On 09/02/14 18:13, Nicolas Boulenguez wrote:
> # pidof thermald
> 20988
> # strace -f -p 20988
> [pid 20991] poll([{fd=7, events=POLLIN}, {fd=0, events=0}, {fd=0, events=0},
> {fd=0, events=0}, {fd=0, events=0}, {fd=0, events=0}, {fd=0, events=0},
> {fd=0, events=0}, {fd=0, events=0}, {fd=5, events=POLLIN}], 10, 4000) = 8
> ([{fd=0, revents=POLLERR|POLLHUP}, {fd=0, revents=POLLERR|POLLHUP}, {fd=0,
> revents=POLLERR|POLLHUP}, {fd=0, revents=POLLERR|POLLHUP}, {fd=0,
> revents=POLLERR|POLLHUP}, {fd=0, revents=POLLERR|POLLHUP}, {fd=0,
> revents=POLLERR|POLLHUP}, {fd=0, revents=POLLERR|POLLHUP}])
> # ls -al /proc/20988/fd
> total 0
> dr-x------ 2 root root 0 févr. 9 19:07 .
> dr-xr-xr-x 9 root root 0 févr. 9 19:06 ..
> lrwx------ 1 root root 64 févr. 9 19:07 0 -> /dev/pts/6 (deleted)
> lrwx------ 1 root root 64 févr. 9 19:07 1 -> /dev/pts/6 (deleted)
> lrwx------ 1 root root 64 févr. 9 19:07 2 -> /dev/pts/6 (deleted)
> lrwx------ 1 root root 64 févr. 9 19:07 3 -> anon_inode:[eventfd]
> lrwx------ 1 root root 64 févr. 9 19:07 4 -> socket:[237532]
> lr-x------ 1 root root 64 févr. 9 19:07 5 -> pipe:[238275]
> l-wx------ 1 root root 64 févr. 9 19:07 6 -> pipe:[238275]
> lrwx------ 1 root root 64 févr. 9 19:07 7 -> socket:[238276]
> # ls -al /proc/20991/fd
> total 0
> dr-x------ 2 root root 0 févr. 9 19:07 .
> dr-xr-xr-x 9 root root 0 févr. 9 19:07 ..
> lrwx------ 1 root root 64 févr. 9 19:07 0 -> /dev/pts/6 (deleted)
> lrwx------ 1 root root 64 févr. 9 19:07 1 -> /dev/pts/6 (deleted)
> lrwx------ 1 root root 64 févr. 9 19:07 2 -> /dev/pts/6 (deleted)
> lrwx------ 1 root root 64 févr. 9 19:07 3 -> anon_inode:[eventfd]
> lrwx------ 1 root root 64 févr. 9 19:07 4 -> socket:[237532]
> lr-x------ 1 root root 64 févr. 9 19:07 5 -> pipe:[238275]
> l-wx------ 1 root root 64 févr. 9 19:07 6 -> pipe:[238275]
> lrwx------ 1 root root 64 févr. 9 19:07 7 -> socket:[238276]
>
Thanks, I can see the problem, and I can now reproduce this. The polling
on the deleted fd 0 is causing the poll to spin. I think I have a fix,
I'll discuss this with the thermald upstream developer on Monday and
hopefully get this sorted out pretty soon.
Thank you for your help.
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
