Hello -- I've also hit this on master as of a few days ago (on
021533194476 ("Kconfig: Disable -Wstringop-overflow for GCC globally"))
just running some short lived docker container in a loop -- the last
process of the mount namespace exiting cleans up the mess docker had set
up and sometimes get caught there:
#5 [ffff9f0fcd927c48] select_collect at ffffffffb2ffa6c2
#6 [ffff9f0fcd927c58] d_walk at ffffffffb2ffb0a8
#7 [ffff9f0fcd927cb0] shrink_dcache_parent at ffffffffb2ffd53e
#8 [ffff9f0fcd927cf8] shrink_dcache_for_umount at ffffffffb2ffd8e7
#9 [ffff9f0fcd927d20] generic_shutdown_super at ffffffffb2fdc39a
#10 [ffff9f0fcd927d38] kill_litter_super at ffffffffb2fdc673
#11 [ffff9f0fcd927d50] deactivate_locked_super at ffffffffb2fdd37f
#12 [ffff9f0fcd927d68] cleanup_mnt at ffffffffb3008d1d
#13 [ffff9f0fcd927d90] task_work_run at ffffffffb2cd97e9
#14 [ffff9f0fcd927db0] do_exit at ffffffffb2cb1c9c
#15 [ffff9f0fcd927e10] do_group_exit at ffffffffb2cb264d
#16 [ffff9f0fcd927e38] __x64_sys_exit_group at ffffffffb2cb26c4
#17 [ffff9f0fcd927e40] do_syscall_64 at ffffffffb376ea47
(I'm trying to reproduce another networking bug with that, it doesn't
happen very often but I noticed docker hangs once in a while, more often
after I reduced how much memory I gave to this VM)
Reverting just 57851607326a ("get rid of DCACHE_GENOCIDE") seems to get
rid of the issue at this point, I'm pretty confident it's no longer
happening by now but I'll reply again tomorrow if I hit the problem
again overnight.
Thanks,
--
Dominique Martinet | Asmadeus
