Closed Bug 550810 Opened 14 years ago Closed 3 years ago

Crash: Fatal IO error 10 (No child processes) on X server :0.0.

Categories

(Core :: Widget: Gtk, defect)

x86
Linux
defect
Not set
critical

Tracking

()

RESOLVED WORKSFORME
Tracking Status
status1.9.1 --- ?

People

(Reporter: tonymec, Unassigned)

References

Details

(Keywords: crash, Whiteboard: [trunk and 1.9.1.x])

Attachments

(2 files)

Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.9pre) Gecko/20100307 SeaMonkey/2.0.4pre - Build ID: 20100307001547

Reproducible: Sometimes

Steps to reproduce:
1. Be away from the computer without having shut it down

Actual results:
- SeaMonkey has been killed, Breakpad hasn't come up
- Here's the stdout/stderr log:

nsHeaderInfo: registerSelf called!
nsHeaderInfo: registerSelf called!
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
seamonkey-bin: Fatal IO error 10 (No child processes) on X server :0.0.

Expected results:
SeaMonkey should still be running

Additional info:
This is the second time this happens to me. The first time (some days ago on a different nightly of the same branch), I noticed it when coming back to the computer after a night's sleep. This time, I was playing Mahjongg (a board game) on a different virtual desktop, and suddently I noticed that all SeaMonkey windows had disappeared from the taskbar. Today it was approximately at the time cron-daily was run, and at a time of high load average (maybe between 2 and 4); the other time it could have been -- don't know if relevant.

The only tabs not totally idle would have been self-reloading tinderbox.mozilla.org tabs (I have animated images set to run only once); and of course the mailer would be periodically polling my POP/NNTP servers and RSS/Atom feed sources. The ChatZilla window was up but not connected (only the "client" tab active).
(In reply to comment #1)
> see bug 548536 comment 4?

IIUC, in bug 548536 it's the X server that "goes away", taking all X applications down with it. Here, the X server seems to go on as normal, but the SeaMonkey process disappears with no Breakpad report, only the "Fatal IO error" message in the sysout/syserr logfile.
P.S. This problem seems to have gone away all by itself, but I cannot be certain because I have no way to make it happen "on demand".
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.10pre) Gecko/20100331 SeaMonkey/2.0.5pre - Build ID: 20100331000511

Happened again tonight (while I was away), after a thousand or so messages:
(seamonkey-bin:4253): Gdk-WARNING **: XID collision, trouble ahead
all identical including the 4253, then one message "deliver mode: 0"
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.10pre) Gecko/20100414 SeaMonkey/2.0.5pre - Build ID: 20100414005300

Happened today while scrolling a feed in the Mailer; here's the log:

nsHeaderInfo: registerSelf called!
nsHeaderInfo: registerSelf called!
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0
deliver mode: 0

(acroread:27427): Gdk-WARNING **: XID collision, trouble ahead
deliver mode: 0
seamonkey-bin: Fatal IO error 10 (No child processes) on X server :0.0.



I explain the "acroread" event by reading and printing a PDF attachment to a mail a couple of hours before, in a browser tab which was then closed -- unrelated to the crash, I suppose?

Breakpad still did not come up.
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.10pre) Gecko/20100422 SeaMonkey/2.0.5pre - Build ID: 20100422003511

happened again after more than 500 "XID collision" warnings for (seamonkey-bin: 11771)
So, this could potentially even be an X or Gdk or XID bug. Fun thing. :(
10 (No child processes) means errno is ECHILD.
I don't know any reason why select, read, or write operations on the connection would cause this value.

Sometimes libxcb raises an error on the connection without setting errno, in which case the value would be meaningless.  The cases that I saw were:
* if there is a problem with an extension request, or
* a request is too long.
I just installed an openSUSE "Online Update" with the following description:

<quote>
xorg-x11-libXext - xorg-x11-libXext: This update fixes a race condition in libXext

XAllocID must only be called with the Display lock held. There's a race condition in libXext that causes apps that use the X shared memory extensions to occasionally crash. It affects lots of X programs, including Evolution, Firefox, various parts of Gnome, etc. Fixed by this update.
References:
578577 (bugzilla) : bug number 578577
</quote>

We'll see if it fixes this bug (or rather, if it doesn't we'll sooner or later see the bug again).
P.S. The bug number in comment #9 is of course at bugzilla.novell.com, not here.
Component: General → Widget: Gtk
Product: SeaMonkey → Core
QA Contact: general → gtk
Version: SeaMonkey 2.0 Branch → 1.9.1 Branch
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.3a5pre) Gecko/20100507 SeaMonkey/2.1a1pre - Build ID: 20100507011137

<sigh /> I got it again a few minutes ago with that libXext fix in place.
status1.9.1: --- → ?
Whiteboard: [trunk and 1.9.1.x]
Version: 1.9.1 Branch → Trunk
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.3a5pre) Gecko/20100510 SeaMonkey/2.1a1pre - Build ID: 20100510003927

Happened while typing an email. (I'm temporarily staying with this nightly until the new addons manager stabilizes.) Here's the end part of the sysout/syserr log (lots of "XID collision, trouble ahead" warnings before that, 788 in the whole log, started maybe 10 hours ago or less):

(seamonkey-bin:18140): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:18140): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:18140): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:18140): Gtk-CRITICAL **: gtk_text_buffer_get_insert: assertion `GTK_IS_TEXT_BUFFER (buffer)' failed

(seamonkey-bin:18140): Gtk-CRITICAL **: gtk_text_buffer_get_iter_at_mark: assertion `GTK_IS_TEXT_MARK (mark)' failed

(seamonkey-bin:18140): Gtk-CRITICAL **: _gtk_text_layout_get_block_cursor: assertion `layout != NULL' failed

(seamonkey-bin:18140): Gtk-CRITICAL **: gtk_text_layout_get_cursor_locations: assertion `layout != NULL' failed

(seamonkey-bin:18140): Gdk-CRITICAL **: gdk_window_invalidate_rect_full: assertion `GDK_IS_WINDOW (window)' failed

(seamonkey-bin:18140): Gdk-CRITICAL **: gdk_window_invalidate_rect_full: assertion `GDK_IS_WINDOW (window)' failed

(seamonkey-bin:18140): Gtk-CRITICAL **: gtk_text_buffer_get_insert: assertion `GTK_IS_TEXT_BUFFER (buffer)' failed

(seamonkey-bin:18140): Gtk-CRITICAL **: gtk_text_buffer_get_iter_at_mark: assertion `GTK_IS_TEXT_MARK (mark)' failed

(seamonkey-bin:18140): Gtk-CRITICAL **: _gtk_text_layout_get_block_cursor: assertion `layout != NULL' failed

(seamonkey-bin:18140): Gtk-CRITICAL **: gtk_text_layout_get_cursor_locations: assertion `layout != NULL' failed

(seamonkey-bin:18140): Gdk-CRITICAL **: gdk_window_invalidate_rect_full: assertion `GDK_IS_WINDOW (window)' failed

(seamonkey-bin:18140): Gdk-CRITICAL **: gdk_window_invalidate_rect_full: assertion `GDK_IS_WINDOW (window)' failed
deliver mode: 0

(seamonkey-bin:18140): Gtk-CRITICAL **: gtk_widget_realize: assertion `GTK_WIDGET_ANCHORED (widget) || GTK_IS_INVISIBLE (widget)' failed

(seamonkey-bin:18140): Gdk-CRITICAL **: gdk_window_set_back_pixmap: assertion `GDK_IS_WINDOW (window)' failed

(acroread:27843): Gdk-WARNING **: XID collision, trouble ahead
seamonkey-bin: Fatal IO error 10 (No child processes) on X server :0.0.
P.S. I had viewed a PDF file in SeaMonkey recently (an attachment to an email, passed by the Mailer to the Browser), printed it, then closed the tab, and continued reading several emails, so that last warning is probably from the Acrobat plugin but maybe 10 to 60 minutes before the crash.
Is it a coincidence that you were using acroread immediately prior to the error on two separate occasions.

And what version of libxcb do you have?
This problem has existed ever since the 2.0 code was released.  This occurs regularly and was never present in the 1.x series of Seamonkey releases.  The Seamonkey executables have to be killed as none of the windows respond.

This hang-up happens regularly and the error messages even occur when only the email client is open.  The email window is open as well as 2 browser page with a total of 10 tabs between them (6 and 4).  

The email system was converted from 1.x version of Seamonkey and it often loses the ability to open email subfolders.  The only way to get the subfolders is to double-click forcing an new email client window to open.

This crash happened, or was discovered, after the program had sat idle over night for around 8 hours.  The acrobat plugin had not been in use overnight and the tabs that used acrobat plugin had been closed earlier in the night.

Executable is running on stock version of Slackware 13.0 Linux.  

Build identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.9) Gecko/20100404 SeaMonkey/2.0.4

This is a copy of the console log.  The first line is repeated some 200 plus times.

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead 

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead
deliver mode: 0

(seamonkey-bin:7882): Gdk-WARNING **: XID collision, trouble ahead

(acroread:8763): Gtk-CRITICAL **: gtk_widget_get_accessible: assertion `GTK_IS_WIDGET (w
idget)' failed

** (acroread:8763): CRITICAL **: atk_object_set_role: assertion `ATK_IS_OBJECT (accessib
le)' failed
seamonkey-bin: Fatal IO error 10 (No child processes) on X server :0.0.

Output from ps before seamonkey was killed so as to be able to use seamonkey.

ps
  PID TTY          TIME CMD
 3438 pts/0    00:00:00 bash
 3546 pts/0    00:00:00 dbus-launch
 7866 pts/0    00:00:00 seamonkey
 7878 pts/0    00:00:00 run-mozilla.sh
 7882 pts/0    04:07:53 seamonkey-bin
 8582 pts/0    00:00:13 acroread <defunct>
 8672 pts/0    00:00:16 acroread <defunct>
 8763 pts/0    00:00:24 acroread <defunct>
 8990 pts/0    00:00:00 ld-linux.so.2
 9291 pts/0    00:00:00 seamonkey-bin
 9462 pts/0    00:00:00 ps
In reply to comment #14: I don't know if acrobat caused it, but most other times there was no PDF viewing anywhere in sight. Similarly, often, *but not always*, there were a number (sometimes hundreds or even thousands) of GTK/GDK "warning" messages (usuain the log before the crash. No times logged, so no idea how long before the crash.

My X server is named Xorg, from openSUSE package xorg-x11-7.4-35.3.i586
libxcb is xorg-x11-libxcb-7.4-11.1.i586
Both package names obtained by rpm -qa | grep

In reply to comment #15: here this error always kills SeaMonkey without user intervention (and without Breakpad coming up either), in both Sm 2.0.* and 2.1a1pre. Do I understand that you have to kill it manually? Maybe your X server is different? (N.B. in the version numbers above, I think the part after the second dash is distro-specific; but I'm not 100% sure.)

Hm, Donald, *three* zombie instances of acroread, isn't that a lot?

This bug has happened when I was away in bed, other times between two letters in an email sentence... I'm totally baffled as to what could cause it.
Hit Enter before I meant to. At line 4 of comment #16:

"warning" messages (usually "XID collision, trouble ahead", like Donald) in the log before the crash. No times logged, (etc.)
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.3a6pre) Gecko/20100615 Lightning/1.1a1pre SeaMonkey/2.1a2pre - Build ID: 20100615020143

Had it again. One PDF file had been viewed during this session, though not near its end. Once again many "trouble ahead" warnings.

I feel as if walking in a fog.
I'm assuming that the "trouble ahead" warnings here are
https://bugzilla.gnome.org/show_bug.cgi?id=590690
so I doubt that they are the cause.
(In reply to comment #19)
> I'm assuming that the "trouble ahead" warnings here are
> https://bugzilla.gnome.org/show_bug.cgi?id=590690
> so I doubt that they are the cause.

You could be right: it has happened (e.g. comment #0) that the crash (well, the "seamonkey-bin was killed" event) occurs without any "trouble ahead" warning. Usually however, there are many, sometimes thousands, before this sporadic problem happens.
I have run into all of the problems listed here in the comments.

I have had the problems happen when only the email window is open and no other windows are open.

My email folder tree is large as well and was converted from the 1.x series of folders. A "du -sh .mozilla" indicated 8.6GB.  Since the email window is always open could the conversion have introduced a problem.  One item I run into is access to subfolders in the mail window.  I have to double-click and open a new mail window to get some subfolders.  The subfolders are at level three.
In reply to comment #21:
I also have large mail folders, but I keep my Inbox empty. I have migrated from Sm1 to Sm2 (or did it still call itself 1.5 at that time?) as soon as it was possible, yet I didn't experience this bug until shortly before comment #0.

I keep my mail folder hierarchy expanded. My biggest mail folder (to which I add but mostly don't remove) is at the moment 262173855 bytes long, or 12.2% of the maximum 32-bit signed integer. "du -sh ~/.mozilla/seamonkey/*.default" gives 1.2G.
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.3a6pre) Gecko/20100625 Lightning/1.1a1pre SeaMonkey/2.1a3pre - Build ID: 20100625012543

Happened again while mouse-scrolling the mailer's Preview Pane. At startup, after displaying for a brief instant, the browser window minimized itself. I first thought it had crashed while starting up, then noticed the dimmed icon (meaning "minimized") on the taskbar.
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:2.0b2pre) Gecko/20100701 Lightning/1.1a1pre SeaMonkey/2.1a3pre - Build ID: 20100701101320

Happened while reading email in the preview pane
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:2.0b2pre) Gecko/20100701 Lightning/1.1a1pre SeaMonkey/2.1a3pre - Build ID: 20100701101320

Happened while away sleeping. A PDF file had been viewed then its tab closed, in the same session but quite some time before I went to bed.

And about the "old" SeaMonkey build: sorry, my bottleneck is the Grand XPCOM Bustage still not solved for Calendar.
Seamonkey died again while typing a text email.  I had just earlier downloaded several files.

Build identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.10) Gecko/20100504 Lightning/1.0b1 SeaMonkey/2.0.5

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead

(seamonkey-bin:28733): Gdk-WARNING **: XID collision, trouble ahead
deliver mode: 0
[ConvConfHandler] isPreferred contentType: application/zip
[ConvConfHandler] isPreferred contentType: application/zip
[ConvConfHandler] isPreferred contentType: application/zip
[ConvConfHandler] isPreferred contentType: application/zip
seamonkey-bin: Fatal IO error 10 (No child processes) on X server :0.0.
Donald: which version of libxcb and xorg-server were in use when you saw that?
I saw something similar the other day but with error 11 (EAGAIN) during or shortly after Firefox startup. libxcb-1.6 xorg-server-1.7.7.
I am running Slackware 13.0 and the two packages are

libxcb 1.3
xorg-server 1.6.3

I run the stock Slackware releases with the 3rd party Gnome builds added in.
FWIW, I'm running openSUSE Linux 11.3 with all updates applied. I can't say more ATM because "zypper up" is running (816th package of 2016, mostly devel/debuginfo, a run started yesterday) so I cannot access the rpm database, not even read-only for "rpm -qa |grep something".
The error that I saw (11) could be reproduced by resizing a Flash plugin.

#0  0x00007ffc9e486f3f in _XIOError () from /usr/lib64/libX11.so.6
#1  0x00007ffc9e48edcb in _XReply () from /usr/lib64/libX11.so.6
#2  0x00007ffc9e4847df in XTranslateCoordinates () from /usr/lib64/libX11.so.6
#3  0x00007ffc9fe01859 in gdk_event_translate (display=<value optimized out>, 
    event=0x7ffc4f43b2b0, xevent=0x7ffff857cc20, 
    return_exposes=<value optimized out>) at gdkevents-x11.c:1874
#4  0x00007ffc9fe01a22 in _gdk_events_queue (display=0x7ffc98e61210)
    at gdkevents-x11.c:2305
#5  0x00007ffc9fe01d7d in gdk_event_dispatch (source=<value optimized out>, 
    callback=0, user_data=0x7ffc98e25390) at gdkevents-x11.c:2366
#6  0x00007ffc9e9bba94 in g_main_dispatch (context=<value optimized out>)
    at gmain.c:1960
#7  IA__g_main_context_dispatch (context=<value optimized out>)
    at gmain.c:2513
#8  0x00007ffc9e9bf1ff in g_main_context_iterate (context=0x7ffc98e167a0, 
    block=<value optimized out>, dispatch=<value optimized out>, 
    self=<value optimized out>) at gmain.c:2591

This was https://bugs.freedesktop.org/show_bug.cgi?id=29412
but I guess that's different because it only affected libX11 1.3.4.
In reply to comment #31: freedesktop.org issue 29412 mentions fatal error 11 (Resource temporarily available); I consistently see fatal error 10 (No child processes) as the last line on the sysout/syserr log. Also, in my case SeaMonkey dies (maybe on a SIGKILL from Xorg, I can't tell) without Breakpoint coming up, so I get no stack trace.
This problem seems to occur after viewing a PDF file where the acrobat reading is embedded in a Seamonkey window.  If the Window or Tab is closed or the back button is used on the Window or tab apparently Seamonkey loses track of the embedded application and is not properly handling notification about the application closing or forgetting the application is gone.  I do have the remember prior sessions active in my preferences.
This problem appears to be specially related to running helper applications, such as the acrobat PDF application, in a Seamonkey Window.  Seamonkey crashed earlier in the day with the same problem and after restarting a PDF file was viewed via the Web.  Seamonkey crashed shortly after that.  No other things were actively being done at the time.  The back button had been used to go back to the prior page.
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2.10) Gecko/20100914 Firefox/3.6.10 ID:20100914123505

Mandriva One 2010.1

xorg-server 1.7.7-1
libxcb 1.6-1

Crashed while my screensaver was on.  During that session I had opened up several PDF files, both via acroread as a helper app and via the plugin in-browser, though neither was active at the time of the crash.

The error was exactly "Crash: Fatal IO error 10 (No child processes) on X server :0.0", though I didn't catch any of the Gdk or Gtk warnings.

Hey Tony,
Does this crash still occur for you?

Flags: needinfo?(antoine.mechelynck)

(In reply to Andrei Purice from comment #36)

Hey Tony,
Does this crash still occur for you?

AFAIK it doesn't but I have a new computer (faster and with more memory). Since I got it there may have been a crash but not this one.

Flags: needinfo?(antoine.mechelynck)

Marking this as Resolved > Worksfor me based on reporter's last comment.
If anyone can still reproduce the issue feel free to re-open the bug or file a new one.

Status: NEW → RESOLVED
Closed: 3 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: