check-sof-logger: disable again dma_nudge() workaround for stuck DMA #4333 #880
Conversation
…#4333 Peter has a tentative fix in thesofproject/linux#3523, remove the workaround so it can better tested and evaluated. Signed-off-by: Marc Herbert <[email protected]> (cherry picked from commit 4b3f847)
|
https://sof-ci.01.org/softestpr/PR880/build1015/devicetest/ is empty because CI is completely broken this week. |
|
@marc-hb , the stuck part still going to need workaround since the firmware will only do the copy on dma burst boundaries, it is still possible that we have stalled trace down in rhe firmware and we need to poke it to give it to us. |
How far apart are DMA burst boundaries? When the firmware boots it prints quite a lot of logs, I would say at least 1 kilobytes. Bug #4333 has always been "a lot of logs or nothing". There's a digression in #4333 about "lagging" logs but please ignore it, this basic test is nowhere near smart enough to catch that. |
I take that back, sorry. In daily run 10879?model=TGLH_SKU0A70_HDA&testcase=check-sof-logger (2002-03-09) the FW logs were not empty but they stopped after only 3 lines, just before receiving the first IPC. TGLH_SKU0A70_HDA seems a bit "special", see thesofproject/sof#4333 (comment) |
| @@ -205,7 +205,12 @@ main() | |||
|
|
|||
| # Workaround for DMA trace bug | |||
| # https://github.com/thesofproject/sof/issues/4333 | |||
| if [ "$f" = data ]; then | |||
| # if [ "$f" = data ]; then | |||
There was a problem hiding this comment.
Is this comment intentionally left in? Also not sure I follow the reasoning for the if false; then line. Guessing that's what the comments are referencing.
There was a problem hiding this comment.
This is just a test DRAFT, don't take it seriously.
The if false is a way to comment out 10 lines with a one-line change.
|
SOFCI TEST |
@ujfalusi has a tentative fix in
thesofproject/linux#3523, remove the
workaround so it can better tested and evaluated.
Signed-off-by: Marc Herbert [email protected]
(cherry picked from commit 4b3f847)