Opened 3 years ago
Closed 3 years ago
#17553 closed bug (fixed)
random freezes after condition_variable changes
Reported by: | 3dEyes | Owned by: | waddlesplash |
---|---|---|---|
Priority: | normal | Milestone: | R1/beta4 |
Component: | System/Kernel | Version: | R1/Development |
Keywords: | condition_variable | Cc: | |
Blocked By: | Blocking: | ||
Platform: | x86-64 |
Description
After the December changes in condition_variable, various programs started to freezes randomly. About 10-15 freezes per day - one of the open windows simply stops refreshing (Tracker, Deskbar, Telegram, Pe and others). In rare cases, there is a complete system freeze. If I revert changes in condition_variables.h and condition_variables.cpp to 1728b8c7778d24e0daed7468f33dadf2f48c15ad commit, the problem disappears completely.
Change History (6)
comment:1 by , 3 years ago
Owner: | changed from | to
---|---|
Status: | new → assigned |
comment:2 by , 3 years ago
comment:4 by , 3 years ago
I built haiku with this patch. I want to check for stability in work during the day
comment:5 by , 3 years ago
Replying to waddlesplash:
Possible patch: https://review.haiku-os.org/c/haiku/+/4890
Uptime is about 8 hours, no freezes. This patch seems to fix my problem.
comment:6 by , 3 years ago
Milestone: | Unscheduled → R1/beta4 |
---|---|
Resolution: | → fixed |
Status: | assigned → closed |
Merged (with some minor modifications to remove now-unneeded atomics) in hrev55841.
Please retrieve a KDL backtrace of a few of the stuck threads (i.e. drop to KDL, run "bt" on a few threads, exit KDL, upload syslog.) I think nephele may have seen some similar symptoms though so maybe you are not the only one who has experienced this. Permanent hangs are very weird, that likely implies a defect in the thread blocking code...