Comments (3)
xairy
commented on May 5, 2024
This is not easy, since percpu_rwsem_down/up_read/write call a lot of functions that can enter scheduler including down_write and down_read.
We could enable events back around those calls (same as we did with rwsem), but in that case we won't ignore synchronization from down_write and down_read.
from kernel-sanitizers.
xairy
commented on May 5, 2024
Hm, I see false positives in tests with percpu-rwsem (without any additional annotations).
386 struct percpu_rw_semaphore pcrws_sync;
387
388 static void pcrws_main(void *arg)
389 {
390 int rv = percpu_init_rwsem(&pcrws_sync);
391 BUG_ON(rv != 0);
392 }
393
394 static void pcrws_write_write(void *arg)
395 {
396 percpu_down_write(&pcrws_sync);
397 *((int *)arg) = 1;
398 percpu_up_write(&pcrws_sync);
399 }
400
401 static void pcrws_read_read(void *arg)
402 {
403 percpu_down_read(&pcrws_sync);
404 *((int *)arg + 4) = *((int *)arg);
405 percpu_up_read(&pcrws_sync);
406 } ==================================================================
ThreadSanitizer: data-race in pcrws_write_write
Write at 0xffff88028a19b400 of size 4 by thread 2838 on CPU 0:
[<ffffffff81252c92>] pcrws_write_write+0x22/0x40 mm/ktsan/tests_inst.c:397
[<ffffffff812523fd>] thr_func+0x2d/0x50 mm/ktsan/tests_inst.c:33
[<ffffffff810b98c1>] kthread+0x161/0x180 kernel/kthread.c:209
[<ffffffff81ea8c8f>] ret_from_fork+0x3f/0x70 arch/x86/entry/entry_64.S:529
Previous read at 0xffff88028a19b400 of size 4 by thread 2839 on CPU 1:
[<ffffffff81252c47>] pcrws_read_read+0x27/0x50 mm/ktsan/tests_inst.c:404
[<ffffffff812523fd>] thr_func+0x2d/0x50 mm/ktsan/tests_inst.c:33
[<ffffffff810b98c1>] kthread+0x161/0x180 kernel/kthread.c:209
[<ffffffff81ea8c8f>] ret_from_fork+0x3f/0x70 arch/x86/entry/entry_64.S:529
Mutexes locked by thread 2838:
Mutex 408151 is locked here:
[<ffffffff81ea5a95>] down_write+0x65/0x80 kernel/locking/rwsem.c:62
[<ffffffff810dfaf5>] percpu_down_write+0x45/0x190 kernel/locking/percpu-rwsem.c:145
[<ffffffff81252c8a>] pcrws_write_write+0x1a/0x40 mm/ktsan/tests_inst.c:396
[<ffffffff812523fd>] thr_func+0x2d/0x50 mm/ktsan/tests_inst.c:33
[<ffffffff810b98c1>] kthread+0x161/0x180 kernel/kthread.c:209
[<ffffffff81ea8c8f>] ret_from_fork+0x3f/0x70 arch/x86/entry/entry_64.S:529
Mutexes locked by thread 2839:
Mutex 408151 is read locked here:
[<ffffffff81ea5a15>] down_read+0x45/0x60 kernel/locking/rwsem.c:28
[<ffffffff810df9ea>] percpu_down_read+0x6a/0xa0 kernel/locking/percpu-rwsem.c:85
[<ffffffff81252c3f>] pcrws_read_read+0x1f/0x50 mm/ktsan/tests_inst.c:403
[<ffffffff812523fd>] thr_func+0x2d/0x50 mm/ktsan/tests_inst.c:33
[<ffffffff810b98c1>] kthread+0x161/0x180 kernel/kthread.c:209
[<ffffffff81ea8c8f>] ret_from_fork+0x3f/0x70 arch/x86/entry/entry_64.S:529
==================================================================
from kernel-sanitizers.
dvyukov
commented on May 5, 2024
Another reason to properly annotate them ;)
from kernel-sanitizers.
Related Issues (20)
- false positives with T0 HOT 3
- Enormous number of sync objects HOT 5
- CPU numbers in reports are presumably wrong
- Detect races on global variables
- Report RCU read critical sections
- False positive in blk_complete_request HOT 1
- kernel BUG at mm/ktsan/thr.c:129!
- Check that allocator does not introduce lots of unnecessary synchronization
- Print information about globals
- Print heap block allocation stack
- false positives on irq stack
- Handle synchronization on struct page
- (Cosmetic) kt_shadow_clear is dead code and seems broken HOT 1
- When does ktsan task destroy? HOT 2
- Does KTSAN support ARM64?
- Code Execution HOT 1
- Move KFENCE internals info to kernel docs HOT 2
- Does kernel hwasan support short granules？ HOT 1
- Is KASAN working properly? HOT 4
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from kernel-sanitizers.