Flush in spi::master::Instance::write before attemting any writes (#1381)

* Flush in `spi::master::Instancee::write` before attemting any writes

Otherwise subsequent calls to `SpiBus::write` will cause corrupted writes
and break `embedded-hal-bus`es transactions. Fixes #1369.

* Update Changelog for #1381

* Fix missing result handling when calling `flush()` in SPI `Instance::write_bytes`

esp-hal/CHANGELOG.md

@@ -30,6 +30,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - USB pullup/pulldown now gets properly cleared and does not interfere anymore on esp32c3 and esp32s3 (#1244)
 - Fixed GPIO counts so that using async code with the higher GPIO number should no longer panic (#1361, #1362)
 - ESP32/ESP32-S2: Wait for I2S getting out of TX_IDLE when starting a transfer (#1375)
+- Fixed writes to SPI not flushing before attemting to write, causing corrupted writes (#1381)
 
 ### Changed
 

esp-hal/src/spi/master.rs

@@ -2618,6 +2618,10 @@ pub trait Instance: crate::private::Sealed {
     fn write_bytes(&mut self, words: &[u8]) -> Result<(), Error> {
         let num_chunks = words.len() / FIFO_SIZE;
 
+        // Flush in case previous writes have not completed yet, required as per
+        // embedded-hal documentation (#1369).
+        self.flush()?;
+
         // The fifo has a limited fixed size, so the data must be chunked and then
         // transmitted
         for (i, chunk) in words.chunks(FIFO_SIZE).enumerate() {