regions_mm: Fix vhm allocator behavior on memory freeing #9539
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
softwarecki
requested review from
abonislawski,
gbernatxintel,
jxstelter,
kv2019i,
lgirdwood,
marcinszkudlinski,
serhiy-katsyuba-intel and
tmleman
Platforms based on xtensa have a non-coherent cache between cores. Before releasing a memory block, it is necessary to invalidate the cache. This memory block can be allocated by another core and performing cache writeback by the previous owner will destroy current content of the main memory. Invalidate cache when freeing allocated memory block. Signed-off-by: Adrian Warecki <adrian.warecki@intel.com>
softwarecki
force-pushed
the
vmh-get-size
branch
from
3d8a561 to
30bb812
Compare
lgirdwood
reviewed
Oct 2, 2024
Member
lgirdwood
left a comment
lgirdwood
left a comment
There was a problem hiding this comment.
@softwarecki LGTM (just have an open on naming) , btw do we need to perform any cache operation on the heap region metadata after the free() ?
@andyross pls review.
Collaborator
Author
Metadata is stored in a separate location, not in a unallocated memory. |
The memory block allocated by the allocator may be larger than a requested one. If the allocated block size exceeds the memory page size, only enough pages are mapped to satisfy the request. Thanks to this optimization, it don't map memory pages that will not be used. When freeing memory, the allocator knows only size of the allocated block. Information about the number of pages actually mapped isn't stored anywhere. Before unmapping memory, determine the number of mapped pages to avoid error when trying to unmap unmapped memory. Signed-off-by: Adrian Warecki <adrian.warecki@intel.com>
softwarecki
force-pushed
the
vmh-get-size
branch
from
30bb812 to
516176b
Compare
lgirdwood
approved these changes
Oct 4, 2024
dabekjakub
approved these changes
Oct 7, 2024
tmleman
approved these changes
Oct 7, 2024
Contributor
tmleman
left a comment
tmleman
left a comment
There was a problem hiding this comment.
LGTM, I would only consider whether it's worth using the CONFIG_INCOHERENT config here.
Collaborator
|
Ready to go when CI passes, @softwarecki can you check the Intel internal CI fail? |
Collaborator
Author
|
SOFCI TEST |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Platforms based on xtensa have a non-coherent cache between cores. Before releasing a memory block, it is necessary to invalidate the cache. This memory block can be allocated by another core and performing cache writeback by the previous owner will destroy current content of the main memory.
Invalidate cache when freeing allocated memory block.
The memory block allocated by the allocator may be larger than a requested one. If the allocated block size exceeds the memory page size, only enough pages are mapped to satisfy the request. Thanks to this optimization, it don't map memory pages that will not be used. When freeing memory, the allocator knows only size of the allocated block. Information about the
number of pages actually mapped isn't stored anywhere.
Before unmapping memory, determine the number of mapped pages to avoid error when trying to unmap unmapped memory.