From Bugzilla Helper: User-Agent: Mozilla/4.76 [en] (X11; U; Linux 2.4.2 i686) Reproducible: Always Steps to Reproduce: 1.Use XFree-4.0.2 from wolverine (on a RH7.0 laptop) 2.Run soffice Actual Results: When I start StarOffice (5.2), the machine first becomes slow (e.g. very slow response on mouse) and then freezes completely after about 30 seconds. Expected Results: When I use the XFree-3.3.6-driver from S3 (by changing the X softlink), StarOffice works fine. I'm running RedHat Linux 7.0 on a laptop with S3 Savage/MX. Since this card is supported in XFree86-4.0.2, I installed the X-rpms (plus the rpms that X required) from wolverine. I haven't tried to reproduce this with a full wolverine installation.
Upgrade to the latest rawhide version of XFree, Mesa and the kernel. The savage driver was recently updated. I do not have StarOffice to test this with, but if the latest RPM's do not work for you, I will get someone in the Durham office to try and reproduce the problem. Does the new stuff fix it for you?
I have tried with XFree86-4.0.2-12.1 and Mesa-3.4-10 from rawhide + kernel 2.4.3-pre4; the problem is the same. So whatever the problem is, it also applies to the latest rawhide rpms. The notebook is an ASUS L8400-C, and lspci reports the S3 card as: 01:00.0 VGA compatible controller: S3 Inc. 86C270-294 Savage/MX-/IX (rev 11) XFree-4.0.2 reports the card as: (--) PCI:*(1:0:0) S3 Savage/MX-MV rev 17, Mem @ 0xf0000000/27 We have reproduced the problem with a collegue that has a similar notebook (same S3-card).
From the homepage of the savage driver, http://www.probo.com/timr/savage40.html, I found that this is a known bug. StarOffice has a work-around; setting SAL_DO_NOT_USE_INVERT50=true "fixes" the problem. I don't know who's to blame, StarOffice or the savage driver.
As far as I'm concerned - *NO* software should ever hang the machine. If it does, whatever is arbitrating access to hardware is responsible for ensuring the stability and proper operation of that hardware. So, if Staroffice hangs the machine, it is an XFree86 bug hands down as it should be impossible for an application to do so. The only exception to this rule IMHO is a fatal hardware flaw, which some hardware has I believe. Nonetheless this issue will likely have to be addressed in a driver update. Thanks for looking into this and providing more details, info. I'll see what I can come up with.
XFree86 4.0.3-4 contains the updated Savage driver version 1.15 from the above URL. Please try it when it is available, with the environment variable fix above. Realistically, the driver author will have to fix this one, and he's aware of the problem but not of a solution yet other than the kludge above. I hope this helps.
*** Bug 32399 has been marked as a duplicate of this bug. ***
1. You probably mean version 1.1.15 of the driver. 2. That version (1.1.15) was also included in XFree86-4.0.2. 3. Thus: The new XFree86-4.0.3-rpms from rawhide changes nothing. StarOffice still crashes without the magic env-variable. The reason can probably be found in the openoffice source code. I have looked at it and talked with the developer of the driver.
Yes I mean 1.1.15, but it was not included with XFree86 4.0.2. Take stock 4.0.2 or 4.0.3 and diff them with either of the authors last 2 patches, and they are not the same. The difference between 1.1.14 (which I was applying) and 1.1.15 is the shrinking of one file. The bug persists, so one will have to work around it as above until the driver author fixes this, and it is available on his website or in the core XFree86 distribution. I've updated to the 1.1.15 driver and a small patch, which I've not released RPM's of yet.
The 4.0.3-9 package in rawhide and on my personal space at ftp://people.redhat.com reverts a savage patch that caused some problems for some people. Not sure if it fixes this problem or not.
Bad news: With XFree86-4.0.3-9 (from ftp://people.redhat.com) on RH71 the machine freezes completely in StarOffice 5.2 install. My video card is a S3 Savage 4 (on board-IBM 300GL).
In OpenOffice install (build 625) the machine freezes. (XFree86-4.0.3-9 / RH71 / S3 Savage 4)
The environment variable set (SAL_DO_NOT_USE_INVERT50=true) DOES work. YES! Woo Hoo! I was able to load Star Office correctly without massive crashification after the environment variable was set. I'll test it out with the openoffice version, too -- since the bug exists on that version of StarOffice, too. http://www.openoffice.org/issues/show_bug.cgi?id=863
Same problem here with an "ATI Technologies Inc 3D Rage Pro" card on a 4.1.0 X-Server on a RedHat 7.1 with a 2.4.3-12 kernel. In my case, the SAL_DO_NOT_USE_INVERT50 variable does NOT work: installation starts, displays glibc version, and then freezes. The freeze is complete, even telnetting in from another box won't work. OpenOffice version is 633.
The very same experience as the 2001-09-25 04:56:28, openoffice-rh.lu comment on different hardware. On Openoffice 638c installation: No hangup, but the X system does not start, glibc version displayed, snd the child process that would display the splash screen crashes with segmentation fault. Right before the fault the process tries to allocate memory (may be frm the video card - can't see details): ============================================================================ [pid 4505] read(6, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0000-\1\000"..., 1024) = 1024 [pid 4505] fstat64(6, {st_dev=makedev(8, 3), st_ino=132142, st_mode=S_IFREG|0755, st_nlink=1, st_uid=500, st_gid=500, st_blksize=4096, st_blocks=512, st_size=254715, st_atime=2001/10/06-22:55:12, st_mtime=2001/09/18-22:44:46, st_ctime=2001/10/06-22:55:02}) = 0 [pid 4505] getcwd("/tmp/sv001.tmp", 128) = 15 [pid 4505] old_mmap(NULL, 199376, PROT_READ|PROT_EXEC, MAP_PRIVATE, 6, 0) = 0x4183e000 [pid 4505] mprotect(0x41868000, 27344, PROT_NONE) = 0 [pid 4505] old_mmap(0x41868000, 28672, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 6, 0x29000) = 0x41868000 [pid 4505] close(6) = 0 [pid 4505] brk(0x80bc000) = 0x80bc000 [pid 4505] --- SIGSEGV (Segmentation fault) ---=========================================================================== Some auxiliary data: [root@senorg /root]# uname -a Linux senorg 2.4.2-2 #1 Sun Apr 8 20:41:30 EDT 2001 i686 unknown [root@senorg /root]# rpm -q glibc glibc-2.2.2-10 [root@senorg /root]# free total used free shared buffers cached Mem: 513284 507340 5944 0 51104 178228 -/+ buffers/cache: 278008 235276 Swap: 525304 0 525304 [root@senorg /root]# rpm -qa | grep XFree XFree86-libs-4.0.3-5 ... XFree86-SVGA-3.3.6-38 ... XFree86-4.0.3-5 ... [root@senorg /root]# The video card is an S3 Trio3D with 4MB video ram. The CPU is a PIII/500Mhz. --------------------- the proposed workaround doesn't cure the bug ----------- [incze@senorg install]$ env | grep SAL SAL_DO_NOT_USE_INVERT50=true [incze@senorg install]$ ./setup glibc version: 2.2.2 [incze@senorg install]$
The Savage driver has been updated in XFree86-4.1.0-5 with Tim Robert's new Savage driver 1.1.20t, which solves this problem. This release will be available within a day or so for testing at: ftp://people.redhat.com/mharris/testing/bleeding-edge
*** Bug 55619 has been marked as a duplicate of this bug. ***
Updated summary line to be a bit to help others when doing queries.
I have verified that the XFree86-4.1.0-5 packages solve the problem (Bug 55619) with my T21. I removed the SAL_DO_NOT_USE_INVERT50=true setting from my environment and was able to do the installation to a freshly created account without any problems.
Verified that this fix works for the savage3 card I've been using. Installed star office successfully and ran the "attraction" screen saver on the root window all weekend without a lockup.