Bug 357285

sax# sax2 -a
SaX: initializing pleast wait ...
SaX: no X-Xserver is running
SaX: will start own server if needed

SaX: something went wrong wile X was called with -probeonly
SaX: try to call 'sax2 -p' and select a singel device ?
abort

# sax -p
[...] cirrus

/var/log/Xorg.0.log
-------------------
[...]
(EE) Unable to locate/open config file
[...]
Fatal server error:
Cannot run in framebuffer mode. Please specify busIDs for all framebuffer devices.

So the issue is that SaX2 even cannot create the temporary xorg.conf for starting the first Xserver. 

I'll attach a screenshot of /var/log/SaX.log.

Created attachment 192565 [details]
var_log_SaX.log.png

SaX2 doesn't work with any driver. Not even with vesa or fbdev driver.

Not sure how to proceed here. It looks like a plain SaX2 issue to me. Latest SaX2 changelog entry is:

* Tue Jan 29 2008 ms@suse.de
- fixed segfault, related to (bnc #354386)

There's even no vi installed on this live CD, which makes it completely unusable.

# vi
You do not have a valid vim binary package installed. Please installl either "vim", "vim-enhanced" or "gvim".

Please provide a LiveCD ISO image with an editor, so I can look into this issue. BTW, setting up network doesn't work either with this LiveCD image.

Matthias, vim is installed - why does it not work?

Only vim-base is installed. vim and vim-data is missing on this Live system.

You need to install the package vim, vim also requires vim-base.
If you want features like syntax highlighting you need additionally vim-data.

This seems to be a funny zypper bug:

+ chroot /usr/src/packages/BUILD/tmp zypper --non-interactive --no-gpg-checks install --auto-agree-with-licenses 3ddiag 915resolution DirectFB Mesa MozillaFirefox MozillaFirefox-translations NetworkManager NetworkManager-glib NetworkManager-gnome OpenOffice_org OpenOffice_org-calc OpenOffice_org-draw ....

Entering non-interactive mode.
Entering no-gpg-checks mode.
Problem downloading the specified RPM file.
File ./SDL not found on media: dir:///
Please check whether the file is accessible.
Problem with the RPM file specified as SDL, skipping.
Problem downloading the specified RPM file.
File ./acl not found on media: dir:///
Please check whether the file is accessible.
Problem with the RPM file specified as acl, skipping.
Problem downloading the specified RPM file.
File ./atk not found on media: dir:///
Please check whether the file is accessible.
...
File ./vim not found on media: dir:///
Please check whether the file is accessible.
Problem with the RPM file specified as vim, skipping.
Problem downloading the specified RPM file.

It does not affect all packages, but 3 letter names - like vim.

the fix will be submitted to STABLE today

*** This bug has been marked as a duplicate of bug 355108 ***

Thanks, but the initial problem with SaX2 is still open.

Coolo, please provide an updated LiveCD ISO image with a working vi once the zyyper issue has been fixed.

the updated ISO has vim install in the log file, but I can't test myself if it fixes the original issue right now.

Thanks. I'll give it a try.

The initial issue remains. A cirrus based configuration created by xorgconfig works, so it's neither a Xserver nor a driver issue. For whatever reasons SaX2 is not able to create the xorg.conf for  the initial Xserver start. Reassigning to Marcus.

Created attachment 192929 [details]
qemu-11.0-alpha1.sh

Marcus, in case you're not familiar with qemu (unlikely).

ok, this is not specific to the live system. I just installed a new monitor and have the same problem on sax2 -r - "something went wrong with X -probeonly", but that command works fine

I guess we need a test machine to debug into the problem. Stefan I think this
looks like a general problem. Could you setup a test machine to reproduce this.
I think Stephan Kulow won't allow us to that on his machine :-)

Coolo, on which kind of system would this be reproducable, e.g. on shannon (ca. 2000 packages) with STABLE from yesterday it doesn't happen.

that's the machine where NX simply works, right? :)

I had the problem on my workstation, when I tried to configure the second monitor.

Yes, that's this machine.

Marcus figured out, that libhd is segfaulting.

Valgrind blames someone else:

Index: sysp/sysp.h
===================================================================
--- sysp/sysp.h (Revision 1502)
+++ sysp/sysp.h (Arbeitskopie)
@@ -780,7 +780,7 @@
        int  size  = 0;

        size = s.length();
-       part = (char*)malloc(size);
+       part = (char*)malloc(size+1);

        for(int n=0;n<size;n++) {
        if (direction == 0) {

there is another problem when looking closer:

Index: lib/pci/lspci.c
===================================================================
--- lib/pci/lspci.c     (Revision 1502)
+++ lib/pci/lspci.c     (Arbeitskopie)
@@ -305,7 +305,10 @@
    first = p;
   last = p;
  }
+ /* we can't clean this up as we use the return data
  pci_cleanup(pacc);
+ */
+ pacc = 0;
  return(first);
 }

Is lspci code only used in sysp or also in the perl bindings loaded into yast2-x11?

ok, not helping the real problem. sysp does no longer crash, but sax2 -a still output the same

OK, new valgrind backtrace:

==32568== Invalid read of size 4
==32568==    at 0x406CEFD: hd_shm_clean (string3.h:52)
==32568==    by 0x406DFCC: hd_shm_init (hd.c:5473)
==32568==    by 0x40725EF: hd_scan (hd.c:1805)
==32568==    by 0x4074409: hd_list (hd.c:3236)
==32568==    by 0x8086C93: MonitorGetData() (monitor.c:65)
==32568==    by 0x807722F: ScanXStuff::Scan() (xstuff.c:176)
==32568==    by 0x804EDE2: ScanModule(char*) (sysp.c:297)
==32568==    by 0x804F2B5: main (sysp.c:211)
==32568==  Address 0x46249d8 is 0 bytes after a block of size 5,936 alloc'd
==32568==    at 0x4022E12: calloc (in /usr/lib/valgrind/x86-linux/vgpreload_memcheck.so)
==32568==    by 0x8086C73: MonitorGetData() (monitor.c:64)
==32568==    by 0x807722F: ScanXStuff::Scan() (xstuff.c:176)
==32568==    by 0x804EDE2: ScanModule(char*) (sysp.c:297)
==32568==    by 0x804F2B5: main (sysp.c:211)

ok, just seen on steffen's screen: there is a dup 

bug 358800

*** This bug has been marked as a duplicate of bug 358800 ***

now I can boot the live CD!