[edgy] usplash segfaults

What graphics hardware is this?

I can confirm this too, using usplash 0.4-12 on AMD64, with a nvidia card. The drivers are the nv ones. My GDB output it slightly different:

(gdb) run
Starting program: /sbin/usplash
(no debugging symbols found)

Program received signal SIGSEGV, Segmentation fault.
0x0000000000402ec6 in ?? ()
(gdb) bt
#0 0x0000000000402ec6 in ?? ()
Cannot access memory at address 0xd90f0690

The relevant output from lspci for my graphics card is:

01:00.0 VGA compatible controller: ATI Technologies Inc Radeon R350 [Radeon 9800 Pro]
01:00.1 Display controller: ATI Technologies Inc Radeon R350 [Radeon 9800 Pro] (Secondary)

usplash 0.4-15 works for me on my amd64 box but I had to compile my own package as the official build for amd64 failed due to some missing -fPIC.

@Michael Bienia:
Where exactly do I have to put this "-fPIC"?

Thanks.

I've attached a diff with the changes I made to compile the package.
I'm not sure if these are the correct places to add -fPIC but it worked for me.

Thanks! This patch worked for me, but I had to fix it to use relative paths. I attached my corrected version.

It appears that the package now FTBFS. From the buildlog (http://librarian.launchpad.net/4135904/buildlog_ubuntu-edgy-amd64.usplash_0.4-16_FAILEDTOBUILD.txt.gz)

/usr/bin/ld: svgalib/staticlib/libvgagl.a(grlib.o): relocation R_X86_64_32 against `currentcontext' can not be used when making a shared object; recompile with -fPIC
svgalib/staticlib/libvgagl.a: could not read symbols: Bad value

Although the resulting binary still segfaults on my machine, here a fix for the FTBFS that works for me.

I ran usplash 0.4-18 through valgrind, I had to do it as root, but it didn't seem to crash. Here is my log anyway.

Sorry, I'll attach it this time. It seems if you forget the attachment "Description", it asks you for it, but it also clears the "file" box.

I confirm the segmentation fault on AMD64 dual core with Nvidia 7600 GS.

Video card :
01:00.0 VGA compatible controller: nVidia Corporation G70 [GeForce 7600 GS] (rev a1)

Usplash version : 0.4-24

dmesg output :
[ 24.217928] usplash[886]: segfault at 00000000ffffffab rip 00002ad46cd61622 rsp 00007fff3de85870 error 4

I have no custom image.

Run from gdb (with sudo) throws a different error though :
(gdb) run
Starting program: /sbin/usplash
(no debugging symbols found)
(no debugging symbols found)

Program received signal SIGSEGV, Segmentation fault.
0x00002b0195870622 in rdw () from /lib/libusplash.so

Hope this helps !

I still have no splash screen even when running from Live cd using the nv35 driver or the nvidia driver. this is as of 16/08/06. This is on an amd 64 bit edgy and the live cd is knot3 amd 64bit

Marking as critical. A significant number of users run the hardware (amd64 + nvidia) to have this problem, and many users won't know what to do - how to get to a working desktop.

...however, the bug doesn't prevent them from getting a working desktop, it just looks confusing.

Targetting for beta (and removing corresponding target from a duplicate)

I've got similar issue as bug #61711 however dunno it's related or not: I can't find any sign for segfault ... The problem also cause very slow booting for me

Attaching a backtrace with the debugging information from:
deb http://people.ubuntu.com/~pitti/ddebs edgy main universe
installed, I hope this is useful. I am running today's edgy, with the processor being an
'AMD Turion(tm) 64 Mobile Technology ML-37' (from /proc/cpuinfo)
and an ati mobility radeon X700 with the radeon driver.

uname -a:
Linux koan 2.6.17-8-generic #2 SMP Tue Sep 19 08:53:32 UTC 2006 x86_64 GNU/Linux

A workaround for this has been committed to bzr, and will be released after beta.

I got impatient and built bzr revision 90, but it still doesn't work. I can't get a GDB backtrace though. Mabye I am being impatient or just doing something wrong?

Confirmed, bzr rev 90 still segfaults on amd64/nvidia. I've attached an apport crash report on it, after unstripping the package. That should be quite a deal of debugging fun :)

x86emu makes me cry. Ok, looking into this.

Hi I have 2 questions related to this bug:

John, how did you exactly reproduced the problem ? [what arguments ?] Did you run usplash from the console directly ? In my case this does not make it crash. Do you manually issue usplash_write commands ? Which ones ?

All, why would the usplash package need to be "unstripped" ? What is the benefit now to strip it in the first place ? Is this common to a lot of ubuntu packages ? How fat does it grow when unstripped ? Anyone here can approciate the time saved when the package ships unstripped. Stripped binaries are required when disk space is an issue AND binaries are fully stable, I don't want to start a debate, just want to know what's the "ubuntu" option on this.

On 9/29/06, François Isabelle <email address hidden> wrote:
>
> Hi I have 2 questions related to this bug:
>
> John, how did you exactly reproduced the problem ? [what arguments ?] Did
> you run usplash from the console directly ? In my case this does not make it
> crash. Do you manually issue usplash_write commands ? Which ones ?

I simply ran "sudo usplash" at a console, and that was enough to make it
segfault. This is on an AMD64 with an NVIDIA video card.... usplash does not
crash on any of my other hardware, nor does it crash in 32-bit mode on the
same system.

All, why would the usplash package need to be "unstripped" ? What is
> the benefit now to strip it in the first place ? Is this common to a lot
> of ubuntu packages ? How fat does it grow when unstripped ? Anyone here
> can approciate the time saved when the package ships unstripped.
> Stripped binaries are required when disk space is an issue AND binaries
> are fully stable, I don't want to start a debate, just want to know
> what's the "ubuntu" option on this.

Virtually all packages are stripped. It does save considerable disk space,
not to mention smaller binaries have a better chance of ending up in the
disk cache for an appreciable amount of time. There's some work towards
automatically-installable debuginfo packages that the Ubuntu crash handler
would install upon a crash, you can read about that in this week's Ubuntu
Weekly News.

Still with usplash 0.4-30 run with "sudo usplash" on amd64 with geforce 6800gt.
[ 519.223299] usplash[7803]: segfault at 00000000ffffff80 rip 00002ae0ca9908e2 rsp 00007fffe0255690 error 4
but it does automatically switch to console 1 and display the blinking cursor, when I switched back to X it segfaulted I think.

After that I ran it with "sudo strace usplash" and it switched to console 1 again with the cursor and when I switched back to X I saw the Desktop in a very bright white tone which got even brighter in a few seconds to total white, so I hit CTRL-C several times in the end, which dropped me back at the regular Desktop. Here is the strace output.

On Notebook, fresh Edgy Beta1 install, no bootsplash. I have an ATI Mobility 7500.
After some seconds, the X Login magically appears :D

Matthew, is there any hope of getting this fixed in the next few days? If not, can we fallback to vga on the affected platforms?

A graceful fallback would be appreciated, even if it is to not activate the 'splash' option at install on this platform.

Booting with 'splash' takes 5 minutes here, when I remove 'splash' it takes 56 seconds. I suspect that the usplash client functions have some sort of a timeout which causes this delay.

I still have the same problem here !
Amd64 edgy with nvidia geforcge 6600gt
i get :
$ dmesg | grep usplash
[ 25.985482] usplash[870]: segfault at 00000000ffffffab rip 00002b2943ce18a2 rsp 00007fff66f058d0 error 4

Would be nice to have usplash for the release !
Anyway keep up the good work ubuntu is really neat :)

Actually, it's because usplash is spinning and using 100% CPU... it's really
apparent that even things like disk access is sluggish during boot (disk not
grinding nearly as fast).

I agree with everyone suggesting a fallback that I'd rather NOT see a
usplash than end up with a slow-as-molasses bootup.

On 10/12/06, Martin Pitt <email address hidden> wrote:
>
> A graceful fallback would be appreciated, even if it is to not activate
> the 'splash' option at install on this platform.
>
> Booting with 'splash' takes 5 minutes here, when I remove 'splash' it
> takes 56 seconds. I suspect that the usplash client functions have some
> sort of a timeout which causes this delay.
>
> --
> [edgy] usplash segfaults
> https://launchpad.net/bugs/56587
>

Fixed by switching to bogl on amd64; usplash 0.4-32 is uploaded and waiting for review and approval.

I'm still getting a corrupt/garbled screen while the system boots up. I can almost make out a progress bar but it is very garbled and shows up on my screen in three different places.

Let me know what information I can provide to help, I'm not new to Linux by any means but I am new to reporting bugs. :-)

On Tue, Oct 17, 2006 at 10:29:50PM -0000, Jeff Hanson wrote:
> I'm still getting a corrupt/garbled screen while the system boots up. I
> can almost make out a progress bar but it is very garbled and shows up
> on my screen in three different places.
>
> Let me know what information I can provide to help, I'm not new to Linux
> by any means but I am new to reporting bugs. :-)

That sounds like an unrelated problem.

--
 - mdz

Fantastic I now have a boot splash screen however when shutting down the system hard crashes this is the problem I had in Dapper that is still specific to the card driver I believe.

* Jeff Hanson

| I'm still getting a corrupt/garbled screen while the system boots up. I
| can almost make out a progress bar but it is very garbled and shows up
| on my screen in three different places.

This should be fixed with the new artwork. The problem was our old
artwork didn't ship a 640x400 (or 640x480) version, which is what
vga16fb needs.

--
Tollef Fog Heen ,''`.
UNIX is user friendly, it's just picky about who its friends are : :' :
                                                                      `. `'
                                                                        `-

This seems to be working correctly after running apt-update on installed Edgy RC. I don't have to remove splash and the system boots from GRUB to kdm login in 39 seconds.

On Edgy Beta the screen was garbled if I ran usplash.

System: AMD64/Nvidia 410 chipset

I still get corruption on my VTs and something new is happening. If I switch to a VT and switch back to X all I see is a blank screen and my mouse, which moves... I'm not sure if this is a usplash issue.

I have an X800

Usplash was not working on my laptop (dell latitude c640, ATI Mobility 7500) but I
had not got any segmentation faults.

The problem was that the /etc/usplash.conf was misconfigured for some unknown reason. Instead of 1024x768 it got entries for 1280x1024. After correcting this entry and reconfiguring the usplash (i.e. sudo dpkg-reconfigure usplash) every seems right.