How long should an xfs_freeze take?

["Chris Pascoe" <c.pascoe@xxxxxxxxxxxxxx>]

Hi,

I'm guessing an xfs_freeze shouldn't take this long, am I correct?

# time xfs_freeze -f /tst1

real    344m40.434s
user    0m0.000s
sys     340m40.810s

The system is a Dual CPU P3 Xeon with 1GB RAM (highmem enabled), running
2.4.17 CVS, taken on 2002/01/23 (just before the -funsigned-char changes
went through), with LVM 1.0.2 added (I see the same behaviour with 1.0.1,
though).

During the time that passed I broke into kdb a dozen or so times - often the
running process on one CPU was the automounter:

0xf71d2000 00000604 00000001  1  001  run   0xf71d2370*automount

[0]kdb> btp 604
    EBP       EIP         Function(args)
           0xc0252221 stext_lock+0x2fcd
                               kernel .text.lock 0xc024f254 0xc024f254
0xc02576c0
0xf71d2000 0xc01447d2 sys_ioctl+0x42 (0x5, 0x810c9365, 0xbffff7b0,
0xbffffae0, 0x4)
                               kernel .text 0xc0100000 0xc0144790 0xc01449d0
           0xc010713b system_call+0x33
                               kernel .text 0xc0100000 0xc0107108 0xc0107140

whilst the rest of the time it was the xfs_freeze process, sample backtrace:

0xe95c2000 00001062 00000904  1  000  run   0xe95c2370*xfs_freeze

Entering kdb (current=0xe95c2000, pid 1062) on processor 0 due to Keyboard
Entry
[0]kdb> bt
    EBP       EIP         Function(args)
0xe95c3b4c 0xc01cb1a7 vn_count+0xb (0xf3d0bcc0, 0xf7e89560, 0xf6d1f000,
0xf6d1f000, 0xf6d1f118)
                               kernel .text 0xc0100000 0xc01cb19c 0xc01cb1ac
0xe95c3b88 0xc01a19f6 xfs_iflush_all+0x56 (0xf6d1f000, 0x1, 0xf6d1f000, 0xc,
0xc039ce80)
                               kernel .text 0xc0100000 0xc01a19a0 0xc01a1b1c
0xe95c3bb8 0xc01995f0 xfs_fs_freeze+0x38 (0xf6d1f000, 0xbffffbac,
0xf6d094a0, 0xf6d095c0, 0x18)
                               kernel .text 0xc0100000 0xc01995b8 0xc0199670
0xe95c3f54 0xc01c5104 xfs_ioctl+0x1650 (0xe95c3c18, 0x2, 0x3b0, 0xec, 0xec)
                               kernel .text 0xc0100000 0xc01c3ab4 0xc01c51f0
           0xc010722c error_code+0x34
                               kernel .text 0xc0100000 0xc01071f8 0xc0107234
Interrupt registers:
eax = 0x08049c50 ebx = 0xe95c3c18 ecx = 0x00000002 edx = 0x000003b0
esi = 0x000000ec edi = 0x000000ec esp = 0x00000010 eip = 0xf6d095c0
ebp = 0x00000000 xss = 0x00010246 xcs = 0xffffffff eflags = 0xc02475db
xds = 0xe95c3c5c xes = 0xf6d1f000 origeax = 0xf6cfcb38 &regs = 0xe95c3c10
Interrupt from user space, end of kernel trace

Alternatively, I'd see a slightly different location in vn_count:

Entering kdb (current=0xe95c2000, pid 1062) on processor 0 due to Keyboard
Entry
[0]kdb> bt
    EBP       EIP         Function(args)
0xe95c3b88 0xc01cb1ab vn_count+0xf (0xe9e811e0, 0xf7e89560, 0xf6d1f000,
0xf6d1f000, 0xf6d1f118)
                               kernel .text 0xc0100000 0xc01cb19c 0xc01cb1ac

Anything that I should try to help figure out what's going on?

Chris