add `itm::write_aligned` for writing 4-byte aligned buffers to an ITM port

LLVM can optimize this better and, for example, unroll the loop into a series of
32-bit writes to the ITM port

3 files changed, 37 insertions, 4 deletions

diff --git a/Cargo.toml b/Cargo.toml
index 540c69c..d030d64 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -10,7 +10,6 @@ repository = "https://github.com/japaric/cortex-m"
 version = "0.2.7"
 
 [dependencies]
-volatile-register = "0.2.0"
-
-[dependencies.cortex-m-semihosting]
-version = "0.1.3"
\ No newline at end of file
+aligned = "0.1.1"
+cortex-m-semihosting = "0.1.3"
+volatile-register = "0.2.0"
\ No newline at end of file
diff --git a/src/itm.rs b/src/itm.rs
index 4c49d14..2aa1499 100644
--- a/src/itm.rs
+++ b/src/itm.rs
@@ -1,6 +1,9 @@
 //! Instrumentation Trace Macrocell
 
 use core::{fmt, ptr, slice};
+
+use aligned::Aligned;
+
 use peripheral::Stim;
 
 fn round_up_to_multiple_of(x: usize, k: usize) -> usize {
@@ -64,6 +67,36 @@ pub fn write_all(port: &Stim, buffer: &[u8]) {
     }
 }
 
+/// Writes a 4-byte aligned `buffer` to the ITM `port`
+pub fn write_aligned(port: &Stim, buffer: &Aligned<u32, [u8]>) {
+    unsafe {
+        let len = buffer.len();
+        let split = len & !0b11;
+        write_words(
+            port,
+            slice::from_raw_parts(buffer.as_ptr() as *const u32, split >> 2),
+        );
+
+        // 3 bytes or less left
+        let mut left = len & 0b11;
+        let mut ptr = buffer.as_ptr().offset(split as isize);
+
+        // at least 2 bytes left
+        if left > 1 {
+            left -= 2;
+            while !port.is_fifo_ready() {}
+            port.write_u16(ptr::read(ptr as *const u16));
+            ptr = ptr.offset(2);
+        }
+
+        // final byte
+        if left == 1 {
+            while !port.is_fifo_ready() {}
+            port.write_u8(*ptr);
+        }
+    }
+}
+
 /// Writes `fmt::Arguments` to the ITM `port`
 pub fn write_fmt(port: &Stim, args: fmt::Arguments) {
     use core::fmt::Write;
diff --git a/src/lib.rs b/src/lib.rs
index 95a1b69..7e9c6fb 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -16,6 +16,7 @@
 #![feature(naked_functions)]
 #![no_std]
 
+extern crate aligned;
 pub extern crate cortex_m_semihosting as semihosting;
 extern crate volatile_register;