Skip to contentSkip to conversation

lvcreate on 2.6.22.1: kernel tried to execute NX-protected page

DISCUSSION Juergen Kreileder 2 participants 4 messages Aug 5, 2007 – Aug 16, 2007 Device mapper

lore
Share & tools
lore
#0 lore raw

I've upgraded devmapper to 1.02.20 and lvm2 to 2.02.26. Didn't help much, I just got a the same BUG again:

kernel tried to execute NX-protected page - exploit attempt? (uid: 0)

BUG: unable to handle kernel paging request at virtual address f492c1f8
 printing eip:

f492c1f8
*pdpt = 0000000000001001
*pde = 80000000348001e3
*pte = ec1c7da0ec1c7da0

Oops: 0011 [#1]
CPU:    0
EIP:    0060:[<f492c1f8>]    Not tainted VLI
EFLAGS: 00010282   (2.6.22.1-jk1-exec-shield #1)

EIP is at 0xf492c1f8
eax: f492c1cc ebx: f492c1cc ecx: 00000000 edx: f492c1f8
esi: f492c1f8 edi: 00000000 ebp: 00000000 esp: d4177db4
ds: 007b es: 007b fs: 0000 gs: 0033 ss: 0068
Process lvcreate (pid: 2303, ti=d4176000 task=d1c88a00 task.ti=d4176000)

Stack: c02088c4 f492c1e4 c02088d0 c03dd95e c2969f60 c0209118 ce684800 e5436280 
       00000287 c03dd952 c03dd952 f6a44548 c018e393 c19ea90c 00000000 c19eb900 
       c03dd952 c18f9780 f45cb300 00000000 c0157664 c18f97cc c18f9780 c01575be 
Call Trace:
 [kobject_cleanup+116/128] kobject_cleanup+0x74/0x80
 [kobject_release+0/16] kobject_release+0x0/0x10
 [kref_put+56/160] kref_put+0x38/0xa0
 [sysfs_hash_and_remove+275/320] sysfs_hash_and_remove+0x113/0x140
 [sysfs_slab_alias+100/128] sysfs_slab_alias+0x64/0x80
 [sysfs_slab_add+174/208] sysfs_slab_add+0xae/0xd0
 [kmem_cache_create+236/320] kmem_cache_create+0xec/0x140
 [jobs_init+46/128] jobs_init+0x2e/0x80
 [kcopyd_init+45/176] kcopyd_init+0x2d/0xb0
 [kcopyd_client_create+28/208] kcopyd_client_create+0x1c/0xd0
 [init_hash_tables+142/192] init_hash_tables+0x8e/0xc0
 [snapshot_ctr+506/752] snapshot_ctr+0x1fa/0x2f0
 [dm_split_args+47/272] dm_split_args+0x2f/0x110
 [dm_table_add_target+252/400] dm_table_add_target+0xfc/0x190
 [vmalloc+32/48] vmalloc+0x20/0x30
 [populate_table+98/192] populate_table+0x62/0xc0
 [table_load+82/240] table_load+0x52/0xf0
 [table_load+0/240] table_load+0x0/0xf0
 [ctl_ioctl+209/288] ctl_ioctl+0xd1/0x120
 [ctl_ioctl+0/288] ctl_ioctl+0x0/0x120
 [do_ioctl+59/96] do_ioctl+0x3b/0x60
 [vfs_ioctl+94/416] vfs_ioctl+0x5e/0x1a0
 [sys_ioctl+61/128] sys_ioctl+0x3d/0x80
 [sysenter_past_esp+95/133] sysenter_past_esp+0x5f/0x85
 =======================
Code: 00 00 00 00 00 00 00 ff ff ff ff 00 00 00 00 00 00 00 00 20 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 <f8> c1 92 f4 f8 c1 92 f4 4c 16 b6 f5 00 00 00 00 00 00 00 00 00 

EIP: [<f492c1f8>] 0xf492c1f8 SS:ESP 0068:d4177db4

Quoted 61 lines

I got the appended BUG from a 32-bit 2.6.22.1 kernel (with exec-shield patch and PAE enabled) on an Athlon64 with dmsetup 1.02.03 and lvm2 v2.02.02.
(Note, the message comes from the vanilla kernel, not from the exec-shiled patch.)

I wasn't able to reproduce the problem so far. The machine creates several snapshot volumes every 4 hours and worked fine with the new kernel for several days. It had 2.6.16.12+exec-shield before and ran flawlessy for over a year.

kernel tried to execute NX-protected page - exploit attempt? (uid: 0)

BUG: unable to handle kernel paging request at virtual address f551df78
 printing eip:

f551df78
*pdpt = 0000000000001001
*pde = 80000000354001e3
*pte = 9293396c5d22e546

Oops: 0011 [#1]
CPU:    0
EIP:    0060:[<f551df78>]    Not tainted VLI
EFLAGS: 00010286   (2.6.22.1-jk1-exec-shield #1)

EIP is at 0xf551df78
eax: f551df4c ebx: f551df4c ecx: 00000000 edx: f551df78
esi: f551df78 edi: 00000000 ebp: 00000000 esp: e8ee5db4
ds: 007b es: 007b fs: 0000 gs: 0033 ss: 0068
Process lvcreate (pid: 25916, ti=e8ee4000 task=f7358a00 task.ti=e8ee4000)

Stack: c02088c4 f551df64 c02088d0 c03dd95e c64ccf00 c0209118 00000287 c03dd95e
       00000287 c018e38b c03dd952 d3e460e8 c018e393 c192a90c 00000000 c192b900
       c03dd952 f557a600 f59bbcc0 00000000 c0157664 f557a64c f557a600 c01575be
Call Trace:
 [kobject_cleanup+116/128] kobject_cleanup+0x74/0x80
 [kobject_release+0/16] kobject_release+0x0/0x10
 [kref_put+56/160] kref_put+0x38/0xa0
 [sysfs_hash_and_remove+267/320] sysfs_hash_and_remove+0x10b/0x140
 [sysfs_hash_and_remove+275/320] sysfs_hash_and_remove+0x113/0x140
 [sysfs_slab_alias+100/128] sysfs_slab_alias+0x64/0x80
 [sysfs_slab_add+174/208] sysfs_slab_add+0xae/0xd0
 [kmem_cache_create+236/320] kmem_cache_create+0xec/0x140
 [jobs_init+46/128] jobs_init+0x2e/0x80
 [kcopyd_init+45/176] kcopyd_init+0x2d/0xb0
 [kcopyd_client_create+28/208] kcopyd_client_create+0x1c/0xd0
 [init_hash_tables+142/192] init_hash_tables+0x8e/0xc0
 [snapshot_ctr+506/752] snapshot_ctr+0x1fa/0x2f0
 [dm_split_args+47/272] dm_split_args+0x2f/0x110
 [dm_table_add_target+252/400] dm_table_add_target+0xfc/0x190
 [vmalloc+32/48] vmalloc+0x20/0x30
 [populate_table+98/192] populate_table+0x62/0xc0
 [table_load+82/240] table_load+0x52/0xf0
 [table_load+0/240] table_load+0x0/0xf0
 [ctl_ioctl+209/288] ctl_ioctl+0xd1/0x120
 [ctl_ioctl+0/288] ctl_ioctl+0x0/0x120
 [do_ioctl+59/96] do_ioctl+0x3b/0x60
 [vfs_ioctl+94/416] vfs_ioctl+0x5e/0x1a0
 [sys_ioctl+61/128] sys_ioctl+0x3d/0x80
 [sysenter_past_esp+95/133] sysenter_past_esp+0x5f/0x85
 =======================
Code: 00 00 00 00 00 00 00 ff ff ff ff 00 00 00 00 00 00 00 00 20 00 00 00 00 00 00 \
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 <78> df 51 f5 78 df 51 f5 74 1b 8b \
                f7 00 00 00 00 00 00 00 00 32
Quoted 1 line

EIP: [<f551df78>] 0xf551df78 SS:ESP 0068:e8ee5db4

Juergen

Signature · 2 lines
Juergen Kreileder, Blackdown Java-Linux Team
http://blog.blackdown.de/
#1 lore raw
Quoted 5 lines · Juergen Kreileder

I've upgraded devmapper to 1.02.20 and lvm2 to 2.02.26. Didn't help much, I just got a the same BUG again:

kernel tried to execute NX-protected page - exploit attempt? (uid: 0)

BUG: unable to handle kernel paging request at virtual address f492c1f8

I suspect this is a module that got unloaded but still had some function pointer registered somewhere.....

do you know if/which module that could be?
(one trick is to compile the kernel such that you don't allow modules to be unloaded at all; if that makes it work it's clearly the type of bug I described above)

#2 lore raw
Quoted 8 lines · Arjan van de Ven
Quoted 5 lines · Juergen Kreileder

I've upgraded devmapper to 1.02.20 and lvm2 to 2.02.26. Didn't help much, I just got a the same BUG again:

kernel tried to execute NX-protected page - exploit attempt? (uid: 0)

BUG: unable to handle kernel paging request at virtual address f492c1f8
Quoted 7 lines

I suspect this is a module that got unloaded but still had some function pointer registered somewhere.....

do you know if/which module that could be?
(one trick is to compile the kernel such that you don't allow modules to be unloaded at all; if that makes it work it's clearly the type of bug I described above)

It's a static kernel, no modules. I've attached the config.

Juergen

#3 lore raw

SLUB problems (was Re: lvcreate on 2.6.22.1: kernel tried to execute NX-protected page)

In reply to Juergen Kreileder

I got a few more oopses like the one in the original message (see below) and one with a different call trace but that happened while creating a snapshot too.

After that I rebuilt the kernel with SLAB instead of SLUB. The system is running fine for a week now.

BUG: unable to handle kernel NULL pointer dereference at virtual address

00000013
printing eip:
c0118ef5
*pdpt = 000000001d044001
*pde = 0000000020946067
*pte = 0000000000000000

Oops: 0002 [#1]
CPU:    0
EIP:    0060:[dup_fd+181/432]    Not tainted VLI
EFLAGS: 00010286   (2.6.22.1-jk1-exec-shield #1)

EIP is at dup_fd+0xb5/0x1b0
eax: f0f30128 ebx: 00000013 ecx: ffffffff edx: 0000000c
esi: f0f300ec edi: f537b7ac ebp: f537b7e4 esp: d94cbef8
ds: 007b es: 007b fs: 0000 gs: 0033 ss: 0068
Process udev (pid: 9491, ti=d94ca000 task=f6b36a00 task.ti=d94ca000)

Stack: 00000001 f537b788 00000020 f0f30128 f537b780 f5784f00 00000000
c18e5370
       01200011 c0119034 fffffff4 f5784f00 c0119360 dd044270 de5469a0
f2b82a80
       0000251a f5784fbc d94cbfb8 bfb1bbe8 c040d0d0 c18e5370 01200011
b7f734a8
Call Trace:
 [copy_files+68/96] copy_files+0x44/0x60
 [copy_process+624/2752] copy_process+0x270/0xac0
 [do_fork+100/496] do_fork+0x64/0x1f0
 [sys_clone+50/64] sys_clone+0x32/0x40
 [sysenter_past_esp+95/133] sysenter_past_esp+0x5f/0x85
 =======================
Code: 08 f3 a5 89 c1 83 e1 03 74 02 f3 a4 8b 5c 24 08 85 db 74 23 89 f6

8b 44 24 0c 8b 08 83 c0 04 89 44 24 0c 85 c9 0f 84 a1 00 00 00 <ff> 41 14 89 4d 00 83 c5 04 4b 75 df 8b 4c 24 04 31 f6 89 ef 8b
EIP: [dup_fd+181/432] dup_fd+0xb5/0x1b0 SS:ESP 0068:d94cbef8

Quoted 61 lines · Juergen Kreileder

I got the appended BUG from a 32-bit 2.6.22.1 kernel (with exec-shield patch and PAE enabled) on an Athlon64 with dmsetup 1.02.03 and lvm2 v2.02.02.
(Note, the message comes from the vanilla kernel, not from the exec-shiled patch.)

I wasn't able to reproduce the problem so far. The machine creates several snapshot volumes every 4 hours and worked fine with the new kernel for several days. It had 2.6.16.12+exec-shield before and ran flawlessy for over a year.

kernel tried to execute NX-protected page - exploit attempt? (uid: 0)

BUG: unable to handle kernel paging request at virtual address f551df78
 printing eip:

f551df78
*pdpt = 0000000000001001
*pde = 80000000354001e3
*pte = 9293396c5d22e546

Oops: 0011 [#1]
CPU:    0
EIP:    0060:[<f551df78>]    Not tainted VLI
EFLAGS: 00010286   (2.6.22.1-jk1-exec-shield #1)

EIP is at 0xf551df78
eax: f551df4c ebx: f551df4c ecx: 00000000 edx: f551df78
esi: f551df78 edi: 00000000 ebp: 00000000 esp: e8ee5db4
ds: 007b es: 007b fs: 0000 gs: 0033 ss: 0068
Process lvcreate (pid: 25916, ti=e8ee4000 task=f7358a00 task.ti=e8ee4000)

Stack: c02088c4 f551df64 c02088d0 c03dd95e c64ccf00 c0209118 00000287 c03dd95e
       00000287 c018e38b c03dd952 d3e460e8 c018e393 c192a90c 00000000 c192b900
       c03dd952 f557a600 f59bbcc0 00000000 c0157664 f557a64c f557a600 c01575be
Call Trace:
 [kobject_cleanup+116/128] kobject_cleanup+0x74/0x80
 [kobject_release+0/16] kobject_release+0x0/0x10
 [kref_put+56/160] kref_put+0x38/0xa0
 [sysfs_hash_and_remove+267/320] sysfs_hash_and_remove+0x10b/0x140
 [sysfs_hash_and_remove+275/320] sysfs_hash_and_remove+0x113/0x140
 [sysfs_slab_alias+100/128] sysfs_slab_alias+0x64/0x80
 [sysfs_slab_add+174/208] sysfs_slab_add+0xae/0xd0
 [kmem_cache_create+236/320] kmem_cache_create+0xec/0x140
 [jobs_init+46/128] jobs_init+0x2e/0x80
 [kcopyd_init+45/176] kcopyd_init+0x2d/0xb0
 [kcopyd_client_create+28/208] kcopyd_client_create+0x1c/0xd0
 [init_hash_tables+142/192] init_hash_tables+0x8e/0xc0
 [snapshot_ctr+506/752] snapshot_ctr+0x1fa/0x2f0
 [dm_split_args+47/272] dm_split_args+0x2f/0x110
 [dm_table_add_target+252/400] dm_table_add_target+0xfc/0x190
 [vmalloc+32/48] vmalloc+0x20/0x30
 [populate_table+98/192] populate_table+0x62/0xc0
 [table_load+82/240] table_load+0x52/0xf0
 [table_load+0/240] table_load+0x0/0xf0
 [ctl_ioctl+209/288] ctl_ioctl+0xd1/0x120
 [ctl_ioctl+0/288] ctl_ioctl+0x0/0x120
 [do_ioctl+59/96] do_ioctl+0x3b/0x60
 [vfs_ioctl+94/416] vfs_ioctl+0x5e/0x1a0
 [sys_ioctl+61/128] sys_ioctl+0x3d/0x80
 [sysenter_past_esp+95/133] sysenter_past_esp+0x5f/0x85
 =======================
Code: 00 00 00 00 00 00 00 ff ff ff ff 00 00 00 00 00 00 00 00 20 00 00 00 00 00 00 \
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 <78> df 51 f5 78 df 51 f5 74 1b 8b \
                f7 00 00 00 00 00 00 00 00 32
Quoted 1 line

EIP: [<f551df78>] 0xf551df78 SS:ESP 0068:e8ee5db4

Juergen

Signature · 2 lines
Juergen Kreileder, Blackdown Java-Linux Team
http://blog.blackdown.de/