mode: OperatingMode,
         tx_buf: &'static TxFdBuf<TX_BUF_SIZE>,
         rx_buf: &'static RxFdBuf<RX_BUF_SIZE>,
     ) -> Self {
         BufferedCanFd {
-            _instance1,
-            _instance2,
+            _phantom: PhantomData,
+            info,
+            state,
             _mode,
             tx_buf,
             rx_buf,
-            properties: Properties::new(),
+            properties: Properties::new(info),
         }
         .setup()
     }
 
     /// Get driver properties
-    pub fn properties(&self) -> &Properties<T> {
+    pub fn properties(&self) -> &Properties {
         &self.properties
     }
 
     fn setup(self) -> Self {
         // We don't want interrupts being processed while we change modes.
-        critical_section::with(|_| unsafe {
+        critical_section::with(|_| {
             let rx_inner = super::common::FdBufferedRxInner {
                 rx_sender: self.rx_buf.sender().into(),
             };
             let tx_inner = super::common::FdBufferedTxInner {
                 tx_receiver: self.tx_buf.receiver().into(),
             };
-            T::mut_state().rx_mode = RxMode::FdBuffered(rx_inner);
-            T::mut_state().tx_mode = TxMode::FdBuffered(tx_inner);
+            let state = self.state as *const State;
+            unsafe {
+                let mut_state = state as *mut State;
+                (*mut_state).rx_mode = RxMode::FdBuffered(rx_inner);
+                (*mut_state).tx_mode = TxMode::FdBuffered(tx_inner);
+            }
         });
         self
     }
@@ -568,7 +597,8 @@ impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>
     /// Async write frame to TX buffer.
     pub async fn write(&mut self, frame: FdFrame) {
         self.tx_buf.send(frame).await;
-        T::IT0Interrupt::pend(); // Wake for Tx
+        self.info.interrupt0.pend(); // Wake for Tx
+                                     //T::IT0Interrupt::pend(); // Wake for Tx
     }
 
     /// Async read frame from RX buffer.
@@ -580,7 +610,7 @@ impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>
     pub fn writer(&self) -> BufferedFdCanSender {
         BufferedFdCanSender {
             tx_buf: self.tx_buf.sender().into(),
-            waker: T::IT0Interrupt::pend,
+            waker: self.info.tx_waker,
         }
     }
 
@@ -590,38 +620,55 @@ impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize>
     }
 }
 
-impl<'c, 'd, T: Instance, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> Drop
-    for BufferedCanFd<'d, T, TX_BUF_SIZE, RX_BUF_SIZE>
-{
+impl<'c, 'd, const TX_BUF_SIZE: usize, const RX_BUF_SIZE: usize> Drop for BufferedCanFd<'d, TX_BUF_SIZE, RX_BUF_SIZE> {
     fn drop(&mut self) {
-        critical_section::with(|_| unsafe {
-            T::mut_state().rx_mode = RxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
-            T::mut_state().tx_mode = TxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+        critical_section::with(|_| {
+            let state = self.state as *const State;
+            unsafe {
+                let mut_state = state as *mut State;
+                (*mut_state).rx_mode = RxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+                (*mut_state).tx_mode = TxMode::NonBuffered(embassy_sync::waitqueue::AtomicWaker::new());
+            }
         });
     }
 }
 
 /// FDCAN Rx only Instance
-pub struct CanRx<'d, T: Instance> {
-    _instance1: PhantomData<T>,
-    _instance2: &'d crate::pac::can::Fdcan,
+pub struct CanRx<'d> {
+    _phantom: PhantomData<&'d ()>,
+    info: &'static Info,
+    state: &'static State,
     _mode: OperatingMode,
 }
 
+impl<'d> CanRx<'d> {
+    /// Returns the next received message frame
+    pub async fn read(&mut self) -> Result<Envelope, BusError> {
+        self.state.rx_mode.read_classic(&self.info, &self.state).await
+    }
+
+    /// Returns the next received message frame
+    pub async fn read_fd(&mut self) -> Result<FdEnvelope, BusError> {
+        self.state.rx_mode.read_fd(&self.info, &self.state).await
+    }
+}
+
 /// FDCAN Tx only Instance
-pub struct CanTx<'d, T: Instance> {
+pub struct CanTx<'d> {
+    _phantom: PhantomData<&'d ()>,
+    info: &'static Info,
+    state: &'static State,
     config: crate::can::fd::config::FdCanConfig,
-    _instance: FdcanInstance<'d, T>, //(PeripheralRef<'a, T>);
     _mode: OperatingMode,
 }
 
-impl<'c, 'd, T: Instance> CanTx<'d, T> {
+impl<'c, 'd> CanTx<'d> {
     /// Queues the message to be sent but exerts backpressure.  If a lower-priority
     /// frame is dropped from the mailbox, it is returned.  If no lower-priority frames
     /// can be replaced, this call asynchronously waits for a frame to be successfully
     /// transmitted, then tries again.
     pub async fn write(&mut self, frame: &Frame) -> Option<Frame> {
-        T::state().tx_mode.write::<T>(frame).await
+        self.state.tx_mode.write(self.info, frame).await
     }
 
     /// Queues the message to be sent but exerts backpressure.  If a lower-priority
@@ -629,19 +676,7 @@ impl<'c, 'd, T: Instance> CanTx<'d, T> {
     /// can be replaced, this call asynchronously waits for a frame to be successfully
     /// transmitted, then tries again.
     pub async fn write_fd(&mut self, frame: &FdFrame) -> Option<FdFrame> {
-        T::state().tx_mode.write_fd::<T>(frame).await
-    }
-}
-
-impl<'c, 'd, T: Instance> CanRx<'d, T> {
-    /// Returns the next received message frame
-    pub async fn read(&mut self) -> Result<Envelope, BusError> {
-        T::state().rx_mode.read_classic::<T>().await
-    }
-
-    /// Returns the next received message frame
-    pub async fn read_fd(&mut self) -> Result<FdEnvelope, BusError> {
-        T::state().rx_mode.read_fd::<T>().await
+        self.state.tx_mode.write_fd(self.info, frame).await
     }
 }
 
@@ -662,7 +697,7 @@ impl RxMode {
     }
 
     fn on_interrupt<T: Instance>(&self, fifonr: usize) {
-        T::regs().ir().write(|w| w.set_rfn(fifonr, true));
+        T::registers().regs.ir().write(|w| w.set_rfn(fifonr, true));
         match self {
             RxMode::NonBuffered(waker) => {
                 waker.wake();
@@ -696,7 +731,6 @@ impl RxMode {
         }
     }
 
-    //async fn read_classic<T: Instance>(&self) -> Result<Envelope, BusError> {
     fn try_read_fd<T: Instance>(&self) -> Option<Result<FdEnvelope, BusError>> {
         if let Some((frame, ts)) = T::registers().read(0) {
             let ts = T::calc_timestamp(T::state().ns_per_timer_tick, ts);
@@ -712,14 +746,18 @@ impl RxMode {
         }
     }
 
-    fn read<T: Instance, F: CanHeader>(&self) -> Option<Result<(F, Timestamp), BusError>> {
-        if let Some((msg, ts)) = T::registers().read(0) {
-            let ts = T::calc_timestamp(T::state().ns_per_timer_tick, ts);
+    fn read<F: CanHeader>(
+        &self,
+        info: &'static Info,
+        state: &'static State,
+    ) -> Option<Result<(F, Timestamp), BusError>> {
+        if let Some((msg, ts)) = info.regs.read(0) {
+            let ts = info.calc_timestamp(state.ns_per_timer_tick, ts);
             Some(Ok((msg, ts)))
-        } else if let Some((msg, ts)) = T::registers().read(1) {
-            let ts = T::calc_timestamp(T::state().ns_per_timer_tick, ts);
+        } else if let Some((msg, ts)) = info.regs.read(1) {
+            let ts = info.calc_timestamp(state.ns_per_timer_tick, ts);
             Some(Ok((msg, ts)))
-        } else if let Some(err) = T::registers().curr_error() {
+        } else if let Some(err) = info.regs.curr_error() {
             // TODO: this is probably wrong
             Some(Err(err))
         } else {
@@ -727,11 +765,16 @@ impl RxMode {
         }
     }
 
-    async fn read_async<T: Instance, F: CanHeader>(&self) -> Result<(F, Timestamp), BusError> {
-        poll_fn(|cx| {
-            T::state().err_waker.register(cx.waker());
+    async fn read_async<F: CanHeader>(
+        &self,
+        info: &'static Info,
+        state: &'static State,
+    ) -> Result<(F, Timestamp), BusError> {
+        //let _ = self.read::<F>(info, state);
+        poll_fn(move |cx| {
+            state.err_waker.register(cx.waker());
             self.register(cx.waker());
-            match self.read::<T, _>() {
+            match self.read::<_>(info, state) {
                 Some(result) => Poll::Ready(result),
                 None => Poll::Pending,
             }
@@ -739,15 +782,15 @@ impl RxMode {
         .await
     }
 
-    async fn read_classic<T: Instance>(&self) -> Result<Envelope, BusError> {
-        match self.read_async::<T, _>().await {
+    async fn read_classic(&self, info: &'static Info, state: &'static State) -> Result<Envelope, BusError> {
+        match self.read_async::<_>(info, state).await {
             Ok((frame, ts)) => Ok(Envelope { ts, frame }),
             Err(e) => Err(e),
         }
     }
 
-    async fn read_fd<T: Instance>(&self) -> Result<FdEnvelope, BusError> {
-        match self.read_async::<T, _>().await {
+    async fn read_fd(&self, info: &'static Info, state: &'static State) -> Result<FdEnvelope, BusError> {
+        match self.read_async::<_>(info, state).await {
             Ok((frame, ts)) => Ok(FdEnvelope { ts, frame }),
             Err(e) => Err(e),
         }
@@ -776,11 +819,11 @@ impl TxMode {
     /// frame is dropped from the mailbox, it is returned.  If no lower-priority frames
     /// can be replaced, this call asynchronously waits for a frame to be successfully
     /// transmitted, then tries again.
-    async fn write_generic<T: Instance, F: embedded_can::Frame + CanHeader>(&self, frame: &F) -> Option<F> {
+    async fn write_generic<F: embedded_can::Frame + CanHeader>(&self, info: &'static Info, frame: &F) -> Option<F> {
         poll_fn(|cx| {
             self.register(cx.waker());
 
-            if let Ok(dropped) = T::registers().write(frame) {
+            if let Ok(dropped) = info.regs.write(frame) {
                 return Poll::Ready(dropped);
             }
 
@@ -795,68 +838,70 @@ impl TxMode {
     /// frame is dropped from the mailbox, it is returned.  If no lower-priority frames
     /// can be replaced, this call asynchronously waits for a frame to be successfully
     /// transmitted, then tries again.
-    async fn write<T: Instance>(&self, frame: &Frame) -> Option<Frame> {
-        self.write_generic::<T, _>(frame).await
+    async fn write(&self, info: &'static Info, frame: &Frame) -> Option<Frame> {
+        self.write_generic::<_>(info, frame).await
     }
 
     /// Queues the message to be sent but exerts backpressure.  If a lower-priority
     /// frame is dropped from the mailbox, it is returned.  If no lower-priority frames
     /// can be replaced, this call asynchronously waits for a frame to be successfully
     /// transmitted, then tries again.
-    async fn write_fd<T: Instance>(&self, frame: &FdFrame) -> Option<FdFrame> {
-        self.write_generic::<T, _>(frame).await
+    async fn write_fd(&self, info: &'static Info, frame: &FdFrame) -> Option<FdFrame> {
+        self.write_generic::<_>(info, frame).await
     }
 }
 
 /// Common driver properties, including filters and error counters
-pub struct Properties<T> {
+pub struct Properties {
+    info: &'static Info,
     // phantom pointer to ensure !Sync
-    instance: PhantomData<*const T>,
+    //instance: PhantomData<*const T>,
 }
 
-impl<T: Instance> Properties<T> {
-    fn new() -> Self {
+impl Properties {
+    fn new(info: &'static Info) -> Self {
         Self {
-            instance: Default::default(),
+            info,
+            //instance: Default::default(),
         }
     }
 
     /// Set a standard address CAN filter in the specified slot in FDCAN memory.
     #[inline]
     pub fn set_standard_filter(&self, slot: StandardFilterSlot, filter: StandardFilter) {
-        T::registers().msg_ram_mut().filters.flssa[slot as usize].activate(filter);
+        self.info.regs.msg_ram_mut().filters.flssa[slot as usize].activate(filter);
     }
 
     /// Set the full array of standard address CAN filters in FDCAN memory.
     /// Overwrites all standard address filters in memory.
     pub fn set_standard_filters(&self, filters: &[StandardFilter; STANDARD_FILTER_MAX as usize]) {
         for (i, f) in filters.iter().enumerate() {
-            T::registers().msg_ram_mut().filters.flssa[i].activate(*f);
+            self.info.regs.msg_ram_mut().filters.flssa[i].activate(*f);
         }
     }
 
     /// Set an extended address CAN filter in the specified slot in FDCAN memory.
     #[inline]
     pub fn set_extended_filter(&self, slot: ExtendedFilterSlot, filter: ExtendedFilter) {
-        T::registers().msg_ram_mut().filters.flesa[slot as usize].activate(filter);
+        self.info.regs.msg_ram_mut().filters.flesa[slot as usize].activate(filter);
     }
 
     /// Set the full array of extended address CAN filters in FDCAN memory.
     /// Overwrites all extended address filters in memory.
     pub fn set_extended_filters(&self, filters: &[ExtendedFilter; EXTENDED_FILTER_MAX as usize]) {
         for (i, f) in filters.iter().enumerate() {
-            T::registers().msg_ram_mut().filters.flesa[i].activate(*f);
+            self.info.regs.msg_ram_mut().filters.flesa[i].activate(*f);
         }
     }
 
     /// Get the CAN RX error counter
     pub fn rx_error_count(&self) -> u8 {
-        T::regs().ecr().read().rec()
+        self.info.regs.regs.ecr().read().rec()
     }
 
     /// Get the CAN TX error counter
     pub fn tx_error_count(&self) -> u8 {
-        T::regs().ecr().read().tec()
+        self.info.regs.regs.ecr().read().tec()
     }
 
     /// Get the current bus error mode
@@ -864,7 +909,7 @@ impl<T: Instance> Properties<T> {
         // This read will clear LEC and DLEC. This is not ideal, but protocol
         // error reporting in this driver should have a big ol' FIXME on it
         // anyway!
-        let psr = T::regs().psr().read();
+        let psr = self.info.regs.regs.psr().read();
         match (psr.bo(), psr.ep()) {
             (false, false) => BusErrorMode::ErrorActive,
             (false, true) => BusErrorMode::ErrorPassive,
@@ -892,10 +937,36 @@ impl State {
     }
 }
 
+struct Info {
+    regs: Registers,
+    interrupt0: crate::interrupt::Interrupt,
+    _interrupt1: crate::interrupt::Interrupt,
+    tx_waker: fn(),
+}
+
+impl Info {
+    #[cfg(feature = "time")]
+    fn calc_timestamp(&self, ns_per_timer_tick: u64, ts_val: u16) -> Timestamp {
+        let now_embassy = embassy_time::Instant::now();
+        if ns_per_timer_tick == 0 {
+            return now_embassy;
+        }
+        let cantime = { self.regs.regs.tscv().read().tsc() };
+        let delta = cantime.overflowing_sub(ts_val).0 as u64;
+        let ns = ns_per_timer_tick * delta as u64;
+        now_embassy - embassy_time::Duration::from_nanos(ns)
+    }
+
+    #[cfg(not(feature = "time"))]
+    fn calc_timestamp(&self, _ns_per_timer_tick: u64, ts_val: u16) -> Timestamp {
+        ts_val
+    }
+}
+
 trait SealedInstance {
     const MSG_RAM_OFFSET: usize;
 
-    fn regs() -> &'static crate::pac::can::Fdcan;
+    fn info() -> &'static Info;
     fn registers() -> crate::can::fd::peripheral::Registers;
     fn state() -> &'static State;
     unsafe fn mut_state() -> &'static mut State;
@@ -915,15 +986,23 @@ pub trait Instance: SealedInstance + RccPeripheral + 'static {
 pub struct FdcanInstance<'a, T>(PeripheralRef<'a, T>);
 
 macro_rules! impl_fdcan {
-    ($inst:ident, $msg_ram_inst:ident, $msg_ram_offset:literal) => {
+    ($inst:ident,
+        //$irq0:ident, $irq1:ident,
+        $msg_ram_inst:ident, $msg_ram_offset:literal) => {
         impl SealedInstance for peripherals::$inst {
             const MSG_RAM_OFFSET: usize = $msg_ram_offset;
 
-            fn regs() -> &'static crate::pac::can::Fdcan {
-                &crate::pac::$inst
+            fn info() -> &'static Info {
+                static INFO: Info = Info {
+                    regs: Registers{regs: crate::pac::$inst, msgram: crate::pac::$msg_ram_inst, msg_ram_offset: $msg_ram_offset},
+                    interrupt0: crate::_generated::peripheral_interrupts::$inst::IT0::IRQ,
+                    _interrupt1: crate::_generated::peripheral_interrupts::$inst::IT1::IRQ,
+                    tx_waker: crate::_generated::peripheral_interrupts::$inst::IT0::pend,
+                };
+                &INFO
             }
             fn registers() -> Registers {
-                Registers{regs: &crate::pac::$inst, msgram: &crate::pac::$msg_ram_inst, msg_ram_offset: Self::MSG_RAM_OFFSET}
+                Registers{regs: crate::pac::$inst, msgram: crate::pac::$msg_ram_inst, msg_ram_offset: Self::MSG_RAM_OFFSET}
             }
             unsafe fn mut_state() -> &'static mut State {
                 static mut STATE: State = State::new();
@@ -939,7 +1018,7 @@ macro_rules! impl_fdcan {
                 if ns_per_timer_tick == 0 {
                     return now_embassy;
                 }
-                let cantime = { Self::regs().tscv().read().tsc() };
+                let cantime = { Self::registers().regs.tscv().read().tsc() };
                 let delta = cantime.overflowing_sub(ts_val).0 as u64;
                 let ns = ns_per_timer_tick * delta as u64;
                 now_embassy - embassy_time::Duration::from_nanos(ns)
diff --git a/embassy-stm32/src/cordic/mod.rs b/embassy-stm32/src/cordic/mod.rs
index 29b11735e..fb342d2e7 100644
--- a/embassy-stm32/src/cordic/mod.rs
+++ b/embassy-stm32/src/cordic/mod.rs
@@ -4,7 +4,7 @@ use embassy_hal_internal::drop::OnDrop;
 use embassy_hal_internal::{into_ref, Peripheral, PeripheralRef};
 
 use crate::pac::cordic::vals;
-use crate::{dma, peripherals};
+use crate::{dma, peripherals, rcc};
 
 mod enums;
 pub use enums::*;
@@ -199,7 +199,7 @@ impl<'d, T: Instance> Cordic<'d, T> {
     /// If you need a peripheral -> CORDIC -> peripheral mode,  
     /// you may want to set Cordic into [Mode::ZeroOverhead] mode, and add extra arguments with [Self::extra_config]
     pub fn new(peri: impl Peripheral<P = T> + 'd, config: Config) -> Self {
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         into_ref!(peri);
 
@@ -259,7 +259,7 @@ impl<'d, T: Instance> Cordic<'d, T> {
 
 impl<'d, T: Instance> Drop for Cordic<'d, T> {
     fn drop(&mut self) {
-        T::disable();
+        rcc::disable::<T>();
     }
 }
 
diff --git a/embassy-stm32/src/crc/v1.rs b/embassy-stm32/src/crc/v1.rs
index e8e0270af..f3d13de7c 100644
--- a/embassy-stm32/src/crc/v1.rs
+++ b/embassy-stm32/src/crc/v1.rs
@@ -2,8 +2,7 @@ use embassy_hal_internal::{into_ref, PeripheralRef};
 
 use crate::pac::CRC as PAC_CRC;
 use crate::peripherals::CRC;
-use crate::rcc::SealedRccPeripheral;
-use crate::Peripheral;
+use crate::{rcc, Peripheral};
 
 /// CRC driver.
 pub struct Crc<'d> {
@@ -17,7 +16,7 @@ impl<'d> Crc<'d> {
 
         // Note: enable and reset come from RccPeripheral.
         // enable CRC clock in RCC.
-        CRC::enable_and_reset();
+        rcc::enable_and_reset::<CRC>();
         // Peripheral the peripheral
         let mut instance = Self { _peri: peripheral };
         instance.reset();
diff --git a/embassy-stm32/src/crc/v2v3.rs b/embassy-stm32/src/crc/v2v3.rs
index ad7c79f12..09d956d7c 100644
--- a/embassy-stm32/src/crc/v2v3.rs
+++ b/embassy-stm32/src/crc/v2v3.rs
@@ -3,8 +3,7 @@ use embassy_hal_internal::{into_ref, PeripheralRef};
 use crate::pac::crc::vals;
 use crate::pac::CRC as PAC_CRC;
 use crate::peripherals::CRC;
-use crate::rcc::SealedRccPeripheral;
-use crate::Peripheral;
+use crate::{rcc, Peripheral};
 
 /// CRC driver.
 pub struct Crc<'d> {
@@ -84,7 +83,7 @@ impl<'d> Crc<'d> {
     pub fn new(peripheral: impl Peripheral<P = CRC> + 'd, config: Config) -> Self {
         // Note: enable and reset come from RccPeripheral.
         // reset to default values and enable CRC clock in RCC.
-        CRC::enable_and_reset();
+        rcc::enable_and_reset::<CRC>();
         into_ref!(peripheral);
         let mut instance = Self {
             _peripheral: peripheral,
diff --git a/embassy-stm32/src/cryp/mod.rs b/embassy-stm32/src/cryp/mod.rs
index f19c94fda..8d600c73c 100644
--- a/embassy-stm32/src/cryp/mod.rs
+++ b/embassy-stm32/src/cryp/mod.rs
@@ -9,7 +9,7 @@ use embassy_sync::waitqueue::AtomicWaker;
 
 use crate::dma::{NoDma, Transfer, TransferOptions};
 use crate::interrupt::typelevel::Interrupt;
-use crate::{interrupt, pac, peripherals, Peripheral};
+use crate::{interrupt, pac, peripherals, rcc, Peripheral};
 
 const DES_BLOCK_SIZE: usize = 8; // 64 bits
 const AES_BLOCK_SIZE: usize = 16; // 128 bits
@@ -51,16 +51,16 @@ pub trait Cipher<'c> {
     fn iv(&self) -> &[u8];
 
     /// Sets the processor algorithm mode according to the associated cipher.
-    fn set_algomode(&self, p: &pac::cryp::Cryp);
+    fn set_algomode(&self, p: pac::cryp::Cryp);
 
     /// Performs any key preparation within the processor, if necessary.
-    fn prepare_key(&self, _p: &pac::cryp::Cryp) {}
+    fn prepare_key(&self, _p: pac::cryp::Cryp) {}
 
     /// Performs any cipher-specific initialization.
-    fn init_phase_blocking<T: Instance, DmaIn, DmaOut>(&self, _p: &pac::cryp::Cryp, _cryp: &Cryp<T, DmaIn, DmaOut>) {}
+    fn init_phase_blocking<T: Instance, DmaIn, DmaOut>(&self, _p: pac::cryp::Cryp, _cryp: &Cryp<T, DmaIn, DmaOut>) {}
 
     /// Performs any cipher-specific initialization.
-    async fn init_phase<T: Instance, DmaIn, DmaOut>(&self, _p: &pac::cryp::Cryp, _cryp: &mut Cryp<'_, T, DmaIn, DmaOut>)
+    async fn init_phase<T: Instance, DmaIn, DmaOut>(&self, _p: pac::cryp::Cryp, _cryp: &mut Cryp<'_, T, DmaIn, DmaOut>)
     where
         DmaIn: crate::cryp::DmaIn<T>,
         DmaOut: crate::cryp::DmaOut<T>,
@@ -68,14 +68,14 @@ pub trait Cipher<'c> {
     }
 
     /// Called prior to processing the last data block for cipher-specific operations.
-    fn pre_final(&self, _p: &pac::cryp::Cryp, _dir: Direction, _padding_len: usize) -> [u32; 4] {
+    fn pre_final(&self, _p: pac::cryp::Cryp, _dir: Direction, _padding_len: usize) -> [u32; 4] {
         return [0; 4];
     }
 
     /// Called after processing the last data block for cipher-specific operations.
     fn post_final_blocking<T: Instance, DmaIn, DmaOut>(
         &self,
-        _p: &pac::cryp::Cryp,
+        _p: pac::cryp::Cryp,
         _cryp: &Cryp<T, DmaIn, DmaOut>,
         _dir: Direction,
         _int_data: &mut [u8; AES_BLOCK_SIZE],
@@ -87,7 +87,7 @@ pub trait Cipher<'c> {
     /// Called after processing the last data block for cipher-specific operations.
     async fn post_final<T: Instance, DmaIn, DmaOut>(
         &self,
-        _p: &pac::cryp::Cryp,
+        _p: pac::cryp::Cryp,
         _cryp: &mut Cryp<'_, T, DmaIn, DmaOut>,
         _dir: Direction,
         _int_data: &mut [u8; AES_BLOCK_SIZE],
@@ -142,7 +142,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for TdesEcb<'c, KEY_SIZE> {
         self.iv
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         #[cfg(cryp_v1)]
         {
             p.cr().modify(|w| w.set_algomode(0));
@@ -184,7 +184,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for TdesCbc<'c, KEY_SIZE> {
         self.iv
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         #[cfg(cryp_v1)]
         {
             p.cr().modify(|w| w.set_algomode(1));
@@ -226,7 +226,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for DesEcb<'c, KEY_SIZE> {
         self.iv
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         #[cfg(cryp_v1)]
         {
             p.cr().modify(|w| w.set_algomode(2));
@@ -267,7 +267,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for DesCbc<'c, KEY_SIZE> {
         self.iv
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         #[cfg(cryp_v1)]
         {
             p.cr().modify(|w| w.set_algomode(3));
@@ -308,7 +308,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesEcb<'c, KEY_SIZE> {
         self.iv
     }
 
-    fn prepare_key(&self, p: &pac::cryp::Cryp) {
+    fn prepare_key(&self, p: pac::cryp::Cryp) {
         #[cfg(cryp_v1)]
         {
             p.cr().modify(|w| w.set_algomode(7));
@@ -322,7 +322,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesEcb<'c, KEY_SIZE> {
         while p.sr().read().busy() {}
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         #[cfg(cryp_v1)]
         {
             p.cr().modify(|w| w.set_algomode(2));
@@ -365,7 +365,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesCbc<'c, KEY_SIZE> {
         self.iv
     }
 
-    fn prepare_key(&self, p: &pac::cryp::Cryp) {
+    fn prepare_key(&self, p: pac::cryp::Cryp) {
         #[cfg(cryp_v1)]
         {
             p.cr().modify(|w| w.set_algomode(7));
@@ -379,7 +379,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesCbc<'c, KEY_SIZE> {
         while p.sr().read().busy() {}
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         #[cfg(cryp_v1)]
         {
             p.cr().modify(|w| w.set_algomode(5));
@@ -421,7 +421,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesCtr<'c, KEY_SIZE> {
         self.iv
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         #[cfg(cryp_v1)]
         {
             p.cr().modify(|w| w.set_algomode(6));
@@ -469,29 +469,25 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesGcm<'c, KEY_SIZE> {
         self.iv.as_slice()
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         p.cr().modify(|w| w.set_algomode0(0));
         p.cr().modify(|w| w.set_algomode3(true));
     }
 
-    fn init_phase_blocking<T: Instance, DmaIn, DmaOut>(&self, p: &pac::cryp::Cryp, _cryp: &Cryp<T, DmaIn, DmaOut>) {
+    fn init_phase_blocking<T: Instance, DmaIn, DmaOut>(&self, p: pac::cryp::Cryp, _cryp: &Cryp<T, DmaIn, DmaOut>) {
         p.cr().modify(|w| w.set_gcm_ccmph(0));
         p.cr().modify(|w| w.set_crypen(true));
         while p.cr().read().crypen() {}
     }
 
-    async fn init_phase<T: Instance, DmaIn, DmaOut>(
-        &self,
-        p: &pac::cryp::Cryp,
-        _cryp: &mut Cryp<'_, T, DmaIn, DmaOut>,
-    ) {
+    async fn init_phase<T: Instance, DmaIn, DmaOut>(&self, p: pac::cryp::Cryp, _cryp: &mut Cryp<'_, T, DmaIn, DmaOut>) {
         p.cr().modify(|w| w.set_gcm_ccmph(0));
         p.cr().modify(|w| w.set_crypen(true));
         while p.cr().read().crypen() {}
     }
 
     #[cfg(cryp_v2)]
-    fn pre_final(&self, p: &pac::cryp::Cryp, dir: Direction, _padding_len: usize) -> [u32; 4] {
+    fn pre_final(&self, p: pac::cryp::Cryp, dir: Direction, _padding_len: usize) -> [u32; 4] {
         //Handle special GCM partial block process.
         if dir == Direction::Encrypt {
             p.cr().modify(|w| w.set_crypen(false));
@@ -505,7 +501,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesGcm<'c, KEY_SIZE> {
     }
 
     #[cfg(any(cryp_v3, cryp_v4))]
-    fn pre_final(&self, p: &pac::cryp::Cryp, _dir: Direction, padding_len: usize) -> [u32; 4] {
+    fn pre_final(&self, p: pac::cryp::Cryp, _dir: Direction, padding_len: usize) -> [u32; 4] {
         //Handle special GCM partial block process.
         p.cr().modify(|w| w.set_npblb(padding_len as u8));
         [0; 4]
@@ -514,7 +510,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesGcm<'c, KEY_SIZE> {
     #[cfg(cryp_v2)]
     fn post_final_blocking<T: Instance, DmaIn, DmaOut>(
         &self,
-        p: &pac::cryp::Cryp,
+        p: pac::cryp::Cryp,
         cryp: &Cryp<T, DmaIn, DmaOut>,
         dir: Direction,
         int_data: &mut [u8; AES_BLOCK_SIZE],
@@ -540,7 +536,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesGcm<'c, KEY_SIZE> {
     #[cfg(cryp_v2)]
     async fn post_final<T: Instance, DmaIn, DmaOut>(
         &self,
-        p: &pac::cryp::Cryp,
+        p: pac::cryp::Cryp,
         cryp: &mut Cryp<'_, T, DmaIn, DmaOut>,
         dir: Direction,
         int_data: &mut [u8; AES_BLOCK_SIZE],
@@ -614,29 +610,25 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesGmac<'c, KEY_SIZE> {
         self.iv.as_slice()
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         p.cr().modify(|w| w.set_algomode0(0));
         p.cr().modify(|w| w.set_algomode3(true));
     }
 
-    fn init_phase_blocking<T: Instance, DmaIn, DmaOut>(&self, p: &pac::cryp::Cryp, _cryp: &Cryp<T, DmaIn, DmaOut>) {
+    fn init_phase_blocking<T: Instance, DmaIn, DmaOut>(&self, p: pac::cryp::Cryp, _cryp: &Cryp<T, DmaIn, DmaOut>) {
         p.cr().modify(|w| w.set_gcm_ccmph(0));
         p.cr().modify(|w| w.set_crypen(true));
         while p.cr().read().crypen() {}
     }
 
-    async fn init_phase<T: Instance, DmaIn, DmaOut>(
-        &self,
-        p: &pac::cryp::Cryp,
-        _cryp: &mut Cryp<'_, T, DmaIn, DmaOut>,
-    ) {
+    async fn init_phase<T: Instance, DmaIn, DmaOut>(&self, p: pac::cryp::Cryp, _cryp: &mut Cryp<'_, T, DmaIn, DmaOut>) {
         p.cr().modify(|w| w.set_gcm_ccmph(0));
         p.cr().modify(|w| w.set_crypen(true));
         while p.cr().read().crypen() {}
     }
 
     #[cfg(cryp_v2)]
-    fn pre_final(&self, p: &pac::cryp::Cryp, dir: Direction, _padding_len: usize) -> [u32; 4] {
+    fn pre_final(&self, p: pac::cryp::Cryp, dir: Direction, _padding_len: usize) -> [u32; 4] {
         //Handle special GCM partial block process.
         if dir == Direction::Encrypt {
             p.cr().modify(|w| w.set_crypen(false));
@@ -650,7 +642,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesGmac<'c, KEY_SIZE> {
     }
 
     #[cfg(any(cryp_v3, cryp_v4))]
-    fn pre_final(&self, p: &pac::cryp::Cryp, _dir: Direction, padding_len: usize) -> [u32; 4] {
+    fn pre_final(&self, p: pac::cryp::Cryp, _dir: Direction, padding_len: usize) -> [u32; 4] {
         //Handle special GCM partial block process.
         p.cr().modify(|w| w.set_npblb(padding_len as u8));
         [0; 4]
@@ -659,7 +651,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesGmac<'c, KEY_SIZE> {
     #[cfg(cryp_v2)]
     fn post_final_blocking<T: Instance, DmaIn, DmaOut>(
         &self,
-        p: &pac::cryp::Cryp,
+        p: pac::cryp::Cryp,
         cryp: &Cryp<T, DmaIn, DmaOut>,
         dir: Direction,
         int_data: &mut [u8; AES_BLOCK_SIZE],
@@ -685,7 +677,7 @@ impl<'c, const KEY_SIZE: usize> Cipher<'c> for AesGmac<'c, KEY_SIZE> {
     #[cfg(cryp_v2)]
     async fn post_final<T: Instance, DmaIn, DmaOut>(
         &self,
-        p: &pac::cryp::Cryp,
+        p: pac::cryp::Cryp,
         cryp: &mut Cryp<'_, T, DmaIn, DmaOut>,
         dir: Direction,
         int_data: &mut [u8; AES_BLOCK_SIZE],
@@ -752,7 +744,7 @@ impl<'c, const KEY_SIZE: usize, const TAG_SIZE: usize, const IV_SIZE: usize> Aes
             } else {
                 aad_header[0] = 0xFF;
                 aad_header[1] = 0xFE;
-                let aad_len_bytes: [u8; 4] = aad_len.to_be_bytes();
+                let aad_len_bytes: [u8; 4] = (aad_len as u32).to_be_bytes();
                 aad_header[2] = aad_len_bytes[0];
                 aad_header[3] = aad_len_bytes[1];
                 aad_header[4] = aad_len_bytes[2];
@@ -773,7 +765,7 @@ impl<'c, const KEY_SIZE: usize, const TAG_SIZE: usize, const IV_SIZE: usize> Aes
         block0[0] |= ((((TAG_SIZE as u8) - 2) >> 1) & 0x07) << 3;
         block0[0] |= ((15 - (iv.len() as u8)) - 1) & 0x07;
         block0[1..1 + iv.len()].copy_from_slice(iv);
-        let payload_len_bytes: [u8; 4] = payload_len.to_be_bytes();
+        let payload_len_bytes: [u8; 4] = (payload_len as u32).to_be_bytes();
         if iv.len() <= 11 {
             block0[12] = payload_len_bytes[0];
         } else if payload_len_bytes[0] > 0 {
@@ -815,12 +807,12 @@ impl<'c, const KEY_SIZE: usize, const TAG_SIZE: usize, const IV_SIZE: usize> Cip
         self.ctr.as_slice()
     }
 
-    fn set_algomode(&self, p: &pac::cryp::Cryp) {
+    fn set_algomode(&self, p: pac::cryp::Cryp) {
         p.cr().modify(|w| w.set_algomode0(1));
         p.cr().modify(|w| w.set_algomode3(true));
     }
 
-    fn init_phase_blocking<T: Instance, DmaIn, DmaOut>(&self, p: &pac::cryp::Cryp, cryp: &Cryp<T, DmaIn, DmaOut>) {
+    fn init_phase_blocking<T: Instance, DmaIn, DmaOut>(&self, p: pac::cryp::Cryp, cryp: &Cryp<T, DmaIn, DmaOut>) {
         p.cr().modify(|w| w.set_gcm_ccmph(0));
 
         cryp.write_bytes_blocking(Self::BLOCK_SIZE, &self.block0);
@@ -829,7 +821,7 @@ impl<'c, const KEY_SIZE: usize, const TAG_SIZE: usize, const IV_SIZE: usize> Cip
         while p.cr().read().crypen() {}
     }
 
-    async fn init_phase<T: Instance, DmaIn, DmaOut>(&self, p: &pac::cryp::Cryp, cryp: &mut Cryp<'_, T, DmaIn, DmaOut>)
+    async fn init_phase<T: Instance, DmaIn, DmaOut>(&self, p: pac::cryp::Cryp, cryp: &mut Cryp<'_, T, DmaIn, DmaOut>)
     where
         DmaIn: crate::cryp::DmaIn<T>,
         DmaOut: crate::cryp::DmaOut<T>,
@@ -847,7 +839,7 @@ impl<'c, const KEY_SIZE: usize, const TAG_SIZE: usize, const IV_SIZE: usize> Cip
     }
 
     #[cfg(cryp_v2)]
-    fn pre_final(&self, p: &pac::cryp::Cryp, dir: Direction, _padding_len: usize) -> [u32; 4] {
+    fn pre_final(&self, p: pac::cryp::Cryp, dir: Direction, _padding_len: usize) -> [u32; 4] {
         //Handle special CCM partial block process.
         let mut temp1 = [0; 4];
         if dir == Direction::Decrypt {
@@ -866,7 +858,7 @@ impl<'c, const KEY_SIZE: usize, const TAG_SIZE: usize, const IV_SIZE: usize> Cip
     }
 
     #[cfg(any(cryp_v3, cryp_v4))]
-    fn pre_final(&self, p: &pac::cryp::Cryp, _dir: Direction, padding_len: usize) -> [u32; 4] {
+    fn pre_final(&self, p: pac::cryp::Cryp, _dir: Direction, padding_len: usize) -> [u32; 4] {
         //Handle special GCM partial block process.
         p.cr().modify(|w| w.set_npblb(padding_len as u8));
         [0; 4]
@@ -875,7 +867,7 @@ impl<'c, const KEY_SIZE: usize, const TAG_SIZE: usize, const IV_SIZE: usize> Cip
     #[cfg(cryp_v2)]
     fn post_final_blocking<T: Instance, DmaIn, DmaOut>(
         &self,
-        p: &pac::cryp::Cryp,
+        p: pac::cryp::Cryp,
         cryp: &Cryp<T, DmaIn, DmaOut>,
         dir: Direction,
         int_data: &mut [u8; AES_BLOCK_SIZE],
@@ -912,7 +904,7 @@ impl<'c, const KEY_SIZE: usize, const TAG_SIZE: usize, const IV_SIZE: usize> Cip
     #[cfg(cryp_v2)]
     async fn post_final<T: Instance, DmaIn, DmaOut>(
         &self,
-        p: &pac::cryp::Cryp,
+        p: pac::cryp::Cryp,
         cryp: &mut Cryp<'_, T, DmaIn, DmaOut>,
         dir: Direction,
         int_data: &mut [u8; AES_BLOCK_SIZE],
@@ -1029,7 +1021,7 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
         outdma: impl Peripheral<P = DmaOut> + 'd,
         _irq: impl interrupt::typelevel::Binding<T::Interrupt, InterruptHandler<T>> + 'd,
     ) -> Self {
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
         into_ref!(peri, indma, outdma);
         let instance = Self {
             _peripheral: peri,
@@ -1083,9 +1075,9 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
         // Set data type to 8-bit. This will match software implementations.
         T::regs().cr().modify(|w| w.set_datatype(2));
 
-        ctx.cipher.prepare_key(&T::regs());
+        ctx.cipher.prepare_key(T::regs());
 
-        ctx.cipher.set_algomode(&T::regs());
+        ctx.cipher.set_algomode(T::regs());
 
         // Set encrypt/decrypt
         if dir == Direction::Encrypt {
@@ -1115,7 +1107,7 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
         // Flush in/out FIFOs
         T::regs().cr().modify(|w| w.fflush());
 
-        ctx.cipher.init_phase_blocking(&T::regs(), self);
+        ctx.cipher.init_phase_blocking(T::regs(), self);
 
         self.store_context(&mut ctx);
 
@@ -1166,9 +1158,9 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
         // Set data type to 8-bit. This will match software implementations.
         T::regs().cr().modify(|w| w.set_datatype(2));
 
-        ctx.cipher.prepare_key(&T::regs());
+        ctx.cipher.prepare_key(T::regs());
 
-        ctx.cipher.set_algomode(&T::regs());
+        ctx.cipher.set_algomode(T::regs());
 
         // Set encrypt/decrypt
         if dir == Direction::Encrypt {
@@ -1198,7 +1190,7 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
         // Flush in/out FIFOs
         T::regs().cr().modify(|w| w.fflush());
 
-        ctx.cipher.init_phase(&T::regs(), self).await;
+        ctx.cipher.init_phase(T::regs(), self).await;
 
         self.store_context(&mut ctx);
 
@@ -1462,7 +1454,7 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
         // Handle the final block, which is incomplete.
         if last_block_remainder > 0 {
             let padding_len = C::BLOCK_SIZE - last_block_remainder;
-            let temp1 = ctx.cipher.pre_final(&T::regs(), ctx.dir, padding_len);
+            let temp1 = ctx.cipher.pre_final(T::regs(), ctx.dir, padding_len);
 
             let mut intermediate_data: [u8; AES_BLOCK_SIZE] = [0; AES_BLOCK_SIZE];
             let mut last_block: [u8; AES_BLOCK_SIZE] = [0; AES_BLOCK_SIZE];
@@ -1478,7 +1470,7 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
             let mut mask: [u8; 16] = [0; 16];
             mask[..last_block_remainder].fill(0xFF);
             ctx.cipher
-                .post_final_blocking(&T::regs(), self, ctx.dir, &mut intermediate_data, temp1, mask);
+                .post_final_blocking(T::regs(), self, ctx.dir, &mut intermediate_data, temp1, mask);
         }
 
         ctx.payload_len += input.len() as u64;
@@ -1559,7 +1551,7 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
         // Handle the final block, which is incomplete.
         if last_block_remainder > 0 {
             let padding_len = C::BLOCK_SIZE - last_block_remainder;
-            let temp1 = ctx.cipher.pre_final(&T::regs(), ctx.dir, padding_len);
+            let temp1 = ctx.cipher.pre_final(T::regs(), ctx.dir, padding_len);
 
             let mut intermediate_data: [u8; AES_BLOCK_SIZE] = [0; AES_BLOCK_SIZE];
             let mut last_block: [u8; AES_BLOCK_SIZE] = [0; AES_BLOCK_SIZE];
@@ -1576,7 +1568,7 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
             let mut mask: [u8; 16] = [0; 16];
             mask[..last_block_remainder].fill(0xFF);
             ctx.cipher
-                .post_final(&T::regs(), self, ctx.dir, &mut intermediate_data, temp1, mask)
+                .post_final(T::regs(), self, ctx.dir, &mut intermediate_data, temp1, mask)
                 .await;
         }
 
@@ -1758,7 +1750,7 @@ impl<'d, T: Instance, DmaIn, DmaOut> Cryp<'d, T, DmaIn, DmaOut> {
         self.load_key(ctx.cipher.key());
 
         // Prepare key if applicable.
-        ctx.cipher.prepare_key(&T::regs());
+        ctx.cipher.prepare_key(T::regs());
         T::regs().cr().write(|w| w.0 = ctx.cr);
 
         // Enable crypto processor.
diff --git a/embassy-stm32/src/dac/mod.rs b/embassy-stm32/src/dac/mod.rs
index 8a748ad72..8bba5ded0 100644
--- a/embassy-stm32/src/dac/mod.rs
+++ b/embassy-stm32/src/dac/mod.rs
@@ -8,7 +8,7 @@ use embassy_hal_internal::{into_ref, PeripheralRef};
 use crate::dma::NoDma;
 #[cfg(any(dac_v3, dac_v4, dac_v5, dac_v6, dac_v7))]
 use crate::pac::dac;
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::{peripherals, Peripheral};
 
 mod tsel;
@@ -131,7 +131,7 @@ impl<'d, T: Instance, const N: u8, DMA> DacChannel<'d, T, N, DMA> {
     ) -> Self {
         into_ref!(dma, pin);
         pin.set_as_analog();
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
         let mut dac = Self {
             phantom: PhantomData,
             dma,
@@ -157,7 +157,7 @@ impl<'d, T: Instance, const N: u8, DMA> DacChannel<'d, T, N, DMA> {
     #[cfg(all(any(dac_v3, dac_v4, dac_v5, dac_v6, dac_v7), not(any(stm32h56x, stm32h57x))))]
     pub fn new_internal(_peri: impl Peripheral<P = T> + 'd, dma: impl Peripheral<P = DMA> + 'd) -> Self {
         into_ref!(dma);
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
         let mut dac = Self {
             phantom: PhantomData,
             dma,
@@ -356,7 +356,7 @@ impl_dma_methods!(2, DacDma2);
 
 impl<'d, T: Instance, const N: u8, DMA> Drop for DacChannel<'d, T, N, DMA> {
     fn drop(&mut self) {
-        T::disable();
+        rcc::disable::<T>();
     }
 }
 
@@ -400,8 +400,8 @@ impl<'d, T: Instance, DMACh1, DMACh2> Dac<'d, T, DMACh1, DMACh2> {
         pin_ch2.set_as_analog();
 
         // Enable twice to increment the DAC refcount for each channel.
-        T::enable_and_reset();
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
+        rcc::enable_and_reset::<T>();
 
         let mut ch1 = DacCh1 {
             phantom: PhantomData,
@@ -444,8 +444,8 @@ impl<'d, T: Instance, DMACh1, DMACh2> Dac<'d, T, DMACh1, DMACh2> {
     ) -> Self {
         into_ref!(dma_ch1, dma_ch2);
         // Enable twice to increment the DAC refcount for each channel.
-        T::enable_and_reset();
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
+        rcc::enable_and_reset::<T>();
 
         let mut ch1 = DacCh1 {
             phantom: PhantomData,
@@ -508,7 +508,7 @@ impl<'d, T: Instance, DMACh1, DMACh2> Dac<'d, T, DMACh1, DMACh2> {
 }
 
 trait SealedInstance {
-    fn regs() -> &'static crate::pac::dac::Dac;
+    fn regs() -> crate::pac::dac::Dac;
 }
 
 /// DAC instance.
@@ -523,8 +523,8 @@ pub trait DacPin<T: Instance, const C: u8>: crate::gpio::Pin + 'static {}
 foreach_peripheral!(
     (dac, $inst:ident) => {
         impl crate::dac::SealedInstance for peripherals::$inst {
-            fn regs() -> &'static crate::pac::dac::Dac {
-                &crate::pac::$inst
+            fn regs() -> crate::pac::dac::Dac {
+                crate::pac::$inst
             }
         }
 
diff --git a/embassy-stm32/src/dcmi.rs b/embassy-stm32/src/dcmi.rs
index 646ee2ce2..858ae49ca 100644
--- a/embassy-stm32/src/dcmi.rs
+++ b/embassy-stm32/src/dcmi.rs
@@ -9,7 +9,7 @@ use embassy_sync::waitqueue::AtomicWaker;
 use crate::dma::Transfer;
 use crate::gpio::{AFType, Speed};
 use crate::interrupt::typelevel::Interrupt;
-use crate::{interrupt, Peripheral};
+use crate::{interrupt, rcc, Peripheral};
 
 /// Interrupt handler.
 pub struct InterruptHandler<T: Instance> {
@@ -350,7 +350,7 @@ where
         use_embedded_synchronization: bool,
         edm: u8,
     ) -> Self {
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         peri.regs().cr().modify(|r| {
             r.set_cm(true); // disable continuous mode (snapshot mode)
diff --git a/embassy-stm32/src/dma/util.rs b/embassy-stm32/src/dma/util.rs
index 962ea2501..5aaca57c9 100644
--- a/embassy-stm32/src/dma/util.rs
+++ b/embassy-stm32/src/dma/util.rs
@@ -48,6 +48,7 @@ impl<'d> ChannelAndRequest<'d> {
         Transfer::new_write_raw(&mut self.channel, self.request, buf, peri_addr, options)
     }
 
+    #[allow(dead_code)]
     pub unsafe fn write_repeated<'a, W: Word>(
         &'a mut self,
         repeated: &'a W,
diff --git a/embassy-stm32/src/dsihost.rs b/embassy-stm32/src/dsihost.rs
index f1c737fdc..e1fb3b0b0 100644
--- a/embassy-stm32/src/dsihost.rs
+++ b/embassy-stm32/src/dsihost.rs
@@ -6,14 +6,14 @@ use embassy_hal_internal::{into_ref, PeripheralRef};
 
 //use crate::gpio::{AnyPin, SealedPin};
 use crate::gpio::{AFType, AnyPin, Pull, Speed};
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::{peripherals, Peripheral};
 
 /// Performs a busy-wait delay for a specified number of microseconds.
 pub fn blocking_delay_ms(ms: u32) {
-    #[cfg(time)]
-    embassy_time::block_for(embassy_time::Duration::from_millis(ms));
-    #[cfg(not(time))]
+    #[cfg(feature = "time")]
+    embassy_time::block_for(embassy_time::Duration::from_millis(ms as u64));
+    #[cfg(not(feature = "time"))]
     cortex_m::asm::delay(unsafe { crate::rcc::get_freqs() }.sys.unwrap().0 / 1_000 * ms);
 }
 
@@ -77,7 +77,7 @@ impl<'d, T: Instance> DsiHost<'d, T> {
     pub fn new(_peri: impl Peripheral<P = T> + 'd, te: impl Peripheral<P = impl TePin<T>> + 'd) -> Self {
         into_ref!(te);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         // Set Tearing Enable pin according to CubeMx example
         te.set_as_af_pull(te.af_num(), AFType::OutputPushPull, Pull::None);
@@ -407,7 +407,7 @@ impl<'d, T: Instance> Drop for DsiHost<'d, T> {
 }
 
 trait SealedInstance: crate::rcc::SealedRccPeripheral {
-    fn regs() -> &'static crate::pac::dsihost::Dsihost;
+    fn regs() -> crate::pac::dsihost::Dsihost;
 }
 
 /// DSI instance trait.
@@ -419,8 +419,8 @@ pin_trait!(TePin, Instance);
 foreach_peripheral!(
     (dsihost, $inst:ident) => {
         impl crate::dsihost::SealedInstance for peripherals::$inst {
-            fn regs() -> &'static crate::pac::dsihost::Dsihost {
-                &crate::pac::$inst
+            fn regs() -> crate::pac::dsihost::Dsihost {
+                crate::pac::$inst
             }
         }
 
diff --git a/embassy-stm32/src/flash/mod.rs b/embassy-stm32/src/flash/mod.rs
index 8c6ca2471..ce2d1a04c 100644
--- a/embassy-stm32/src/flash/mod.rs
+++ b/embassy-stm32/src/flash/mod.rs
@@ -96,7 +96,7 @@ pub enum FlashBank {
 #[cfg_attr(any(flash_f1, flash_f3), path = "f1f3.rs")]
 #[cfg_attr(flash_f4, path = "f4.rs")]
 #[cfg_attr(flash_f7, path = "f7.rs")]
-#[cfg_attr(any(flash_g0, flash_g4), path = "g.rs")]
+#[cfg_attr(any(flash_g0, flash_g4c2, flash_g4c3, flash_g4c4), path = "g.rs")]
 #[cfg_attr(flash_h7, path = "h7.rs")]
 #[cfg_attr(flash_h7ab, path = "h7.rs")]
 #[cfg_attr(flash_u5, path = "u5.rs")]
@@ -105,7 +105,7 @@ pub enum FlashBank {
 #[cfg_attr(
     not(any(
         flash_l0, flash_l1, flash_l4, flash_wl, flash_wb, flash_f0, flash_f1, flash_f3, flash_f4, flash_f7, flash_g0,
-        flash_g4, flash_h7, flash_h7ab, flash_u5, flash_h50, flash_u0
+        flash_g4c2, flash_g4c3, flash_g4c4, flash_h7, flash_h7ab, flash_u5, flash_h50, flash_u0
     )),
     path = "other.rs"
 )]
diff --git a/embassy-stm32/src/fmc.rs b/embassy-stm32/src/fmc.rs
index aced69878..82d8089f4 100644
--- a/embassy-stm32/src/fmc.rs
+++ b/embassy-stm32/src/fmc.rs
@@ -4,7 +4,7 @@ use core::marker::PhantomData;
 use embassy_hal_internal::into_ref;
 
 use crate::gpio::{AFType, Pull, Speed};
-use crate::Peripheral;
+use crate::{rcc, Peripheral};
 
 /// FMC driver
 pub struct Fmc<'d, T: Instance> {
@@ -27,7 +27,7 @@ where
 
     /// Enable the FMC peripheral and reset it.
     pub fn enable(&mut self) {
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
     }
 
     /// Enable the memory controller on applicable chips.
@@ -35,7 +35,7 @@ where
         // fmc v1 and v2 does not have the fmcen bit
         // fsmc v1, v2 and v3 does not have the fmcen bit
         // This is a "not" because it is expected that all future versions have this bit
-        #[cfg(not(any(fmc_v1x3, fmc_v2x1, fsmc_v1x0, fsmc_v1x3, fsmc_v2x3, fsmc_v3x1, fmc_v4)))]
+        #[cfg(not(any(fmc_v1x3, fmc_v2x1, fsmc_v1x0, fsmc_v1x3, fmc_v4)))]
         T::REGS.bcr1().modify(|r| r.set_fmcen(true));
         #[cfg(any(fmc_v4))]
         T::REGS.nor_psram().bcr1().modify(|r| r.set_fmcen(true));
@@ -54,14 +54,14 @@ where
     const REGISTERS: *const () = T::REGS.as_ptr() as *const _;
 
     fn enable(&mut self) {
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
     }
 
     fn memory_controller_enable(&mut self) {
         // fmc v1 and v2 does not have the fmcen bit
         // fsmc v1, v2 and v3 does not have the fmcen bit
         // This is a "not" because it is expected that all future versions have this bit
-        #[cfg(not(any(fmc_v1x3, fmc_v2x1, fsmc_v1x0, fsmc_v1x3, fsmc_v2x3, fsmc_v3x1, fmc_v4)))]
+        #[cfg(not(any(fmc_v1x3, fmc_v2x1, fsmc_v1x0, fsmc_v1x3, fmc_v4)))]
         T::REGS.bcr1().modify(|r| r.set_fmcen(true));
         #[cfg(any(fmc_v4))]
         T::REGS.nor_psram().bcr1().modify(|r| r.set_fmcen(true));
@@ -200,7 +200,7 @@ impl<'d, T: Instance> Fmc<'d, T> {
     ));
 }
 
-trait SealedInstance: crate::rcc::SealedRccPeripheral {
+trait SealedInstance: crate::rcc::RccPeripheral {
     const REGS: crate::pac::fmc::Fmc;
 }
 
diff --git a/embassy-stm32/src/gpio.rs b/embassy-stm32/src/gpio.rs
index d60a857ef..de08127cf 100644
--- a/embassy-stm32/src/gpio.rs
+++ b/embassy-stm32/src/gpio.rs
@@ -822,7 +822,7 @@ foreach_pin!(
 
 pub(crate) unsafe fn init(_cs: CriticalSection) {
     #[cfg(afio)]
-    <crate::peripherals::AFIO as crate::rcc::SealedRccPeripheral>::enable_and_reset_with_cs(_cs);
+    crate::rcc::enable_and_reset_with_cs::<crate::peripherals::AFIO>(_cs);
 
     crate::_generated::init_gpio();
 }
diff --git a/embassy-stm32/src/hash/mod.rs b/embassy-stm32/src/hash/mod.rs
index 787d5b1c9..4d4a8ec5b 100644
--- a/embassy-stm32/src/hash/mod.rs
+++ b/embassy-stm32/src/hash/mod.rs
@@ -17,8 +17,7 @@ use crate::dma::NoDma;
 use crate::dma::Transfer;
 use crate::interrupt::typelevel::Interrupt;
 use crate::peripherals::HASH;
-use crate::rcc::SealedRccPeripheral;
-use crate::{interrupt, pac, peripherals, Peripheral};
+use crate::{interrupt, pac, peripherals, rcc, Peripheral};
 
 #[cfg(hash_v1)]
 const NUM_CONTEXT_REGS: usize = 51;
@@ -130,7 +129,7 @@ impl<'d, T: Instance, D> Hash<'d, T, D> {
         dma: impl Peripheral<P = D> + 'd,
         _irq: impl interrupt::typelevel::Binding<T::Interrupt, InterruptHandler<T>> + 'd,
     ) -> Self {
-        HASH::enable_and_reset();
+        rcc::enable_and_reset::<HASH>();
         into_ref!(peripheral, dma);
         let instance = Self {
             _peripheral: peripheral,
diff --git a/embassy-stm32/src/hrtim/mod.rs b/embassy-stm32/src/hrtim/mod.rs
index 02e45819c..c9d5bff17 100644
--- a/embassy-stm32/src/hrtim/mod.rs
+++ b/embassy-stm32/src/hrtim/mod.rs
@@ -9,7 +9,7 @@ pub use traits::Instance;
 
 use crate::gpio::{AFType, AnyPin};
 use crate::time::Hertz;
-use crate::Peripheral;
+use crate::{rcc, Peripheral};
 
 /// HRTIM burst controller instance.
 pub struct BurstController<T: Instance> {
@@ -172,7 +172,7 @@ impl<'d, T: Instance> AdvancedPwm<'d, T> {
     fn new_inner(tim: impl Peripheral<P = T> + 'd) -> Self {
         into_ref!(tim);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         #[cfg(stm32f334)]
         if crate::pac::RCC.cfgr3().read().hrtim1sw() == crate::pac::rcc::vals::Timsw::PLL1_P {
diff --git a/embassy-stm32/src/i2c/mod.rs b/embassy-stm32/src/i2c/mod.rs
index ef5fd0972..6d12af2cc 100644
--- a/embassy-stm32/src/i2c/mod.rs
+++ b/embassy-stm32/src/i2c/mod.rs
@@ -9,16 +9,16 @@ use core::future::Future;
 use core::iter;
 use core::marker::PhantomData;
 
-use embassy_hal_internal::{into_ref, Peripheral};
+use embassy_hal_internal::{Peripheral, PeripheralRef};
 use embassy_sync::waitqueue::AtomicWaker;
 #[cfg(feature = "time")]
 use embassy_time::{Duration, Instant};
 
 use crate::dma::ChannelAndRequest;
-use crate::gpio::{AFType, Pull};
+use crate::gpio::{AFType, AnyPin, Pull, SealedPin as _, Speed};
 use crate::interrupt::typelevel::Interrupt;
 use crate::mode::{Async, Blocking, Mode};
-use crate::rcc::{ClockEnableBit, SealedRccPeripheral};
+use crate::rcc::{RccInfo, SealedRccPeripheral};
 use crate::time::Hertz;
 use crate::{interrupt, peripherals};
 
@@ -72,11 +72,29 @@ impl Default for Config {
     }
 }
 
+impl Config {
+    fn scl_pull_mode(&self) -> Pull {
+        match self.scl_pullup {
+            true => Pull::Up,
+            false => Pull::Down,
+        }
+    }
+
+    fn sda_pull_mode(&self) -> Pull {
+        match self.sda_pullup {
+            true => Pull::Up,
+            false => Pull::Down,
+        }
+    }
+}
+
 /// I2C driver.
 pub struct I2c<'d, M: Mode> {
     info: &'static Info,
     state: &'static State,
     kernel_clock: Hertz,
+    scl: Option<PeripheralRef<'d, AnyPin>>,
+    sda: Option<PeripheralRef<'d, AnyPin>>,
     tx_dma: Option<ChannelAndRequest<'d>>,
     rx_dma: Option<ChannelAndRequest<'d>>,
     #[cfg(feature = "time")]
@@ -98,7 +116,15 @@ impl<'d> I2c<'d, Async> {
         freq: Hertz,
         config: Config,
     ) -> Self {
-        Self::new_inner(peri, scl, sda, new_dma!(tx_dma), new_dma!(rx_dma), freq, config)
+        Self::new_inner(
+            peri,
+            new_pin!(scl, AFType::OutputOpenDrain, Speed::Medium, config.scl_pull_mode()),
+            new_pin!(sda, AFType::OutputOpenDrain, Speed::Medium, config.sda_pull_mode()),
+            new_dma!(tx_dma),
+            new_dma!(rx_dma),
+            freq,
+            config,
+        )
     }
 }
 
@@ -111,7 +137,15 @@ impl<'d> I2c<'d, Blocking> {
         freq: Hertz,
         config: Config,
     ) -> Self {
-        Self::new_inner(peri, scl, sda, None, None, freq, config)
+        Self::new_inner(
+            peri,
+            new_pin!(scl, AFType::OutputOpenDrain, Speed::Medium, config.scl_pull_mode()),
+            new_pin!(sda, AFType::OutputOpenDrain, Speed::Medium, config.sda_pull_mode()),
+            None,
+            None,
+            freq,
+            config,
+        )
     }
 }
 
@@ -119,34 +153,13 @@ impl<'d, M: Mode> I2c<'d, M> {
     /// Create a new I2C driver.
     fn new_inner<T: Instance>(
         _peri: impl Peripheral<P = T> + 'd,
-        scl: impl Peripheral<P = impl SclPin<T>> + 'd,
-        sda: impl Peripheral<P = impl SdaPin<T>> + 'd,
+        scl: Option<PeripheralRef<'d, AnyPin>>,
+        sda: Option<PeripheralRef<'d, AnyPin>>,
         tx_dma: Option<ChannelAndRequest<'d>>,
         rx_dma: Option<ChannelAndRequest<'d>>,
         freq: Hertz,
         config: Config,
     ) -> Self {
-        into_ref!(scl, sda);
-
-        T::enable_and_reset();
-
-        scl.set_as_af_pull(
-            scl.af_num(),
-            AFType::OutputOpenDrain,
-            match config.scl_pullup {
-                true => Pull::Up,
-                false => Pull::None,
-            },
-        );
-        sda.set_as_af_pull(
-            sda.af_num(),
-            AFType::OutputOpenDrain,
-            match config.sda_pullup {
-                true => Pull::Up,
-                false => Pull::None,
-            },
-        );
-
         unsafe { T::EventInterrupt::enable() };
         unsafe { T::ErrorInterrupt::enable() };
 
@@ -154,18 +167,23 @@ impl<'d, M: Mode> I2c<'d, M> {
             info: T::info(),
             state: T::state(),
             kernel_clock: T::frequency(),
+            scl,
+            sda,
             tx_dma,
             rx_dma,
             #[cfg(feature = "time")]
             timeout: config.timeout,
             _phantom: PhantomData,
         };
-
-        this.init(freq, config);
-
+        this.enable_and_init(freq, config);
         this
     }
 
+    fn enable_and_init(&mut self, freq: Hertz, config: Config) {
+        self.info.rcc.enable_and_reset();
+        self.init(freq, config);
+    }
+
     fn timeout(&self) -> Timeout {
         Timeout {
             #[cfg(feature = "time")]
@@ -174,6 +192,15 @@ impl<'d, M: Mode> I2c<'d, M> {
     }
 }
 
+impl<'d, M: Mode> Drop for I2c<'d, M> {
+    fn drop(&mut self) {
+        self.scl.as_ref().map(|x| x.set_as_disconnected());
+        self.sda.as_ref().map(|x| x.set_as_disconnected());
+
+        self.info.rcc.disable()
+    }
+}
+
 #[derive(Copy, Clone)]
 struct Timeout {
     #[cfg(feature = "time")]
@@ -224,7 +251,7 @@ impl State {
 
 struct Info {
     regs: crate::pac::i2c::I2c,
-    pub(crate) enable_bit: ClockEnableBit,
+    rcc: RccInfo,
 }
 
 peri_trait!(
@@ -265,7 +292,7 @@ foreach_peripheral!(
             fn info() -> &'static Info {
                 static INFO: Info = Info{
                     regs: crate::pac::$inst,
-                    enable_bit: crate::peripherals::$inst::ENABLE_BIT,
+                    rcc: crate::peripherals::$inst::RCC_INFO,
                 };
                 &INFO
             }
diff --git a/embassy-stm32/src/i2c/v1.rs b/embassy-stm32/src/i2c/v1.rs
index 0269e53aa..28026f83c 100644
--- a/embassy-stm32/src/i2c/v1.rs
+++ b/embassy-stm32/src/i2c/v1.rs
@@ -700,12 +700,6 @@ impl<'d> I2c<'d, Async> {
     }
 }
 
-impl<'d, M: PeriMode> Drop for I2c<'d, M> {
-    fn drop(&mut self) {
-        self.info.enable_bit.disable()
-    }
-}
-
 enum Mode {
     Fast,
     Standard,
diff --git a/embassy-stm32/src/i2c/v2.rs b/embassy-stm32/src/i2c/v2.rs
index aa6daf786..80163c287 100644
--- a/embassy-stm32/src/i2c/v2.rs
+++ b/embassy-stm32/src/i2c/v2.rs
@@ -671,12 +671,6 @@ impl<'d> I2c<'d, Async> {
     }
 }
 
-impl<'d, M: Mode> Drop for I2c<'d, M> {
-    fn drop(&mut self) {
-        self.info.enable_bit.disable();
-    }
-}
-
 /// I2C Stop Configuration
 ///
 /// Peripheral options for generating the STOP condition
diff --git a/embassy-stm32/src/i2s.rs b/embassy-stm32/src/i2s.rs
index c78810a38..f893dd235 100644
--- a/embassy-stm32/src/i2s.rs
+++ b/embassy-stm32/src/i2s.rs
@@ -1,7 +1,9 @@
 //! Inter-IC Sound (I2S)
+
 use embassy_hal_internal::into_ref;
 
-use crate::gpio::{AFType, AnyPin, SealedPin};
+use crate::dma::ChannelAndRequest;
+use crate::gpio::{AFType, AnyPin, SealedPin, Speed};
 use crate::mode::Async;
 use crate::pac::spi::vals;
 use crate::spi::{Config as SpiConfig, *};
@@ -17,15 +19,6 @@ pub enum Mode {
     Slave,
 }
 
-/// I2S function
-#[derive(Copy, Clone)]
-pub enum Function {
-    /// Transmit audio data
-    Transmit,
-    /// Receive audio data
-    Receive,
-}
-
 /// I2C standard
 #[derive(Copy, Clone)]
 pub enum Standard {
@@ -42,7 +35,7 @@ pub enum Standard {
 }
 
 impl Standard {
-    #[cfg(any(spi_v1, spi_f1))]
+    #[cfg(any(spi_v1, spi_v3, spi_f1))]
     const fn i2sstd(&self) -> vals::I2sstd {
         match self {
             Standard::Philips => vals::I2sstd::PHILIPS,
@@ -53,7 +46,7 @@ impl Standard {
         }
     }
 
-    #[cfg(any(spi_v1, spi_f1))]
+    #[cfg(any(spi_v1, spi_v3, spi_f1))]
     const fn pcmsync(&self) -> vals::Pcmsync {
         match self {
             Standard::PcmLongSync => vals::Pcmsync::LONG,
@@ -76,7 +69,7 @@ pub enum Format {
 }
 
 impl Format {
-    #[cfg(any(spi_v1, spi_f1))]
+    #[cfg(any(spi_v1, spi_v3, spi_f1))]
     const fn datlen(&self) -> vals::Datlen {
         match self {
             Format::Data16Channel16 => vals::Datlen::BITS16,
@@ -86,7 +79,7 @@ impl Format {
         }
     }
 
-    #[cfg(any(spi_v1, spi_f1))]
+    #[cfg(any(spi_v1, spi_v3, spi_f1))]
     const fn chlen(&self) -> vals::Chlen {
         match self {
             Format::Data16Channel16 => vals::Chlen::BITS16,
@@ -107,7 +100,7 @@ pub enum ClockPolarity {
 }
 
 impl ClockPolarity {
-    #[cfg(any(spi_v1, spi_f1))]
+    #[cfg(any(spi_v1, spi_v3, spi_f1))]
     const fn ckpol(&self) -> vals::Ckpol {
         match self {
             ClockPolarity::IdleHigh => vals::Ckpol::IDLEHIGH,
@@ -127,15 +120,13 @@ impl ClockPolarity {
 pub struct Config {
     /// Mode
     pub mode: Mode,
-    /// Function (transmit, receive)
-    pub function: Function,
     /// Which I2S standard to use.
     pub standard: Standard,
     /// Data format.
     pub format: Format,
     /// Clock polarity.
     pub clock_polarity: ClockPolarity,
-    /// True to eanble master clock output from this instance.
+    /// True to enable master clock output from this instance.
     pub master_clock: bool,
 }
 
@@ -143,7 +134,6 @@ impl Default for Config {
     fn default() -> Self {
         Self {
             mode: Mode::Master,
-            function: Function::Transmit,
             standard: Standard::Philips,
             format: Format::Data16Channel16,
             clock_polarity: ClockPolarity::IdleLow,
@@ -155,43 +145,169 @@ impl Default for Config {
 /// I2S driver.
 pub struct I2S<'d> {
     _peri: Spi<'d, Async>,
-    sd: Option<PeripheralRef<'d, AnyPin>>,
+    txsd: Option<PeripheralRef<'d, AnyPin>>,
+    rxsd: Option<PeripheralRef<'d, AnyPin>>,
     ws: Option<PeripheralRef<'d, AnyPin>>,
     ck: Option<PeripheralRef<'d, AnyPin>>,
     mck: Option<PeripheralRef<'d, AnyPin>>,
 }
 
+/// I2S function
+#[derive(Copy, Clone)]
+#[allow(dead_code)]
+enum Function {
+    /// Transmit audio data
+    Transmit,
+    /// Receive audio data
+    Receive,
+    #[cfg(spi_v3)]
+    /// Transmit and Receive audio data
+    FullDuplex,
+}
+
 impl<'d> I2S<'d> {
-    /// Note: Full-Duplex modes are not supported at this time
-    pub fn new<T: Instance>(
+    /// Create a transmitter driver
+    pub fn new_txonly<T: Instance>(
         peri: impl Peripheral<P = T> + 'd,
         sd: impl Peripheral<P = impl MosiPin<T>> + 'd,
         ws: impl Peripheral<P = impl WsPin<T>> + 'd,
         ck: impl Peripheral<P = impl CkPin<T>> + 'd,
         mck: impl Peripheral<P = impl MckPin<T>> + 'd,
         txdma: impl Peripheral<P = impl TxDma<T>> + 'd,
+        freq: Hertz,
+        config: Config,
+    ) -> Self {
+        into_ref!(sd);
+        Self::new_inner(
+            peri,
+            new_pin!(sd, AFType::OutputPushPull, Speed::VeryHigh),
+            None,
+            ws,
+            ck,
+            mck,
+            new_dma!(txdma),
+            None,
+            freq,
+            config,
+            Function::Transmit,
+        )
+    }
+
+    /// Create a receiver driver
+    pub fn new_rxonly<T: Instance>(
+        peri: impl Peripheral<P = T> + 'd,
+        sd: impl Peripheral<P = impl MisoPin<T>> + 'd,
+        ws: impl Peripheral<P = impl WsPin<T>> + 'd,
+        ck: impl Peripheral<P = impl CkPin<T>> + 'd,
+        mck: impl Peripheral<P = impl MckPin<T>> + 'd,
+        #[cfg(not(spi_v3))] txdma: impl Peripheral<P = impl TxDma<T>> + 'd,
+        rxdma: impl Peripheral<P = impl RxDma<T>> + 'd,
+        freq: Hertz,
+        config: Config,
+    ) -> Self {
+        into_ref!(sd);
+        Self::new_inner(
+            peri,
+            None,
+            new_pin!(sd, AFType::OutputPushPull, Speed::VeryHigh),
+            ws,
+            ck,
+            mck,
+            #[cfg(not(spi_v3))]
+            new_dma!(txdma),
+            #[cfg(spi_v3)]
+            None,
+            new_dma!(rxdma),
+            freq,
+            config,
+            #[cfg(not(spi_v3))]
+            Function::Transmit,
+            #[cfg(spi_v3)]
+            Function::Receive,
+        )
+    }
+
+    #[cfg(spi_v3)]
+    /// Create a full duplex transmitter driver
+    pub fn new_full_duplex<T: Instance>(
+        peri: impl Peripheral<P = T> + 'd,
+        txsd: impl Peripheral<P = impl MosiPin<T>> + 'd,
+        rxsd: impl Peripheral<P = impl MisoPin<T>> + 'd,
+        ws: impl Peripheral<P = impl WsPin<T>> + 'd,
+        ck: impl Peripheral<P = impl CkPin<T>> + 'd,
+        mck: impl Peripheral<P = impl MckPin<T>> + 'd,
+        txdma: impl Peripheral<P = impl TxDma<T>> + 'd,
         rxdma: impl Peripheral<P = impl RxDma<T>> + 'd,
         freq: Hertz,
         config: Config,
     ) -> Self {
-        into_ref!(sd, ws, ck, mck);
+        into_ref!(txsd, rxsd);
+        Self::new_inner(
+            peri,
+            new_pin!(txsd, AFType::OutputPushPull, Speed::VeryHigh),
+            new_pin!(rxsd, AFType::OutputPushPull, Speed::VeryHigh),
+            ws,
+            ck,
+            mck,
+            new_dma!(txdma),
+            new_dma!(rxdma),
+            freq,
+            config,
+            Function::FullDuplex,
+        )
+    }
+
+    /// Write audio data.
+    pub async fn read<W: Word>(&mut self, data: &mut [W]) -> Result<(), Error> {
+        self._peri.read(data).await
+    }
+
+    /// Write audio data.
+    pub async fn write<W: Word>(&mut self, data: &[W]) -> Result<(), Error> {
+        self._peri.write(data).await
+    }
+
+    /// Transfer audio data.
+    pub async fn transfer<W: Word>(&mut self, read: &mut [W], write: &[W]) -> Result<(), Error> {
+        self._peri.transfer(read, write).await
+    }
 
-        sd.set_as_af(sd.af_num(), AFType::OutputPushPull);
-        sd.set_speed(crate::gpio::Speed::VeryHigh);
+    /// Transfer audio data in place.
+    pub async fn transfer_in_place<W: Word>(&mut self, data: &mut [W]) -> Result<(), Error> {
+        self._peri.transfer_in_place(data).await
+    }
+
+    fn new_inner<T: Instance>(
+        peri: impl Peripheral<P = T> + 'd,
+        txsd: Option<PeripheralRef<'d, AnyPin>>,
+        rxsd: Option<PeripheralRef<'d, AnyPin>>,
+        ws: impl Peripheral<P = impl WsPin<T>> + 'd,
+        ck: impl Peripheral<P = impl CkPin<T>> + 'd,
+        mck: impl Peripheral<P = impl MckPin<T>> + 'd,
+        txdma: Option<ChannelAndRequest<'d>>,
+        rxdma: Option<ChannelAndRequest<'d>>,
+        freq: Hertz,
+        config: Config,
+        function: Function,
+    ) -> Self {
+        into_ref!(ws, ck, mck);
 
         ws.set_as_af(ws.af_num(), AFType::OutputPushPull);
-        ws.set_speed(crate::gpio::Speed::VeryHigh);
+        ws.set_speed(Speed::VeryHigh);
 
         ck.set_as_af(ck.af_num(), AFType::OutputPushPull);
-        ck.set_speed(crate::gpio::Speed::VeryHigh);
+        ck.set_speed(Speed::VeryHigh);
 
         mck.set_as_af(mck.af_num(), AFType::OutputPushPull);
-        mck.set_speed(crate::gpio::Speed::VeryHigh);
+        mck.set_speed(Speed::VeryHigh);
 
         let mut spi_cfg = SpiConfig::default();
         spi_cfg.frequency = freq;
+
         let spi = Spi::new_internal(peri, txdma, rxdma, spi_cfg);
 
+        let regs = T::info().regs;
+
         // TODO move i2s to the new mux infra.
         //#[cfg(all(rcc_f4, not(stm32f410)))]
         //let pclk = unsafe { get_freqs() }.plli2s1_q.unwrap();
@@ -200,26 +316,23 @@ impl<'d> I2S<'d> {
 
         let (odd, div) = compute_baud_rate(pclk, freq, config.master_clock, config.format);
 
-        #[cfg(any(spi_v1, spi_f1))]
+        #[cfg(any(spi_v1, spi_v3, spi_f1))]
         {
-            use stm32_metapac::spi::vals::{I2scfg, Odd};
+            #[cfg(spi_v3)]
+            {
+                regs.cr1().modify(|w| w.set_spe(false));
 
-            // 1. Select the I2SDIV[7:0] bits in the SPI_I2SPR register to define the serial clock baud
-            // rate to reach the proper audio sample frequency. The ODD bit in the SPI_I2SPR
-            // register also has to be defined.
+                reset_incompatible_bitfields::<T>();
+            }
 
-            spi.info.regs.i2spr().modify(|w| {
-                w.set_i2sdiv(div);
-                w.set_odd(match odd {
-                    true => Odd::ODD,
-                    false => Odd::EVEN,
-                });
+            use stm32_metapac::spi::vals::{I2scfg, Odd};
 
-                w.set_mckoe(config.master_clock);
-            });
+            // 1. Select the I2SDIV[7:0] bits in the SPI_I2SPR/SPI_I2SCFGR register to define the serial clock baud
+            // rate to reach the proper audio sample frequency. The ODD bit in the
+            // SPI_I2SPR/SPI_I2SCFGR register also has to be defined.
 
             // 2. Select the CKPOL bit to define the steady level for the communication clock. Set the
-            // MCKOE bit in the SPI_I2SPR register if the master clock MCK needs to be provided to
+            // MCKOE bit in the SPI_I2SPR/SPI_I2SCFGR register if the master clock MCK needs to be provided to
             // the external DAC/ADC audio component (the I2SDIV and ODD values should be
             // computed depending on the state of the MCK output, for more details refer to
             // Section 28.4.4: Clock generator).
@@ -235,50 +348,72 @@ impl<'d> I2S<'d> {
 
             // 5. The I2SE bit in SPI_I2SCFGR register must be set.
 
-            spi.info.regs.i2scfgr().modify(|w| {
+            let clk_reg = {
+                #[cfg(any(spi_v1, spi_f1))]
+                {
+                    regs.i2spr()
+                }
+                #[cfg(spi_v3)]
+                {
+                    regs.i2scfgr()
+                }
+            };
+
+            clk_reg.modify(|w| {
+                w.set_i2sdiv(div);
+                w.set_odd(match odd {
+                    true => Odd::ODD,
+                    false => Odd::EVEN,
+                });
+
+                w.set_mckoe(config.master_clock);
+            });
+
+            regs.i2scfgr().modify(|w| {
                 w.set_ckpol(config.clock_polarity.ckpol());
 
                 w.set_i2smod(true);
+
                 w.set_i2sstd(config.standard.i2sstd());
                 w.set_pcmsync(config.standard.pcmsync());
 
                 w.set_datlen(config.format.datlen());
                 w.set_chlen(config.format.chlen());
 
-                w.set_i2scfg(match (config.mode, config.function) {
+                w.set_i2scfg(match (config.mode, function) {
                     (Mode::Master, Function::Transmit) => I2scfg::MASTERTX,
                     (Mode::Master, Function::Receive) => I2scfg::MASTERRX,
+                    #[cfg(spi_v3)]
+                    (Mode::Master, Function::FullDuplex) => I2scfg::MASTERFULLDUPLEX,
                     (Mode::Slave, Function::Transmit) => I2scfg::SLAVETX,
                     (Mode::Slave, Function::Receive) => I2scfg::SLAVERX,
+                    #[cfg(spi_v3)]
+                    (Mode::Slave, Function::FullDuplex) => I2scfg::SLAVEFULLDUPLEX,
                 });
 
-                w.set_i2se(true)
+                #[cfg(any(spi_v1, spi_f1))]
+                w.set_i2se(true);
             });
+
+            #[cfg(spi_v3)]
+            regs.cr1().modify(|w| w.set_spe(true));
         }
 
         Self {
             _peri: spi,
-            sd: Some(sd.map_into()),
+            txsd: txsd.map(|w| w.map_into()),
+            rxsd: rxsd.map(|w| w.map_into()),
             ws: Some(ws.map_into()),
             ck: Some(ck.map_into()),
             mck: Some(mck.map_into()),
         }
     }
-
-    /// Write audio data.
-    pub async fn write<W: Word>(&mut self, data: &[W]) -> Result<(), Error> {
-        self._peri.write(data).await
-    }
-
-    /// Read audio data.
-    pub async fn read<W: Word>(&mut self, data: &mut [W]) -> Result<(), Error> {
-        self._peri.read(data).await
-    }
 }
 
 impl<'d> Drop for I2S<'d> {
     fn drop(&mut self) {
-        self.sd.as_ref().map(|x| x.set_as_disconnected());
+        self.txsd.as_ref().map(|x| x.set_as_disconnected());
+        self.rxsd.as_ref().map(|x| x.set_as_disconnected());
         self.ws.as_ref().map(|x| x.set_as_disconnected());
         self.ck.as_ref().map(|x| x.set_as_disconnected());
         self.mck.as_ref().map(|x| x.set_as_disconnected());
@@ -320,3 +455,71 @@ fn compute_baud_rate(i2s_clock: Hertz, request_freq: Hertz, mclk: bool, data_for
         ((division & 1) == 1, (division >> 1) as u8)
     }
 }
+
+#[cfg(spi_v3)]
+// The STM32H7 reference manual specifies that any incompatible bitfields should be reset
+// to their reset values while operating in I2S mode.
+fn reset_incompatible_bitfields<T: Instance>() {
+    let regs = T::info().regs;
+    regs.cr1().modify(|w| {
+        let iolock = w.iolock();
+        let csusp = w.csusp();
+        let spe = w.cstart();
+        let cstart = w.cstart();
+        w.0 = 0;
+        w.set_iolock(iolock);
+        w.set_csusp(csusp);
+        w.set_spe(spe);
+        w.set_cstart(cstart);
+    });
+
+    regs.cr2().write(|w| w.0 = 0);
+
+    regs.cfg1().modify(|w| {
+        let txdmaen = w.txdmaen();
+        let rxdmaen = w.rxdmaen();
+        let fthlv = w.fthlv();
+        w.0 = 0;
+        w.set_txdmaen(txdmaen);
+        w.set_rxdmaen(rxdmaen);
+        w.set_fthlv(fthlv);
+    });
+
+    regs.cfg2().modify(|w| {
+        let afcntr = w.afcntr();
+        let lsbfirst = w.lsbfirst();
+        let ioswp = w.ioswp();
+        w.0 = 0;
+        w.set_afcntr(afcntr);
+        w.set_lsbfirst(lsbfirst);
+        w.set_ioswp(ioswp);
+    });
+
+    regs.ier().modify(|w| {
+        let tifreie = w.tifreie();
+        let ovrie = w.ovrie();
+        let udrie = w.udrie();
+        let txpie = w.txpie();
+        let rxpie = w.rxpie();
+
+        w.0 = 0;
+
+        w.set_tifreie(tifreie);
+        w.set_ovrie(ovrie);
+        w.set_udrie(udrie);
+        w.set_txpie(txpie);
+        w.set_rxpie(rxpie);
+    });
+
+    regs.ifcr().write(|w| {
+        w.set_suspc(true);
+        w.set_tifrec(true);
+        w.set_ovrc(true);
+        w.set_udrc(true);
+    });
+
+    regs.crcpoly().write(|w| w.0 = 0x107);
+    regs.txcrc().write(|w| w.0 = 0);
+    regs.rxcrc().write(|w| w.0 = 0);
+    regs.udrdr().write(|w| w.0 = 0);
+}
diff --git a/embassy-stm32/src/ipcc.rs b/embassy-stm32/src/ipcc.rs
index 4d535cce2..6c8347311 100644
--- a/embassy-stm32/src/ipcc.rs
+++ b/embassy-stm32/src/ipcc.rs
@@ -6,10 +6,9 @@ use core::task::Poll;
 
 use embassy_sync::waitqueue::AtomicWaker;
 
-use crate::interrupt;
 use crate::interrupt::typelevel::Interrupt;
 use crate::peripherals::IPCC;
-use crate::rcc::SealedRccPeripheral;
+use crate::{interrupt, rcc};
 
 /// Interrupt handler.
 pub struct ReceiveInterruptHandler {}
@@ -102,7 +101,7 @@ pub struct Ipcc;
 impl Ipcc {
     /// Enable IPCC.
     pub fn enable(_config: Config) {
-        IPCC::enable_and_reset();
+        rcc::enable_and_reset::<IPCC>();
         IPCC::set_cpu2(true);
 
         // set RF wake-up clock = LSE
diff --git a/embassy-stm32/src/lib.rs b/embassy-stm32/src/lib.rs
index 81ee60c1c..95f59360a 100644
--- a/embassy-stm32/src/lib.rs
+++ b/embassy-stm32/src/lib.rs
@@ -83,7 +83,7 @@ pub mod hrtim;
 pub mod hsem;
 #[cfg(i2c)]
 pub mod i2c;
-#[cfg(all(spi_v1, rcc_f4))]
+#[cfg(any(all(spi_v1, rcc_f4), spi_v3))]
 pub mod i2s;
 #[cfg(stm32wb)]
 pub mod ipcc;
@@ -194,7 +194,6 @@ pub(crate) use stm32_metapac as pac;
 use crate::interrupt::Priority;
 #[cfg(feature = "rt")]
 pub use crate::pac::NVIC_PRIO_BITS;
-use crate::rcc::SealedRccPeripheral;
 
 /// `embassy-stm32` global configuration.
 #[non_exhaustive]
@@ -310,11 +309,11 @@ pub fn init(config: Config) -> Peripherals {
         });
 
         #[cfg(not(any(stm32f1, stm32wb, stm32wl)))]
-        peripherals::SYSCFG::enable_and_reset_with_cs(cs);
+        rcc::enable_and_reset_with_cs::<peripherals::SYSCFG>(cs);
         #[cfg(not(any(stm32h5, stm32h7, stm32h7rs, stm32wb, stm32wl)))]
-        peripherals::PWR::enable_and_reset_with_cs(cs);
+        rcc::enable_and_reset_with_cs::<peripherals::PWR>(cs);
         #[cfg(not(any(stm32f2, stm32f4, stm32f7, stm32l0, stm32h5, stm32h7, stm32h7rs)))]
-        peripherals::FLASH::enable_and_reset_with_cs(cs);
+        rcc::enable_and_reset_with_cs::<peripherals::FLASH>(cs);
 
         // Enable the VDDIO2 power supply on chips that have it.
         // Note that this requires the PWR peripheral to be enabled first.
diff --git a/embassy-stm32/src/ltdc.rs b/embassy-stm32/src/ltdc.rs
index 0cc8a0557..c262e7a0c 100644
--- a/embassy-stm32/src/ltdc.rs
+++ b/embassy-stm32/src/ltdc.rs
@@ -1,7 +1,7 @@
 //! LTDC
 use core::marker::PhantomData;
 
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::{peripherals, Peripheral};
 
 /// LTDC driver.
@@ -60,7 +60,7 @@ impl<'d, T: Instance> Ltdc<'d, T> {
                 .modify(|w| w.set_pllsaidivr(stm32_metapac::rcc::vals::Pllsaidivr::DIV2));
         });
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         //new_pin!(clk, AFType::OutputPushPull, Speed::VeryHigh,  Pull::None);
 
@@ -93,7 +93,7 @@ impl<'d, T: Instance> Drop for Ltdc<'d, T> {
 }
 
 trait SealedInstance: crate::rcc::SealedRccPeripheral {
-    fn regs() -> &'static crate::pac::ltdc::Ltdc;
+    fn regs() -> crate::pac::ltdc::Ltdc;
 }
 
 /// DSI instance trait.
@@ -132,8 +132,8 @@ pin_trait!(B7Pin, Instance);
 foreach_peripheral!(
     (ltdc, $inst:ident) => {
         impl crate::ltdc::SealedInstance for peripherals::$inst {
-            fn regs() -> &'static crate::pac::ltdc::Ltdc {
-                &crate::pac::$inst
+            fn regs() -> crate::pac::ltdc::Ltdc {
+                crate::pac::$inst
             }
         }
 
diff --git a/embassy-stm32/src/opamp.rs b/embassy-stm32/src/opamp.rs
index 487902959..ca94a573d 100644
--- a/embassy-stm32/src/opamp.rs
+++ b/embassy-stm32/src/opamp.rs
@@ -110,6 +110,32 @@ impl<'d, T: Instance> OpAmp<'d, T> {
 
         OpAmpOutput { _inner: self }
     }
+    /// Configure the OpAmp as a buffer for the DAC it is connected to,
+    /// outputting to the provided output pin, and enable the opamp.
+    ///
+    /// The output pin is held within the returned [`OpAmpOutput`] struct,
+    /// preventing it being used elsewhere. The `OpAmpOutput` can then be
+    /// directly used as an ADC input. The opamp will be disabled when the
+    /// [`OpAmpOutput`] is dropped.
+    #[cfg(opamp_g4)]
+    pub fn buffer_dac(
+        &'d mut self,
+        out_pin: impl Peripheral<P = impl OutputPin<T> + crate::gpio::Pin> + 'd,
+    ) -> OpAmpOutput<'d, T> {
+        into_ref!(out_pin);
+        out_pin.set_as_analog();
+
+        T::regs().csr().modify(|w| {
+            use crate::pac::opamp::vals::*;
+
+            w.set_vm_sel(VmSel::OUTPUT);
+            w.set_vp_sel(VpSel::DAC3_CH1);
+            w.set_opaintoen(Opaintoen::OUTPUTPIN);
+            w.set_opampen(true);
+        });
+
+        OpAmpOutput { _inner: self }
+    }
 
     /// Configure the OpAmp as a buffer for the provided input pin,
     /// with the output only used internally, and enable the opamp.
diff --git a/embassy-stm32/src/ospi/mod.rs b/embassy-stm32/src/ospi/mod.rs
index 536da4ca0..882781cce 100644
--- a/embassy-stm32/src/ospi/mod.rs
+++ b/embassy-stm32/src/ospi/mod.rs
@@ -16,7 +16,7 @@ use crate::dma::{word, ChannelAndRequest};
 use crate::gpio::{AFType, AnyPin, Pull, SealedPin as _, Speed};
 use crate::mode::{Async, Blocking, Mode as PeriMode};
 use crate::pac::octospi::{vals, Octospi as Regs};
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::{peripherals, Peripheral};
 
 /// OPSI driver config.
@@ -198,7 +198,7 @@ impl<'d, T: Instance, M: PeriMode> Ospi<'d, T, M> {
         into_ref!(peri);
 
         // System configuration
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
         while T::REGS.sr().read().busy() {}
 
         // Device configuration
@@ -1013,7 +1013,7 @@ impl<'d, T: Instance, M: PeriMode> Drop for Ospi<'d, T, M> {
         self.nss.as_ref().map(|x| x.set_as_disconnected());
         self.dqs.as_ref().map(|x| x.set_as_disconnected());
 
-        T::disable();
+        rcc::disable::<T>();
     }
 }
 
diff --git a/embassy-stm32/src/qspi/mod.rs b/embassy-stm32/src/qspi/mod.rs
index a82e93b5b..06c8f4812 100644
--- a/embassy-stm32/src/qspi/mod.rs
+++ b/embassy-stm32/src/qspi/mod.rs
@@ -13,7 +13,7 @@ use crate::dma::ChannelAndRequest;
 use crate::gpio::{AFType, AnyPin, Pull, Speed};
 use crate::mode::{Async, Blocking, Mode as PeriMode};
 use crate::pac::quadspi::Quadspi as Regs;
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::{peripherals, Peripheral};
 
 /// QSPI transfer configuration.
@@ -102,7 +102,7 @@ impl<'d, T: Instance, M: PeriMode> Qspi<'d, T, M> {
     ) -> Self {
         into_ref!(peri);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         while T::REGS.sr().read().busy() {}
 
diff --git a/embassy-stm32/src/rcc/bd.rs b/embassy-stm32/src/rcc/bd.rs
index f3ac4fdbe..4e9c18594 100644
--- a/embassy-stm32/src/rcc/bd.rs
+++ b/embassy-stm32/src/rcc/bd.rs
@@ -33,7 +33,7 @@ pub enum LseDrive {
 }
 
 // All families but these have the LSEDRV register
-#[cfg(not(any(rcc_f1, rcc_f1cl, rcc_f100, rcc_f2, rcc_f4, rcc_f400, rcc_f410, rcc_l1)))]
+#[cfg(not(any(rcc_f1, rcc_f1cl, rcc_f100, rcc_f2, rcc_f4, rcc_f410, rcc_l1)))]
 impl From<LseDrive> for crate::pac::rcc::vals::Lsedrv {
     fn from(value: LseDrive) -> Self {
         use crate::pac::rcc::vals::Lsedrv;
@@ -186,7 +186,7 @@ impl LsConfig {
         }
         ok &= reg.lseon() == lse_en;
         ok &= reg.lsebyp() == lse_byp;
-        #[cfg(not(any(rcc_f1, rcc_f1cl, rcc_f100, rcc_f2, rcc_f4, rcc_f400, rcc_f410, rcc_l1)))]
+        #[cfg(not(any(rcc_f1, rcc_f1cl, rcc_f100, rcc_f2, rcc_f4, rcc_f410, rcc_l1)))]
         if let Some(lse_drv) = lse_drv {
             ok &= reg.lsedrv() == lse_drv.into();
         }
@@ -224,7 +224,7 @@ impl LsConfig {
 
         if lse_en {
             bdcr().modify(|w| {
-                #[cfg(not(any(rcc_f1, rcc_f1cl, rcc_f100, rcc_f2, rcc_f4, rcc_f400, rcc_f410, rcc_l1)))]
+                #[cfg(not(any(rcc_f1, rcc_f1cl, rcc_f100, rcc_f2, rcc_f4, rcc_f410, rcc_l1)))]
                 if let Some(lse_drv) = lse_drv {
                     w.set_lsedrv(lse_drv.into());
                 }
diff --git a/embassy-stm32/src/rcc/f013.rs b/embassy-stm32/src/rcc/f013.rs
index 0946287ea..63dc27bdd 100644
--- a/embassy-stm32/src/rcc/f013.rs
+++ b/embassy-stm32/src/rcc/f013.rs
@@ -95,7 +95,7 @@ pub struct Config {
 
     #[cfg(all(stm32f3, not(rcc_f37)))]
     pub adc: AdcClockSource,
-    #[cfg(all(stm32f3, not(rcc_f37), adc3_common))]
+    #[cfg(all(stm32f3, not(rcc_f37), any(peri_adc3_common, peri_adc34_common)))]
     pub adc34: AdcClockSource,
 
     /// Per-peripheral kernel clock selection muxes
@@ -125,7 +125,7 @@ impl Default for Config {
 
             #[cfg(all(stm32f3, not(rcc_f37)))]
             adc: AdcClockSource::Hclk(AdcHclkPrescaler::Div1),
-            #[cfg(all(stm32f3, not(rcc_f37), adc3_common))]
+            #[cfg(all(stm32f3, not(rcc_f37), any(peri_adc3_common, peri_adc34_common)))]
             adc34: AdcClockSource::Hclk(AdcHclkPrescaler::Div1),
 
             mux: Default::default(),
@@ -276,7 +276,7 @@ pub(crate) unsafe fn init(config: Config) {
 
     // Set prescalers
     // CFGR has been written before (PLL, PLL48) don't overwrite these settings
-    RCC.cfgr().modify(|w: &mut stm32_metapac::rcc::regs::Cfgr| {
+    RCC.cfgr().modify(|w| {
         #[cfg(not(stm32f0))]
         {
             w.set_ppre1(config.apb1_pre);
@@ -339,7 +339,7 @@ pub(crate) unsafe fn init(config: Config) {
         }
     };
 
-    #[cfg(all(stm32f3, not(rcc_f37), adc3_common))]
+    #[cfg(all(stm32f3, not(rcc_f37), any(peri_adc3_common, peri_adc34_common)))]
     let adc34 = {
         #[cfg(peri_adc3_common)]
         let common = crate::pac::ADC3_COMMON;
@@ -404,7 +404,7 @@ pub(crate) unsafe fn init(config: Config) {
         hclk1: Some(hclk),
         #[cfg(all(stm32f3, not(rcc_f37)))]
         adc: Some(adc),
-        #[cfg(all(stm32f3, not(rcc_f37), adc3_common))]
+        #[cfg(all(stm32f3, not(rcc_f37), any(peri_adc3_common, peri_adc34_common)))]
         adc34: Some(adc34),
         rtc: rtc,
         hsi48: hsi48,
diff --git a/embassy-stm32/src/rcc/hsi48.rs b/embassy-stm32/src/rcc/hsi48.rs
index da81abc34..efabd059f 100644
--- a/embassy-stm32/src/rcc/hsi48.rs
+++ b/embassy-stm32/src/rcc/hsi48.rs
@@ -2,7 +2,7 @@
 
 use crate::pac::crs::vals::Syncsrc;
 use crate::pac::{CRS, RCC};
-use crate::rcc::SealedRccPeripheral;
+use crate::rcc::{self, SealedRccPeripheral};
 use crate::time::Hertz;
 
 /// HSI48 speed
@@ -44,7 +44,7 @@ pub(crate) fn init_hsi48(config: Hsi48Config) -> Hertz {
     while r.read().hsi48rdy() == false {}
 
     if config.sync_from_usb {
-        crate::peripherals::CRS::enable_and_reset();
+        rcc::enable_and_reset::<crate::peripherals::CRS>();
 
         CRS.cfgr().modify(|w| {
             w.set_syncsrc(Syncsrc::USB);
diff --git a/embassy-stm32/src/rcc/mod.rs b/embassy-stm32/src/rcc/mod.rs
index 28816256c..0bf344c40 100644
--- a/embassy-stm32/src/rcc/mod.rs
+++ b/embassy-stm32/src/rcc/mod.rs
@@ -67,23 +67,185 @@ pub(crate) unsafe fn get_freqs() -> &'static Clocks {
 }
 
 pub(crate) trait SealedRccPeripheral {
-    const ENABLE_BIT: ClockEnableBit;
-
     fn frequency() -> Hertz;
-    fn enable_and_reset_with_cs(cs: CriticalSection);
-    fn disable_with_cs(cs: CriticalSection);
-
-    fn enable_and_reset() {
-        critical_section::with(|cs| Self::enable_and_reset_with_cs(cs))
-    }
-    fn disable() {
-        critical_section::with(|cs| Self::disable_with_cs(cs))
-    }
+    const RCC_INFO: RccInfo;
 }
 
 #[allow(private_bounds)]
 pub trait RccPeripheral: SealedRccPeripheral + 'static {}
 
+/// Runtime information necessary to reset, enable and disable a peripheral.
+pub(crate) struct RccInfo {
+    /// Offset in 32-bit words of the xxxRSTR register into the RCC register block, or 0xff if the
+    /// peripheral has no reset bit (we don't use an `Option` to save one byte of storage).
+    reset_offset_or_0xff: u8,
+    /// Position of the xxxRST bit within the xxxRSTR register (0..=31).
+    reset_bit: u8,
+    /// Offset in 32-bit words of the xxxENR register into the RCC register block.
+    enable_offset: u8,
+    /// Position of the xxxEN bit within the xxxENR register (0..=31).
+    enable_bit: u8,
+    /// If this peripheral shares the same xxxRSTR bit and xxxEN bit with other peripherals, we
+    /// maintain a refcount in `crate::_generated::REFCOUNTS` at this index. If the bit is not
+    /// shared, this is 0xff (we don't use an `Option` to save one byte of storage).
+    refcount_idx_or_0xff: u8,
+    /// Stop mode of the peripheral, used to maintain `REFCOUNT_STOP1` and `REFCOUNT_STOP2`.
+    #[cfg(feature = "low-power")]
+    stop_mode: StopMode,
+}
+
+#[cfg(feature = "low-power")]
+#[allow(dead_code)]
+pub(crate) enum StopMode {
+    Standby,
+    Stop2,
+    Stop1,
+}
+
+impl RccInfo {
+    /// Safety:
+    /// - `reset_offset_and_bit`, if set, must correspond to valid xxxRST bit
+    /// - `enable_offset_and_bit` must correspond to valid xxxEN bit
+    /// - `refcount_idx`, if set, must correspond to valid refcount in `_generated::REFCOUNTS`
+    /// - `stop_mode` must be valid
+    pub(crate) const unsafe fn new(
+        reset_offset_and_bit: Option<(u8, u8)>,
+        enable_offset_and_bit: (u8, u8),
+        refcount_idx: Option<u8>,
+        #[cfg(feature = "low-power")] stop_mode: StopMode,
+    ) -> Self {
+        let (reset_offset_or_0xff, reset_bit) = match reset_offset_and_bit {
+            Some((offset, bit)) => (offset, bit),
+            None => (0xff, 0xff),
+        };
+        let (enable_offset, enable_bit) = enable_offset_and_bit;
+        let refcount_idx_or_0xff = match refcount_idx {
+            Some(idx) => idx,
+            None => 0xff,
+        };
+        Self {
+            reset_offset_or_0xff,
+            reset_bit,
+            enable_offset,
+            enable_bit,
+            refcount_idx_or_0xff,
+            #[cfg(feature = "low-power")]
+            stop_mode,
+        }
+    }
+
+    // TODO: should this be `unsafe`?
+    pub(crate) fn enable_and_reset_with_cs(&self, _cs: CriticalSection) {
+        if self.refcount_idx_or_0xff != 0xff {
+            let refcount_idx = self.refcount_idx_or_0xff as usize;
+            unsafe {
+                crate::_generated::REFCOUNTS[refcount_idx] += 1;
+            }
+            if unsafe { crate::_generated::REFCOUNTS[refcount_idx] } > 1 {
+                return;
+            }
+        }
+
+        #[cfg(feature = "low-power")]
+        match self.stop_mode {
+            StopMode::Standby => {}
+            StopMode::Stop2 => unsafe {
+                REFCOUNT_STOP2 += 1;
+            },
+            StopMode::Stop1 => unsafe {
+                REFCOUNT_STOP1 += 1;
+            },
+        }
+
+        // set the xxxRST bit
+        let reset_ptr = self.reset_ptr();
+        if let Some(reset_ptr) = reset_ptr {
+            unsafe {
+                let val = reset_ptr.read_volatile();
+                reset_ptr.write_volatile(val | 1u32 << self.reset_bit);
+            }
+        }
+
+        // set the xxxEN bit
+        let enable_ptr = self.enable_ptr();
+        unsafe {
+            let val = enable_ptr.read_volatile();
+            enable_ptr.write_volatile(val | 1u32 << self.enable_bit);
+        }
+
+        // we must wait two peripheral clock cycles before the clock is active
+        // this seems to work, but might be incorrect
+        // see http://efton.sk/STM32/gotcha/g183.html
+
+        // dummy read (like in the ST HALs)
+        let _ = unsafe { enable_ptr.read_volatile() };
+
+        // DSB for good measure
+        cortex_m::asm::dsb();
+
+        // clear the xxxRST bit
+        if let Some(reset_ptr) = reset_ptr {
+            unsafe {
+                let val = reset_ptr.read_volatile();
+                reset_ptr.write_volatile(val & !(1u32 << self.reset_bit));
+            }
+        }
+    }
+
+    // TODO: should this be `unsafe`?
+    pub(crate) fn disable_with_cs(&self, _cs: CriticalSection) {
+        if self.refcount_idx_or_0xff != 0xff {
+            let refcount_idx = self.refcount_idx_or_0xff as usize;
+            unsafe {
+                crate::_generated::REFCOUNTS[refcount_idx] -= 1;
+            }
+            if unsafe { crate::_generated::REFCOUNTS[refcount_idx] } > 0 {
+                return;
+            }
+        }
+
+        #[cfg(feature = "low-power")]
+        match self.stop_mode {
+            StopMode::Standby => {}
+            StopMode::Stop2 => unsafe {
+                REFCOUNT_STOP2 -= 1;
+            },
+            StopMode::Stop1 => unsafe {
+                REFCOUNT_STOP1 -= 1;
+            },
+        }
+
+        // clear the xxxEN bit
+        let enable_ptr = self.enable_ptr();
+        unsafe {
+            let val = enable_ptr.read_volatile();
+            enable_ptr.write_volatile(val & !(1u32 << self.enable_bit));
+        }
+    }
+
+    // TODO: should this be `unsafe`?
+    pub(crate) fn enable_and_reset(&self) {
+        critical_section::with(|cs| self.enable_and_reset_with_cs(cs))
+    }
+
+    // TODO: should this be `unsafe`?
+    pub(crate) fn disable(&self) {
+        critical_section::with(|cs| self.disable_with_cs(cs))
+    }
+
+    fn reset_ptr(&self) -> Option<*mut u32> {
+        if self.reset_offset_or_0xff != 0xff {
+            Some(unsafe { (RCC.as_ptr() as *mut u32).add(self.reset_offset_or_0xff as _) })
+        } else {
+            None
+        }
+    }
+
+    fn enable_ptr(&self) -> *mut u32 {
+        unsafe { (RCC.as_ptr() as *mut u32).add(self.enable_offset as _) }
+    }
+}
+
 #[allow(unused)]
 mod util {
     use crate::time::Hertz;
@@ -128,8 +290,9 @@ pub fn frequency<T: RccPeripheral>() -> Hertz {
 /// # Safety
 ///
 /// Peripheral must not be in use.
-pub unsafe fn enable_and_reset<T: RccPeripheral>() {
-    T::enable_and_reset();
+// TODO: should this be `unsafe`?
+pub fn enable_and_reset_with_cs<T: RccPeripheral>(cs: CriticalSection) {
+    T::RCC_INFO.enable_and_reset_with_cs(cs);
 }
 
 /// Disables peripheral `T`.
@@ -137,52 +300,27 @@ pub unsafe fn enable_and_reset<T: RccPeripheral>() {
 /// # Safety
 ///
 /// Peripheral must not be in use.
-pub unsafe fn disable<T: RccPeripheral>() {
-    T::disable();
+// TODO: should this be `unsafe`?
+pub fn disable_with_cs<T: RccPeripheral>(cs: CriticalSection) {
+    T::RCC_INFO.disable_with_cs(cs);
 }
 
-/// Struct representing some clock enable bit (xxxENR.xxEN), only known at runtime.
-#[derive(Clone, Copy)]
-pub(crate) struct ClockEnableBit {
-    /// offset in 32bit words of the xxxENR register into the RCC register block.
-    offset: u8,
-    /// bit within the register (0..=31)
-    bit: u8,
+/// Enables and resets peripheral `T`.
+///
+/// # Safety
+///
+/// Peripheral must not be in use.
+// TODO: should this be `unsafe`?
+pub fn enable_and_reset<T: RccPeripheral>() {
+    T::RCC_INFO.enable_and_reset();
 }
 
-impl ClockEnableBit {
-    /// Safety: offset+bit must correspond to a valid xxxEN bit.
-    pub(crate) const unsafe fn new(offset: u8, bit: u8) -> Self {
-        Self { offset, bit }
-    }
-
-    fn ptr(self) -> *mut u32 {
-        unsafe { (RCC.as_ptr() as *mut u32).add(self.offset as _) }
-    }
-
-    #[allow(unused)]
-    pub(crate) fn enable_with_cs(self, _cs: CriticalSection) {
-        let p = self.ptr();
-        unsafe {
-            let val = p.read_volatile();
-            p.write_volatile(val | 1u32 << self.bit);
-        }
-    }
-
-    pub(crate) fn disable_with_cs(self, _cs: CriticalSection) {
-        let p = self.ptr();
-        unsafe {
-            let val = p.read_volatile();
-            p.write_volatile(val & !(1u32 << self.bit));
-        }
-    }
-
-    #[allow(unused)]
-    pub(crate) fn enable(self) {
-        critical_section::with(|cs| self.enable_with_cs(cs))
-    }
-
-    pub(crate) fn disable(self) {
-        critical_section::with(|cs| self.disable_with_cs(cs))
-    }
+/// Disables peripheral `T`.
+///
+/// # Safety
+///
+/// Peripheral must not be in use.
+// TODO: should this be `unsafe`?
+pub fn disable<T: RccPeripheral>() {
+    T::RCC_INFO.disable();
 }
diff --git a/embassy-stm32/src/rng.rs b/embassy-stm32/src/rng.rs
index 7a228e4a4..94491c32f 100644
--- a/embassy-stm32/src/rng.rs
+++ b/embassy-stm32/src/rng.rs
@@ -10,7 +10,7 @@ use embassy_sync::waitqueue::AtomicWaker;
 use rand_core::{CryptoRng, RngCore};
 
 use crate::interrupt::typelevel::Interrupt;
-use crate::{interrupt, pac, peripherals, Peripheral};
+use crate::{interrupt, pac, peripherals, rcc, Peripheral};
 
 static RNG_WAKER: AtomicWaker = AtomicWaker::new();
 
@@ -52,7 +52,7 @@ impl<'d, T: Instance> Rng<'d, T> {
         inner: impl Peripheral<P = T> + 'd,
         _irq: impl interrupt::typelevel::Binding<T::Interrupt, InterruptHandler<T>> + 'd,
     ) -> Self {
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
         into_ref!(inner);
         let mut random = Self { _inner: inner };
         random.reset();
diff --git a/embassy-stm32/src/rtc/datetime.rs b/embassy-stm32/src/rtc/datetime.rs
index 77d89293d..32732e96e 100644
--- a/embassy-stm32/src/rtc/datetime.rs
+++ b/embassy-stm32/src/rtc/datetime.rs
@@ -1,65 +1,6 @@
 #[cfg(feature = "chrono")]
 use chrono::{Datelike, NaiveDate, Timelike, Weekday};
 
-#[cfg(any(feature = "defmt", feature = "time"))]
-use crate::peripherals::RTC;
-#[cfg(any(feature = "defmt", feature = "time"))]
-use crate::rtc::SealedInstance;
-
-/// Represents an instant in time that can be substracted to compute a duration
-pub struct RtcInstant {
-    /// 0..59
-    pub second: u8,
-    /// 0..256
-    pub subsecond: u16,
-}
-
-impl RtcInstant {
-    #[cfg(not(rtc_v2f2))]
-    pub(super) const fn from(second: u8, subsecond: u16) -> Result<Self, Error> {
-        if second > 59 {
-            Err(Error::InvalidSecond)
-        } else {
-            Ok(Self { second, subsecond })
-        }
-    }
-}
-
-#[cfg(feature = "defmt")]
-impl defmt::Format for RtcInstant {
-    fn format(&self, fmt: defmt::Formatter) {
-        defmt::write!(
-            fmt,
-            "{}:{}",
-            self.second,
-            RTC::regs().prer().read().prediv_s() - self.subsecond,
-        )
-    }
-}
-
-#[cfg(feature = "time")]
-impl core::ops::Sub for RtcInstant {
-    type Output = embassy_time::Duration;
-
-    fn sub(self, rhs: Self) -> Self::Output {
-        use embassy_time::{Duration, TICK_HZ};
-
-        let second = if self.second < rhs.second {
-            self.second + 60
-        } else {
-            self.second
-        };
-
-        let psc = RTC::regs().prer().read().prediv_s() as u32;
-
-        let self_ticks = second as u32 * (psc + 1) + (psc - self.subsecond as u32);
-        let other_ticks = rhs.second as u32 * (psc + 1) + (psc - rhs.subsecond as u32);
-        let rtc_ticks = self_ticks - other_ticks;
-
-        Duration::from_ticks(((rtc_ticks * TICK_HZ as u32) / (psc + 1)) as u64)
-    }
-}
-
 /// Errors regarding the [`DateTime`] struct.
 #[derive(Clone, Debug, PartialEq, Eq)]
 pub enum Error {
diff --git a/embassy-stm32/src/rtc/low_power.rs b/embassy-stm32/src/rtc/low_power.rs
new file mode 100644
index 000000000..a4ff1acbc
--- /dev/null
+++ b/embassy-stm32/src/rtc/low_power.rs
@@ -0,0 +1,230 @@
+use super::{bcd2_to_byte, DateTimeError, Rtc, RtcError};
+use crate::peripherals::RTC;
+use crate::rtc::SealedInstance;
+
+/// Represents an instant in time that can be substracted to compute a duration
+pub(super) struct RtcInstant {
+    /// 0..59
+    second: u8,
+    /// 0..256
+    subsecond: u16,
+}
+
+impl RtcInstant {
+    #[cfg(not(rtc_v2f2))]
+    const fn from(second: u8, subsecond: u16) -> Result<Self, DateTimeError> {
+        if second > 59 {
+            Err(DateTimeError::InvalidSecond)
+        } else {
+            Ok(Self { second, subsecond })
+        }
+    }
+}
+
+#[cfg(feature = "defmt")]
+impl defmt::Format for RtcInstant {
+    fn format(&self, fmt: defmt::Formatter) {
+        defmt::write!(
+            fmt,
+            "{}:{}",
+            self.second,
+            RTC::regs().prer().read().prediv_s() - self.subsecond,
+        )
+    }
+}
+
+#[cfg(feature = "time")]
+impl core::ops::Sub for RtcInstant {
+    type Output = embassy_time::Duration;
+
+    fn sub(self, rhs: Self) -> Self::Output {
+        use embassy_time::{Duration, TICK_HZ};
+
+        let second = if self.second < rhs.second {
+            self.second + 60
+        } else {
+            self.second
+        };
+
+        let psc = RTC::regs().prer().read().prediv_s() as u32;
+
+        let self_ticks = second as u32 * (psc + 1) + (psc - self.subsecond as u32);
+        let other_ticks = rhs.second as u32 * (psc + 1) + (psc - rhs.subsecond as u32);
+        let rtc_ticks = self_ticks - other_ticks;
+
+        Duration::from_ticks(((rtc_ticks * TICK_HZ as u32) / (psc + 1)) as u64)
+    }
+}
+
+#[repr(u8)]
+#[derive(Clone, Copy, Debug)]
+pub(crate) enum WakeupPrescaler {
+    Div2 = 2,
+    Div4 = 4,
+    Div8 = 8,
+    Div16 = 16,
+}
+
+#[cfg(any(stm32f4, stm32l0, stm32g4, stm32l5, stm32wb, stm32h5, stm32g0))]
+impl From<WakeupPrescaler> for crate::pac::rtc::vals::Wucksel {
+    fn from(val: WakeupPrescaler) -> Self {
+        use crate::pac::rtc::vals::Wucksel;
+
+        match val {
+            WakeupPrescaler::Div2 => Wucksel::DIV2,
+            WakeupPrescaler::Div4 => Wucksel::DIV4,
+            WakeupPrescaler::Div8 => Wucksel::DIV8,
+            WakeupPrescaler::Div16 => Wucksel::DIV16,
+        }
+    }
+}
+
+#[cfg(any(stm32f4, stm32l0, stm32g4, stm32l5, stm32wb, stm32h5, stm32g0))]
+impl From<crate::pac::rtc::vals::Wucksel> for WakeupPrescaler {
+    fn from(val: crate::pac::rtc::vals::Wucksel) -> Self {
+        use crate::pac::rtc::vals::Wucksel;
+
+        match val {
+            Wucksel::DIV2 => WakeupPrescaler::Div2,
+            Wucksel::DIV4 => WakeupPrescaler::Div4,
+            Wucksel::DIV8 => WakeupPrescaler::Div8,
+            Wucksel::DIV16 => WakeupPrescaler::Div16,
+            _ => unreachable!(),
+        }
+    }
+}
+
+impl WakeupPrescaler {
+    pub fn compute_min(val: u32) -> Self {
+        *[
+            WakeupPrescaler::Div2,
+            WakeupPrescaler::Div4,
+            WakeupPrescaler::Div8,
+            WakeupPrescaler::Div16,
+        ]
+        .iter()
+        .find(|psc| **psc as u32 > val)
+        .unwrap_or(&WakeupPrescaler::Div16)
+    }
+}
+
+impl Rtc {
+    /// Return the current instant.
+    fn instant(&self) -> Result<RtcInstant, RtcError> {
+        self.time_provider().read(|_, tr, ss| {
+            let second = bcd2_to_byte((tr.st(), tr.su()));
+
+            RtcInstant::from(second, ss).map_err(RtcError::InvalidDateTime)
+        })
+    }
+
+    /// start the wakeup alarm and with a duration that is as close to but less than
+    /// the requested duration, and record the instant the wakeup alarm was started
+    pub(crate) fn start_wakeup_alarm(
+        &self,
+        requested_duration: embassy_time::Duration,
+        cs: critical_section::CriticalSection,
+    ) {
+        use embassy_time::{Duration, TICK_HZ};
+
+        #[cfg(any(rtc_v3, rtc_v3u5, rtc_v3l5))]
+        use crate::pac::rtc::vals::Calrf;
+
+        // Panic if the rcc mod knows we're not using low-power rtc
+        #[cfg(any(rcc_wb, rcc_f4, rcc_f410))]
+        unsafe { crate::rcc::get_freqs() }.rtc.unwrap();
+
+        let requested_duration = requested_duration.as_ticks().clamp(0, u32::MAX as u64);
+        let rtc_hz = Self::frequency().0 as u64;
+        let rtc_ticks = requested_duration * rtc_hz / TICK_HZ;
+        let prescaler = WakeupPrescaler::compute_min((rtc_ticks / u16::MAX as u64) as u32);
+
+        // adjust the rtc ticks to the prescaler and subtract one rtc tick
+        let rtc_ticks = rtc_ticks / prescaler as u64;
+        let rtc_ticks = rtc_ticks.clamp(0, (u16::MAX - 1) as u64).saturating_sub(1) as u16;
+
+        self.write(false, |regs| {
+            regs.cr().modify(|w| w.set_wute(false));
+
+            #[cfg(any(
+                rtc_v2f0, rtc_v2f2, rtc_v2f3, rtc_v2f4, rtc_v2f7, rtc_v2h7, rtc_v2l0, rtc_v2l1, rtc_v2l4, rtc_v2wb
+            ))]
+            {
+                regs.isr().modify(|w| w.set_wutf(false));
+                while !regs.isr().read().wutwf() {}
+            }
+
+            #[cfg(any(rtc_v3, rtc_v3u5, rtc_v3l5))]
+            {
+                regs.scr().write(|w| w.set_cwutf(Calrf::CLEAR));
+                while !regs.icsr().read().wutwf() {}
+            }
+
+            regs.cr().modify(|w| w.set_wucksel(prescaler.into()));
+            regs.wutr().write(|w| w.set_wut(rtc_ticks));
+            regs.cr().modify(|w| w.set_wute(true));
+            regs.cr().modify(|w| w.set_wutie(true));
+        });
+
+        let instant = self.instant().unwrap();
+        trace!(
+            "rtc: start wakeup alarm for {} ms (psc: {}, ticks: {}) at {}",
+            Duration::from_ticks(rtc_ticks as u64 * TICK_HZ * prescaler as u64 / rtc_hz).as_millis(),
+            prescaler as u32,
+            rtc_ticks,
+            instant,
+        );
+
+        assert!(self.stop_time.borrow(cs).replace(Some(instant)).is_none())
+    }
+
+    /// stop the wakeup alarm and return the time elapsed since `start_wakeup_alarm`
+    /// was called, otherwise none
+    pub(crate) fn stop_wakeup_alarm(&self, cs: critical_section::CriticalSection) -> Option<embassy_time::Duration> {
+        use crate::interrupt::typelevel::Interrupt;
+        #[cfg(any(rtc_v3, rtc_v3u5, rtc_v3l5))]
+        use crate::pac::rtc::vals::Calrf;
+
+        let instant = self.instant().unwrap();
+        if RTC::regs().cr().read().wute() {
+            trace!("rtc: stop wakeup alarm at {}", instant);
+
+            self.write(false, |regs| {
+                regs.cr().modify(|w| w.set_wutie(false));
+                regs.cr().modify(|w| w.set_wute(false));
+
+                #[cfg(any(
+                    rtc_v2f0, rtc_v2f2, rtc_v2f3, rtc_v2f4, rtc_v2f7, rtc_v2h7, rtc_v2l0, rtc_v2l1, rtc_v2l4, rtc_v2wb
+                ))]
+                regs.isr().modify(|w| w.set_wutf(false));
+
+                #[cfg(any(rtc_v3, rtc_v3u5, rtc_v3l5))]
+                regs.scr().write(|w| w.set_cwutf(Calrf::CLEAR));
+
+                // Check RM for EXTI and/or NVIC section, "Event event input mapping" or "EXTI interrupt/event mapping" or something similar,
+                // there is a table for every "Event input" / "EXTI Line".
+                // If you find the EXTI line related to "RTC wakeup" marks as "Configurable" (not "Direct"),
+                // then write 1 to related field of Pending Register, to clean it's pending state.
+                #[cfg(any(exti_v1, stm32h7, stm32wb))]
+                crate::pac::EXTI
+                    .pr(0)
+                    .modify(|w| w.set_line(RTC::EXTI_WAKEUP_LINE, true));
+
+                <RTC as crate::rtc::SealedInstance>::WakeupInterrupt::unpend();
+            });
+        }
+
+        self.stop_time.borrow(cs).take().map(|stop_time| instant - stop_time)
+    }
+
+    pub(crate) fn enable_wakeup_line(&self) {
+        use crate::interrupt::typelevel::Interrupt;
+        use crate::pac::EXTI;
+
+        <RTC as crate::rtc::SealedInstance>::WakeupInterrupt::unpend();
+        unsafe { <RTC as crate::rtc::SealedInstance>::WakeupInterrupt::enable() };
+
+        EXTI.rtsr(0).modify(|w| w.set_line(RTC::EXTI_WAKEUP_LINE, true));
+        EXTI.imr(0).modify(|w| w.set_line(RTC::EXTI_WAKEUP_LINE, true));
+    }
+}
diff --git a/embassy-stm32/src/rtc/mod.rs b/embassy-stm32/src/rtc/mod.rs
index b12a0db66..a7f70b153 100644
--- a/embassy-stm32/src/rtc/mod.rs
+++ b/embassy-stm32/src/rtc/mod.rs
@@ -2,6 +2,9 @@
 mod datetime;
 
 #[cfg(feature = "low-power")]
+mod low_power;
+
+#[cfg(feature = "low-power")]
 use core::cell::Cell;
 
 #[cfg(feature = "low-power")]
@@ -9,8 +12,6 @@ use embassy_sync::blocking_mutex::raw::CriticalSectionRawMutex;
 #[cfg(feature = "low-power")]
 use embassy_sync::blocking_mutex::Mutex;
 
-#[cfg(not(rtc_v2f2))]
-use self::datetime::RtcInstant;
 use self::datetime::{day_of_week_from_u8, day_of_week_to_u8};
 pub use self::datetime::{DateTime, DayOfWeek, Error as DateTimeError};
 use crate::pac::rtc::regs::{Dr, Tr};
@@ -32,60 +33,6 @@ use embassy_hal_internal::Peripheral;
 
 use crate::peripherals::RTC;
 
-#[allow(dead_code)]
-#[repr(u8)]
-#[derive(Clone, Copy, Debug)]
-pub(crate) enum WakeupPrescaler {
-    Div2 = 2,
-    Div4 = 4,
-    Div8 = 8,
-    Div16 = 16,
-}
-
-#[cfg(any(stm32f4, stm32l0, stm32g4, stm32l5, stm32wb, stm32h5, stm32g0))]
-impl From<WakeupPrescaler> for crate::pac::rtc::vals::Wucksel {
-    fn from(val: WakeupPrescaler) -> Self {
-        use crate::pac::rtc::vals::Wucksel;
-
-        match val {
-            WakeupPrescaler::Div2 => Wucksel::DIV2,
-            WakeupPrescaler::Div4 => Wucksel::DIV4,
-            WakeupPrescaler::Div8 => Wucksel::DIV8,
-            WakeupPrescaler::Div16 => Wucksel::DIV16,
-        }
-    }
-}
-
-#[cfg(any(stm32f4, stm32l0, stm32g4, stm32l5, stm32wb, stm32h5, stm32g0))]
-impl From<crate::pac::rtc::vals::Wucksel> for WakeupPrescaler {
-    fn from(val: crate::pac::rtc::vals::Wucksel) -> Self {
-        use crate::pac::rtc::vals::Wucksel;
-
-        match val {
-            Wucksel::DIV2 => WakeupPrescaler::Div2,
-            Wucksel::DIV4 => WakeupPrescaler::Div4,
-            Wucksel::DIV8 => WakeupPrescaler::Div8,
-            Wucksel::DIV16 => WakeupPrescaler::Div16,
-            _ => unreachable!(),
-        }
-    }
-}
-
-#[cfg(feature = "low-power")]
-impl WakeupPrescaler {
-    pub fn compute_min(val: u32) -> Self {
-        *[
-            WakeupPrescaler::Div2,
-            WakeupPrescaler::Div4,
-            WakeupPrescaler::Div8,
-            WakeupPrescaler::Div16,
-        ]
-        .iter()
-        .find(|psc| **psc as u32 > val)
-        .unwrap_or(&WakeupPrescaler::Div16)
-    }
-}
-
 /// Errors that can occur on methods on [RtcClock]
 #[non_exhaustive]
 #[derive(Clone, Debug, PartialEq, Eq)]
@@ -106,15 +53,6 @@ pub struct RtcTimeProvider {
 }
 
 impl RtcTimeProvider {
-    #[cfg(not(rtc_v2f2))]
-    pub(crate) fn instant(&self) -> Result<RtcInstant, RtcError> {
-        self.read(|_, tr, ss| {
-            let second = bcd2_to_byte((tr.st(), tr.su()));
-
-            RtcInstant::from(second, ss).map_err(RtcError::InvalidDateTime)
-        })
-    }
-
     /// Return the current datetime.
     ///
     /// # Errors
@@ -165,8 +103,7 @@ impl RtcTimeProvider {
 /// RTC driver.
 pub struct Rtc {
     #[cfg(feature = "low-power")]
-    stop_time: Mutex<CriticalSectionRawMutex, Cell<Option<RtcInstant>>>,
-    #[cfg(not(feature = "low-power"))]
+    stop_time: Mutex<CriticalSectionRawMutex, Cell<Option<low_power::RtcInstant>>>,
     _private: (),
 }
 
@@ -205,12 +142,11 @@ impl Rtc {
     /// Create a new RTC instance.
     pub fn new(_rtc: impl Peripheral<P = RTC>, rtc_config: RtcConfig) -> Self {
         #[cfg(not(any(stm32l0, stm32f3, stm32l1, stm32f0, stm32f2)))]
-        <RTC as crate::rcc::SealedRccPeripheral>::enable_and_reset();
+        crate::rcc::enable_and_reset::<RTC>();
 
         let mut this = Self {
             #[cfg(feature = "low-power")]
             stop_time: Mutex::const_new(CriticalSectionRawMutex::new(), Cell::new(None)),
-            #[cfg(not(feature = "low-power"))]
             _private: (),
         };
 
@@ -223,9 +159,8 @@ impl Rtc {
         // Wait for the clock to update after initialization
         #[cfg(not(rtc_v2f2))]
         {
-            let now = this.instant().unwrap();
-
-            while this.instant().unwrap().subsecond == now.subsecond {}
+            let now = this.time_provider().read(|_, _, ss| Ok(ss)).unwrap();
+            while now == this.time_provider().read(|_, _, ss| Ok(ss)).unwrap() {}
         }
 
         this
@@ -284,12 +219,6 @@ impl Rtc {
         Ok(())
     }
 
-    #[cfg(not(rtc_v2f2))]
-    /// Return the current instant.
-    fn instant(&self) -> Result<RtcInstant, RtcError> {
-        self.time_provider().instant()
-    }
-
     /// Return the current datetime.
     ///
     /// # Errors
@@ -320,7 +249,7 @@ impl Rtc {
     /// The registers retain their values during wakes from standby mode or system resets. They also
     /// retain their value when Vdd is switched off as long as V_BAT is powered.
     pub fn read_backup_register(&self, register: usize) -> Option<u32> {
-        RTC::read_backup_register(&RTC::regs(), register)
+        RTC::read_backup_register(RTC::regs(), register)
     }
 
     /// Set content of the backup register.
@@ -328,120 +257,7 @@ impl Rtc {
     /// The registers retain their values during wakes from standby mode or system resets. They also
     /// retain their value when Vdd is switched off as long as V_BAT is powered.
     pub fn write_backup_register(&self, register: usize, value: u32) {
-        RTC::write_backup_register(&RTC::regs(), register, value)
-    }
-
-    #[cfg(feature = "low-power")]
-    /// start the wakeup alarm and with a duration that is as close to but less than
-    /// the requested duration, and record the instant the wakeup alarm was started
-    pub(crate) fn start_wakeup_alarm(
-        &self,
-        requested_duration: embassy_time::Duration,
-        cs: critical_section::CriticalSection,
-    ) {
-        use embassy_time::{Duration, TICK_HZ};
-
-        #[cfg(any(rtc_v3, rtc_v3u5, rtc_v3l5))]
-        use crate::pac::rtc::vals::Calrf;
-
-        // Panic if the rcc mod knows we're not using low-power rtc
-        #[cfg(any(rcc_wb, rcc_f4, rcc_f410))]
-        unsafe { crate::rcc::get_freqs() }.rtc.unwrap();
-
-        let requested_duration = requested_duration.as_ticks().clamp(0, u32::MAX as u64);
-        let rtc_hz = Self::frequency().0 as u64;
-        let rtc_ticks = requested_duration * rtc_hz / TICK_HZ;
-        let prescaler = WakeupPrescaler::compute_min((rtc_ticks / u16::MAX as u64) as u32);
-
-        // adjust the rtc ticks to the prescaler and subtract one rtc tick
-        let rtc_ticks = rtc_ticks / prescaler as u64;
-        let rtc_ticks = rtc_ticks.clamp(0, (u16::MAX - 1) as u64).saturating_sub(1) as u16;
-
-        self.write(false, |regs| {
-            regs.cr().modify(|w| w.set_wute(false));
-
-            #[cfg(any(
-                rtc_v2f0, rtc_v2f2, rtc_v2f3, rtc_v2f4, rtc_v2f7, rtc_v2h7, rtc_v2l0, rtc_v2l1, rtc_v2l4, rtc_v2wb
-            ))]
-            {
-                regs.isr().modify(|w| w.set_wutf(false));
-                while !regs.isr().read().wutwf() {}
-            }
-
-            #[cfg(any(rtc_v3, rtc_v3u5, rtc_v3l5))]
-            {
-                regs.scr().write(|w| w.set_cwutf(Calrf::CLEAR));
-                while !regs.icsr().read().wutwf() {}
-            }
-
-            regs.cr().modify(|w| w.set_wucksel(prescaler.into()));
-            regs.wutr().write(|w| w.set_wut(rtc_ticks));
-            regs.cr().modify(|w| w.set_wute(true));
-            regs.cr().modify(|w| w.set_wutie(true));
-        });
-
-        let instant = self.instant().unwrap();
-        trace!(
-            "rtc: start wakeup alarm for {} ms (psc: {}, ticks: {}) at {}",
-            Duration::from_ticks(rtc_ticks as u64 * TICK_HZ * prescaler as u64 / rtc_hz).as_millis(),
-            prescaler as u32,
-            rtc_ticks,
-            instant,
-        );
-
-        assert!(self.stop_time.borrow(cs).replace(Some(instant)).is_none())
-    }
-
-    #[cfg(feature = "low-power")]
-    /// stop the wakeup alarm and return the time elapsed since `start_wakeup_alarm`
-    /// was called, otherwise none
-    pub(crate) fn stop_wakeup_alarm(&self, cs: critical_section::CriticalSection) -> Option<embassy_time::Duration> {
-        use crate::interrupt::typelevel::Interrupt;
-        #[cfg(any(rtc_v3, rtc_v3u5, rtc_v3l5))]
-        use crate::pac::rtc::vals::Calrf;
-
-        let instant = self.instant().unwrap();
-        if RTC::regs().cr().read().wute() {
-            trace!("rtc: stop wakeup alarm at {}", instant);
-
-            self.write(false, |regs| {
-                regs.cr().modify(|w| w.set_wutie(false));
-                regs.cr().modify(|w| w.set_wute(false));
-
-                #[cfg(any(
-                    rtc_v2f0, rtc_v2f2, rtc_v2f3, rtc_v2f4, rtc_v2f7, rtc_v2h7, rtc_v2l0, rtc_v2l1, rtc_v2l4, rtc_v2wb
-                ))]
-                regs.isr().modify(|w| w.set_wutf(false));
-
-                #[cfg(any(rtc_v3, rtc_v3u5, rtc_v3l5))]
-                regs.scr().write(|w| w.set_cwutf(Calrf::CLEAR));
-
-                // Check RM for EXTI and/or NVIC section, "Event event input mapping" or "EXTI interrupt/event mapping" or something similar,
-                // there is a table for every "Event input" / "EXTI Line".
-                // If you find the EXTI line related to "RTC wakeup" marks as "Configurable" (not "Direct"),
-                // then write 1 to related field of Pending Register, to clean it's pending state.
-                #[cfg(any(exti_v1, stm32h7, stm32wb))]
-                crate::pac::EXTI
-                    .pr(0)
-                    .modify(|w| w.set_line(RTC::EXTI_WAKEUP_LINE, true));
-
-                <RTC as crate::rtc::SealedInstance>::WakeupInterrupt::unpend();
-            });
-        }
-
-        self.stop_time.borrow(cs).take().map(|stop_time| instant - stop_time)
-    }
-
-    #[cfg(feature = "low-power")]
-    pub(crate) fn enable_wakeup_line(&self) {
-        use crate::interrupt::typelevel::Interrupt;
-        use crate::pac::EXTI;
-
-        <RTC as crate::rtc::SealedInstance>::WakeupInterrupt::unpend();
-        unsafe { <RTC as crate::rtc::SealedInstance>::WakeupInterrupt::enable() };
-
-        EXTI.rtsr(0).modify(|w| w.set_line(RTC::EXTI_WAKEUP_LINE, true));
-        EXTI.imr(0).modify(|w| w.set_line(RTC::EXTI_WAKEUP_LINE, true));
+        RTC::write_backup_register(RTC::regs(), register, value)
     }
 }
 
@@ -482,13 +298,13 @@ trait SealedInstance {
     ///
     /// The registers retain their values during wakes from standby mode or system resets. They also
     /// retain their value when Vdd is switched off as long as V_BAT is powered.
-    fn read_backup_register(rtc: &crate::pac::rtc::Rtc, register: usize) -> Option<u32>;
+    fn read_backup_register(rtc: crate::pac::rtc::Rtc, register: usize) -> Option<u32>;
 
     /// Set content of the backup register.
     ///
     /// The registers retain their values during wakes from standby mode or system resets. They also
     /// retain their value when Vdd is switched off as long as V_BAT is powered.
-    fn write_backup_register(rtc: &crate::pac::rtc::Rtc, register: usize, value: u32);
+    fn write_backup_register(rtc: crate::pac::rtc::Rtc, register: usize, value: u32);
 
     // fn apply_config(&mut self, rtc_config: RtcConfig);
 }
diff --git a/embassy-stm32/src/rtc/v2.rs b/embassy-stm32/src/rtc/v2.rs
index 92f9de846..cdc1cb299 100644
--- a/embassy-stm32/src/rtc/v2.rs
+++ b/embassy-stm32/src/rtc/v2.rs
@@ -95,7 +95,7 @@ impl super::Rtc {
 
     pub(super) fn write<F, R>(&self, init_mode: bool, f: F) -> R
     where
-        F: FnOnce(&crate::pac::rtc::Rtc) -> R,
+        F: FnOnce(crate::pac::rtc::Rtc) -> R,
     {
         let r = RTC::regs();
         // Disable write protection.
@@ -112,7 +112,7 @@ impl super::Rtc {
             while !r.isr().read().initf() {}
         }
 
-        let result = f(&r);
+        let result = f(r);
 
         if init_mode {
             r.isr().modify(|w| w.set_init(false)); // Exits init mode
@@ -140,7 +140,7 @@ impl SealedInstance for crate::peripherals::RTC {
     #[cfg(all(feature = "low-power", stm32l0))]
     type WakeupInterrupt = crate::interrupt::typelevel::RTC;
 
-    fn read_backup_register(rtc: &Rtc, register: usize) -> Option<u32> {
+    fn read_backup_register(rtc: Rtc, register: usize) -> Option<u32> {
         if register < Self::BACKUP_REGISTER_COUNT {
             Some(rtc.bkpr(register).read().bkp())
         } else {
@@ -148,7 +148,7 @@ impl SealedInstance for crate::peripherals::RTC {
         }
     }
 
-    fn write_backup_register(rtc: &Rtc, register: usize, value: u32) {
+    fn write_backup_register(rtc: Rtc, register: usize, value: u32) {
         if register < Self::BACKUP_REGISTER_COUNT {
             rtc.bkpr(register).write(|w| w.set_bkp(value));
         }
diff --git a/embassy-stm32/src/rtc/v3.rs b/embassy-stm32/src/rtc/v3.rs
index e51e09e7c..02fd5272e 100644
--- a/embassy-stm32/src/rtc/v3.rs
+++ b/embassy-stm32/src/rtc/v3.rs
@@ -97,7 +97,7 @@ impl super::Rtc {
 
     pub(super) fn write<F, R>(&self, init_mode: bool, f: F) -> R
     where
-        F: FnOnce(&crate::pac::rtc::Rtc) -> R,
+        F: FnOnce(crate::pac::rtc::Rtc) -> R,
     {
         let r = RTC::regs();
         // Disable write protection.
@@ -112,7 +112,7 @@ impl super::Rtc {
             while !r.icsr().read().initf() {}
         }
 
-        let result = f(&r);
+        let result = f(r);
 
         if init_mode {
             r.icsr().modify(|w| w.set_init(false)); // Exits init mode
@@ -143,7 +143,7 @@ impl SealedInstance for crate::peripherals::RTC {
         }
     );
 
-    fn read_backup_register(_rtc: &Rtc, register: usize) -> Option<u32> {
+    fn read_backup_register(_rtc: Rtc, register: usize) -> Option<u32> {
         #[allow(clippy::if_same_then_else)]
         if register < Self::BACKUP_REGISTER_COUNT {
             //Some(rtc.bkpr()[register].read().bits())
@@ -153,7 +153,7 @@ impl SealedInstance for crate::peripherals::RTC {
         }
     }
 
-    fn write_backup_register(_rtc: &Rtc, register: usize, _value: u32) {
+    fn write_backup_register(_rtc: Rtc, register: usize, _value: u32) {
         if register < Self::BACKUP_REGISTER_COUNT {
             // RTC3 backup registers come from the TAMP peripheral, not RTC. Not() even in the L412 PAC
             //self.rtc.bkpr()[register].write(|w| w.bits(value))
diff --git a/embassy-stm32/src/sai/mod.rs b/embassy-stm32/src/sai/mod.rs
index 54dd81524..3faecdc33 100644
--- a/embassy-stm32/src/sai/mod.rs
+++ b/embassy-stm32/src/sai/mod.rs
@@ -11,7 +11,7 @@ pub use crate::dma::word;
 use crate::dma::{ringbuffer, Channel, ReadableRingBuffer, Request, TransferOptions, WritableRingBuffer};
 use crate::gpio::{AFType, AnyPin, SealedPin as _};
 use crate::pac::sai::{vals, Sai as Regs};
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::{peripherals, Peripheral};
 
 /// SAI error
@@ -722,7 +722,7 @@ pub struct SubBlock<'d, T, S: SubBlockInstance> {
 /// You can then create a [`Sai`] driver for each each half.
 pub fn split_subblocks<'d, T: Instance>(peri: impl Peripheral<P = T> + 'd) -> (SubBlock<'d, T, A>, SubBlock<'d, T, B>) {
     into_ref!(peri);
-    T::enable_and_reset();
+    rcc::enable_and_reset::<T>();
 
     (
         SubBlock {
@@ -978,7 +978,7 @@ impl<'d, T: Instance, W: word::Word> Sai<'d, T, W> {
 
     /// Reset SAI operation.
     pub fn reset() {
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
     }
 
     /// Flush.
diff --git a/embassy-stm32/src/sdmmc/mod.rs b/embassy-stm32/src/sdmmc/mod.rs
index f79a11606..9c14837e1 100644
--- a/embassy-stm32/src/sdmmc/mod.rs
+++ b/embassy-stm32/src/sdmmc/mod.rs
@@ -16,7 +16,7 @@ use crate::dma::NoDma;
 use crate::gpio::{AFType, AnyPin, Pull, SealedPin, Speed};
 use crate::interrupt::typelevel::Interrupt;
 use crate::pac::sdmmc::Sdmmc as RegBlock;
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::time::Hertz;
 use crate::{interrupt, peripherals, Peripheral};
 
@@ -468,7 +468,7 @@ impl<'d, T: Instance, Dma: SdmmcDma<T> + 'd> Sdmmc<'d, T, Dma> {
     ) -> Self {
         into_ref!(sdmmc, dma);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         T::Interrupt::unpend();
         unsafe { T::Interrupt::enable() };
diff --git a/embassy-stm32/src/spi/mod.rs b/embassy-stm32/src/spi/mod.rs
index 5fc8691ac..4eaf7777f 100644
--- a/embassy-stm32/src/spi/mod.rs
+++ b/embassy-stm32/src/spi/mod.rs
@@ -13,7 +13,7 @@ use crate::dma::{slice_ptr_parts, word, ChannelAndRequest};
 use crate::gpio::{AFType, AnyPin, Pull, SealedPin as _, Speed};
 use crate::mode::{Async, Blocking, Mode as PeriMode};
 use crate::pac::spi::{regs, vals, Spi as Regs};
-use crate::rcc::{ClockEnableBit, SealedRccPeripheral};
+use crate::rcc::{RccInfo, SealedRccPeripheral};
 use crate::time::Hertz;
 use crate::Peripheral;
 
@@ -120,17 +120,30 @@ impl<'d, M: PeriMode> Spi<'d, M> {
         rx_dma: Option<ChannelAndRequest<'d>>,
         config: Config,
     ) -> Self {
-        let regs = T::info().regs;
-        let kernel_clock = T::frequency();
-        let br = compute_baud_rate(kernel_clock, config.frequency);
+        let mut this = Self {
+            info: T::info(),
+            kernel_clock: T::frequency(),
+            sck,
+            mosi,
+            miso,
+            tx_dma,
+            rx_dma,
+            current_word_size: <u8 as SealedWord>::CONFIG,
+            _phantom: PhantomData,
+        };
+        this.enable_and_init(config);
+        this
+    }
 
+    fn enable_and_init(&mut self, config: Config) {
+        let br = compute_baud_rate(self.kernel_clock, config.frequency);
         let cpha = config.raw_phase();
         let cpol = config.raw_polarity();
-
         let lsbfirst = config.raw_byte_order();
 
-        T::enable_and_reset();
+        self.info.rcc.enable_and_reset();
 
+        let regs = self.info.regs;
         #[cfg(any(spi_v1, spi_f1))]
         {
             regs.cr2().modify(|w| {
@@ -148,9 +161,10 @@ impl<'d, M: PeriMode> Spi<'d, M> {
                 w.set_ssm(true);
                 w.set_crcen(false);
                 w.set_bidimode(vals::Bidimode::UNIDIRECTIONAL);
-                if mosi.is_none() {
-                    w.set_rxonly(vals::Rxonly::OUTPUTDISABLED);
-                }
+                // we're doing "fake rxonly", by actually writing one
+                // byte to TXDR for each byte we want to receive. if we
+                // set OUTPUTDISABLED here, this hangs.
+                w.set_rxonly(vals::Rxonly::FULLDUPLEX);
                 w.set_dff(<u8 as SealedWord>::CONFIG)
             });
         }
@@ -208,18 +222,6 @@ impl<'d, M: PeriMode> Spi<'d, M> {
                 w.set_spe(true);
             });
         }
-
-        Self {
-            info: T::info(),
-            kernel_clock,
-            sck,
-            mosi,
-            miso,
-            tx_dma,
-            rx_dma,
-            current_word_size: <u8 as SealedWord>::CONFIG,
-            _phantom: PhantomData,
-        }
     }
 
     /// Reconfigures it with the supplied config.
@@ -350,17 +352,46 @@ impl<'d, M: PeriMode> Spi<'d, M> {
 
     /// Blocking write.
     pub fn blocking_write<W: Word>(&mut self, words: &[W]) -> Result<(), Error> {
+        // needed in v3+ to avoid overrun causing the SPI RX state machine to get stuck...?
+        #[cfg(any(spi_v3, spi_v4, spi_v5))]
+        self.info.regs.cr1().modify(|w| w.set_spe(false));
         self.info.regs.cr1().modify(|w| w.set_spe(true));
         flush_rx_fifo(self.info.regs);
         self.set_word_size(W::CONFIG);
         for word in words.iter() {
-            let _ = transfer_word(self.info.regs, *word)?;
+            // this cannot use `transfer_word` because on SPIv2 and higher,
+            // the SPI RX state machine hangs if no physical pin is connected to the SCK AF.
+            // This is the case when the SPI has been created with `new_(blocking_?)txonly_nosck`.
+            // See https://github.com/embassy-rs/embassy/issues/2902
+            // This is not documented as an errata by ST, and I've been unable to find anything online...
+            #[cfg(not(any(spi_v1, spi_f1)))]
+            write_word(self.info.regs, *word)?;
+
+            // if we're doing tx only, after writing the last byte to FIFO we have to wait
+            // until it's actually sent. On SPIv1 you're supposed to use the BSY flag for this
+            // but apparently it's broken, it clears too soon. Workaround is to wait for RXNE:
+            // when it gets set you know the transfer is done, even if you don't care about rx.
+            // Luckily this doesn't affect SPIv2+.
+            // See http://efton.sk/STM32/gotcha/g68.html
+            // ST doesn't seem to document this in errata sheets (?)
+            #[cfg(any(spi_v1, spi_f1))]
+            transfer_word(self.info.regs, *word)?;
         }
+
+        // wait until last word is transmitted. (except on v1, see above)
+        #[cfg(not(any(spi_v1, spi_f1, spi_v2)))]
+        while !self.info.regs.sr().read().txc() {}
+        #[cfg(spi_v2)]
+        while self.info.regs.sr().read().bsy() {}
+
         Ok(())
     }
 
     /// Blocking read.
     pub fn blocking_read<W: Word>(&mut self, words: &mut [W]) -> Result<(), Error> {
+        // needed in v3+ to avoid overrun causing the SPI RX state machine to get stuck...?
+        #[cfg(any(spi_v3, spi_v4, spi_v5))]
+        self.info.regs.cr1().modify(|w| w.set_spe(false));
         self.info.regs.cr1().modify(|w| w.set_spe(true));
         flush_rx_fifo(self.info.regs);
         self.set_word_size(W::CONFIG);
@@ -374,6 +405,9 @@ impl<'d, M: PeriMode> Spi<'d, M> {
     ///
     /// This writes the contents of `data` on MOSI, and puts the received data on MISO in `data`, at the same time.
     pub fn blocking_transfer_in_place<W: Word>(&mut self, words: &mut [W]) -> Result<(), Error> {
+        // needed in v3+ to avoid overrun causing the SPI RX state machine to get stuck...?
+        #[cfg(any(spi_v3, spi_v4, spi_v5))]
+        self.info.regs.cr1().modify(|w| w.set_spe(false));
         self.info.regs.cr1().modify(|w| w.set_spe(true));
         flush_rx_fifo(self.info.regs);
         self.set_word_size(W::CONFIG);
@@ -390,6 +424,9 @@ impl<'d, M: PeriMode> Spi<'d, M> {
     /// The transfer runs for `max(read.len(), write.len())` bytes. If `read` is shorter extra bytes are ignored.
     /// If `write` is shorter it is padded with zero bytes.
     pub fn blocking_transfer<W: Word>(&mut self, read: &mut [W], write: &[W]) -> Result<(), Error> {
+        // needed in v3+ to avoid overrun causing the SPI RX state machine to get stuck...?
+        #[cfg(any(spi_v3, spi_v4, spi_v5))]
+        self.info.regs.cr1().modify(|w| w.set_spe(false));
         self.info.regs.cr1().modify(|w| w.set_spe(true));
         flush_rx_fifo(self.info.regs);
         self.set_word_size(W::CONFIG);
@@ -508,6 +545,7 @@ impl<'d> Spi<'d, Async> {
         peri: impl Peripheral<P = T> + 'd,
         sck: impl Peripheral<P = impl SckPin<T>> + 'd,
         miso: impl Peripheral<P = impl MisoPin<T>> + 'd,
+        #[cfg(any(spi_v1, spi_f1, spi_v2))] tx_dma: impl Peripheral<P = impl TxDma<T>> + 'd,
         rx_dma: impl Peripheral<P = impl RxDma<T>> + 'd,
         config: Config,
     ) -> Self {
@@ -516,6 +554,9 @@ impl<'d> Spi<'d, Async> {
             new_pin!(sck, AFType::OutputPushPull, Speed::VeryHigh, config.sck_pull_mode()),
             None,
             new_pin!(miso, AFType::Input, Speed::VeryHigh),
+            #[cfg(any(spi_v1, spi_f1, spi_v2))]
+            new_dma!(tx_dma),
+            #[cfg(any(spi_v3, spi_v4, spi_v5))]
             None,
             new_dma!(rx_dma),
             config,
@@ -571,7 +612,7 @@ impl<'d> Spi<'d, Async> {
         // see RM0453 rev 1 section 7.2.13 page 291
         // The SUBGHZSPI_SCK frequency is obtained by PCLK3 divided by two.
         // The SUBGHZSPI_SCK clock maximum speed must not exceed 16 MHz.
-        let pclk3_freq = <crate::peripherals::SUBGHZSPI as crate::rcc::SealedRccPeripheral>::frequency().0;
+        let pclk3_freq = <crate::peripherals::SUBGHZSPI as SealedRccPeripheral>::frequency().0;
         let freq = Hertz(core::cmp::min(pclk3_freq / 2, 16_000_000));
         let mut config = Config::default();
         config.mode = MODE_0;
@@ -584,11 +625,11 @@ impl<'d> Spi<'d, Async> {
     #[allow(dead_code)]
     pub(crate) fn new_internal<T: Instance>(
         peri: impl Peripheral<P = T> + 'd,
-        tx_dma: impl Peripheral<P = impl TxDma<T>> + 'd,
-        rx_dma: impl Peripheral<P = impl RxDma<T>> + 'd,
+        tx_dma: Option<ChannelAndRequest<'d>>,
+        rx_dma: Option<ChannelAndRequest<'d>>,
         config: Config,
     ) -> Self {
-        Self::new_inner(peri, None, None, None, new_dma!(tx_dma), new_dma!(rx_dma), config)
+        Self::new_inner(peri, None, None, None, tx_dma, rx_dma, config)
     }
 
     /// SPI write, using DMA.
@@ -622,12 +663,100 @@ impl<'d> Spi<'d, Async> {
     }
 
     /// SPI read, using DMA.
+    #[cfg(any(spi_v3, spi_v4, spi_v5))]
+    pub async fn read<W: Word>(&mut self, data: &mut [W]) -> Result<(), Error> {
+        if data.is_empty() {
+            return Ok(());
+        }
+
+        let regs = self.info.regs;
+
+        regs.cr1().modify(|w| {
+            w.set_spe(false);
+        });
+
+        let comm = regs.cfg2().modify(|w| {
+            let prev = w.comm();
+            w.set_comm(vals::Comm::RECEIVER);
+            prev
+        });
+
+        #[cfg(spi_v3)]
+        let i2scfg = regs.i2scfgr().modify(|w| {
+            w.i2smod().then(|| {
+                let prev = w.i2scfg();
+                w.set_i2scfg(match prev {
+                    vals::I2scfg::SLAVERX | vals::I2scfg::SLAVEFULLDUPLEX => vals::I2scfg::SLAVERX,
+                    vals::I2scfg::MASTERRX | vals::I2scfg::MASTERFULLDUPLEX => vals::I2scfg::MASTERRX,
+                    _ => panic!("unsupported configuration"),
+                });
+                prev
+            })
+        });
+
+        let rx_src = regs.rx_ptr();
+
+        for mut chunk in data.chunks_mut(u16::max_value().into()) {
+            self.set_word_size(W::CONFIG);
+            set_rxdmaen(regs, true);
+
+            let tsize = chunk.len();
+
+            let transfer = unsafe {
+                self.rx_dma
+                    .as_mut()
+                    .unwrap()
+                    .read(rx_src, &mut chunk, Default::default())
+            };
+
+            regs.cr2().modify(|w| {
+                w.set_tsize(tsize as u16);
+            });
+
+            regs.cr1().modify(|w| {
+                w.set_spe(true);
+            });
+
+            regs.cr1().modify(|w| {
+                w.set_cstart(true);
+            });
+
+            transfer.await;
+
+            finish_dma(regs);
+        }
+
+        regs.cr1().modify(|w| {
+            w.set_spe(false);
+        });
+
+        regs.cfg2().modify(|w| {
+            w.set_comm(comm);
+        });
+
+        regs.cr2().modify(|w| {
+            w.set_tsize(0);
+        });
+
+        #[cfg(spi_v3)]
+        if let Some(i2scfg) = i2scfg {
+            regs.i2scfgr().modify(|w| {
+                w.set_i2scfg(i2scfg);
+            });
+        }
+
+        Ok(())
+    }
+
+    /// SPI read, using DMA.
+    #[cfg(any(spi_v1, spi_f1, spi_v2))]
     pub async fn read<W: Word>(&mut self, data: &mut [W]) -> Result<(), Error> {
         if data.is_empty() {
             return Ok(());
         }
 
         self.set_word_size(W::CONFIG);
+
         self.info.regs.cr1().modify(|w| {
             w.set_spe(false);
         });
@@ -738,7 +867,7 @@ impl<'d, M: PeriMode> Drop for Spi<'d, M> {
         self.mosi.as_ref().map(|x| x.set_as_disconnected());
         self.miso.as_ref().map(|x| x.set_as_disconnected());
 
-        self.info.enable_bit.disable();
+        self.info.rcc.disable();
     }
 }
 
@@ -805,8 +934,8 @@ impl RegsExt for Regs {
     }
 }
 
-fn check_error_flags(sr: regs::Sr) -> Result<(), Error> {
-    if sr.ovr() {
+fn check_error_flags(sr: regs::Sr, ovr: bool) -> Result<(), Error> {
+    if sr.ovr() && ovr {
         return Err(Error::Overrun);
     }
     #[cfg(not(any(spi_f1, spi_v3, spi_v4, spi_v5)))]
@@ -832,11 +961,11 @@ fn check_error_flags(sr: regs::Sr) -> Result<(), Error> {
     Ok(())
 }
 
-fn spin_until_tx_ready(regs: Regs) -> Result<(), Error> {
+fn spin_until_tx_ready(regs: Regs, ovr: bool) -> Result<(), Error> {
     loop {
         let sr = regs.sr().read();
 
-        check_error_flags(sr)?;
+        check_error_flags(sr, ovr)?;
 
         #[cfg(not(any(spi_v3, spi_v4, spi_v5)))]
         if sr.txe() {
@@ -853,7 +982,7 @@ fn spin_until_rx_ready(regs: Regs) -> Result<(), Error> {
     loop {
         let sr = regs.sr().read();
 
-        check_error_flags(sr)?;
+        check_error_flags(sr, true)?;
 
         #[cfg(not(any(spi_v3, spi_v4, spi_v5)))]
         if sr.rxne() {
@@ -907,7 +1036,13 @@ fn finish_dma(regs: Regs) {
     while regs.sr().read().ftlvl().to_bits() > 0 {}
 
     #[cfg(any(spi_v3, spi_v4, spi_v5))]
-    while !regs.sr().read().txc() {}
+    {
+        if regs.cr2().read().tsize() == 0 {
+            while !regs.sr().read().txc() {}
+        } else {
+            while !regs.sr().read().eot() {}
+        }
+    }
     #[cfg(not(any(spi_v3, spi_v4, spi_v5)))]
     while regs.sr().read().bsy() {}
 
@@ -931,7 +1066,7 @@ fn finish_dma(regs: Regs) {
 }
 
 fn transfer_word<W: Word>(regs: Regs, tx_word: W) -> Result<W, Error> {
-    spin_until_tx_ready(regs)?;
+    spin_until_tx_ready(regs, true)?;
 
     unsafe {
         ptr::write_volatile(regs.tx_ptr(), tx_word);
@@ -946,6 +1081,21 @@ fn transfer_word<W: Word>(regs: Regs, tx_word: W) -> Result<W, Error> {
     Ok(rx_word)
 }
 
+#[allow(unused)] // unused in SPIv1
+fn write_word<W: Word>(regs: Regs, tx_word: W) -> Result<(), Error> {
+    // for write, we intentionally ignore the rx fifo, which will cause
+    // overrun errors that we have to ignore.
+    spin_until_tx_ready(regs, false)?;
+
+    unsafe {
+        ptr::write_volatile(regs.tx_ptr(), tx_word);
+
+        #[cfg(any(spi_v3, spi_v4, spi_v5))]
+        regs.cr1().modify(|reg| reg.set_cstart(true));
+    }
+    Ok(())
+}
+
 // Note: It is not possible to impl these traits generically in embedded-hal 0.2 due to a conflict with
 // some marker traits. For details, see https://github.com/rust-embedded/embedded-hal/pull/289
 macro_rules! impl_blocking {
@@ -1118,7 +1268,7 @@ mod word_impl {
 
 pub(crate) struct Info {
     pub(crate) regs: Regs,
-    pub(crate) enable_bit: ClockEnableBit,
+    pub(crate) rcc: RccInfo,
 }
 
 struct State {}
@@ -1145,7 +1295,7 @@ foreach_peripheral!(
     (spi, $inst:ident) => {
         peri_trait_impl!($inst, Info {
             regs: crate::pac::$inst,
-            enable_bit: crate::peripherals::$inst::ENABLE_BIT,
+            rcc: crate::peripherals::$inst::RCC_INFO,
         });
     };
 );
diff --git a/embassy-stm32/src/time_driver.rs b/embassy-stm32/src/time_driver.rs
index e592fbf7d..f8041bf1e 100644
--- a/embassy-stm32/src/time_driver.rs
+++ b/embassy-stm32/src/time_driver.rs
@@ -12,7 +12,7 @@ use stm32_metapac::timer::{regs, TimGp16};
 
 use crate::interrupt::typelevel::Interrupt;
 use crate::pac::timer::vals;
-use crate::rcc::SealedRccPeripheral;
+use crate::rcc::{self, SealedRccPeripheral};
 #[cfg(feature = "low-power")]
 use crate::rtc::Rtc;
 use crate::timer::{CoreInstance, GeneralInstance1Channel};
@@ -276,7 +276,7 @@ impl RtcDriver {
     fn init(&'static self, cs: critical_section::CriticalSection) {
         let r = regs_gp16();
 
-        <T as SealedRccPeripheral>::enable_and_reset_with_cs(cs);
+        rcc::enable_and_reset_with_cs::<T>(cs);
 
         let timer_freq = T::frequency();
 
diff --git a/embassy-stm32/src/timer/input_capture.rs b/embassy-stm32/src/timer/input_capture.rs
index a1c1486f9..0258d4077 100644
--- a/embassy-stm32/src/timer/input_capture.rs
+++ b/embassy-stm32/src/timer/input_capture.rs
@@ -7,7 +7,7 @@ use core::task::{Context, Poll};
 
 use embassy_hal_internal::{into_ref, PeripheralRef};
 
-use super::low_level::{CountingMode, InputCaptureMode, InputTISelection, Timer};
+use super::low_level::{CountingMode, FilterValue, InputCaptureMode, InputTISelection, Timer};
 use super::{
     CaptureCompareInterruptHandler, Channel, Channel1Pin, Channel2Pin, Channel3Pin, Channel4Pin,
     GeneralInstance4Channel,
@@ -40,11 +40,9 @@ macro_rules! channel_impl {
             #[doc = concat!("Create a new ", stringify!($channel), " capture pin instance.")]
             pub fn $new_chx(pin: impl Peripheral<P = impl $pin_trait<T>> + 'd, pull_type: Pull) -> Self {
                 into_ref!(pin);
-                critical_section::with(|_| {
-                    pin.set_as_af_pull(pin.af_num(), AFType::Input, pull_type);
-                    #[cfg(gpio_v2)]
-                    pin.set_speed(crate::gpio::Speed::VeryHigh);
-                });
+
+                pin.set_as_af_pull(pin.af_num(), AFType::Input, pull_type);
+
                 CapturePin {
                     _pin: pin.map_into(),
                     phantom: PhantomData,
@@ -83,7 +81,7 @@ impl<'d, T: GeneralInstance4Channel> InputCapture<'d, T> {
         let mut this = Self { inner: Timer::new(tim) };
 
         this.inner.set_counting_mode(counting_mode);
-        this.set_tick_freq(freq);
+        this.inner.set_tick_freq(freq);
         this.inner.enable_outputs(); // Required for advanced timers, see GeneralInstance4Channel for details
         this.inner.start();
 
@@ -109,24 +107,6 @@ impl<'d, T: GeneralInstance4Channel> InputCapture<'d, T> {
         self.inner.get_channel_enable_state(channel)
     }
 
-    /// Set tick frequency.
-    ///
-    /// Note: when you call this, the max period value changes
-    pub fn set_tick_freq(&mut self, freq: Hertz) {
-        let f = freq;
-        assert!(f.0 > 0);
-        let timer_f = self.inner.get_clock_frequency();
-
-        let pclk_ticks_per_timer_period = timer_f / f;
-        let psc: u16 = unwrap!((pclk_ticks_per_timer_period - 1).try_into());
-
-        let regs = self.inner.regs_core();
-        regs.psc().write_value(psc);
-
-        // Generate an Update Request
-        regs.egr().write(|r| r.set_ug(true));
-    }
-
     /// Set the input capture mode for a given channel.
     pub fn set_input_capture_mode(&mut self, channel: Channel, mode: InputCaptureMode) {
         self.inner.set_input_capture_mode(channel, mode);
@@ -148,10 +128,13 @@ impl<'d, T: GeneralInstance4Channel> InputCapture<'d, T> {
     }
 
     fn new_future(&self, channel: Channel, mode: InputCaptureMode, tisel: InputTISelection) -> InputCaptureFuture<T> {
-        self.inner.enable_channel(channel, true);
-        self.inner.set_input_capture_mode(channel, mode);
+        // Configuration steps from ST RM0390 (STM32F446) chapter 17.3.5
+        // or ST RM0008 (STM32F103) chapter 15.3.5 Input capture mode
         self.inner.set_input_ti_selection(channel, tisel);
-        self.inner.clear_input_interrupt(channel);
+        self.inner.set_input_capture_filter(channel, FilterValue::NOFILTER);
+        self.inner.set_input_capture_mode(channel, mode);
+        self.inner.set_input_capture_prescaler(channel, 0);
+        self.inner.enable_channel(channel, true);
         self.inner.enable_input_interrupt(channel, true);
 
         InputCaptureFuture {
diff --git a/embassy-stm32/src/timer/low_level.rs b/embassy-stm32/src/timer/low_level.rs
index 7f533b75c..e643722aa 100644
--- a/embassy-stm32/src/timer/low_level.rs
+++ b/embassy-stm32/src/timer/low_level.rs
@@ -7,9 +7,12 @@
 //! The available functionality depends on the timer type.
 
 use embassy_hal_internal::{into_ref, Peripheral, PeripheralRef};
+// Re-export useful enums
+pub use stm32_metapac::timer::vals::{FilterValue, Sms as SlaveMode, Ts as TriggerSource};
 
 use super::*;
 use crate::pac::timer::vals;
+use crate::rcc;
 use crate::time::Hertz;
 
 /// Input capture mode.
@@ -181,7 +184,7 @@ pub struct Timer<'d, T: CoreInstance> {
 
 impl<'d, T: CoreInstance> Drop for Timer<'d, T> {
     fn drop(&mut self) {
-        T::disable()
+        rcc::disable::<T>();
     }
 }
 
@@ -190,7 +193,7 @@ impl<'d, T: CoreInstance> Timer<'d, T> {
     pub fn new(tim: impl Peripheral<P = T> + 'd) -> Self {
         into_ref!(tim);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         Self { tim }
     }
@@ -273,6 +276,22 @@ impl<'d, T: CoreInstance> Timer<'d, T> {
         }
     }
 
+    /// Set tick frequency.
+    pub fn set_tick_freq(&mut self, freq: Hertz) {
+        let f = freq;
+        assert!(f.0 > 0);
+        let timer_f = self.get_clock_frequency();
+
+        let pclk_ticks_per_timer_period = timer_f / f;
+        let psc: u16 = unwrap!((pclk_ticks_per_timer_period - 1).try_into());
+
+        let regs = self.regs_core();
+        regs.psc().write_value(psc);
+
+        // Generate an Update Request
+        regs.egr().write(|r| r.set_ug(true));
+    }
+
     /// Clear update interrupt.
     ///
     /// Returns whether the update interrupt flag was set.
@@ -572,6 +591,16 @@ impl<'d, T: GeneralInstance4Channel> Timer<'d, T> {
     pub fn set_cc_dma_enable_state(&self, channel: Channel, ccde: bool) {
         self.regs_gp16().dier().modify(|w| w.set_ccde(channel.index(), ccde))
     }
+
+    /// Set Timer Slave Mode
+    pub fn set_slave_mode(&self, sms: SlaveMode) {
+        self.regs_gp16().smcr().modify(|r| r.set_sms(sms));
+    }
+
+    /// Set Timer Trigger Source
+    pub fn set_trigger_source(&self, ts: TriggerSource) {
+        self.regs_gp16().smcr().modify(|r| r.set_ts(ts));
+    }
 }
 
 #[cfg(not(stm32l0))]
diff --git a/embassy-stm32/src/timer/mod.rs b/embassy-stm32/src/timer/mod.rs
index 314b6006b..25782ee13 100644
--- a/embassy-stm32/src/timer/mod.rs
+++ b/embassy-stm32/src/timer/mod.rs
@@ -8,6 +8,7 @@ use embassy_sync::waitqueue::AtomicWaker;
 pub mod complementary_pwm;
 pub mod input_capture;
 pub mod low_level;
+pub mod pwm_input;
 pub mod qei;
 pub mod simple_pwm;
 
diff --git a/embassy-stm32/src/timer/pwm_input.rs b/embassy-stm32/src/timer/pwm_input.rs
new file mode 100644
index 000000000..dcf098a78
--- /dev/null
+++ b/embassy-stm32/src/timer/pwm_input.rs
@@ -0,0 +1,114 @@
+//! PWM Input driver.
+
+use embassy_hal_internal::into_ref;
+
+use super::low_level::{CountingMode, InputCaptureMode, InputTISelection, SlaveMode, Timer, TriggerSource};
+use super::{Channel, Channel1Pin, Channel2Pin, GeneralInstance4Channel};
+use crate::gpio::{AFType, Pull};
+use crate::time::Hertz;
+use crate::Peripheral;
+
+/// PWM Input driver.
+pub struct PwmInput<'d, T: GeneralInstance4Channel> {
+    channel: Channel,
+    inner: Timer<'d, T>,
+}
+
+impl<'d, T: GeneralInstance4Channel> PwmInput<'d, T> {
+    /// Create a new PWM input driver.
+    pub fn new(
+        tim: impl Peripheral<P = T> + 'd,
+        pin: impl Peripheral<P = impl Channel1Pin<T>> + 'd,
+        pull_type: Pull,
+        freq: Hertz,
+    ) -> Self {
+        into_ref!(pin);
+
+        pin.set_as_af_pull(pin.af_num(), AFType::Input, pull_type);
+
+        Self::new_inner(tim, freq, Channel::Ch1, Channel::Ch2)
+    }
+
+    /// Create a new PWM input driver.
+    pub fn new_alt(
+        tim: impl Peripheral<P = T> + 'd,
+        pin: impl Peripheral<P = impl Channel2Pin<T>> + 'd,
+        pull_type: Pull,
+        freq: Hertz,
+    ) -> Self {
+        into_ref!(pin);
+
+        pin.set_as_af_pull(pin.af_num(), AFType::Input, pull_type);
+
+        Self::new_inner(tim, freq, Channel::Ch2, Channel::Ch1)
+    }
+
+    fn new_inner(tim: impl Peripheral<P = T> + 'd, freq: Hertz, ch1: Channel, ch2: Channel) -> Self {
+        let mut inner = Timer::new(tim);
+
+        inner.set_counting_mode(CountingMode::EdgeAlignedUp);
+        inner.set_tick_freq(freq);
+        inner.enable_outputs(); // Required for advanced timers, see GeneralInstance4Channel for details
+        inner.start();
+
+        // Configuration steps from ST RM0390 (STM32F446) chapter 17.3.6
+        // or ST RM0008 (STM32F103) chapter 15.3.6 Input capture mode
+        inner.set_input_ti_selection(ch1, InputTISelection::Normal);
+        inner.set_input_capture_mode(ch1, InputCaptureMode::Rising);
+
+        inner.set_input_ti_selection(ch2, InputTISelection::Alternate);
+        inner.set_input_capture_mode(ch2, InputCaptureMode::Falling);
+
+        inner.set_trigger_source(match ch1 {
+            Channel::Ch1 => TriggerSource::TI1FP1,
+            Channel::Ch2 => TriggerSource::TI2FP2,
+            _ => panic!("Invalid channel for PWM input"),
+        });
+
+        inner.set_slave_mode(SlaveMode::RESET_MODE);
+
+        // Must call the `enable` function after
+
+        Self { channel: ch1, inner }
+    }
+
+    /// Enable the given channel.
+    pub fn enable(&mut self) {
+        self.inner.enable_channel(Channel::Ch1, true);
+        self.inner.enable_channel(Channel::Ch2, true);
+    }
+
+    /// Disable the given channel.
+    pub fn disable(&mut self) {
+        self.inner.enable_channel(Channel::Ch1, false);
+        self.inner.enable_channel(Channel::Ch2, false);
+    }
+
+    /// Check whether given channel is enabled
+    pub fn is_enabled(&self) -> bool {
+        self.inner.get_channel_enable_state(Channel::Ch1)
+    }
+
+    /// Get the period tick count
+    pub fn get_period_ticks(&self) -> u32 {
+        self.inner.get_capture_value(self.channel)
+    }
+
+    /// Get the pulse width tick count
+    pub fn get_width_ticks(&self) -> u32 {
+        self.inner.get_capture_value(match self.channel {
+            Channel::Ch1 => Channel::Ch2,
+            Channel::Ch2 => Channel::Ch1,
+            _ => panic!("Invalid channel for PWM input"),
+        })
+    }
+
+    /// Get the duty cycle in 100%
+    pub fn get_duty_cycle(&self) -> f32 {
+        let period = self.get_period_ticks();
+        if period == 0 {
+            return 0.;
+        }
+        100. * (self.get_width_ticks() as f32) / (period as f32)
+    }
+}
diff --git a/embassy-stm32/src/tsc/mod.rs b/embassy-stm32/src/tsc/mod.rs
index bf583f04c..045d6317c 100644
--- a/embassy-stm32/src/tsc/mod.rs
+++ b/embassy-stm32/src/tsc/mod.rs
@@ -72,7 +72,7 @@ pub use enums::*;
 
 use crate::gpio::{AFType, AnyPin};
 use crate::pac::tsc::Tsc as Regs;
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 use crate::{peripherals, Peripheral};
 
 #[cfg(tsc_v1)]
@@ -649,7 +649,7 @@ impl<'d, T: Instance> Tsc<'d, T> {
     ) -> Self {
         into_ref!(peri);
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
 
         T::REGS.cr().modify(|w| {
             w.set_tsce(true);
@@ -880,7 +880,7 @@ impl<'d, T: Instance> Tsc<'d, T> {
 
 impl<'d, T: Instance> Drop for Tsc<'d, T> {
     fn drop(&mut self) {
-        T::disable();
+        rcc::disable::<T>();
     }
 }
 
diff --git a/embassy-stm32/src/ucpd.rs b/embassy-stm32/src/ucpd.rs
index d6d0682b9..89e2f5d49 100644
--- a/embassy-stm32/src/ucpd.rs
+++ b/embassy-stm32/src/ucpd.rs
@@ -28,7 +28,7 @@ use crate::interrupt;
 use crate::interrupt::typelevel::Interrupt;
 use crate::pac::ucpd::vals::{Anamode, Ccenable, PscUsbpdclk, Txmode};
 pub use crate::pac::ucpd::vals::{Phyccsel as CcSel, TypecVstateCc as CcVState};
-use crate::rcc::RccPeripheral;
+use crate::rcc::{self, RccPeripheral};
 
 pub(crate) fn init(
     _cs: critical_section::CriticalSection,
@@ -103,7 +103,7 @@ impl<'d, T: Instance> Ucpd<'d, T> {
         cc1.set_as_analog();
         cc2.set_as_analog();
 
-        T::enable_and_reset();
+        rcc::enable_and_reset::<T>();
         T::Interrupt::unpend();
         unsafe { T::Interrupt::enable() };
 
@@ -212,7 +212,7 @@ impl<'d, T: Instance> Drop for CcPhy<'d, T> {
             drop_not_ready.store(true, Ordering::Relaxed);
         } else {
             r.cfgr1().write(|w| w.set_ucpden(false));
-            T::disable();
+            rcc::disable::<T>();
             T::Interrupt::disable();
         }
     }
@@ -325,7 +325,7 @@ impl<'d, T: Instance> Drop for PdPhy<'d, T> {
             drop_not_ready.store(true, Ordering::Relaxed);
         } else {
             T::REGS.cfgr1().write(|w| w.set_ucpden(false));
-            T::disable();
+            rcc::disable::<T>();
             T::Interrupt::disable();
         }
     }
diff --git a/embassy-stm32/src/usart/buffered.rs b/embassy-stm32/src/usart/buffered.rs
index 492ad334b..fd79e035e 100644
--- a/embassy-stm32/src/usart/buffered.rs
+++ b/embassy-stm32/src/usart/buffered.rs
@@ -6,7 +6,7 @@ use core::task::Poll;
 
 use embassy_embedded_hal::SetConfig;
 use embassy_hal_internal::atomic_ring_buffer::RingBuffer;
-use embassy_hal_internal::{into_ref, Peripheral};
+use embassy_hal_internal::{Peripheral, PeripheralRef};
 use embassy_sync::waitqueue::AtomicWaker;
 
 #[cfg(not(any(usart_v1, usart_v2)))]
@@ -15,8 +15,7 @@ use super::{
     clear_interrupt_flags, configure, rdr, reconfigure, sr, tdr, Config, ConfigError, CtsPin, Error, Info, Instance,
     Regs, RtsPin, RxPin, TxPin,
 };
-use crate::gpio::AFType;
-use crate::interrupt::typelevel::Interrupt as _;
+use crate::gpio::{AFType, AnyPin, SealedPin as _};
 use crate::interrupt::{self, InterruptExt};
 use crate::time::Hertz;
 
@@ -155,7 +154,9 @@ pub struct BufferedUartTx<'d> {
     info: &'static Info,
     state: &'static State,
     kernel_clock: Hertz,
-    _phantom: PhantomData<&'d mut ()>,
+    tx: Option<PeripheralRef<'d, AnyPin>>,
+    cts: Option<PeripheralRef<'d, AnyPin>>,
+    de: Option<PeripheralRef<'d, AnyPin>>,
 }
 
 /// Rx-only buffered UART
@@ -165,7 +166,8 @@ pub struct BufferedUartRx<'d> {
     info: &'static Info,
     state: &'static State,
     kernel_clock: Hertz,
-    _phantom: PhantomData<&'d mut ()>,
+    rx: Option<PeripheralRef<'d, AnyPin>>,
+    rts: Option<PeripheralRef<'d, AnyPin>>,
 }
 
 impl<'d> SetConfig for BufferedUart<'d> {
@@ -206,9 +208,17 @@ impl<'d> BufferedUart<'d> {
         rx_buffer: &'d mut [u8],
         config: Config,
     ) -> Result<Self, ConfigError> {
-        T::enable_and_reset();
-
-        Self::new_inner(peri, rx, tx, tx_buffer, rx_buffer, config)
+        Self::new_inner(
+            peri,
+            new_pin!(rx, AFType::Input),
+            new_pin!(tx, AFType::OutputPushPull),
+            None,
+            None,
+            None,
+            tx_buffer,
+            rx_buffer,
+            config,
+        )
     }
 
     /// Create a new bidirectional buffered UART driver with request-to-send and clear-to-send pins
@@ -223,18 +233,17 @@ impl<'d> BufferedUart<'d> {
         rx_buffer: &'d mut [u8],
         config: Config,
     ) -> Result<Self, ConfigError> {
-        into_ref!(cts, rts);
-
-        T::enable_and_reset();
-
-        rts.set_as_af(rts.af_num(), AFType::OutputPushPull);
-        cts.set_as_af(cts.af_num(), AFType::Input);
-        T::info().regs.cr3().write(|w| {
-            w.set_rtse(true);
-            w.set_ctse(true);
-        });
-
-        Self::new_inner(peri, rx, tx, tx_buffer, rx_buffer, config)
+        Self::new_inner(
+            peri,
+            new_pin!(rx, AFType::Input),
+            new_pin!(tx, AFType::OutputPushPull),
+            new_pin!(rts, AFType::OutputPushPull),
+            new_pin!(cts, AFType::Input),
+            None,
+            tx_buffer,
+            rx_buffer,
+            config,
+        )
     }
 
     /// Create a new bidirectional buffered UART driver with a driver-enable pin
@@ -249,66 +258,89 @@ impl<'d> BufferedUart<'d> {
         rx_buffer: &'d mut [u8],
         config: Config,
     ) -> Result<Self, ConfigError> {
-        into_ref!(de);
-
-        T::enable_and_reset();
-
-        de.set_as_af(de.af_num(), AFType::OutputPushPull);
-        T::info().regs.cr3().write(|w| {
-            w.set_dem(true);
-        });
-
-        Self::new_inner(peri, rx, tx, tx_buffer, rx_buffer, config)
+        Self::new_inner(
+            peri,
+            new_pin!(rx, AFType::Input),
+            new_pin!(tx, AFType::OutputPushPull),
+            None,
+            None,
+            new_pin!(de, AFType::OutputPushPull),
+            tx_buffer,
+            rx_buffer,
+            config,
+        )
     }
 
     fn new_inner<T: Instance>(
         _peri: impl Peripheral<P = T> + 'd,
-        rx: impl Peripheral<P = impl RxPin<T>> + 'd,
-        tx: impl Peripheral<P = impl TxPin<T>> + 'd,
+        rx: Option<PeripheralRef<'d, AnyPin>>,
+        tx: Option<PeripheralRef<'d, AnyPin>>,
+        rts: Option<PeripheralRef<'d, AnyPin>>,
+        cts: Option<PeripheralRef<'d, AnyPin>>,
+        de: Option<PeripheralRef<'d, AnyPin>>,
         tx_buffer: &'d mut [u8],
         rx_buffer: &'d mut [u8],
         config: Config,
     ) -> Result<Self, ConfigError> {
-        into_ref!(_peri, rx, tx);
-
         let info = T::info();
         let state = T::buffered_state();
         let kernel_clock = T::frequency();
-        let len = tx_buffer.len();
-        unsafe { state.tx_buf.init(tx_buffer.as_mut_ptr(), len) };
-        let len = rx_buffer.len();
-        unsafe { state.rx_buf.init(rx_buffer.as_mut_ptr(), len) };
-
-        let r = info.regs;
-        rx.set_as_af(rx.af_num(), AFType::Input);
-        tx.set_as_af(tx.af_num(), AFType::OutputPushPull);
-
-        configure(info, kernel_clock, &config, true, true)?;
-
-        r.cr1().modify(|w| {
-            w.set_rxneie(true);
-            w.set_idleie(true);
-        });
-
-        T::Interrupt::unpend();
-        unsafe { T::Interrupt::enable() };
 
-        state.tx_rx_refcount.store(2, Ordering::Relaxed);
-
-        Ok(Self {
+        let mut this = Self {
             rx: BufferedUartRx {
                 info,
                 state,
                 kernel_clock,
-                _phantom: PhantomData,
+                rx,
+                rts,
             },
             tx: BufferedUartTx {
                 info,
                 state,
                 kernel_clock,
-                _phantom: PhantomData,
+                tx,
+                cts,
+                de,
             },
-        })
+        };
+        this.enable_and_configure(tx_buffer, rx_buffer, &config)?;
+        Ok(this)
+    }
+
+    fn enable_and_configure(
+        &mut self,
+        tx_buffer: &'d mut [u8],
+        rx_buffer: &'d mut [u8],
+        config: &Config,
+    ) -> Result<(), ConfigError> {
+        let info = self.rx.info;
+        let state = self.rx.state;
+        state.tx_rx_refcount.store(2, Ordering::Relaxed);
+
+        info.rcc.enable_and_reset();
+
+        let len = tx_buffer.len();
+        unsafe { state.tx_buf.init(tx_buffer.as_mut_ptr(), len) };
+        let len = rx_buffer.len();
+        unsafe { state.rx_buf.init(rx_buffer.as_mut_ptr(), len) };
+
+        info.regs.cr3().write(|w| {
+            w.set_rtse(self.rx.rts.is_some());
+            w.set_ctse(self.tx.cts.is_some());
+            #[cfg(not(any(usart_v1, usart_v2)))]
+            w.set_dem(self.tx.de.is_some());
+        });
+        configure(info, self.rx.kernel_clock, &config, true, true)?;
+
+        info.regs.cr1().modify(|w| {
+            w.set_rxneie(true);
+            w.set_idleie(true);
+        });
+
+        info.interrupt.unpend();
+        unsafe { info.interrupt.enable() };
+
+        Ok(())
     }
 
     /// Split the driver into a Tx and Rx part (useful for sending to separate tasks)
@@ -515,6 +547,8 @@ impl<'d> Drop for BufferedUartRx<'d> {
             }
         }
 
+        self.rx.as_ref().map(|x| x.set_as_disconnected());
+        self.rts.as_ref().map(|x| x.set_as_disconnected());
         drop_tx_rx(self.info, state);
     }
 }
@@ -532,6 +566,9 @@ impl<'d> Drop for BufferedUartTx<'d> {
             }
         }
 
+        self.tx.as_ref().map(|x| x.set_as_disconnected());
+        self.cts.as_ref().map(|x| x.set_as_disconnected());
+        self.de.as_ref().map(|x| x.set_as_disconnected());
         drop_tx_rx(self.info, state);
     }
 }
@@ -545,7 +582,7 @@ fn drop_tx_rx(info: &Info, state: &State) {
         refcount == 1
     });
     if is_last_drop {
-        info.enable_bit.disable();
+        info.rcc.disable();
     }
 }
 
diff --git a/embassy-stm32/src/usart/mod.rs b/embassy-stm32/src/usart/mod.rs
index b24335f3a..53321391a 100644
--- a/embassy-stm32/src/usart/mod.rs
+++ b/embassy-stm32/src/usart/mod.rs
@@ -14,7 +14,7 @@ use embassy_sync::waitqueue::AtomicWaker;
 use futures_util::future::{select, Either};
 
 use crate::dma::ChannelAndRequest;
-use crate::gpio::{AFType, AnyPin, SealedPin};
+use crate::gpio::{AFType, AnyPin, SealedPin as _};
 use crate::interrupt::typelevel::Interrupt as _;
 use crate::interrupt::{self, Interrupt, InterruptExt};
 use crate::mode::{Async, Blocking, Mode};
@@ -28,7 +28,7 @@ use crate::pac::usart::Lpuart as Regs;
 #[cfg(any(usart_v1, usart_v2))]
 use crate::pac::usart::Usart as Regs;
 use crate::pac::usart::{regs, vals};
-use crate::rcc::{ClockEnableBit, SealedRccPeripheral};
+use crate::rcc::{RccInfo, SealedRccPeripheral};
 use crate::time::Hertz;
 use crate::Peripheral;
 
@@ -429,29 +429,33 @@ impl<'d, M: Mode> UartTx<'d, M> {
         tx_dma: Option<ChannelAndRequest<'d>>,
         config: Config,
     ) -> Result<Self, ConfigError> {
-        T::enable_and_reset();
-
-        let info = T::info();
-        let state = T::state();
-        let kernel_clock = T::frequency();
-        let r = info.regs;
-        r.cr3().modify(|w| {
-            w.set_ctse(cts.is_some());
-        });
-        configure(info, kernel_clock, &config, false, true)?;
-
-        state.tx_rx_refcount.store(1, Ordering::Relaxed);
-
-        Ok(Self {
-            info,
-            state,
-            kernel_clock,
+        let mut this = Self {
+            info: T::info(),
+            state: T::state(),
+            kernel_clock: T::frequency(),
             tx,
             cts,
             de: None,
             tx_dma,
             _phantom: PhantomData,
-        })
+        };
+        this.enable_and_configure(&config)?;
+        Ok(this)
+    }
+
+    fn enable_and_configure(&mut self, config: &Config) -> Result<(), ConfigError> {
+        let info = self.info;
+        let state = self.state;
+        state.tx_rx_refcount.store(1, Ordering::Relaxed);
+
+        info.rcc.enable_and_reset();
+
+        info.regs.cr3().modify(|w| {
+            w.set_ctse(self.cts.is_some());
+        });
+        configure(info, self.kernel_clock, config, false, true)?;
+
+        Ok(())
     }
 
     /// Reconfigure the driver
@@ -775,34 +779,38 @@ impl<'d, M: Mode> UartRx<'d, M> {
         rx_dma: Option<ChannelAndRequest<'d>>,
         config: Config,
     ) -> Result<Self, ConfigError> {
-        T::enable_and_reset();
-
-        let info = T::info();
-        let state = T::state();
-        let kernel_clock = T::frequency();
-        let r = info.regs;
-        r.cr3().write(|w| {
-            w.set_rtse(rts.is_some());
-        });
-        configure(info, kernel_clock, &config, true, false)?;
-
-        T::Interrupt::unpend();
-        unsafe { T::Interrupt::enable() };
-
-        state.tx_rx_refcount.store(1, Ordering::Relaxed);
-
-        Ok(Self {
+        let mut this = Self {
             _phantom: PhantomData,
-            info,
-            state,
-            kernel_clock,
+            info: T::info(),
+            state: T::state(),
+            kernel_clock: T::frequency(),
             rx,
             rts,
             rx_dma,
             detect_previous_overrun: config.detect_previous_overrun,
             #[cfg(any(usart_v1, usart_v2))]
             buffered_sr: stm32_metapac::usart::regs::Sr(0),
-        })
+        };
+        this.enable_and_configure(&config)?;
+        Ok(this)
+    }
+
+    fn enable_and_configure(&mut self, config: &Config) -> Result<(), ConfigError> {
+        let info = self.info;
+        let state = self.state;
+        state.tx_rx_refcount.store(1, Ordering::Relaxed);
+
+        info.rcc.enable_and_reset();
+
+        info.regs.cr3().write(|w| {
+            w.set_rtse(self.rts.is_some());
+        });
+        configure(info, self.kernel_clock, &config, true, false)?;
+
+        info.interrupt.unpend();
+        unsafe { info.interrupt.enable() };
+
+        Ok(())
     }
 
     /// Reconfigure the driver
@@ -916,7 +924,7 @@ fn drop_tx_rx(info: &Info, state: &State) {
         refcount == 1
     });
     if is_last_drop {
-        info.enable_bit.disable();
+        info.rcc.disable();
     }
 }
 
@@ -1228,27 +1236,11 @@ impl<'d, M: Mode> Uart<'d, M> {
         rx_dma: Option<ChannelAndRequest<'d>>,
         config: Config,
     ) -> Result<Self, ConfigError> {
-        T::enable_and_reset();
-
         let info = T::info();
         let state = T::state();
         let kernel_clock = T::frequency();
-        let r = info.regs;
-
-        r.cr3().write(|w| {
-            w.set_rtse(rts.is_some());
-            w.set_ctse(cts.is_some());
-            #[cfg(not(any(usart_v1, usart_v2)))]
-            w.set_dem(de.is_some());
-        });
-        configure(info, kernel_clock, &config, true, true)?;
-
-        T::Interrupt::unpend();
-        unsafe { T::Interrupt::enable() };
 
-        state.tx_rx_refcount.store(2, Ordering::Relaxed);
-
-        Ok(Self {
+        let mut this = Self {
             tx: UartTx {
                 _phantom: PhantomData,
                 info,
@@ -1271,7 +1263,30 @@ impl<'d, M: Mode> Uart<'d, M> {
                 #[cfg(any(usart_v1, usart_v2))]
                 buffered_sr: stm32_metapac::usart::regs::Sr(0),
             },
-        })
+        };
+        this.enable_and_configure(&config)?;
+        Ok(this)
+    }
+
+    fn enable_and_configure(&mut self, config: &Config) -> Result<(), ConfigError> {
+        let info = self.rx.info;
+        let state = self.rx.state;
+        state.tx_rx_refcount.store(2, Ordering::Relaxed);
+
+        info.rcc.enable_and_reset();
+
+        info.regs.cr3().write(|w| {
+            w.set_rtse(self.rx.rts.is_some());
+            w.set_ctse(self.tx.cts.is_some());
+            #[cfg(not(any(usart_v1, usart_v2)))]
+            w.set_dem(self.tx.de.is_some());
+        });
+        configure(info, self.rx.kernel_clock, config, true, true)?;
+
+        info.interrupt.unpend();
+        unsafe { info.interrupt.enable() };
+
+        Ok(())
     }
 
     /// Perform a blocking write
@@ -1718,7 +1733,7 @@ impl State {
 
 struct Info {
     regs: Regs,
-    enable_bit: ClockEnableBit,
+    rcc: RccInfo,
     interrupt: Interrupt,
     kind: Kind,
 }
@@ -1754,7 +1769,7 @@ macro_rules! impl_usart {
             fn info() -> &'static Info {
                 static INFO: Info = Info {
                     regs: unsafe { Regs::from_ptr(crate::pac::$inst.as_ptr()) },
-                    enable_bit: crate::peripherals::$inst::ENABLE_BIT,
+                    rcc: crate::peripherals::$inst::RCC_INFO,
                     interrupt: crate::interrupt::typelevel::$irq::IRQ,
                     kind: $kind,
                 };
diff --git a/embassy-stm32/src/usart/ringbuffered.rs b/embassy-stm32/src/usart/ringbuffered.rs
index f3a88b93f..8cf75933a 100644
--- a/embassy-stm32/src/usart/ringbuffered.rs
+++ b/embassy-stm32/src/usart/ringbuffered.rs
@@ -4,10 +4,12 @@ use core::sync::atomic::{compiler_fence, Ordering};
 use core::task::Poll;
 
 use embassy_embedded_hal::SetConfig;
+use embassy_hal_internal::PeripheralRef;
 use futures_util::future::{select, Either};
 
 use super::{clear_interrupt_flags, rdr, reconfigure, sr, Config, ConfigError, Error, Info, State, UartRx};
 use crate::dma::ReadableRingBuffer;
+use crate::gpio::{AnyPin, SealedPin as _};
 use crate::mode::Async;
 use crate::time::Hertz;
 use crate::usart::{Regs, Sr};
@@ -19,6 +21,8 @@ pub struct RingBufferedUartRx<'d> {
     info: &'static Info,
     state: &'static State,
     kernel_clock: Hertz,
+    rx: Option<PeripheralRef<'d, AnyPin>>,
+    rts: Option<PeripheralRef<'d, AnyPin>>,
     ring_buf: ReadableRingBuffer<'d, u8>,
 }
 
@@ -49,6 +53,8 @@ impl<'d> UartRx<'d, Async> {
         let state = self.state;
         let kernel_clock = self.kernel_clock;
         let ring_buf = unsafe { ReadableRingBuffer::new(rx_dma, request, rdr(info.regs), dma_buf, opts) };
+        let rx = unsafe { self.rx.as_ref().map(|x| x.clone_unchecked()) };
+        let rts = unsafe { self.rts.as_ref().map(|x| x.clone_unchecked()) };
 
         // Don't disable the clock
         mem::forget(self);
@@ -57,6 +63,8 @@ impl<'d> UartRx<'d, Async> {
             info,
             state,
             kernel_clock,
+            rx,
+            rts,
             ring_buf,
         }
     }
@@ -221,6 +229,8 @@ impl<'d> RingBufferedUartRx<'d> {
 impl Drop for RingBufferedUartRx<'_> {
     fn drop(&mut self) {
         self.teardown_uart();
+        self.rx.as_ref().map(|x| x.set_as_disconnected());
+        self.rts.as_ref().map(|x| x.set_as_disconnected());
         super::drop_tx_rx(self.info, self.state);
     }
 }
diff --git a/embassy-stm32/src/usb/mod.rs b/embassy-stm32/src/usb/mod.rs
index 349438ec5..ce9fe0a9b 100644
--- a/embassy-stm32/src/usb/mod.rs
+++ b/embassy-stm32/src/usb/mod.rs
@@ -6,7 +6,7 @@ mod _version;
 pub use _version::*;
 
 use crate::interrupt::typelevel::Interrupt;
-use crate::rcc::SealedRccPeripheral;
+use crate::rcc;
 
 /// clock, power initialization stuff that's common for USB and OTG.
 fn common_init<T: Instance>() {
@@ -65,5 +65,5 @@ fn common_init<T: Instance>() {
     T::Interrupt::unpend();
     unsafe { T::Interrupt::enable() };
 
-    <T as SealedRccPeripheral>::enable_and_reset();
+    rcc::enable_and_reset::<T>();
 }
diff --git a/embassy-stm32/src/usb/otg.rs b/embassy-stm32/src/usb/otg.rs
index 3debd5079..e5131250a 100644
--- a/embassy-stm32/src/usb/otg.rs
+++ b/embassy-stm32/src/usb/otg.rs
@@ -13,7 +13,7 @@ use embassy_usb_synopsys_otg::{
 use crate::gpio::AFType;
 use crate::interrupt;
 use crate::interrupt::typelevel::Interrupt;
-use crate::rcc::{RccPeripheral, SealedRccPeripheral};
+use crate::rcc::{self, RccPeripheral};
 
 const MAX_EP_COUNT: usize = 9;
 
@@ -246,7 +246,7 @@ impl<'d, T: Instance> Bus<'d, T> {
     fn disable(&mut self) {
         T::Interrupt::disable();
 
-        <T as SealedRccPeripheral>::disable();
+        rcc::disable::<T>();
         self.inited = false;
 
         #[cfg(stm32l4)]
diff --git a/embassy-stm32/src/usb/usb.rs b/embassy-stm32/src/usb/usb.rs
index 81a2d2623..b6c88ac9a 100644
--- a/embassy-stm32/src/usb/usb.rs
+++ b/embassy-stm32/src/usb/usb.rs
@@ -267,9 +267,9 @@ impl<'d, T: Instance> Driver<'d, T> {
             w.set_fres(true);
         });
 
-        #[cfg(time)]
+        #[cfg(feature = "time")]
         embassy_time::block_for(embassy_time::Duration::from_millis(100));
-        #[cfg(not(time))]
+        #[cfg(not(feature = "time"))]
         cortex_m::asm::delay(unsafe { crate::rcc::get_freqs() }.sys.unwrap().0 / 10);
 
         #[cfg(not(usb_v4))]
diff --git a/embassy-sync/build.rs b/embassy-sync/build.rs
index afd76dad1..ecd2c0c9f 100644
--- a/embassy-sync/build.rs
+++ b/embassy-sync/build.rs
@@ -1,31 +1,7 @@
-use std::env;
+#[path = "./build_common.rs"]
+mod common;
 
 fn main() {
-    println!("cargo:rerun-if-changed=build.rs");
-
-    let target = env::var("TARGET").unwrap();
-
-    if target.starts_with("thumbv6m-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv6m");
-    } else if target.starts_with("thumbv7m-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv7m");
-    } else if target.starts_with("thumbv7em-") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv7m");
-        println!("cargo:rustc-cfg=armv7em"); // (not currently used)
-    } else if target.starts_with("thumbv8m.base") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv8m");
-        println!("cargo:rustc-cfg=armv8m_base");
-    } else if target.starts_with("thumbv8m.main") {
-        println!("cargo:rustc-cfg=cortex_m");
-        println!("cargo:rustc-cfg=armv8m");
-        println!("cargo:rustc-cfg=armv8m_main");
-    }
-
-    if target.ends_with("-eabihf") {
-        println!("cargo:rustc-cfg=has_fpu");
-    }
+    let mut cfgs = common::CfgSet::new();
+    common::set_target_cfgs(&mut cfgs);
 }
diff --git a/embassy-sync/build_common.rs b/embassy-sync/build_common.rs
new file mode 100644
index 000000000..2c65f8529
--- /dev/null
+++ b/embassy-sync/build_common.rs
@@ -0,0 +1,109 @@
+// NOTE: this file is copy-pasted between several Embassy crates, because there is no
+// straightforward way to share this code:
+// - it cannot be placed into the root of the repo and linked from each build.rs using `#[path =
+// "../build_common.rs"]`, because `cargo publish` requires that all files published with a crate
+// reside in the crate's directory,
+// - it cannot be symlinked from `embassy-xxx/build_common.rs` to `../build_common.rs`, because
+// symlinks don't work on Windows.
+
+use std::collections::HashSet;
+use std::env;
+use std::ffi::OsString;
+use std::process::Command;
+
+/// Helper for emitting cargo instruction for enabling configs (`cargo:rustc-cfg=X`) and declaring
+/// them (`cargo:rust-check-cfg=cfg(X)`).
+#[derive(Debug)]
+pub struct CfgSet {
+    enabled: HashSet<String>,
+    declared: HashSet<String>,
+    emit_declared: bool,
+}
+
+impl CfgSet {
+    pub fn new() -> Self {
+        Self {
+            enabled: HashSet::new(),
+            declared: HashSet::new(),
+            emit_declared: is_rustc_nightly(),
+        }
+    }
+
+    /// Enable a config, which can then be used in `#[cfg(...)]` for conditional compilation.
+    ///
+    /// All configs that can potentially be enabled should be unconditionally declared using
+    /// [`Self::declare()`].
+    pub fn enable(&mut self, cfg: impl AsRef<str>) {
+        if self.enabled.insert(cfg.as_ref().to_owned()) {
+            println!("cargo:rustc-cfg={}", cfg.as_ref());
+        }
+    }
+
+    pub fn enable_all(&mut self, cfgs: &[impl AsRef<str>]) {
+        for cfg in cfgs.iter() {
+            self.enable(cfg.as_ref());
+        }
+    }
+
+    /// Declare a valid config for conditional compilation, without enabling it.
+    ///
+    /// This enables rustc to check that the configs in `#[cfg(...)]` attributes are valid.
+    pub fn declare(&mut self, cfg: impl AsRef<str>) {
+        if self.declared.insert(cfg.as_ref().to_owned()) && self.emit_declared {
+            println!("cargo:rustc-check-cfg=cfg({})", cfg.as_ref());
+        }
+    }
+
+    pub fn declare_all(&mut self, cfgs: &[impl AsRef<str>]) {
+        for cfg in cfgs.iter() {
+            self.declare(cfg.as_ref());
+        }
+    }
+
+    pub fn set(&mut self, cfg: impl Into<String>, enable: bool) {
+        let cfg = cfg.into();
+        if enable {
+            self.enable(cfg.clone());
+        }
+        self.declare(cfg);
+    }
+}
+
+fn is_rustc_nightly() -> bool {
+    let rustc = env::var_os("RUSTC").unwrap_or_else(|| OsString::from("rustc"));
+
+    let output = Command::new(rustc)
+        .arg("--version")
+        .output()
+        .expect("failed to run `rustc --version`");
+
+    String::from_utf8_lossy(&output.stdout).contains("nightly")
+}
+
+/// Sets configs that describe the target platform.
+pub fn set_target_cfgs(cfgs: &mut CfgSet) {
+    let target = env::var("TARGET").unwrap();
+
+    if target.starts_with("thumbv6m-") {
+        cfgs.enable_all(&["cortex_m", "armv6m"]);
+    } else if target.starts_with("thumbv7m-") {
+        cfgs.enable_all(&["cortex_m", "armv7m"]);
+    } else if target.starts_with("thumbv7em-") {
+        cfgs.enable_all(&["cortex_m", "armv7m", "armv7em"]);
+    } else if target.starts_with("thumbv8m.base") {
+        cfgs.enable_all(&["cortex_m", "armv8m", "armv8m_base"]);
+    } else if target.starts_with("thumbv8m.main") {
+        cfgs.enable_all(&["cortex_m", "armv8m", "armv8m_main"]);
+    }
+    cfgs.declare_all(&[
+        "cortex_m",
+        "armv6m",
+        "armv7m",
+        "armv7em",
+        "armv8m",
+        "armv8m_base",
+        "armv8m_main",
+    ]);
+
+    cfgs.set("has_fpu", target.ends_with("-eabihf"));
+}
diff --git a/embassy-sync/src/lib.rs b/embassy-sync/src/lib.rs
index 1873483f9..a5eee8d02 100644
--- a/embassy-sync/src/lib.rs
+++ b/embassy-sync/src/lib.rs
@@ -1,4 +1,4 @@
-#![cfg_attr(not(any(feature = "std", feature = "wasm")), no_std)]
+#![cfg_attr(not(feature = "std"), no_std)]
 #![allow(async_fn_in_trait)]
 #![allow(clippy::new_without_default)]
 #![doc = include_str!("../README.md")]
diff --git a/embassy-time/Cargo.toml b/embassy-time/Cargo.toml
index ca7ad2d09..ce3f3d8c2 100644
--- a/embassy-time/Cargo.toml
+++ b/embassy-time/Cargo.toml
@@ -27,9 +27,17 @@ features = ["defmt", "std"]
 std = ["tick-hz-1_000_000", "critical-section/std"]
 wasm = ["dep:wasm-bindgen", "dep:js-sys", "dep:wasm-timer", "tick-hz-1_000_000"]
 
-## Display a timestamp of the number of seconds since startup next to defmt log messages
+## Display the time since startup next to defmt log messages.
+## At most 1 `defmt-timestamp-uptime-*` feature can be used.
+## `defmt-timestamp-uptime` is provided for backwards compatibility (provides the same format as `uptime-us`).
 ## To use this you must have a time driver provided.
 defmt-timestamp-uptime = ["defmt"]
+defmt-timestamp-uptime-s = ["defmt"]
+defmt-timestamp-uptime-ms = ["defmt"]
+defmt-timestamp-uptime-us = ["defmt"]
+defmt-timestamp-uptime-ts = ["defmt"]
+defmt-timestamp-uptime-tms = ["defmt"]
+defmt-timestamp-uptime-tus = ["defmt"]
 
 ## Create a `MockDriver` that can be manually advanced for testing purposes.
 mock-driver = ["tick-hz-1_000_000"]
diff --git a/embassy-time/src/lib.rs b/embassy-time/src/lib.rs
index 3c8575ee9..24ee51be7 100644
--- a/embassy-time/src/lib.rs
+++ b/embassy-time/src/lib.rs
@@ -46,5 +46,20 @@ pub(crate) const GCD_1K: u64 = gcd(TICK_HZ, 1_000);
 pub(crate) const GCD_1M: u64 = gcd(TICK_HZ, 1_000_000);
 pub(crate) const GCD_1G: u64 = gcd(TICK_HZ, 1_000_000_000);
 
-#[cfg(feature = "defmt-timestamp-uptime")]
+#[cfg(feature = "defmt-timestamp-uptime-s")]
+defmt::timestamp! {"{=u64}", Instant::now().as_secs() }
+
+#[cfg(feature = "defmt-timestamp-uptime-ms")]
+defmt::timestamp! {"{=u64:ms}", Instant::now().as_millis() }
+
+#[cfg(any(feature = "defmt-timestamp-uptime", feature = "defmt-timestamp-uptime-us"))]
 defmt::timestamp! {"{=u64:us}", Instant::now().as_micros() }
+
+#[cfg(feature = "defmt-timestamp-uptime-ts")]
+defmt::timestamp! {"{=u64:ts}", Instant::now().as_secs() }
+
+#[cfg(feature = "defmt-timestamp-uptime-tms")]
+defmt::timestamp! {"{=u64:tms}", Instant::now().as_millis() }
+
+#[cfg(feature = "defmt-timestamp-uptime-tus")]
+defmt::timestamp! {"{=u64:tus}", Instant::now().as_micros() }
diff --git a/embassy-usb-dfu/Cargo.toml b/embassy-usb-dfu/Cargo.toml
index 267e412c0..ca87bcb62 100644
--- a/embassy-usb-dfu/Cargo.toml
+++ b/embassy-usb-dfu/Cargo.toml
@@ -26,9 +26,11 @@ flavors = [
 features = ["defmt", "cortex-m", "dfu"]
 
 [dependencies]
+defmt = { version = "0.3.5", optional = true }
+log = { version = "0.4.17", optional = true }
+
 bitflags = "2.4.1"
 cortex-m = { version = "0.7.7", features = ["inline-asm"], optional = true }
-defmt = { version = "0.3.5", optional = true }
 embassy-boot = { version = "0.2.0", path = "../embassy-boot" }
 embassy-futures = { version = "0.1.1", path = "../embassy-futures" }
 embassy-sync = { version = "0.6.0", path = "../embassy-sync" }
diff --git a/examples/boot/application/nrf/Cargo.toml b/examples/boot/application/nrf/Cargo.toml
index f0a710335..cf34bb7cf 100644
--- a/examples/boot/application/nrf/Cargo.toml
+++ b/examples/boot/application/nrf/Cargo.toml
@@ -25,3 +25,10 @@ cortex-m-rt = "0.7.0"
 ed25519-dalek = ["embassy-boot/ed25519-dalek"]
 ed25519-salty = ["embassy-boot/ed25519-salty"]
 skip-include = []
+defmt = [
+      "dep:defmt",
+      "dep:defmt-rtt",
+      "embassy-nrf/defmt",
+      "embassy-boot-nrf/defmt",
+      "embassy-sync/defmt",
+]
diff --git a/examples/boot/application/stm32f3/Cargo.toml b/examples/boot/application/stm32f3/Cargo.toml
index fe1a6f5b1..9b79b01ce 100644
--- a/examples/boot/application/stm32f3/Cargo.toml
+++ b/examples/boot/application/stm32f3/Cargo.toml
@@ -23,6 +23,7 @@ cortex-m-rt = "0.7.0"
 [features]
 defmt = [
       "dep:defmt",
+      "dep:defmt-rtt",
       "embassy-stm32/defmt",
       "embassy-boot-stm32/defmt",
       "embassy-sync/defmt",
diff --git a/examples/boot/application/stm32f3/src/bin/a.rs b/examples/boot/application/stm32f3/src/bin/a.rs
index 8858ae3da..b608b2e01 100644
--- a/examples/boot/application/stm32f3/src/bin/a.rs
+++ b/examples/boot/application/stm32f3/src/bin/a.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_boot_stm32::{AlignedBuffer, FirmwareUpdater, FirmwareUpdaterConfig};
 use embassy_embedded_hal::adapter::BlockingAsync;
diff --git a/examples/boot/application/stm32f3/src/bin/b.rs b/examples/boot/application/stm32f3/src/bin/b.rs
index 22ba82d5e..b1a505631 100644
--- a/examples/boot/application/stm32f3/src/bin/b.rs
+++ b/examples/boot/application/stm32f3/src/bin/b.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::gpio::{Level, Output, Speed};
diff --git a/examples/boot/application/stm32f7/Cargo.toml b/examples/boot/application/stm32f7/Cargo.toml
index 37e362824..0167dfb76 100644
--- a/examples/boot/application/stm32f7/Cargo.toml
+++ b/examples/boot/application/stm32f7/Cargo.toml
@@ -24,6 +24,7 @@ cortex-m-rt = "0.7.0"
 [features]
 defmt = [
       "dep:defmt",
+      "dep:defmt-rtt",
       "embassy-stm32/defmt",
       "embassy-boot-stm32/defmt",
       "embassy-sync/defmt",
diff --git a/examples/boot/application/stm32f7/src/bin/a.rs b/examples/boot/application/stm32f7/src/bin/a.rs
index d3df11fe4..172b4c235 100644
--- a/examples/boot/application/stm32f7/src/bin/a.rs
+++ b/examples/boot/application/stm32f7/src/bin/a.rs
@@ -3,7 +3,7 @@
 
 use core::cell::RefCell;
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_boot_stm32::{AlignedBuffer, BlockingFirmwareUpdater, FirmwareUpdaterConfig};
 use embassy_executor::Spawner;
diff --git a/examples/boot/application/stm32f7/src/bin/b.rs b/examples/boot/application/stm32f7/src/bin/b.rs
index 190477204..6bc9c9ab8 100644
--- a/examples/boot/application/stm32f7/src/bin/b.rs
+++ b/examples/boot/application/stm32f7/src/bin/b.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::gpio::{Level, Output, Speed};
diff --git a/examples/boot/application/stm32h7/Cargo.toml b/examples/boot/application/stm32h7/Cargo.toml
index 52cd0b546..61643d485 100644
--- a/examples/boot/application/stm32h7/Cargo.toml
+++ b/examples/boot/application/stm32h7/Cargo.toml
@@ -24,6 +24,7 @@ cortex-m-rt = "0.7.0"
 [features]
 defmt = [
       "dep:defmt",
+      "dep:defmt-rtt",
       "embassy-stm32/defmt",
       "embassy-boot-stm32/defmt",
       "embassy-sync/defmt",
diff --git a/examples/boot/application/stm32h7/src/bin/a.rs b/examples/boot/application/stm32h7/src/bin/a.rs
index f61ac1f71..c1b1a267a 100644
--- a/examples/boot/application/stm32h7/src/bin/a.rs
+++ b/examples/boot/application/stm32h7/src/bin/a.rs
@@ -3,7 +3,7 @@
 
 use core::cell::RefCell;
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_boot_stm32::{AlignedBuffer, BlockingFirmwareUpdater, FirmwareUpdaterConfig};
 use embassy_executor::Spawner;
diff --git a/examples/boot/application/stm32h7/src/bin/b.rs b/examples/boot/application/stm32h7/src/bin/b.rs
index 5f3f35207..13bdae1f1 100644
--- a/examples/boot/application/stm32h7/src/bin/b.rs
+++ b/examples/boot/application/stm32h7/src/bin/b.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::gpio::{Level, Output, Speed};
diff --git a/examples/boot/application/stm32l0/Cargo.toml b/examples/boot/application/stm32l0/Cargo.toml
index 0f3cbe654..2990089ac 100644
--- a/examples/boot/application/stm32l0/Cargo.toml
+++ b/examples/boot/application/stm32l0/Cargo.toml
@@ -23,6 +23,7 @@ cortex-m-rt = "0.7.0"
 [features]
 defmt = [
       "dep:defmt",
+      "dep:defmt-rtt",
       "embassy-stm32/defmt",
       "embassy-boot-stm32/defmt",
       "embassy-sync/defmt",
diff --git a/examples/boot/application/stm32l0/src/bin/a.rs b/examples/boot/application/stm32l0/src/bin/a.rs
index f066c1139..dcc10e5c6 100644
--- a/examples/boot/application/stm32l0/src/bin/a.rs
+++ b/examples/boot/application/stm32l0/src/bin/a.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_boot_stm32::{AlignedBuffer, FirmwareUpdater, FirmwareUpdaterConfig};
 use embassy_embedded_hal::adapter::BlockingAsync;
diff --git a/examples/boot/application/stm32l0/src/bin/b.rs b/examples/boot/application/stm32l0/src/bin/b.rs
index 6bf00f41a..a59c6f540 100644
--- a/examples/boot/application/stm32l0/src/bin/b.rs
+++ b/examples/boot/application/stm32l0/src/bin/b.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::gpio::{Level, Output, Speed};
diff --git a/examples/boot/application/stm32l1/Cargo.toml b/examples/boot/application/stm32l1/Cargo.toml
index 3e964df9c..c07d71591 100644
--- a/examples/boot/application/stm32l1/Cargo.toml
+++ b/examples/boot/application/stm32l1/Cargo.toml
@@ -23,6 +23,7 @@ cortex-m-rt = "0.7.0"
 [features]
 defmt = [
       "dep:defmt",
+      "dep:defmt-rtt",
       "embassy-stm32/defmt",
       "embassy-boot-stm32/defmt",
       "embassy-sync/defmt",
diff --git a/examples/boot/application/stm32l1/src/bin/a.rs b/examples/boot/application/stm32l1/src/bin/a.rs
index f066c1139..dcc10e5c6 100644
--- a/examples/boot/application/stm32l1/src/bin/a.rs
+++ b/examples/boot/application/stm32l1/src/bin/a.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_boot_stm32::{AlignedBuffer, FirmwareUpdater, FirmwareUpdaterConfig};
 use embassy_embedded_hal::adapter::BlockingAsync;
diff --git a/examples/boot/application/stm32l1/src/bin/b.rs b/examples/boot/application/stm32l1/src/bin/b.rs
index 6bf00f41a..a59c6f540 100644
--- a/examples/boot/application/stm32l1/src/bin/b.rs
+++ b/examples/boot/application/stm32l1/src/bin/b.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::gpio::{Level, Output, Speed};
diff --git a/examples/boot/application/stm32l4/Cargo.toml b/examples/boot/application/stm32l4/Cargo.toml
index b154403ac..72b937ca7 100644
--- a/examples/boot/application/stm32l4/Cargo.toml
+++ b/examples/boot/application/stm32l4/Cargo.toml
@@ -23,6 +23,7 @@ cortex-m-rt = "0.7.0"
 [features]
 defmt = [
       "dep:defmt",
+      "dep:defmt-rtt",
       "embassy-stm32/defmt",
       "embassy-boot-stm32/defmt",
       "embassy-sync/defmt",
diff --git a/examples/boot/application/stm32l4/src/bin/a.rs b/examples/boot/application/stm32l4/src/bin/a.rs
index a0079ee33..7f8015c04 100644
--- a/examples/boot/application/stm32l4/src/bin/a.rs
+++ b/examples/boot/application/stm32l4/src/bin/a.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_boot_stm32::{AlignedBuffer, FirmwareUpdater, FirmwareUpdaterConfig};
 use embassy_embedded_hal::adapter::BlockingAsync;
diff --git a/examples/boot/application/stm32l4/src/bin/b.rs b/examples/boot/application/stm32l4/src/bin/b.rs
index 22ba82d5e..b1a505631 100644
--- a/examples/boot/application/stm32l4/src/bin/b.rs
+++ b/examples/boot/application/stm32l4/src/bin/b.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::gpio::{Level, Output, Speed};
diff --git a/examples/boot/application/stm32wb-dfu/src/main.rs b/examples/boot/application/stm32wb-dfu/src/main.rs
index 929d6802c..0ab99ff90 100644
--- a/examples/boot/application/stm32wb-dfu/src/main.rs
+++ b/examples/boot/application/stm32wb-dfu/src/main.rs
@@ -3,7 +3,7 @@
 
 use core::cell::RefCell;
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_boot_stm32::{AlignedBuffer, BlockingFirmwareState, FirmwareUpdaterConfig};
 use embassy_executor::Spawner;
diff --git a/examples/boot/application/stm32wl/Cargo.toml b/examples/boot/application/stm32wl/Cargo.toml
index 93ead617c..a5160b797 100644
--- a/examples/boot/application/stm32wl/Cargo.toml
+++ b/examples/boot/application/stm32wl/Cargo.toml
@@ -23,6 +23,7 @@ cortex-m-rt = "0.7.0"
 [features]
 defmt = [
       "dep:defmt",
+      "dep:defmt-rtt",
       "embassy-stm32/defmt",
       "embassy-boot-stm32/defmt",
       "embassy-sync/defmt",
diff --git a/examples/boot/application/stm32wl/src/bin/a.rs b/examples/boot/application/stm32wl/src/bin/a.rs
index 2fb16bdc4..9f4f0b238 100644
--- a/examples/boot/application/stm32wl/src/bin/a.rs
+++ b/examples/boot/application/stm32wl/src/bin/a.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_boot_stm32::{AlignedBuffer, FirmwareUpdater, FirmwareUpdaterConfig};
 use embassy_embedded_hal::adapter::BlockingAsync;
diff --git a/examples/boot/application/stm32wl/src/bin/b.rs b/examples/boot/application/stm32wl/src/bin/b.rs
index 8dd15d8cd..e954d8b91 100644
--- a/examples/boot/application/stm32wl/src/bin/b.rs
+++ b/examples/boot/application/stm32wl/src/bin/b.rs
@@ -1,7 +1,7 @@
 #![no_std]
 #![no_main]
 
-#[cfg(feature = "defmt-rtt")]
+#[cfg(feature = "defmt")]
 use defmt_rtt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::gpio::{Level, Output, Speed};
diff --git a/examples/boot/bootloader/nrf/Cargo.toml b/examples/boot/bootloader/nrf/Cargo.toml
index 980149bea..9d5d51a13 100644
--- a/examples/boot/bootloader/nrf/Cargo.toml
+++ b/examples/boot/bootloader/nrf/Cargo.toml
@@ -19,13 +19,13 @@ cfg-if = "1.0.0"
 [features]
 defmt = [
     "dep:defmt",
+    "dep:defmt-rtt",
     "embassy-boot-nrf/defmt",
     "embassy-nrf/defmt",
 ]
 softdevice = [
     "embassy-boot-nrf/softdevice",
 ]
-debug = ["defmt-rtt", "defmt"]
 
 [profile.dev]
 debug = 2
diff --git a/examples/boot/bootloader/rp/Cargo.toml b/examples/boot/bootloader/rp/Cargo.toml
index 7eec3df1b..c15c980ca 100644
--- a/examples/boot/bootloader/rp/Cargo.toml
+++ b/examples/boot/bootloader/rp/Cargo.toml
@@ -23,10 +23,10 @@ cfg-if = "1.0.0"
 [features]
 defmt = [
     "dep:defmt",
+    "dep:defmt-rtt",
     "embassy-boot-rp/defmt",
     "embassy-rp/defmt",
 ]
-debug = ["defmt-rtt", "defmt"]
 
 [profile.release]
 debug = true
diff --git a/examples/boot/bootloader/stm32-dual-bank/Cargo.toml b/examples/boot/bootloader/stm32-dual-bank/Cargo.toml
index 55adf84d7..b91b05412 100644
--- a/examples/boot/bootloader/stm32-dual-bank/Cargo.toml
+++ b/examples/boot/bootloader/stm32-dual-bank/Cargo.toml
@@ -22,8 +22,7 @@ embedded-storage-async = "0.4.0"
 cfg-if = "1.0.0"
 
 [features]
-defmt = ["dep:defmt", "embassy-boot-stm32/defmt", "embassy-stm32/defmt"]
-debug = ["defmt-rtt", "defmt"]
+defmt = ["dep:defmt", "dep:defmt-rtt", "embassy-boot-stm32/defmt", "embassy-stm32/defmt"]
 
 [profile.dev]
 debug = 2
diff --git a/examples/boot/bootloader/stm32/Cargo.toml b/examples/boot/bootloader/stm32/Cargo.toml
index ef2b99404..541186949 100644
--- a/examples/boot/bootloader/stm32/Cargo.toml
+++ b/examples/boot/bootloader/stm32/Cargo.toml
@@ -21,10 +21,10 @@ cfg-if = "1.0.0"
 [features]
 defmt = [
     "dep:defmt",
+    "dep:defmt-rtt",
     "embassy-boot-stm32/defmt",
     "embassy-stm32/defmt",
 ]
-debug = ["defmt-rtt", "defmt"]
 
 [profile.dev]
 debug = 2
diff --git a/examples/boot/bootloader/stm32wb-dfu/Cargo.toml b/examples/boot/bootloader/stm32wb-dfu/Cargo.toml
index 93b5d8b34..9950ed7b6 100644
--- a/examples/boot/bootloader/stm32wb-dfu/Cargo.toml
+++ b/examples/boot/bootloader/stm32wb-dfu/Cargo.toml
@@ -24,12 +24,12 @@ embassy-futures = { version = "0.1.1", path = "../../../../embassy-futures" }
 [features]
 defmt = [
     "dep:defmt",
+    "dep:defmt-rtt",
     "embassy-boot-stm32/defmt",
     "embassy-stm32/defmt",
     "embassy-usb/defmt",
     "embassy-usb-dfu/defmt"
 ]
-debug = ["defmt-rtt", "defmt"]
 
 [profile.dev]
 debug = 2
diff --git a/examples/rp/src/bin/zerocopy.rs b/examples/rp/src/bin/zerocopy.rs
new file mode 100644
index 000000000..39f03c8e4
--- /dev/null
+++ b/examples/rp/src/bin/zerocopy.rs
@@ -0,0 +1,94 @@
+//! This example shows how to use `zerocopy_channel` from `embassy_sync` for
+//! sending large values between two tasks without copying.
+//! The example also shows how to use the RP2040 ADC with DMA.
+#![no_std]
+#![no_main]
+
+use core::sync::atomic::{AtomicU16, Ordering};
+
+use defmt::*;
+use embassy_executor::Spawner;
+use embassy_rp::adc::{self, Adc, Async, Config, InterruptHandler};
+use embassy_rp::bind_interrupts;
+use embassy_rp::gpio::Pull;
+use embassy_rp::peripherals::DMA_CH0;
+use embassy_sync::blocking_mutex::raw::NoopRawMutex;
+use embassy_sync::zerocopy_channel::{Channel, Receiver, Sender};
+use embassy_time::{Duration, Ticker, Timer};
+use static_cell::StaticCell;
+use {defmt_rtt as _, panic_probe as _};
+
+type SampleBuffer = [u16; 512];
+
+bind_interrupts!(struct Irqs {
+    ADC_IRQ_FIFO => InterruptHandler;
+});
+
+const BLOCK_SIZE: usize = 512;
+const NUM_BLOCKS: usize = 2;
+static MAX: AtomicU16 = AtomicU16::new(0);
+
+struct AdcParts {
+    adc: Adc<'static, Async>,
+    pin: adc::Channel<'static>,
+    dma: DMA_CH0,
+}
+
+#[embassy_executor::main]
+async fn main(spawner: Spawner) {
+    let p = embassy_rp::init(Default::default());
+    info!("Here we go!");
+
+    let adc_parts = AdcParts {
+        adc: Adc::new(p.ADC, Irqs, Config::default()),
+        pin: adc::Channel::new_pin(p.PIN_29, Pull::None),
+        dma: p.DMA_CH0,
+    };
+
+    static BUF: StaticCell<[SampleBuffer; NUM_BLOCKS]> = StaticCell::new();
+    let buf = BUF.init([[0; BLOCK_SIZE]; NUM_BLOCKS]);
+
+    static CHANNEL: StaticCell<Channel<'_, NoopRawMutex, SampleBuffer>> = StaticCell::new();
+    let channel = CHANNEL.init(Channel::new(buf));
+    let (sender, receiver) = channel.split();
+
+    spawner.must_spawn(consumer(receiver));
+    spawner.must_spawn(producer(sender, adc_parts));
+
+    let mut ticker = Ticker::every(Duration::from_secs(1));
+    loop {
+        ticker.next().await;
+        let max = MAX.load(Ordering::Relaxed);
+        info!("latest block's max value: {:?}", max);
+    }
+}
+
+#[embassy_executor::task]
+async fn producer(mut sender: Sender<'static, NoopRawMutex, SampleBuffer>, mut adc: AdcParts) {
+    loop {
+        // Obtain a free buffer from the channel
+        let buf = sender.send().await;
+
+        // Fill it with data
+        adc.adc.read_many(&mut adc.pin, buf, 1, &mut adc.dma).await.unwrap();
+
+        // Notify the channel that the buffer is now ready to be received
+        sender.send_done();
+    }
+}
+
+#[embassy_executor::task]
+async fn consumer(mut receiver: Receiver<'static, NoopRawMutex, SampleBuffer>) {
+    loop {
+        // Receive a buffer from the channel
+        let buf = receiver.receive().await;
+
+        // Simulate using the data, while the producer is filling up the next buffer
+        Timer::after_micros(1000).await;
+        let max = buf.iter().max().unwrap();
+        MAX.store(*max, Ordering::Relaxed);
+
+        // Notify the channel that the buffer is now ready to be reused
+        receiver.receive_done();
+    }
+}
diff --git a/examples/stm32f1/src/bin/input_capture.rs b/examples/stm32f1/src/bin/input_capture.rs
new file mode 100644
index 000000000..5e2dab9e6
--- /dev/null
+++ b/examples/stm32f1/src/bin/input_capture.rs
@@ -0,0 +1,52 @@
+#![no_std]
+#![no_main]
+
+use defmt::*;
+use embassy_executor::Spawner;
+use embassy_stm32::gpio::{Level, Output, Pull, Speed};
+use embassy_stm32::time::khz;
+use embassy_stm32::timer::input_capture::{CapturePin, InputCapture};
+use embassy_stm32::timer::{self, Channel};
+use embassy_stm32::{bind_interrupts, peripherals};
+use embassy_time::Timer;
+use {defmt_rtt as _, panic_probe as _};
+
+/// Connect PA2 and PC13 with a 1k Ohm resistor
+
+#[embassy_executor::task]
+async fn blinky(led: peripherals::PC13) {
+    let mut led = Output::new(led, Level::High, Speed::Low);
+
+    loop {
+        info!("high");
+        led.set_high();
+        Timer::after_millis(300).await;
+
+        info!("low");
+        led.set_low();
+        Timer::after_millis(300).await;
+    }
+}
+
+bind_interrupts!(struct Irqs {
+    TIM2 => timer::CaptureCompareInterruptHandler<peripherals::TIM2>;
+});
+
+#[embassy_executor::main]
+async fn main(spawner: Spawner) {
+    let p = embassy_stm32::init(Default::default());
+    info!("Hello World!");
+
+    unwrap!(spawner.spawn(blinky(p.PC13)));
+
+    let ch3 = CapturePin::new_ch3(p.PA2, Pull::None);
+    let mut ic = InputCapture::new(p.TIM2, None, None, Some(ch3), None, Irqs, khz(1000), Default::default());
+
+    loop {
+        info!("wait for rising edge");
+        ic.wait_for_rising_edge(Channel::Ch3).await;
+
+        let capture_value = ic.get_capture_value(Channel::Ch3);
+        info!("new capture! {}", capture_value);
+    }
+}
diff --git a/examples/stm32f1/src/bin/pwm_input.rs b/examples/stm32f1/src/bin/pwm_input.rs
new file mode 100644
index 000000000..f74853d4e
--- /dev/null
+++ b/examples/stm32f1/src/bin/pwm_input.rs
@@ -0,0 +1,54 @@
+#![no_std]
+#![no_main]
+
+use defmt::*;
+use embassy_executor::Spawner;
+use embassy_stm32::gpio::{Level, Output, Pull, Speed};
+use embassy_stm32::time::khz;
+use embassy_stm32::timer::pwm_input::PwmInput;
+use embassy_stm32::{bind_interrupts, peripherals, timer};
+use embassy_time::Timer;
+use {defmt_rtt as _, panic_probe as _};
+
+/// Connect PA0 and PC13 with a 1k Ohm resistor
+
+#[embassy_executor::task]
+async fn blinky(led: peripherals::PC13) {
+    let mut led = Output::new(led, Level::High, Speed::Low);
+
+    loop {
+        info!("high");
+        led.set_high();
+        Timer::after_millis(300).await;
+
+        info!("low");
+        led.set_low();
+        Timer::after_millis(300).await;
+    }
+}
+
+bind_interrupts!(struct Irqs {
+    TIM2 => timer::CaptureCompareInterruptHandler<peripherals::TIM2>;
+});
+
+#[embassy_executor::main]
+async fn main(spawner: Spawner) {
+    let p = embassy_stm32::init(Default::default());
+    info!("Hello World!");
+
+    unwrap!(spawner.spawn(blinky(p.PC13)));
+
+    let mut pwm_input = PwmInput::new(p.TIM2, p.PA0, Pull::None, khz(10));
+    pwm_input.enable();
+
+    loop {
+        Timer::after_millis(500).await;
+        let period = pwm_input.get_period_ticks();
+        let width = pwm_input.get_width_ticks();
+        let duty_cycle = pwm_input.get_duty_cycle();
+        info!(
+            "period ticks: {} width ticks: {} duty cycle: {}",
+            period, width, duty_cycle
+        );
+    }
+}
diff --git a/examples/stm32f4/src/bin/i2s_dma.rs b/examples/stm32f4/src/bin/i2s_dma.rs
index 97a04b2aa..27b165f1b 100644
--- a/examples/stm32f4/src/bin/i2s_dma.rs
+++ b/examples/stm32f4/src/bin/i2s_dma.rs
@@ -15,14 +15,13 @@ async fn main(_spawner: Spawner) {
     let p = embassy_stm32::init(Default::default());
     info!("Hello World!");
 
-    let mut i2s = I2S::new(
+    let mut i2s = I2S::new_txonly(
         p.SPI2,
         p.PC3,  // sd
         p.PB12, // ws
         p.PB10, // ck
         p.PC6,  // mck
         p.DMA1_CH4,
-        p.DMA1_CH3,
         Hertz(1_000_000),
         Config::default(),
     );
diff --git a/examples/stm32f4/src/bin/pwm_input.rs b/examples/stm32f4/src/bin/pwm_input.rs
new file mode 100644
index 000000000..ce200549d
--- /dev/null
+++ b/examples/stm32f4/src/bin/pwm_input.rs
@@ -0,0 +1,54 @@
+#![no_std]
+#![no_main]
+
+use defmt::*;
+use embassy_executor::Spawner;
+use embassy_stm32::gpio::{Level, Output, Pull, Speed};
+use embassy_stm32::time::khz;
+use embassy_stm32::timer::pwm_input::PwmInput;
+use embassy_stm32::{bind_interrupts, peripherals, timer};
+use embassy_time::Timer;
+use {defmt_rtt as _, panic_probe as _};
+
+/// Connect PB2 and PA6 with a 1k Ohm resistor
+
+#[embassy_executor::task]
+async fn blinky(led: peripherals::PB2) {
+    let mut led = Output::new(led, Level::High, Speed::Low);
+
+    loop {
+        info!("high");
+        led.set_high();
+        Timer::after_millis(300).await;
+
+        info!("low");
+        led.set_low();
+        Timer::after_millis(300).await;
+    }
+}
+
+bind_interrupts!(struct Irqs {
+    TIM2 => timer::CaptureCompareInterruptHandler<peripherals::TIM2>;
+});
+
+#[embassy_executor::main]
+async fn main(spawner: Spawner) {
+    let p = embassy_stm32::init(Default::default());
+    info!("Hello World!");
+
+    unwrap!(spawner.spawn(blinky(p.PB2)));
+
+    let mut pwm_input = PwmInput::new(p.TIM3, p.PA6, Pull::None, khz(10));
+    pwm_input.enable();
+
+    loop {
+        Timer::after_millis(500).await;
+        let period = pwm_input.get_period_ticks();
+        let width = pwm_input.get_width_ticks();
+        let duty_cycle = pwm_input.get_duty_cycle();
+        info!(
+            "period ticks: {} width ticks: {} duty cycle: {}",
+            period, width, duty_cycle
+        );
+    }
+}
diff --git a/examples/stm32f4/src/bin/usb_ethernet.rs b/examples/stm32f4/src/bin/usb_ethernet.rs
index 19ae16e8b..b398c35da 100644
--- a/examples/stm32f4/src/bin/usb_ethernet.rs
+++ b/examples/stm32f4/src/bin/usb_ethernet.rs
@@ -43,7 +43,7 @@ bind_interrupts!(struct Irqs {
 // If you are trying this and your USB device doesn't connect, the most
 // common issues are the RCC config and vbus_detection
 //
-// See https://embassy.dev/book/dev/faq.html#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
+// See https://embassy.dev/book/#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
 // for more information.
 #[embassy_executor::main]
 async fn main(spawner: Spawner) {
diff --git a/examples/stm32f4/src/bin/usb_hid_keyboard.rs b/examples/stm32f4/src/bin/usb_hid_keyboard.rs
index 537ff63ea..1270995c4 100644
--- a/examples/stm32f4/src/bin/usb_hid_keyboard.rs
+++ b/examples/stm32f4/src/bin/usb_hid_keyboard.rs
@@ -24,7 +24,7 @@ bind_interrupts!(struct Irqs {
 // If you are trying this and your USB device doesn't connect, the most
 // common issues are the RCC config and vbus_detection
 //
-// See https://embassy.dev/book/dev/faq.html#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
+// See https://embassy.dev/book/#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
 // for more information.
 #[embassy_executor::main]
 async fn main(_spawner: Spawner) {
diff --git a/examples/stm32f4/src/bin/usb_hid_mouse.rs b/examples/stm32f4/src/bin/usb_hid_mouse.rs
index df4b7426c..45136f965 100644
--- a/examples/stm32f4/src/bin/usb_hid_mouse.rs
+++ b/examples/stm32f4/src/bin/usb_hid_mouse.rs
@@ -21,7 +21,7 @@ bind_interrupts!(struct Irqs {
 // If you are trying this and your USB device doesn't connect, the most
 // common issues are the RCC config and vbus_detection
 //
-// See https://embassy.dev/book/dev/faq.html#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
+// See https://embassy.dev/book/#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
 // for more information.
 #[embassy_executor::main]
 async fn main(_spawner: Spawner) {
diff --git a/examples/stm32f4/src/bin/usb_raw.rs b/examples/stm32f4/src/bin/usb_raw.rs
index 1452e7c5f..b2d706208 100644
--- a/examples/stm32f4/src/bin/usb_raw.rs
+++ b/examples/stm32f4/src/bin/usb_raw.rs
@@ -72,7 +72,7 @@ bind_interrupts!(struct Irqs {
 // If you are trying this and your USB device doesn't connect, the most
 // common issues are the RCC config and vbus_detection
 //
-// See https://embassy.dev/book/dev/faq.html#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
+// See https://embassy.dev/book/#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
 // for more information.
 #[embassy_executor::main]
 async fn main(_spawner: Spawner) {
diff --git a/examples/stm32f4/src/bin/usb_serial.rs b/examples/stm32f4/src/bin/usb_serial.rs
index b2bd390b6..328b5effe 100644
--- a/examples/stm32f4/src/bin/usb_serial.rs
+++ b/examples/stm32f4/src/bin/usb_serial.rs
@@ -19,7 +19,7 @@ bind_interrupts!(struct Irqs {
 // If you are trying this and your USB device doesn't connect, the most
 // common issues are the RCC config and vbus_detection
 //
-// See https://embassy.dev/book/dev/faq.html#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
+// See https://embassy.dev/book/#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
 // for more information.
 #[embassy_executor::main]
 async fn main(_spawner: Spawner) {
diff --git a/examples/stm32f7/src/bin/can.rs b/examples/stm32f7/src/bin/can.rs
index e32b4d3df..a82e335a9 100644
--- a/examples/stm32f7/src/bin/can.rs
+++ b/examples/stm32f7/src/bin/can.rs
@@ -24,7 +24,7 @@ bind_interrupts!(struct Irqs {
 });
 
 #[embassy_executor::task]
-pub async fn send_can_message(tx: &'static mut CanTx<'static, CAN3>) {
+pub async fn send_can_message(tx: &'static mut CanTx<'static>) {
     loop {
         let frame = Frame::new_data(unwrap!(StandardId::new(0 as _)), &[0]).unwrap();
         tx.write(&frame).await;
@@ -45,7 +45,7 @@ async fn main(spawner: Spawner) {
     let rx_pin = Input::new(&mut p.PA15, Pull::Up);
     core::mem::forget(rx_pin);
 
-    static CAN: StaticCell<Can<'static, CAN3>> = StaticCell::new();
+    static CAN: StaticCell<Can<'static>> = StaticCell::new();
     let can = CAN.init(Can::new(p.CAN3, p.PA8, p.PA15, Irqs));
     can.modify_filters().enable_bank(0, Fifo::Fifo0, Mask32::accept_all());
 
@@ -62,7 +62,7 @@ async fn main(spawner: Spawner) {
 
     let (tx, mut rx) = can.split();
 
-    static CAN_TX: StaticCell<CanTx<'static, CAN3>> = StaticCell::new();
+    static CAN_TX: StaticCell<CanTx<'static>> = StaticCell::new();
     let tx = CAN_TX.init(tx);
     spawner.spawn(send_can_message(tx)).unwrap();
 
diff --git a/examples/stm32f7/src/bin/usb_serial.rs b/examples/stm32f7/src/bin/usb_serial.rs
index 0e5cc7c5c..1906b28ed 100644
--- a/examples/stm32f7/src/bin/usb_serial.rs
+++ b/examples/stm32f7/src/bin/usb_serial.rs
@@ -19,7 +19,7 @@ bind_interrupts!(struct Irqs {
 // If you are trying this and your USB device doesn't connect, the most
 // common issues are the RCC config and vbus_detection
 //
-// See https://embassy.dev/book/dev/faq.html#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
+// See https://embassy.dev/book/#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
 // for more information.
 #[embassy_executor::main]
 async fn main(_spawner: Spawner) {
diff --git a/examples/stm32g0/src/bin/spi_neopixel.rs b/examples/stm32g0/src/bin/spi_neopixel.rs
index 2deee271d..edcae74f7 100644
--- a/examples/stm32g0/src/bin/spi_neopixel.rs
+++ b/examples/stm32g0/src/bin/spi_neopixel.rs
@@ -76,7 +76,7 @@ async fn main(_spawner: Spawner) {
 
     let mut config = Config::default();
     config.frequency = Hertz(4_000_000);
-    let mut spi = Spi::new_txonly_nosck(p.SPI1, p.PB5, p.DMA1_CH3, config);
+    let mut spi = Spi::new_txonly(p.SPI1, p.PB3, p.PB5, p.DMA1_CH3, config); // SCK is unused.
 
     let mut neopixels = Ws2812::new();
 
diff --git a/examples/stm32h7/src/bin/usb_serial.rs b/examples/stm32h7/src/bin/usb_serial.rs
index 1c50fc1c8..65ae597d4 100644
--- a/examples/stm32h7/src/bin/usb_serial.rs
+++ b/examples/stm32h7/src/bin/usb_serial.rs
@@ -18,7 +18,7 @@ bind_interrupts!(struct Irqs {
 // If you are trying this and your USB device doesn't connect, the most
 // common issues are the RCC config and vbus_detection
 //
-// See https://embassy.dev/book/dev/faq.html#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
+// See https://embassy.dev/book/#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
 // for more information.
 #[embassy_executor::main]
 async fn main(_spawner: Spawner) {
diff --git a/examples/stm32l4/src/bin/usb_serial.rs b/examples/stm32l4/src/bin/usb_serial.rs
index ed9671d0f..c3b1211d8 100644
--- a/examples/stm32l4/src/bin/usb_serial.rs
+++ b/examples/stm32l4/src/bin/usb_serial.rs
@@ -19,7 +19,7 @@ bind_interrupts!(struct Irqs {
 // If you are trying this and your USB device doesn't connect, the most
 // common issues are the RCC config and vbus_detection
 //
-// See https://embassy.dev/book/dev/faq.html#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
+// See https://embassy.dev/book/#_the_usb_examples_are_not_working_on_my_board_is_there_anything_else_i_need_to_configure
 // for more information.
 #[embassy_executor::main]
 async fn main(_spawner: Spawner) {
diff --git a/rust-toolchain-nightly.toml b/rust-toolchain-nightly.toml
index ac160b995..2143c1b2d 100644
--- a/rust-toolchain-nightly.toml
+++ b/rust-toolchain-nightly.toml
@@ -1,5 +1,5 @@
 [toolchain]
-channel = "nightly-2024-04-14"
+channel = "nightly-2024-05-20"
 components = [ "rust-src", "rustfmt", "llvm-tools", "miri" ]
 targets = [
     "thumbv7em-none-eabi",
diff --git a/tests/stm32/Cargo.toml b/tests/stm32/Cargo.toml
index f6d1d98eb..de390ee38 100644
--- a/tests/stm32/Cargo.toml
+++ b/tests/stm32/Cargo.toml
@@ -7,34 +7,36 @@ autobins = false
 
 [features]
 stm32c031c6 = ["embassy-stm32/stm32c031c6", "cm0", "not-gpdma"]
-stm32f103c8 = ["embassy-stm32/stm32f103c8", "not-gpdma"]
-stm32f207zg = ["embassy-stm32/stm32f207zg", "chrono", "not-gpdma", "eth", "rng"]
+stm32f103c8 = ["embassy-stm32/stm32f103c8", "spi-v1", "not-gpdma"]
+stm32f207zg = ["embassy-stm32/stm32f207zg", "spi-v1", "chrono", "not-gpdma", "eth", "rng"]
 stm32f303ze = ["embassy-stm32/stm32f303ze", "chrono", "not-gpdma"]
-stm32f429zi = ["embassy-stm32/stm32f429zi", "chrono", "eth", "stop", "can", "not-gpdma", "dac", "rng"]
-stm32f446re = ["embassy-stm32/stm32f446re", "chrono", "stop", "can", "not-gpdma", "dac", "sdmmc"]
+stm32f429zi = ["embassy-stm32/stm32f429zi", "spi-v1", "chrono", "eth", "stop", "can", "not-gpdma", "dac", "rng"]
+stm32f446re = ["embassy-stm32/stm32f446re", "spi-v1", "chrono", "stop", "can", "not-gpdma", "dac", "sdmmc"]
 stm32f767zi = ["embassy-stm32/stm32f767zi", "chrono", "not-gpdma", "eth", "rng"]
 stm32g071rb = ["embassy-stm32/stm32g071rb", "cm0", "not-gpdma", "dac", "ucpd"]
 stm32g491re = ["embassy-stm32/stm32g491re", "chrono", "stop", "not-gpdma", "rng", "fdcan", "cordic"]
-stm32h563zi = ["embassy-stm32/stm32h563zi", "chrono", "eth", "rng", "fdcan", "hash", "cordic", "stop"]
-stm32h753zi = ["embassy-stm32/stm32h753zi", "chrono", "not-gpdma", "eth", "rng", "fdcan", "hash", "cryp"]
-stm32h755zi = ["embassy-stm32/stm32h755zi-cm7", "chrono", "not-gpdma", "eth", "dac", "rng", "fdcan", "hash", "cryp"]
-stm32h7a3zi = ["embassy-stm32/stm32h7a3zi", "not-gpdma", "rng", "fdcan"]
+stm32h563zi = ["embassy-stm32/stm32h563zi", "spi-v345", "chrono", "eth", "rng", "fdcan", "hash", "cordic", "stop"]
+stm32h753zi = ["embassy-stm32/stm32h753zi", "spi-v345", "chrono", "not-gpdma", "eth", "rng", "fdcan", "hash", "cryp"]
+stm32h755zi = ["embassy-stm32/stm32h755zi-cm7", "spi-v345", "chrono", "not-gpdma", "eth", "dac", "rng", "fdcan", "hash", "cryp"]
+stm32h7a3zi = ["embassy-stm32/stm32h7a3zi", "spi-v345", "not-gpdma", "rng", "fdcan"]
 stm32l073rz = ["embassy-stm32/stm32l073rz", "cm0", "not-gpdma", "rng"]
-stm32l152re = ["embassy-stm32/stm32l152re", "chrono", "not-gpdma"]
+stm32l152re = ["embassy-stm32/stm32l152re", "spi-v1", "chrono", "not-gpdma"]
 stm32l496zg = ["embassy-stm32/stm32l496zg", "not-gpdma", "rng"]
 stm32l4a6zg = ["embassy-stm32/stm32l4a6zg", "chrono", "not-gpdma", "rng", "hash"]
 stm32l4r5zi = ["embassy-stm32/stm32l4r5zi", "chrono", "not-gpdma", "rng"]
 stm32l552ze = ["embassy-stm32/stm32l552ze", "not-gpdma", "rng", "hash"]
-stm32u585ai = ["embassy-stm32/stm32u585ai", "chrono", "rng", "hash", "cordic"]
-stm32u5a5zj = ["embassy-stm32/stm32u5a5zj", "chrono", "rng", "hash"] # FIXME: cordic test cause it crash
+stm32u585ai = ["embassy-stm32/stm32u585ai", "spi-v345", "chrono", "rng", "hash", "cordic"]
+stm32u5a5zj = ["embassy-stm32/stm32u5a5zj", "spi-v345", "chrono", "rng", "hash"] # FIXME: cordic test cause it crash
 stm32wb55rg = ["embassy-stm32/stm32wb55rg", "chrono", "not-gpdma", "ble", "mac" , "rng"]
-stm32wba52cg = ["embassy-stm32/stm32wba52cg", "chrono", "rng", "hash"]
+stm32wba52cg = ["embassy-stm32/stm32wba52cg", "spi-v345", "chrono", "rng", "hash"]
 stm32wl55jc = ["embassy-stm32/stm32wl55jc-cm4", "not-gpdma", "rng", "chrono"]
 stm32f091rc = ["embassy-stm32/stm32f091rc", "cm0", "not-gpdma", "chrono"]
-stm32h503rb = ["embassy-stm32/stm32h503rb", "rng", "stop"]
-stm32h7s3l8 = ["embassy-stm32/stm32h7s3l8", "rng", "cordic", "hash"] # TODO: fdcan crashes, cryp dma hangs.
+stm32h503rb = ["embassy-stm32/stm32h503rb", "spi-v345", "rng", "stop"]
+stm32h7s3l8 = ["embassy-stm32/stm32h7s3l8", "spi-v345", "rng", "cordic", "hash"] # TODO: fdcan crashes, cryp dma hangs.
 stm32u083rc = ["embassy-stm32/stm32u083rc", "cm0", "rng", "chrono"]
 
+spi-v1 = []
+spi-v345 = []
 cryp = []
 hash = []
 eth = ["embassy-executor/task-arena-size-16384"]
diff --git a/tests/stm32/build.rs b/tests/stm32/build.rs
index 176adff62..675115568 100644
--- a/tests/stm32/build.rs
+++ b/tests/stm32/build.rs
@@ -14,6 +14,7 @@ fn main() -> Result<(), Box<dyn Error>> {
         feature = "stm32c031c6",
         feature = "stm32wb55rg",
         feature = "stm32l073rz",
+        feature = "stm32h503rb",
         // wrong ram size in stm32-data
         feature = "stm32wl55jc",
         feature = "stm32u5a5zj",
diff --git a/tests/stm32/src/bin/can.rs b/tests/stm32/src/bin/can.rs
index 551764458..ba8a33e34 100644
--- a/tests/stm32/src/bin/can.rs
+++ b/tests/stm32/src/bin/can.rs
@@ -9,9 +9,7 @@ use common::*;
 use embassy_executor::Spawner;
 use embassy_stm32::bind_interrupts;
 use embassy_stm32::can::filter::Mask32;
-use embassy_stm32::can::{
-    Can, Fifo, Rx0InterruptHandler, Rx1InterruptHandler, SceInterruptHandler, TxInterruptHandler,
-};
+use embassy_stm32::can::{Fifo, Rx0InterruptHandler, Rx1InterruptHandler, SceInterruptHandler, TxInterruptHandler};
 use embassy_stm32::gpio::{Input, Pull};
 use embassy_stm32::peripherals::CAN1;
 use embassy_time::Duration;
@@ -48,7 +46,7 @@ async fn main(_spawner: Spawner) {
     let rx_pin = Input::new(&mut rx, Pull::Up);
     core::mem::forget(rx_pin);
 
-    let mut can = Can::new(can, rx, tx, Irqs);
+    let mut can = embassy_stm32::can::Can::new(can, rx, tx, Irqs);
 
     info!("Configuring can...");
 
diff --git a/tests/stm32/src/bin/can_common.rs b/tests/stm32/src/bin/can_common.rs
index 4b39269cc..4e1740ad5 100644
--- a/tests/stm32/src/bin/can_common.rs
+++ b/tests/stm32/src/bin/can_common.rs
@@ -8,7 +8,8 @@ pub struct TestOptions {
     pub max_buffered: u8,
 }
 
-pub async fn run_can_tests<'d, T: can::Instance>(can: &mut can::Can<'d, T>, options: &TestOptions) {
+pub async fn run_can_tests<'d>(can: &mut can::Can<'d>, options: &TestOptions) {
+    //pub async fn run_can_tests<'d, T: can::Instance>(can: &mut can::Can<'d, T>, options: &TestOptions) {
     let mut i: u8 = 0;
     loop {
         //let tx_frame = can::frame::Frame::new_standard(0x123, &[i, 0x12 as u8, 0x34 as u8, 0x56 as u8, 0x78 as u8, 0x9A as u8, 0xBC as u8 ]).unwrap();
@@ -79,11 +80,7 @@ pub async fn run_can_tests<'d, T: can::Instance>(can: &mut can::Can<'d, T>, opti
     }
 }
 
-pub async fn run_split_can_tests<'d, T: can::Instance>(
-    tx: &mut can::CanTx<'d, T>,
-    rx: &mut can::CanRx<'d, T>,
-    options: &TestOptions,
-) {
+pub async fn run_split_can_tests<'d>(tx: &mut can::CanTx<'d>, rx: &mut can::CanRx<'d>, options: &TestOptions) {
     for i in 0..options.max_buffered {
         // Try filling up the RX FIFO0 buffers
         //let tx_frame = if 0 != (i & 0x01) {
diff --git a/tests/stm32/src/bin/fdcan.rs b/tests/stm32/src/bin/fdcan.rs
index 20bd3f7e3..bc2b7edd4 100644
--- a/tests/stm32/src/bin/fdcan.rs
+++ b/tests/stm32/src/bin/fdcan.rs
@@ -88,7 +88,7 @@ fn options() -> (Config, TestOptions) {
     )
 }
 
-#[cfg(any(feature = "stm32g491re", feature = "stm32g431cb"))]
+#[cfg(any(feature = "stm32g491re"))]
 fn options() -> (Config, TestOptions) {
     info!("G4 config");
     (
diff --git a/tests/stm32/src/bin/spi.rs b/tests/stm32/src/bin/spi.rs
index c1576bfeb..0ffd0f653 100644
--- a/tests/stm32/src/bin/spi.rs
+++ b/tests/stm32/src/bin/spi.rs
@@ -6,6 +6,7 @@ mod common;
 use common::*;
 use defmt::assert_eq;
 use embassy_executor::Spawner;
+use embassy_stm32::gpio::{Level, Output, Speed};
 use embassy_stm32::spi::{self, Spi};
 use embassy_stm32::time::Hertz;
 
@@ -14,18 +15,19 @@ async fn main(_spawner: Spawner) {
     let p = embassy_stm32::init(config());
     info!("Hello World!");
 
-    let spi = peri!(p, SPI);
-    let sck = peri!(p, SPI_SCK);
-    let mosi = peri!(p, SPI_MOSI);
-    let miso = peri!(p, SPI_MISO);
+    let mut spi_peri = peri!(p, SPI);
+    let mut sck = peri!(p, SPI_SCK);
+    let mut mosi = peri!(p, SPI_MOSI);
+    let mut miso = peri!(p, SPI_MISO);
 
     let mut spi_config = spi::Config::default();
     spi_config.frequency = Hertz(1_000_000);
 
     let mut spi = Spi::new_blocking(
-        spi, sck,  // Arduino D13
-        mosi, // Arduino D11
-        miso, // Arduino D12
+        &mut spi_peri,
+        &mut sck,  // Arduino D13
+        &mut mosi, // Arduino D11
+        &mut miso, // Arduino D12
         spi_config,
     );
 
@@ -60,11 +62,43 @@ async fn main(_spawner: Spawner) {
 
     // Assert the RCC bit gets disabled on drop.
     #[cfg(feature = "stm32f429zi")]
-    {
-        defmt::assert!(embassy_stm32::pac::RCC.apb2enr().read().spi1en());
-        drop(spi);
-        defmt::assert!(!embassy_stm32::pac::RCC.apb2enr().read().spi1en());
-    }
+    defmt::assert!(embassy_stm32::pac::RCC.apb2enr().read().spi1en());
+    drop(spi);
+    #[cfg(feature = "stm32f429zi")]
+    defmt::assert!(!embassy_stm32::pac::RCC.apb2enr().read().spi1en());
+
+    // test rx-only configuration
+    let mut spi = Spi::new_blocking_rxonly(&mut spi_peri, &mut sck, &mut miso, spi_config);
+    let mut mosi_out = Output::new(&mut mosi, Level::Low, Speed::VeryHigh);
+    mosi_out.set_high();
+    spi.blocking_read(&mut buf).unwrap();
+    assert_eq!(buf, [0xff; 9]);
+    mosi_out.set_low();
+    spi.blocking_read(&mut buf).unwrap();
+    assert_eq!(buf, [0x00; 9]);
+    spi.blocking_read::<u8>(&mut []).unwrap();
+    spi.blocking_read::<u8>(&mut []).unwrap();
+    drop(mosi_out);
+    drop(spi);
+
+    // Test tx-only. Just check it doesn't hang, not much else we can do without using SPI slave.
+    let mut spi = Spi::new_blocking_txonly(&mut spi_peri, &mut sck, &mut mosi, spi_config);
+    spi.blocking_transfer(&mut buf, &data).unwrap();
+    spi.blocking_transfer_in_place(&mut buf).unwrap();
+    spi.blocking_write(&buf).unwrap();
+    spi.blocking_read(&mut buf).unwrap();
+    spi.blocking_transfer::<u8>(&mut [], &[]).unwrap();
+    spi.blocking_transfer_in_place::<u8>(&mut []).unwrap();
+    spi.blocking_read::<u8>(&mut []).unwrap();
+    spi.blocking_write::<u8>(&[]).unwrap();
+    drop(spi);
+
+    // Test tx-only nosck.
+    let mut spi = Spi::new_blocking_txonly_nosck(&mut spi_peri, &mut mosi, spi_config);
+    spi.blocking_write(&buf).unwrap();
+    spi.blocking_write::<u8>(&[]).unwrap();
+    spi.blocking_write(&buf).unwrap();
+    drop(spi);
 
     info!("Test OK");
     cortex_m::asm::bkpt();
diff --git a/tests/stm32/src/bin/spi_dma.rs b/tests/stm32/src/bin/spi_dma.rs
index 5d46726dd..fd26d3f71 100644
--- a/tests/stm32/src/bin/spi_dma.rs
+++ b/tests/stm32/src/bin/spi_dma.rs
@@ -6,6 +6,7 @@ mod common;
 use common::*;
 use defmt::assert_eq;
 use embassy_executor::Spawner;
+use embassy_stm32::gpio::{Level, Output, Speed};
 use embassy_stm32::spi::{self, Spi};
 use embassy_stm32::time::Hertz;
 
@@ -14,21 +15,24 @@ async fn main(_spawner: Spawner) {
     let p = embassy_stm32::init(config());
     info!("Hello World!");
 
-    let spi = peri!(p, SPI);
-    let sck = peri!(p, SPI_SCK);
-    let mosi = peri!(p, SPI_MOSI);
-    let miso = peri!(p, SPI_MISO);
-    let tx_dma = peri!(p, SPI_TX_DMA);
-    let rx_dma = peri!(p, SPI_RX_DMA);
+    let mut spi_peri = peri!(p, SPI);
+    let mut sck = peri!(p, SPI_SCK);
+    let mut mosi = peri!(p, SPI_MOSI);
+    let mut miso = peri!(p, SPI_MISO);
+    let mut tx_dma = peri!(p, SPI_TX_DMA);
+    let mut rx_dma = peri!(p, SPI_RX_DMA);
 
     let mut spi_config = spi::Config::default();
     spi_config.frequency = Hertz(1_000_000);
 
     let mut spi = Spi::new(
-        spi, sck,  // Arduino D13
-        mosi, // Arduino D11
-        miso, // Arduino D12
-        tx_dma, rx_dma, spi_config,
+        &mut spi_peri,
+        &mut sck,  // Arduino D13
+        &mut mosi, // Arduino D11
+        &mut miso, // Arduino D12
+        &mut tx_dma,
+        &mut rx_dma,
+        spi_config,
     );
 
     let data: [u8; 9] = [0x00, 0xFF, 0xAA, 0x55, 0xC0, 0xFF, 0xEE, 0xC0, 0xDE];
@@ -59,8 +63,12 @@ async fn main(_spawner: Spawner) {
     spi.transfer_in_place::<u8>(&mut []).await.unwrap();
     spi.read::<u8>(&mut []).await.unwrap();
     spi.write::<u8>(&[]).await.unwrap();
+    spi.blocking_transfer::<u8>(&mut [], &[]).unwrap();
+    spi.blocking_transfer_in_place::<u8>(&mut []).unwrap();
+    spi.blocking_read::<u8>(&mut []).unwrap();
+    spi.blocking_write::<u8>(&[]).unwrap();
 
-    // === Check mixing blocking with async.
+    // Check mixing blocking with async.
     spi.blocking_transfer(&mut buf, &data).unwrap();
     assert_eq!(buf, data);
     spi.transfer(&mut buf, &data).await.unwrap();
@@ -76,6 +84,65 @@ async fn main(_spawner: Spawner) {
     spi.blocking_read(&mut buf).unwrap();
     spi.write(&buf).await.unwrap();
 
+    core::mem::drop(spi);
+
+    // test rx-only configuration
+    let mut spi = Spi::new_rxonly(
+        &mut spi_peri,
+        &mut sck,
+        &mut miso,
+        // SPIv1/f1 requires txdma even if rxonly.
+        #[cfg(not(feature = "spi-v345"))]
+        &mut tx_dma,
+        &mut rx_dma,
+        spi_config,
+    );
+    let mut mosi_out = Output::new(&mut mosi, Level::Low, Speed::VeryHigh);
+    mosi_out.set_high();
+    spi.read(&mut buf).await.unwrap();
+    assert_eq!(buf, [0xff; 9]);
+    spi.blocking_read(&mut buf).unwrap();
+    assert_eq!(buf, [0xff; 9]);
+    spi.read(&mut buf).await.unwrap();
+    assert_eq!(buf, [0xff; 9]);
+    spi.read(&mut buf).await.unwrap();
+    assert_eq!(buf, [0xff; 9]);
+    spi.blocking_read(&mut buf).unwrap();
+    assert_eq!(buf, [0xff; 9]);
+    spi.blocking_read(&mut buf).unwrap();
+    assert_eq!(buf, [0xff; 9]);
+    mosi_out.set_low();
+    spi.read(&mut buf).await.unwrap();
+    assert_eq!(buf, [0x00; 9]);
+    spi.read::<u8>(&mut []).await.unwrap();
+    spi.blocking_read::<u8>(&mut []).unwrap();
+    drop(mosi_out);
+    drop(spi);
+
+    // Test tx-only. Just check it doesn't hang, not much else we can do without using SPI slave.
+    let mut spi = Spi::new_txonly(&mut spi_peri, &mut sck, &mut mosi, &mut tx_dma, spi_config);
+    spi.blocking_write(&buf).unwrap();
+    spi.write(&buf).await.unwrap();
+    spi.blocking_write(&buf).unwrap();
+    spi.blocking_write(&buf).unwrap();
+    spi.write(&buf).await.unwrap();
+    spi.write(&buf).await.unwrap();
+    spi.write::<u8>(&[]).await.unwrap();
+    spi.blocking_write::<u8>(&[]).unwrap();
+    drop(spi);
+
+    // Test tx-only nosck.
+    let mut spi = Spi::new_txonly_nosck(&mut spi_peri, &mut mosi, &mut tx_dma, spi_config);
+    spi.blocking_write(&buf).unwrap();
+    spi.write(&buf).await.unwrap();
+    spi.blocking_write(&buf).unwrap();
+    spi.blocking_write(&buf).unwrap();
+    spi.write(&buf).await.unwrap();
+    spi.write(&buf).await.unwrap();
+    spi.write::<u8>(&[]).await.unwrap();
+    spi.blocking_write::<u8>(&[]).unwrap();
+    drop(spi);
+
     info!("Test OK");
     cortex_m::asm::bkpt();
 }