lkml.org 
[lkml]   [2011]   [Jan]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [GIT] Networking
From
Date
Le jeudi 20 janvier 2011 à 13:12 -0800, Linus Torvalds a écrit :
> On Wed, Jan 19, 2011 at 6:04 PM, David Miller <davem@davemloft.net> wrote:
> >
> > 1) Revert a netlink flag sanity check that is causing regressions in
> > existing applications.
> ...
>
> This is a long-shot, but I thought I'd ask before I start trying to
> bisect the fourth independent suspend/resume related issue in this
> merge window..
>
> When I suspend/resume while logged in by closing the lid on my laptop
> on FC14, it causes the gnome-screensaver-dialog to start up. So far so
> fine, that's what I want, and it all works fin in 2.6.37.
>
> But in current -git (and in -rc8, so it's not changed by your latest
> pull request), gnome-screensaver-dialog gets stuck after I type in my
> password, making the box basically useless.
>
> So I straced it over the network, and if I attach _when_ it is already
> stuck, it immediately becomes unstuck. But if I attach to it before
> typing my password, I can see the hang in strace, and it looks like
> this:
>
> ...
> read(3, 0x9806500, 4096) = -1 EAGAIN (Resource
> temporarily unavailable)
> poll([{fd=4, events=POLLIN}, {fd=3, events=POLLIN}, {fd=12,
> events=POLLIN|POLLPRI}, {fd=14, events=POLLIN|POLLPRI}, {fd=9,
> events=POLLIN|POLLPRI}, {fd=10, events=POLLIN|POLLPRI}, {fd=15,
> events=POLLIN}, {fd=16, events=POLLIN}, {fd=17, events=0}, {fd=19,
> events=POLLIN}], 10, -1) = ? ERESTART_RESTARTBLOCK (To be restarted)
> restart_syscall(
>
> and that's it - it's now hung. So why did it work when I straced it
> while hung? And why is it doing that ERESTART_RESTARTBLOCK in the
> first place, I'm not seeing any signals there?
>
> So I tried sending it a useless signal, which will re-animate the
> strace, and now I get:
>
> restart_syscall(<... resuming interrupted call ...>) = 1
> --- SIGWINCH (Window changed) @ 0 (0) ---
> poll([{fd=3, events=POLLIN|POLLOUT}], 1, -1) = 1 ([{fd=3, revents=POLLOUT}])
>
> Whee. That signal got it started again, and the poll finished immediately.
>
> And how/why did the input to the poll apparently change? That looks
> suspicious too. Might be some odd strace artifact, but whatever.
>
> So I'm contacting you because that fd=3 is a socket (I didn't check
> details), and because anything I find in the git logs that discusses
> "poll" seems to be network-related. So I'm wondering it this rings any
> bells, because bisecting this is going to be painful as hell (since I
> have to carefully work around all the _other_ problems I've bisected
> on that machine while doing so).
>

Do you know the type of socket ? UNIX or INET ?

You could try a revert of 2c6607c611cb7bf0a6750bcea3
(net: add POLLPRI to sock_def_readable())

But I dont understand how it could hurt...




--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2011-01-20 22:29    [W:0.042 / U:13.180 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site