liusdu
liusdu
Probably found what cause this problem but do not have proper solution.. @hbhasker What happened in this issue: 1. thread is in sentry context to send package using write/send 2....
> @uravas could you review this CL? Does it break #1813? Hi @uravas My test failed with your latest patch. When you deal with this case(procees in T state and...
> Did you intend to mention to @uravas ? Sorry for this mistake~
> After kernel fixes: [torvalds/linux@7c3b00e06d731](https://github.com/torvalds/linux/commit/7c3b00e06d731) [torvalds/linux@570ac9337b5c1](https://github.com/torvalds/linux/commit/570ac9337b5c1) the original problem should be fixed and probably we don't need pid_is_stopped() hunks. > > But I believe we still need the hunk in...
> After kernel fixes: [torvalds/linux@7c3b00e06d731](https://github.com/torvalds/linux/commit/7c3b00e06d731) [torvalds/linux@570ac9337b5c1](https://github.com/torvalds/linux/commit/570ac9337b5c1) the original problem should be fixed and probably we don't need pid_is_stopped() hunks. > > But I believe we still need the hunk in...
> > freezer will fail if it races attach-to-stopped-task. > > That seems completely ok for me from CRIU point of view, if freezer can't freeze processes for any reason,...
> python test/zdtm.py run -t zdtm/static/stopped --ignore-taint --freezecg zdtm:f Hi @avagin This tests hangs forever in my environment. ``` [root@my criu]# python test/zdtm.py run -t zdtm/static/stopped --ignore-taint --freezecg zdtm:f userns...
> After kernel fixes: [torvalds/linux@7c3b00e06d731](https://github.com/torvalds/linux/commit/7c3b00e06d731) [torvalds/linux@570ac9337b5c1](https://github.com/torvalds/linux/commit/570ac9337b5c1) the original problem should be fixed and probably we don't need pid_is_stopped() hunks. > > But I believe we still need the hunk in...
> > If deduplication is enabled, the last process to trigger the page fault will fail. > > Would it be possible to add a ZDTM test for this use-case?...
@rst0git @rppt I found it very difficult to disable background fetch. I cannot reliably reproduce this issue in the tests because the memory may be fetched by the background process...