Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Fix buffered/mmap I/O race #14498

Merged
merged 1 commit into from
Feb 23, 2023
Merged

Fix buffered/mmap I/O race #14498

merged 1 commit into from
Feb 23, 2023

Conversation

behlendorf
Copy link
Contributor

@behlendorf behlendorf commented Feb 15, 2023 •
edited
Loading

Motivation and Context

Issue #13608.

Description

When a page is faulted in for memory mapped I/O the page lock may be dropped before it has been read and marked up to date. If a buffered read encounters such a page in mappedread() it must wait until the page has been updated. Failure to do so will result in a panic on debug builds and incorrect data on production builds.

The critical part of this change is in mappedread() where pages which are not up to date are now handled. Additionally, it includes the following simplifications.

  • zfs_getpage() and zfs_fillpage() could be passed an array of pages. This could be more efficient if it was used but in practice only a single page was ever provided. These interfaces were simplified to acknowledge that.

  • update_pages() was modified to correctly set the PG_error bit on a page when it cannot be read by dmu_read().

  • Setting PG_error and PG_uptodate was moved to zfs_fillpage() from zpl_readpage_common(). This is consistent with the handling in update_pages() and mappedread().

  • Minor additional refactoring to comments and variable declarations to improve readability.

  • Add a test case to exercise concurrent buffered, direct, and mmap IO to the same file.

  • Reduce the mmap_sync test case default run time.

How Has This Been Tested?

Added a new mmap_mixed test which triggers the bug and manually instrumented to code to verify it was correctly handled.

Types of changes

  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Performance enhancement (non-breaking change which improves efficiency)
  • Code cleanup (non-breaking change which makes code smaller or more readable)
  • Breaking change (fix or feature that would cause existing functionality to change)
  • Library ABI change (libzfs, libzfs_core, libnvpair, libuutil and libzfsbootenv)
  • Documentation (a change to man pages or other documentation)

Checklist:

@behlendorf behlendorf added the Status: Code Review Needed Ready for review and testing label Feb 15, 2023
@behlendorf behlendorf force-pushed the mappedread-uptodate branch 2 times, most recently from 7ef06cb to cd8c482 Compare February 15, 2023 18:55
@robn
Copy link
Member

robn commented Feb 16, 2023

@behlendorf did you mean #13608?

@behlendorf
Copy link
Contributor Author

@robn yes I did, I'll fix up that typo. Thanks.

@ryao
Copy link
Contributor

ryao commented Feb 17, 2023

While these changes look good to me, the CentOS 8 and CentOS Stream 8 ZTS failures need investigation before this can be merged. The two had the same exact failure. It is not immediately clear to me why without digging deeper, but I see the OOM killer activating in the console logs, which might be a clue.

@behlendorf
Fix buffered/direct/mmap I/O race
e0b989e 
When a page is faulted in for memory mapped I/O the page lock
may be dropped before it has been read and marked up to date.
If a buffered read encounters such a page in mappedread() it
must wait until the page has been updated. Failure to do so
will result in a panic on debug builds and incorrect data on
production builds.

The critical part of this change is in mappedread() where pages
which are not up to date are now handled. Additionally, it
includes the following simplifications.

- zfs_getpage() and zfs_fillpage() could be passed an array of
  pages. This could be more efficient if it was used but in
  practice only a single page was ever provided. These
  interfaces were simplified to acknowledge that.

- update_pages() was modified to correctly set the PG_error bit
  on a page when it cannot be read by dmu_read().

- Setting PG_error and PG_uptodate was moved to zfs_fillpage()
  from zpl_readpage_common(). This is consistent with the
  handling in update_pages() and mappedread().

- Minor additional refactoring to comments and variable
  declarations to improve readability.

- Add a test case to exercise concurrent buffered, direct,
  and mmap IO to the same file.

- Reduce the mmap_sync test case default run time.

Signed-off-by: Brian Behlendorf <[email protected]>
Closes openzfs#13604
@behlendorf behlendorf added Status: Accepted Ready to integrate (reviewed, tested) and removed Status: Code Review Needed Ready for review and testing labels Feb 22, 2023
@behlendorf behlendorf merged commit 89cd219 into openzfs:master Feb 23, 2023
lundman pushed a commit to openzfsonwindows/openzfs that referenced this pull request Mar 3, 2023
@behlendorf @lundman
Fix buffered/direct/mmap I/O race
cfc6db8 
When a page is faulted in for memory mapped I/O the page lock
may be dropped before it has been read and marked up to date.
If a buffered read encounters such a page in mappedread() it
must wait until the page has been updated. Failure to do so
will result in a panic on debug builds and incorrect data on
production builds.

The critical part of this change is in mappedread() where pages
which are not up to date are now handled. Additionally, it
includes the following simplifications.

- zfs_getpage() and zfs_fillpage() could be passed an array of
  pages. This could be more efficient if it was used but in
  practice only a single page was ever provided. These
  interfaces were simplified to acknowledge that.

- update_pages() was modified to correctly set the PG_error bit
  on a page when it cannot be read by dmu_read().

- Setting PG_error and PG_uptodate was moved to zfs_fillpage()
  from zpl_readpage_common(). This is consistent with the
  handling in update_pages() and mappedread().

- Minor additional refactoring to comments and variable
  declarations to improve readability.

- Add a test case to exercise concurrent buffered, direct,
  and mmap IO to the same file.

- Reduce the mmap_sync test case default run time.

Reviewed-by: Richard Yao <[email protected]>
Reviewed-by: Brian Atkinson <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes openzfs#13608 
Closes openzfs#14498
robn pushed a commit to robn/zfs that referenced this pull request Apr 15, 2023
@behlendorf @robn
Fix buffered/direct/mmap I/O race
be1cb7c 
When a page is faulted in for memory mapped I/O the page lock
may be dropped before it has been read and marked up to date.
If a buffered read encounters such a page in mappedread() it
must wait until the page has been updated. Failure to do so
will result in a panic on debug builds and incorrect data on
production builds.

The critical part of this change is in mappedread() where pages
which are not up to date are now handled. Additionally, it
includes the following simplifications.

- zfs_getpage() and zfs_fillpage() could be passed an array of
  pages. This could be more efficient if it was used but in
  practice only a single page was ever provided. These
  interfaces were simplified to acknowledge that.

- update_pages() was modified to correctly set the PG_error bit
  on a page when it cannot be read by dmu_read().

- Setting PG_error and PG_uptodate was moved to zfs_fillpage()
  from zpl_readpage_common(). This is consistent with the
  handling in update_pages() and mappedread().

- Minor additional refactoring to comments and variable
  declarations to improve readability.

- Add a test case to exercise concurrent buffered, direct,
  and mmap IO to the same file.

- Reduce the mmap_sync test case default run time.

Reviewed-by: Richard Yao <[email protected]>
Reviewed-by: Brian Atkinson <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes openzfs#13608
Closes openzfs#14498
robn pushed a commit to robn/zfs that referenced this pull request Apr 15, 2023
@behlendorf @robn
Fix buffered/direct/mmap I/O race
c25e59f 
When a page is faulted in for memory mapped I/O the page lock
may be dropped before it has been read and marked up to date.
If a buffered read encounters such a page in mappedread() it
must wait until the page has been updated. Failure to do so
will result in a panic on debug builds and incorrect data on
production builds.

The critical part of this change is in mappedread() where pages
which are not up to date are now handled. Additionally, it
includes the following simplifications.

- zfs_getpage() and zfs_fillpage() could be passed an array of
  pages. This could be more efficient if it was used but in
  practice only a single page was ever provided. These
  interfaces were simplified to acknowledge that.

- update_pages() was modified to correctly set the PG_error bit
  on a page when it cannot be read by dmu_read().

- Setting PG_error and PG_uptodate was moved to zfs_fillpage()
  from zpl_readpage_common(). This is consistent with the
  handling in update_pages() and mappedread().

- Minor additional refactoring to comments and variable
  declarations to improve readability.

- Add a test case to exercise concurrent buffered, direct,
  and mmap IO to the same file.

- Reduce the mmap_sync test case default run time.

Reviewed-by: Richard Yao <[email protected]>
Reviewed-by: Brian Atkinson <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes openzfs#13608
Closes openzfs#14498
@robn robn mentioned this pull request Apr 15, 2023
Merged
13 tasks
robn pushed a commit to robn/zfs that referenced this pull request Apr 20, 2023
@behlendorf @robn
Fix buffered/direct/mmap I/O race
630df11 
When a page is faulted in for memory mapped I/O the page lock
may be dropped before it has been read and marked up to date.
If a buffered read encounters such a page in mappedread() it
must wait until the page has been updated. Failure to do so
will result in a panic on debug builds and incorrect data on
production builds.

The critical part of this change is in mappedread() where pages
which are not up to date are now handled. Additionally, it
includes the following simplifications.

- zfs_getpage() and zfs_fillpage() could be passed an array of
  pages. This could be more efficient if it was used but in
  practice only a single page was ever provided. These
  interfaces were simplified to acknowledge that.

- update_pages() was modified to correctly set the PG_error bit
  on a page when it cannot be read by dmu_read().

- Setting PG_error and PG_uptodate was moved to zfs_fillpage()
  from zpl_readpage_common(). This is consistent with the
  handling in update_pages() and mappedread().

- Minor additional refactoring to comments and variable
  declarations to improve readability.

- Add a test case to exercise concurrent buffered, direct,
  and mmap IO to the same file.

- Reduce the mmap_sync test case default run time.

Reviewed-by: Richard Yao <[email protected]>
Reviewed-by: Brian Atkinson <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes openzfs#13608
Closes openzfs#14498
behlendorf added a commit that referenced this pull request Apr 21, 2023
@behlendorf
Fix buffered/direct/mmap I/O race
c7db374 
When a page is faulted in for memory mapped I/O the page lock
may be dropped before it has been read and marked up to date.
If a buffered read encounters such a page in mappedread() it
must wait until the page has been updated. Failure to do so
will result in a panic on debug builds and incorrect data on
production builds.

The critical part of this change is in mappedread() where pages
which are not up to date are now handled. Additionally, it
includes the following simplifications.

- zfs_getpage() and zfs_fillpage() could be passed an array of
  pages. This could be more efficient if it was used but in
  practice only a single page was ever provided. These
  interfaces were simplified to acknowledge that.

- update_pages() was modified to correctly set the PG_error bit
  on a page when it cannot be read by dmu_read().

- Setting PG_error and PG_uptodate was moved to zfs_fillpage()
  from zpl_readpage_common(). This is consistent with the
  handling in update_pages() and mappedread().

- Minor additional refactoring to comments and variable
  declarations to improve readability.

- Add a test case to exercise concurrent buffered, direct,
  and mmap IO to the same file.

- Reduce the mmap_sync test case default run time.

Reviewed-by: Richard Yao <[email protected]>
Reviewed-by: Brian Atkinson <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes #13608
Closes #14498
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@bwatkinson bwatkinson bwatkinson approved these changes

@ryao ryao ryao approved these changes

Assignees
No one assigned
Labels
Status: Accepted Ready to integrate (reviewed, tested)
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@behlendorf @robn @ryao @bwatkinson