On Thu, Nov 20, 2014 at 07:08:15PM +0100, Toralf Förster wrote: > With latest git tree of trinity at a user mode linux image it stays here forever: > [child0:5249] <timed out> > [main] Bailing main loop because Completed maximum number of operations.. > [watchdog] [5096] Watchdog exiting because Completed maximum number of operations.. So that [main] line is the last line in main_loop() On return, we do this.. 159 main_loop(); 160 161 shm->mainpid = 0; 162 _exit(EXIT_SUCCESS); and yet.. > The proces list shows: > > $ ps fx -eo pid,start_time,command | grep -e trinity -e sleep | grep -v grep > 4878 18:30 | \_ bash -c logger "2#-1, M=/mnt/hostfs"; cd ~; sudo su -c 'if [[ -d ./t3 ]]; then sudo chmod -R a+rwx ./t3; sudo rm -rf ./t3; fi'; mkdir ./t3 && cd ./t3 || exit; if [[ -n /mnt/hostfs ]]; then if [[ -d /mnt/hostfs/victims/v1 ]]; then sudo chmod -R a+rwx /mnt/hostfs/victims/v1; sudo rm -rf /mnt/hostfs/victims/v1 || exit; fi; mkdir -p /mnt/hostfs/victims/v1/v2; for i in $(seq -w 0 99); do touch /mnt/hostfs/victims/v1/v2/f$i; mkdir /mnt/hostfs/victims/v1/v2/d$i; done; fi; MALLOC_CHECK_=2 trinity -C 2 -N 25000 -q -V /mnt/hostfs/victims/v1/v2 > 5095 18:30 | \_ trinity -C 2 -N 25000 -q -V /mnt/hostfs/victims/v1/v2 > 5096 18:30 | \_ [trinity-watchdo] <defunct> > 5097 18:30 | \_ [trinity-main] Somehow it's still around. > Here are the stacks: > > $ sudo cat /proc/5097/stack > > [<0805f8b4>] __switch_to+0x44/0x70 > [<0850b194>] __schedule+0x2f4/0x3a0 > [<08097b8a>] __cond_resched+0x1a/0x30 > [<0850b371>] _cond_resched+0x31/0x50 > [<080dbbb2>] truncate_inode_pages_range+0x192/0x650 > [<080dc102>] truncate_inode_pages_final+0x52/0x60 > [<08275f18>] hostfs_evict_inode+0x18/0x40 > [<08126e8d>] evict+0xdd/0x1b0 > [<08127b0d>] iput+0x16d/0x180 > [<08123538>] __dentry_kill+0x138/0x200 > [<08123f66>] dput+0x156/0x180 > [<0810fa15>] __fput+0x175/0x190 > [<0810fa6b>] ____fput+0xb/0x10 > [<08092956>] task_work_run+0x76/0x90 > [<0807e92d>] do_exit+0x32d/0x940 > [<0807f022>] do_group_exit+0xa2/0xf0 > [<0807f087>] SyS_exit_group+0x17/0x20 > [<08062980>] handle_syscall+0x60/0x80 > [<080746fc>] userspace+0x46c/0x5e0 > [<0805f720>] fork_handler+0x60/0x70 > [<ffffffff>] 0xffffffff This is the interesting part. The process is about to exit, but hostfs is doing.. something. It might just be taking a really long time, or it might be stuck. If it happens again, you might be able to use ftrace to figure out if hostfs is actually making forward progress or not. Perhaps the UML folks have some ideas. > Maybe it helps you to improve trinity, if not, ignore this mail > ;-) afaics, there's nothing here that trinity can do, once we've called _exit(), we're done. Anything that happens afterwards is the kernels fault :) Dave -- To unsubscribe from this list: send the line "unsubscribe trinity" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
