[ScopedSpinGuard] Use std::atomic::compare_exchange_strong() for spinlock [crashpad/crashpad : main]
4 views
Skip to first unread message
Ben Hamilton (Gerrit)
May 16, 2024, 4:47:22 PMMay 16
to Mark Mentovai, Justin Cohen, Tim Sergeant, Crashpad LUCI CQ, crashp...@chromium.org
Attention needed from Justin Cohen and Mark Mentovai
Ben Hamilton added 1 comment![]( "Open in Gerrit")
Patchset-level comments
File-level comment, Patchset 2 (Latest):
Ben Hamilton . resolved
Mark, I'm trying to write an arm64 test to reproduce the spurious failure from
`std::atomic::compare_exchange_weak()`, but I'm not able to reproduce it.
Do you have any thoughts on how I can test this?
Related details
Attention is currently required from:
- Justin Cohen
- Mark Mentovai
Submit Requirements:
Code-Review
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
Mark Mentovai (Gerrit)
May 16, 2024, 6:03:49 PMMay 16
to Ben Hamilton, Justin Cohen, Tim Sergeant, Crashpad LUCI CQ, crashp...@chromium.org
Attention needed from Ben Hamilton and Justin Cohen
Mark Mentovai voted Code-Review+1![]( "Open in Gerrit")
Attention is currently required from:
- Ben Hamilton
- Justin Cohen
Submit Requirements:
Code-Review
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
Mark Mentovai (Gerrit)
May 16, 2024, 6:05:29 PMMay 16
to Ben Hamilton, Justin Cohen, Tim Sergeant, Crashpad LUCI CQ, crashp...@chromium.org
Attention needed from Ben Hamilton and Justin Cohen
Mark Mentovai added 1 comment![]( "Open in Gerrit")
Patchset-level comments
Ben Hamilton . resolved
Mark, I'm trying to write an arm64 test to reproduce the spurious failure from
`std::atomic::compare_exchange_weak()`, but I'm not able to reproduce it.Do you have any thoughts on how I can test this?
Mark Mentovai
Mark, I'm trying to write an arm64 test to reproduce the spurious failure from
`std::atomic::compare_exchange_weak()`, but I'm not able to reproduce it.Do you have any thoughts on how I can test this?
Nothing immediately comes to mind.
Related details
Attention is currently required from:
- Ben Hamilton
- Justin Cohen
Submit Requirements:
Code-Review
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
Ben Hamilton (Gerrit)
May 16, 2024, 6:16:55 PMMay 16
to Mark Mentovai, Justin Cohen, Tim Sergeant, Crashpad LUCI CQ, crashp...@chromium.org
Attention needed from Justin Cohen
Ben Hamilton voted Commit-Queue+2![]( "Open in Gerrit")
Attention is currently required from:
- Justin Cohen
Submit Requirements:
Code-Review
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
Crashpad LUCI CQ (Gerrit)
May 16, 2024, 6:17:05 PMMay 16
to Ben Hamilton, Mark Mentovai, Justin Cohen, Tim Sergeant, crashp...@chromium.org
Crashpad LUCI CQ submitted the change![]( "Open in Gerrit")
Change information
Commit message:
[ScopedSpinGuard] Use std::atomic::compare_exchange_strong() for spinlock
Previously, ScopedSpinGuard used std::atomic::compare_exchange_weak()
in a loop to implement a spinlock. After looping for the specified
number of nanoseconds, it would give up and return an error.
A few bugs have come in on ARM platforms (https://crbug.com/340980960,
http://b/296082201) which indicate that this can fail even in
single-threaded cases where nothing else has the spinlock.
From https://cbloomrants.blogspot.com/2011/07/07-14-11-compareexchangestrong-vs.html :
> compare_exchange_weak exists for LL-SC (load linked/store
> conditional) type architectures (Power, ARM, basically everything
> except x86), because on them compare_exchange_strong must be
> implemented as a loop, while compare_exchange_weak can be
> non-looping.
and:
https://en.cppreference.com/w/cpp/atomic/atomic/compare_exchange#Notes
> compare_exchange_weak is allowed to fail spuriously, that is, acts
> as if *this != expected even if they are equal. When a
> compare-and-exchange is in a loop, compare_exchange_weak will yield
> better performance on some platforms.
>
> When compare_exchange_weak would require a loop and
> compare_exchange_strong would not, compare_exchange_strong is
> preferable [...]
My conclusion is that this logic needs to use
`compare_exchange_strong` to avoid spurious failures on ARM in the
common case when there's no other thread holding the spinlock.
Change-Id: I2a08031db6b219d7d14a5cd02b3634985f81ab06
Bug: b:340980960
Change-Id: I2a08031db6b219d7d14a5cd02b3634985f81ab06
Reviewed-by: Mark Mentovai <ma...@chromium.org>
Commit-Queue: Ben Hamilton <benha...@google.com>
Files:
- M util/synchronization/scoped_spin_guard.h
Change size: S
Delta: 1 file changed, 11 insertions(+), 4 deletions(-)
Branch: refs/heads/main
Submit Requirements:
Code-Review: +1 by Mark Mentovai
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
Reply all
Reply to author
Forward
0 new messages