make modules fails with kernel 2.6.24-rc4 prepatch

Logged In: YES
user_id=567099
Originator: NO

I could probably mention that this is block inclusion in Ubuntu Hardy..

Logged In: YES
user_id=18318
Originator: NO

So I know that someone in VMware has a patch for this issue, but I don't have it right now in a form that's easily digestible. One problem that we're slowly working on is making it easier to get open-vm-tools code out, since waiting three months between releases isn't really acceptable turn-around time.

shawarma, do you have any specific deadlines that you need a fix for this by?

Logged In: YES
user_id=567099
Originator: NO

Alpha 3 is set for January 10th, so as early as possible after new year's would be really good. Any later than Friday the 4th of January would be stretching it a bit.

Thanks for looking into this!

Patch to fix this issue in the latest open-vm-tools release

Logged In: YES
user_id=18318
Originator: NO

File Added: open-vm-tools-kernel_stdint.patch

Logged In: YES
user_id=18318
Originator: NO

File Added: open-vm-tools-kernel_stdint-soren.patch

Soren's updated version of the patch that actually works

Logged In: YES
user_id=95186
Originator: NO

This is also a block for Fedora 9

Logged In: YES
user_id=567099
Originator: NO

Er.. Use the patch, then?

Logged In: YES
user_id=95186
Originator: NO

Thank you for that truly insightful comment :-)

Yes I'll bundle the patch, but I'm seeing major stability issues, after a kernel hang and a Workstation crash in less than 5 minutes :-( (with a 2.6.24 kernel...).

Logged In: YES
user_id=230582
Originator: NO

Packaging this for Gentoo, I've also encountered multiple repeating CPU soft-lockups, seemingly involving the vmblock module, when using the soren patch. I'm happy to provide more information or run any tests required to try and resolve this, if it'll help?

Logged In: YES
user_id=825608
Originator: NO

Do you have any additional information about the soft-lockup you're seeing? What about the lockup points towards the vmblock module and do you have a stack trace? Were you doing anything specific when the soft-lockup occurred, perhaps dropping or copying a file into the guest?

Logged In: YES
user_id=230582
Originator: NO

The host is running linux, 2.6.23 kernel, with vmblock loaded and running workstation-6.0.2.
The client is running linux, 2.6.24 kernel.

The softlockups happen immediately after boot and seem to repeat, the scrollback buffer tends not to let me get more than a screen's worth of output, and I can't find the lockups recorded in the logs. To avoid transcription errors, the best I've got are screenshots which I'll attach following this post.

The init script we ended up using attempts to load vmblock if it's not already loaded, wait's quarter of a second just ensure the fs is all registered, creates any necessary directories, and then mounts the vmblock filesystem, after which it finally loads the guestd daemon. I think it's during the mount that the softlockups occur. This occurs whether I've grabbed focus or not at the time, there's no dragging or dropping of files involved, and X isn't started. Under 2.6.23 the init script all works fine.

If there's any further tests or patches I can try out to help, please let me know...

Logged In: YES
user_id=230582
Originator: NO

Hmmm, seemingly no attach facility since I'm not the bug owner... 5:(
[<........>] address's have been left out for my typing sanity. Please email me directly if you want the screenshots.

note: mount[2417] exited with preempt_count 1
BUG: scheduling while atomic: mount/2417/0x10000002
Pid: 2417, comm: mount Tainted: G D 2.6.24-gentoo #1

schedule+0x5a3/0x6c0
smp_apic_timer_interrupt+0x56/0x90
fbcon_putcs+0x0/0x2c0
quicklist_trim+0xf8/0x150
apic_timer_interruot+0x28/0x30
pgd_dtor+0x0/0x40
__cond_resched+0x27/0x30
cond_resched+0x27/0x30
unmap_vmas+0x55a/0x590
exit_mmap+0x7d/0x120
mmput+0x1e/0x80
do_exit+0x14c/0x7e0
printk+0x1b/0x20
die+0x22f/0x230
do_page_fault+0x29b/0x660
_atomic_dec_and_lock+0x47/0x70
_spin_lock+0x1e/0x80
do_page_fault+0x0/0x660
error_code+0x72/0x80
d_instantiate+0x22/0x50
d_alloc_root+0x47/0x60
VMBlockCleanupFileSystem+0xe2/0x130 [vmblock]
get_sb_nodev+0x56/0xb0
VMBlockCleanupFileSystem+0x60/0x130 [vmblock]
vfs_kern_mount+0x37/0x90
VMBlockCleanupFileSystem+0x60/0x130 [vmblock]
do_kern_mount+0x3d/0xe0
do_mount+0x4db/0x670
__alloc_pages+0x5a/0x370
__inc_zone_state+0x17/0x70
_spin_unlock+0x11/0x30
handle_mm_fault+0x1b8/0x600
get_page_from_freelist+0x1c0/0x470
tick_program_event+0x38/0x60
__alloc_pages+0x5a/0x370
do_page_fault+0x0/0x660
__get_free_pages+0x2e/0x50
copy_mount_options+0x40/0x140
sys_mount+0x77/0xb0
sysenter_past+esp+0x5f/0x89

Everything from mmput and below is repeated exactly in the repeating softlock (bug addresses and everything).

Logged In: YES
user_id=95186
Originator: NO

I'm seeing something similar on Fedora 9 (rawhide) (kernel 2.6.24), locks up on boot during the init script (which also loads vmblock, waits one second then mounts it). I see both CPU cores going to 100% and system is unresponsive.

Logged In: YES
user_id=230582
Originator: NO

In case it's of any consequence, the host I've done all my testing on is also a dual core machine (Core Duo), the guest, however, has been allocated only one processor.

Logged In: YES
user_id=825608
Originator: NO

Thanks for the information. Obviously we haven't seen this problem on any of the kernels we've tested. I'll get a 2.6.24 guest and see if I can pin down the problem.

Logged In: YES
user_id=230582
Originator: NO

Well, one of our users informed me of the 2008.01.23.74039 package, and I now can't seem to recreate the 2.6.24 issues.

Logged In: YES
user_id=1887497
Originator: NO

I tried this with build 2008.01.23.74039 in a 64bit vm with a fresh 2.6.24-rc8 kernel and everything built fine.

This might just have been an issue with the older rc sources.

Logged In: YES
user_id=18318
Originator: NO

Think it's safe to say this one is taken care of.