Re: [alsa-devel] [PATCH 1/5] ARM: tegra: config: enable dmaengine based APB DMA driver
On 08/17/2012 12:38 AM, Laxman Dewangan wrote:
On Thursday 16 August 2012 11:23 PM, Stephen Warren wrote:
On 08/16/2012 08:13 AM, Laxman Dewangan wrote:
Enable config for dmaengine based Tegra APB DMA driver and disable the legacy APB DMA driver (SYSTEM_DMA).
Laxman, if I apply this series to next-20120816 (plus a few patches in my local work branch plus the CPU hotplug patches from Joseph, although I suspect none of that matters), then audio playback on Tegra20 is broken; the pitch is far too high. Audio playback on Tegra30 works as expected.
I run the test again on ventana and did not see any issue. I play one song and saw similar behavior with/without this series.
OK, I'll test some other boards. I tested Whistler, although all the clocking logic is identical between all Tegra20 boards.
With this series:
ubuntu@tegra-ubuntu:/home$ aplay "1-04 Hungry Like The Wolf.wav" Playing WAVE '1-04 Hungry Like The Wolf.wav' : Signed 16 bit Little Endian, Rate 44100 Hz, Stereo underrun!!! (at least 5039.388 ms long)
Why are you getting all these underruns? I almost never see this, except when something else is hammering the SD card containing the .wav file. If I saw those, I'd count it as a test failure.
On 08/17/2012 09:17 AM, Stephen Warren wrote:
On 08/17/2012 12:38 AM, Laxman Dewangan wrote:
On Thursday 16 August 2012 11:23 PM, Stephen Warren wrote:
On 08/16/2012 08:13 AM, Laxman Dewangan wrote:
Enable config for dmaengine based Tegra APB DMA driver and disable the legacy APB DMA driver (SYSTEM_DMA).
Laxman, if I apply this series to next-20120816 (plus a few patches in my local work branch plus the CPU hotplug patches from Joseph, although I suspect none of that matters), then audio playback on Tegra20 is broken; the pitch is far too high. Audio playback on Tegra30 works as expected.
I run the test again on ventana and did not see any issue. I play one song and saw similar behavior with/without this series.
OK, I'll test some other boards. I tested Whistler, although all the clocking logic is identical between all Tegra20 boards.
I tested both 3.6-rc2 and next-20120816, and they both work fine on Ventana.
I then applied this patch series to both (plus your "dma: tegra: enable/disable dma clock" for 3.6-rc2) and it causes (or at least exposes) the problem.
All testing was with a full git clean, using tegra_defconfig, and with no other patches applied.
If you're not seeing this problem, are you sure you're executing the kernel you think you are, and that it got correctly switched to the new dmaengine driver, and rebuilt after applying your patches?
participants (1)
-
Stephen Warren