[BUG] If ChainDMA is enabled FW crashes during multiple-pause-resume test #7084
Description
Describe the bug
If ChainDMA is enabled in topology the multiple-pause-resume test fails in CI:
https://sof-ci.01.org/sofpr/PR7073/build3851/devicetest/index.html
To Reproduce
Enable ChainDMA for TGL by reverting c3807ae and run on TGL upx-i11:
TPLG=/lib/firmware/intel/avs-tplg/sof-hda-generic.tplg MODEL=TGLU_UP_HDA_IPC4ZPH ~/work/sof-test/test-case/multiple-pause-resume.sh -r 50
Reproduction Rate
Does not always fail immediately, but sometimes takes over ten seconds. However, the FW always crashes well before the test completes.
Expected behavior
multiple-pause-resume test would pass.
Impact
Can not enable ChainDMA when its broken and no multichannel audio without ChainDMA.
Environment
- Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).
- Name of the topology file
- Topology: {sof-hda-generic.tplg}
- Name of the platform(s) on which the bug is observed.
- Platform: {TGL upx-i11}
Screenshots or console output
Please refer to: https://sof-ci.01.org/sofpr/PR7073/build3851/devicetest/index.html
I have executed the test several times and there is no clear pattern to be seen on what happened on FW logs before the crash. However, from the Linux side logs its visible that the crash always happens while handling ChainDMA messages:
Feb 13 15:35:58 jhary kernel: sof-audio-pci-intel-tgl 0000:00:1f.3: ipc timed out for 0xe040000|0x0
...
Feb 13 16:19:46 jhary kernel: sof-audio-pci-intel-tgl 0000:00:1f.3: ipc timed out for 0xe050101|0x0
...
Feb 13 16:26:02 jhary kernel: sof-audio-pci-intel-tgl 0000:00:1f.3: ipc timed out for 0xe070101|0x0
...
Feb 13 22:13:53 jhary kernel: sof-audio-pci-intel-tgl 0000:00:1f.3: ipc timed out for 0xe050000|0x0