Bug #50399 (was: Re: Release-critical Bugreport for November 26, 1999)
- To: firstname.lastname@example.org, Torsten Lanschoff <email@example.com>, Timshel Knoll <firstname.lastname@example.org>, Bug #50399 <email@example.com>, Debian QA Group <firstname.lastname@example.org>
- Subject: Bug #50399 (was: Re: Release-critical Bugreport for November 26, 1999)
- From: Fabrice Gautier <email@example.com>
- Date: Wed, 1 Dec 1999 20:26:26 +0100
- Message-id: <19991201202626.A1912@droopy.rezel.enst.fr>
- Mail-followup-to: firstname.lastname@example.org, Torsten Lanschoff <email@example.com>, Timshel Knoll <firstname.lastname@example.org>, Bug #50399 <email@example.com>, Debian QA Group <firstname.lastname@example.org>
- In-reply-to: <19991201070916.C751@jupiter.rhein-neckar.de>
- References: <19991126031503.A6112@master.debian.org> <19991127172145.C21939@jupiter.rhein-neckar.de> <19991129213807.E22390@wormhole.galaxy> <19991130195100.V751@jupiter.rhein-neckar.de> <19991201013931.A17992@droopy.rezel.enst.fr> <19991201070916.C751@jupiter.rhein-neckar.de>
[I'm replying to all because i don't really know what i should do...]
On Wed, Dec 01, 1999 at 07:09:16AM +0100, Christian Kurz wrote:
> On 99-12-01 Fabrice Gautier wrote:
> > On Tue, Nov 30, 1999 at 07:51:00PM +0100, Christian Kurz wrote:
> > > > > > Package: gnomba (main)
> > > > > > Maintainer: Timshel Knoll <email@example.com>
> > > > > > 50399 XFree eat all the memory when i launch gnomba
> > > > >
> > > > > Needs to be examined to see what's happening and how to fix.
> > >
> > > > Timshel asked for more info and it seems he did not get any replies. I would
> > > > say it is a bug in the windowmanager and/or the x server...
> > >
> > > May I then suggest closing this bug report. Or has someone seen this bug
> > > too and can debug it?
> > No!
> > I reported this bug, and it still occurs...
> It's bad to here, that the bug is still there.
> > In fact i didn't receive the Timshel e-mail because there is an error in
> > the From: header in my bugreport (a bug/feature in reportbug i should report
> > oneday)
> > But i just scan the bug webpage and saw his mail.
> And what do learn from it? Don't trust the reporting software. ;) But
> thanks to you for checking the webpage and sending out more information.
> > So i use XF86_SVGA, and this is this server that eats all the memory.
> > I don't think it's a window-manager bug because it's appear both with
> > sawmill and IceWM, but i tried with twm too and it doesn't eat memory
> > but it doesn't scan anything, so it maybe a gnome bug...
> Hm, looks like a gnome-bug. Did you use the gnome-version of icewm?
> so, I think, this is a bug in the gnome-program and not xfree86. Could
> someone try to examine (strace) the bug, to see what's happening?
I have an strace but it don't know to whom i shall send it - i don't
want to send à 64k gziped file to debian-devel so if someone wants it
I just noticed some repeted call to shmget with something like this:
shmget(IPC_PRIVATE, 259616, IPC_CREAT|0x1ff|0666) = 637959
shmat(637959, 0, 0) = 0x405ba000
shmget(IPC_PRIVATE, 8784, IPC_CREAT|0x1ff|0666) = 637960
shmat(637960, 0, 0) = 0x405fa000
write(5, "\201\1\4\0\16\1\200\4\7\274\t\0\0\1\200\4\201\5\7\0\17"..., 252) = 252
read(5, "\1\2\t\4\0\0\0\0\24\0\300\3\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 32) = 32
shmdt(0x405ba000) = 0
shmctl(637959, IPC_RMID, 0) = 0
shmdt(0x405fa000) = 0
shmctl(637960, IPC_RMID, 0) = 0
write(5, "\201\2\2\0\16\1\200\4\201\2\2\0\21\1\200\4+\5\1\0", 20) = 20
read(5, "\1\2\f\4\0\0\0\0\24\0\300\3\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 32) = 32
shmget(IPC_PRIVATE, 259616, IPC_CREAT|0x1ff|0666) = 638215
shmat(638215, 0, 0) = 0x405ba000
shmget(IPC_PRIVATE, 8784, IPC_CREAT|0x1ff|0666) = 638216
But i re-compiled gnomba from source package and ran it with gdb
and i isolated in gnomba the section which causes the problem.
It's in the function DoScan in smbwrap.c in the while loop from line 715
to line 813. At this point the samba scan is done, and i think this loop
is meant to prepare the gtk tree structure. Each time this loop is executed
my XFREE_SVGA eat approximately 3 more Megs. In fact, each call to
the gtk_ctree_node_set_cell_style function cost 1 Meg.
So this may be a gtk/gdk/glib bug but i'm definitely not a gtk guru...
> > Sorry for the adress problem, this one is correct
> Alright, you also noted the wrong email-adress and fixed it now. There
> are people who don't notice this and this worse.
In fact, each time i report a bug using reportbug -M , it use
firstname.lastname@example.org as the from email adress, while with mutt it only use
login which is fine. I've to read some man or README to see who i can
set a correct email adress for reportbug.