Repository navigation
memory: remove align to memory bank in APL - #1557
Conversation
|
@abonislawski I don't know if it's coincidence, but both APL configurations are failing. |
|
SOFCI TEST |
Because of "BYT_MB_RT5651 fail" its probably a CI issue but APL fails looks strange |
acc0869 to
fd92b33
Compare
There was a problem hiding this comment.
Why do we align on page size here ?
There was a problem hiding this comment.
@lgirdwood I have started with 32 but there was strange errors in sof-ci/jenkins/pr-device-test: "verify_pcm_list" and "check_aplay_function". So I tried with page size and it works.
@tlauda fyi
There was a problem hiding this comment.
@abonislawski Can we find the root cause? Right now it works, but one change in memory map and we can potentially break it again.
There was a problem hiding this comment.
@tlauda take a look at logs: https://sof-ci.01.org/sofpr/PR1557/build2263/
Im not sure if I can do something with this.
Note that in the past these buffers were always aligned with fixed size so maybe there was a reason for this or just no one has ever tried it in any other way
There was a problem hiding this comment.
@abonislawski We just need FW logs to see what happened.
There was a problem hiding this comment.
@tlauda looks like @lgirdwood advice with CACHE_LINE_SIZE (64) is also working so we will stay with it for this PR and debug later 32 if needed
fd92b33 to
8effa4e
Compare
|
@lgirdwood @tlauda please finish review and merge, this PR is needed for #1567 |
Due to small number of memory banks in APL it will be not possible to power down some of them so there is no point in align to memory banks, this would significantly limit free memory for buffers Signed-off-by: Adrian Bonislawski <adrian.bonislawski@linux.intel.com>
8effa4e to
4c7a3e9
Compare
|
#1568 merged so this one is ready now (used HEAP_BUF_ALIGNMENT) |
Due to small number of memory banks in APL it will be not possible to power down some of them
so there is no point in align to memory banks, this would significantly limit free memory for buffers
Signed-off-by: Adrian Bonislawski adrian.bonislawski@linux.intel.com