linux 2.6.23

[csmithtx]

slax.org/ss/user/image20071010055317.jpg

I upgraded my kernel to 2.6.23, updated with aufs and squashfs from cvs ...

all appears to initially work well without issues, except with both vmware 6.0.0 and 6.0.1 (which both work well in the 2.6.22 kernel)

I can run a virtual machine, but upon poweroff I get this BAD error ... anyone got any pointers on deciphering this? it appears to me to be aufs related ... perhaps it is kernel, vmware, or squashfs related?

Code:

Message from syslogd at Fri Oct 12 20:32 2007 ...

invalid opcode: 0000

SMP

CPU: 0

EFLAGS: 00213046 (2.6.23 #1)

EIP is at __remove_from_page_cache+0x32/0x40

eax: 00000014 ebx: 00000000 ecx: 0000000c edx: ffffffdb

ds: 007b es: 007b fs: 00d8 gs: 0033 ss: 0068

Process vmware-vmx (pid: 9963, ti=e8574000 task=f444aaa0 task.ti=e8574000)

Stack: e9e7e4a8 c270c660 c014edec c270c660 00000013 c015477a c270c660 c015485f

00001000 00000000 00000000 00000450 c018cf33 00000000 00000000 00000000

ffffffff 00000000 0000000e 00000000 c270c600 c270c620 c270c640 c270fe40

Call Trace:

truncate_complete_page+0x2a/0x50

truncate_inode_pages_range+0xbf/0x330

__getblk+0x23/0x2b0

vmtruncate+0x8d/0x170

inode_setattr+0x6b/0x190

shmem_notify_change+0x61/0xf0

notify_change+0x2c5/0x3a0

aufs_setattr+0x199/0x2d0

unmap_vmas+0x299/0x530

do_truncate+0x6a/0x90

check_pgt_cache+0x1b/0x20

do_sys_ftruncate+0x102/0x170

syscall_call+0x7/0xb

=======================

Code: 8b 50 14 8d 43 04 e8 0e bd 1b 00 ba 05 00 00 00 89 f0 c7 46 10 00 00 00 00 ff 4b 30 e8 f8 86 00 00 8b 5e 08 85 db 79 03 5b 5e c3 <0f> 0b eb fe 8d 76 00 8d bc 27 00 00 00 00 53 83 ec 04 8b 5c 24

au_h_iptr_i+0x9/0x20

sys_ftruncate64+0x1b/0x20

notify_change+0x2c5/0x3a0

truncate_inode_pages+0x17/0x20

EIP: __remove_from_page_cache+0x32/0x40 SS:ESP 0068:e8575d88

remove_from_page_cache+0x1c/0x30

esi: c270c660 edi: 00000004 ebp: e9e7e4a8 esp: e8575d88

EIP: 0060:[<c014ed32>] Tainted: P VLI

Message was edited by: csmithtx to clean up syslog message

[rsa911]

I would first re-run the "vmware-config.pl" script to ensure the vmware modules are recompiled with the new kernel (make sure you also have your new kernel sources installed)

[csmithtx]

Thanks, but I can run a vm machine (it works), it's just on shutdown of a virtual machine, that I get the error ... I tested different virtual machines, I think I even tested two at the same time to see that the error occurs with the first vm shutdown ... the error is severe enough to require manually powering off the machine (hold powerbutton 4+ seconds) as it will not auto reboot ... It's not a complete system lock up, just no further vmware functionality until rebooting.

I've tried to use vmware-any-any updates 112 and 113 for both 6.0.0 and 6.0.1 in 2.6.23 with no success (each time with a clean install ... one of the many benefits of a modular o/s like slax using aufs) I have yet to try the cvs aufs and squashfs that I used in 2.6.22 vs the10/10/07 cvs update I used in 2.6.23...

I am hoping to find some linux kernel savvy debuggers to assist with their perspective / interpretation of the syslog message that seems to have it's own wish to be sent somewhere (as seen by -- cut here -- at the end of the message) as if there is relevant information to be conveyed ... With any luck, I'll get enough feedback to help me figure out what my next efforts will be ... Perhaps it may assist VMWare with their 2.6.23 research ...

[striderjw]

This may have a bearing:

http://servervirtualization.blogs.techtarget.com/2007/10/10/linux-kernel-2623-win-for-xen-and-kvm-an...

HTH

[csmithtx]

AWESOME Link!!!! Thanks for passing this on ...

this may leave me regressed for a bit with the linux 2.6.22 kernel for my vmware usage

I am confused that vmware 6.0.1 compiled in 2.6.23 without warning (without any of the vmware-any-any updates) seemingly as if the kernel would support it ...

I am also confused then, why vmware actually works --- if the crash with reboot can be tolerated a bit ...

[csmithtx]

there are (12) comments to the author's article that have relevance too ... it does appear aufs/squashfs related as another person has indicated that they have vmware in a debian 2.6.23

petr also posted that the vmware-any-any updates 111, 112, 113 address these issues ...

Still, if others have insight to understanding the message ... ie what exactly generated the invalid opcode, I would appreciate more insights!

Thanks!

[csmithtx]

I finally tested this with a slackware 12 based slax real installation to a hard drive and a plain vanilla 2.6.23.1 kernel ... I do NOT get errors ... vmware-workstation 6.0.1 works very well!

I think the errors were cvs aufs related ... or my incorrect implementation of this source into my previous unsuccessful 2.6.23 and 2.6.23.1 kernel tests.

I've opened a discussion thread on this topic in this forum as well http://slax.org/forum/viewtopic.php?p=91019