security ima: Kernel BUG in ima_file_free -- bisected to commit 6c21a7fb492bf7e2c4985937082ce58ddeca84bd
Hi, Mimi
Used the latest LTP to test 2.6.33-rc6, a Kernel BUG occured on my x86_64 (OS: Fedora 12). The message from dmesg is as following.
The LTP case is testcases/kernel/syscalls/pipe/pipe06.c. For seeing code easily, I recreated a simple code to reproduce this BUG. please check the code in the attached.
I bisected a commit 6c21a7fb492bf7e2c4985937082ce58ddeca84bd,
--------
commit 6c21a7fb492bf7e2c4985937082ce58ddeca84bd
Author: Mimi Zohar <[email protected]>
Date: Thu Oct 22 17:30:13 2009 -0400
LSM: imbed ima calls in the security hooks
--------Maybe you should fix it ;-)
--------------------------------------------------------------------------
BUG: unable to handle kernel NULL pointer dereference at 00000000000000ae IP: [<ffffffff811e3057>] ima_file_free+0x2e/0x1fb PGD 13a08e067 PUD 139fdf067 PMD 0 Oops: 0000 [#2] SMP last sysfs file: /sys/devices/pci0000:00/0000:00:1c.1/0000:19:00.0/0000:1a:00.0/irq CPU 1 Pid: 1868, comm: a.out Tainted: G D 2.6.33-rc6 #1 D2671/PRIMERGY RIP: 0010:[<ffffffff811e3057>] [<ffffffff811e3057>] ima_file_free+0x2e/0x1fb RSP: 0018:ffff88013a713e48 EFLAGS: 00010202 RAX: ffff8801325a70c0 RBX: ffff88013a1c1f00 RCX: 0000041500000415 RDX: 000004143a713e68 RSI: ffffffff81a3a080 RDI: ffff88013a1c1f00 RBP: ffff88013a713e88 R08: ffff88013b75c8c0 R09: ffff88013a713ec8 R10: ffff88013a713df8 R11: ffff88013a713e08 R12: ffff88013a1c1f00 R13: 0000000000000000 R14: 00000000ffffffe8 R15: ffff88013a713f38 FS: 00007f6e5ea48700(0000) GS:ffff880028280000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000000ae CR3: 000000013a101000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process a.out (pid: 1868, threadinfo ffff88013a712000, task ffff880137e20000) Stack: 000004143a713e68 0000041500000415 ffff88013fc03500 ffff88013a1c1f00 <0> ffff880133d3a000 ffff88013a1c1f00 00000000ffffffe8 ffff88013a713f38 <0> ffff88013a713ea8 ffffffff811c87f9 ffff88013a713ea8 ffff88013a1c1f00 Call Trace: [<ffffffff811c87f9>] security_file_free+0x2d/0x31 [<ffffffff81106767>] put_filp+0x22/0x36 [<ffffffff8110c53c>] free_write_pipe+0x2f/0x34 [<ffffffff8110d161>] do_pipe_flags+0xdc/0xf6 [<ffffffff8110d19c>] sys_pipe2+0x21/0x63 [<ffffffff8110d1ee>] sys_pipe+0x10/0x12 [<ffffffff81009c72>] system_call_fastpath+0x16/0x1b Code: e5 41 57 41 56 41 55 41 54 53 48 83 ec 18 0f 1f 44 00 00 83 3d a2 50 bb 00 00 48 8b 47 18 49 89 fc 4c 8b 68 10 0f 84 bb 01 00 00 <41> 0f b7 85 ae 00 00 00 25 00 f0 00 00 3d 00 80 00 00 0f 85 a3 RIP [<ffffffff811e3057>] ima_file_free+0x2e/0x1fb RSP <ffff88013a713e48> CR2: 00000000000000ae ---[ end trace 1d8416bb1c67accb ]--- --------------------------------------------------------------------------
Shi Weihua
#include <fcntl.h>
#include <errno.h>
int pipe_ret, pipes[2];
int main(int ac, char **av)
{
int i, numb_fds;
numb_fds = getdtablesize();
for (i = 0; i < numb_fds; i++) {
pipe_ret = pipe(pipes);
if (pipe_ret < 0) {
if (errno != EMFILE)
printf ("got unexpected error - %d", errno);
else
printf ("got expected error - %d", errno);
break;
}
}
return 0;
}Shi Weihua <[email protected]> wrote on 02/01/2010 12:06:44 AM:
Quoted 3 lines
Hi, Mimi
Used the latest LTP to test 2.6.33-rc6, a Kernel BUG occured on my
x86_64 (OS:
Quoted 3 lines
Fedora 12).
The message from dmesg is as following.
The LTP case istestcases/kernel/syscalls/pipe/pipe06.c. For seeing code
easily, I
Quoted 1 line
recreated a simple code to reproduce this BUG. please check the code in
the attached.
Quoted 10 lines
I bisected a commit
6c21a7fb492bf7e2c4985937082ce58ddeca84bd,-------- commit 6c21a7fb492bf7e2c4985937082ce58ddeca84bd Author: Mimi Zohar <[email protected]> Date: Thu Oct 22 17:30:13 2009 -0400 LSM: imbed ima calls in the security hooks --------Maybe you should fix it ;-)
Thanks for isolating the problem. The problem is caused by free_write_pipe()
calling path_put(), which puts the dentry and mnt, before it calls put_filp().
The ordering should be like in __fput(), which puts the dentry and mnt as the
last thing it does.
Mimi
Shi Weihua:
Quoted 6 lines
Used the latest LTP to test 2.6.33-rc6, a Kernel BUG occured on my
x86_64(OS: Fedora 12). The message from dmesg is as following.
The LTP case istestcases/kernel/syscalls/pipe/pipe06.c. For seeing code easily, I recreated a simple code to reproduce this BUG. please check the code in the attached.I bisected a commit
6c21a7fb492bf7e2c4985937082ce58ddeca84bd,
The problem is already known and a patch is posted in last year.
http://marc.info/?l=linux-kernel&m=126206539702176&w=2
J. R. Okajima
[email protected] wrote on 02/01/2010 01:03:33 PM:
Quoted 2 lines
Shi Weihua:
Quoted 1 line
Used the latest LTP to test 2.6.33-rc6, a Kernel BUG occured on my
x86_64
Quoted 3 lines
(OS: Fedora 12).
Quoted 2 lines
The message from dmesg is as following.
The LTP case istestcases/kernel/syscalls/pipe/pipe06.c. For seeing
code easily, I
Quoted 1 line
Quoted 1 line
recreated a simple code to reproduce this BUG. please check the code
in the attached.
Quoted 8 lines
Quoted 2 lines
I bisected a commit
6c21a7fb492bf7e2c4985937082ce58ddeca84bd,The problem is already known and a patch is posted in last year.
http://marc.info/?l=linux-kernel&m=126206539702176&w=2J. R. Okajima
Thanks for the pointer. Al, can you please push this patch forward?
Thanks!
Mimi