Suspiciously high number of crashes with high number (>300) of threads |
||||
Issue descriptionSome Chrome platforms (Windows and CrOS) seem to have a very high number of crashes where the number of threads is incredibly high (>300). This somehow suggests me that we are leaking threads somewhere. https://crash.corp.google.com/browse?q=product.name%3D%27Chrome%27%20AND%20ThreadCount%3E300&ignore_case=false&enable_rewrite=true&omit_field_name=&omit_field_value=&omit_field_opt=%3D#samplereports You can see the thread count in the "Threads" page.
,
Aug 3 2017
Can we capture the stack when creating a new thread to see what's going on here?
,
Aug 3 2017
I looked historically back and there doesn't appear to be a regression here - the % of crashes with >300 threads has remained relatively constant, at least not in the last year or so. It would be good to do more analysis on what the ~300 threads are actually doing here...
,
Aug 3 2017
Unfortunately, no minidumps are available (they were all discarded), so we can't verify the results of the symbolizer. Jumping through a random set, there are likely some interesting ones For example... https://crash.corp.google.com/browse?q=product.name%3D%27Chrome%27%20AND%20ThreadCount%3E300%20AND%20ReportID%3D%273af948c788000000%27&ignore_case=false&enable_rewrite=true&omit_field_name=&omit_field_value=&omit_field_opt=%3D&stbtiq=&reportid=&index=0 spins up tons of threads manually (They all have ThreadMain + MessagePumpDefault) with many suspicious rtc:PLatformThreads running around.
,
Aug 3 2017
+oysteine I wonder if this is something where slow-reports could come handy.
,
Sep 25 2017
,
Jan 10
Archiving P3s older than 1 year with no owner or component. |
||||
►
Sign in to add a comment |
||||
Comment 1 by dcheng@chromium.org
, Aug 3 2017