Linux 3.15 at91 fixes

[alexandermorozov]

There are several distinct things in this merge request:

1. It brings back audio codec support on at91sama5x EK. It was lost somewhere near migration to 3.6.9. Just reapplying old patches by Bo Shen is not enough now, and several more fixes in DT and wm8904 driver are required. In addition to provided commits, i2c0 and sound DT entries should be manually enabled, since i2c0 uses some common pins with ISI and is codec is disabled by default. Also in current config all DMA0 channels are already used but are strictly required for SSC operation, so we also need to free two DMA channels. After e.g. following extra patch and proper menuconfig settings audio works (tested with mplayer -ao oss), but it's not very stable.

diff --git a/arch/arm/boot/dts/sama5d35ek.dts b/arch/arm/boot/dts/sama5d35ek.dts
index d58d2f4..b84d250 100644
--- a/arch/arm/boot/dts/sama5d35ek.dts
+++ b/arch/arm/boot/dts/sama5d35ek.dts
@@ -28,6 +28,11 @@
                                status = "okay";
                        };

+                       i2c0: i2c@f0014000 {
+                               dmas = <>; /* free two DMA channels for ssc0 */
+                               status = "okay";
+                       };
+
                        i2c1: i2c@f0018000 {
                                status = "okay";
                        };
@@ -58,4 +63,8 @@
                        gpio-key,wakeup;
                };
        };
+
+       sound {
+               status = "okay";
+       };
 };

2. Added several missing DT definitions for various perepherals.

3. Fixes in clock code, UART and SPI drivers. UART fix deals with a rather obscure race condition. We run into problems with it while using high baud rates (>=230400) and "packet"-based protocol. Sometimes several last bytes of the packet were received and placed in ring buffer but not reported to TTY layer until more bytes arrived. In request-response protocols that means request timeout.

[alexandermorozov]

Hi,

I cannot confirm it as I have no hardware and is unfamiliar with this chip family. I've added wm9818 to of_match list because it had been already present in i2c_id list. It was added there by Mark Brown in commit df1553c, here is corresponding commit message:

ASoC: Add basic WM8918 support

The WM8918 is register compatible with the WM8904 with a subset of the
functionality. Add the device ID, a subsequent patch will ensure that only
the relevant functionality is exported to userspace.

Signed-off-by: Mark Brown <[email protected]>
Acked-by: Liam Girdwood <[email protected]>

So as I understand basic functions should be supported.

Best regards,
Alexander

[voiceshen]

Hi Alexander,
Thanks for your information.
Related patches has been sent to mainline mail list.

Best Regards,
Bo Shen

[noglitch]

This old branch only accepts bug fixes. Update like this one can only enter a newer AT91 branch like linux-3.18-at91 or linux-4.1-at91.
Thanks for your patch anyway.
Best regards,