issue-1751: Make behavior for read/write requests with O_DIRECT consistent with Linux behavior for write-back cache #4723
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
e673
commented
Nov 26, 2025
|
|
||
| struct TReadRequest | ||
| : public TRequestBase<fuse_read_in, ui32, ui32> | ||
| : public TRequestBase<fuse_read_in, void, ui32> |
Collaborator
Author
There was a problem hiding this comment.
Странно, но этот код вообще нигде не тестировался.
Операция FUSE_READ не возвращает число прочитанных байт, а сразу содержит буфер переменного размера.
https://github.com/libfuse/libfuse/blob/ab5033f5d1bd1630870dc725eb3eae6ef3ef4859/doc/libfuse-operations.txt#L115
Contributor
neihar
requested changes
Nov 26, 2025
|
|
||
| TFuture<NProto::TReadDataResponse> future; | ||
| if (WriteBackCache) { | ||
| if (useWriteBackCache) { |
Collaborator
There was a problem hiding this comment.
shouldUseServerWriteBackCache or ShouldUseServerWriteBackCache(fi) in place
|
|
||
| bootstrap.Service->ReadDataHandler = [&] (auto, auto) { | ||
| NProto::TReadDataResponse result; | ||
| *result.MutableBuffer() = TString("direct_data"); |
Collaborator
There was a problem hiding this comment.
suggest to use as in other places "result.MutableBuffer()->assign", just for consistency
|
|
||
| { | ||
| // write request without O_DIRECT should go to write-back cache | ||
| auto reqWrite = std::make_shared<TWriteRequest>( |
Collaborator
There was a problem hiding this comment.
can we add counter to writedatahandler and check it explicitly in test?
| reinterpret_cast<const char*>(reqRead->Out->Data()), | ||
| rspRead.GetValue()); | ||
|
|
||
| UNIT_ASSERT_VALUES_EQUAL("cached_data", readData); |
Collaborator
There was a problem hiding this comment.
check counter in handler? blow as well
e673
changed the title
This was referenced Nov 26, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
#1751
Linux behavior
https://github.com/torvalds/linux/blob/30f09200cc4aefbd8385b01e41bde2e4565a6f0e/mm/filemap.c#L2907
https://github.com/torvalds/linux/blob/30f09200cc4aefbd8385b01e41bde2e4565a6f0e/mm/filemap.c#L4181
Note: O_DIRECT flag is a non-POSIX feature so we should follow Linux behavior.
Current behavior in filestore
Proposed behavior
Add new state for
WriteDataEntry—PendingDirect, indicating that the request hasO_DIRECTflag and cannot become cached. These requests share the same pending queue with non-direct requests. Queue rules make processing direct and non-direct requests mutually exclusive:Note: cache invalidation is not an acceptable option here as it may result in data corruption.
For ReadData requests — the same logic as for write requests:
Additional: presence of direct requests in the pending queue trigger Flush immediately for the node.