On Tue, Feb 17, 2009 at 1:22 PM, Alexander Griesser
<[email protected]> wrote:
> Denys Vlasenko wrote:
>>>> Try attached patch, it makes read failures to abort the probing
>>>> and go to the next device.
>>> Thanks, but the patch doesn't cleanly apply to 1.13.2 for me, see
>>> attached files util.c.rej and volume_id_internal.h.rej).
>>
>> Just replace the whole util-linux/volume_id/* with this archive's contents.
>
> BTDT, not getting better.
So it still takes 4m 30s?
> Attached is a new strace run of the updated blkid tree.
Please add -tt options to make strace timestamp the log.
New log has this:
1860 lstat64("/dev/fd0", {st_mode=S_IFBLK|0660, st_rdev=makedev(2,
0), ...}) = 0
1860 open("/dev/fd0", O_RDONLY) = 4
1860 ioctl(4, BLKGETSIZE64, 0xbf9f6e78) = 0
1860 lseek(4, 0, SEEK_SET) = 0
1860 read(4, 0x92581e0, 512) = -1 EIO (Input/output error)
1860 close(4) = 0
IOW: it aborts at the very first error, whereas old log was doing this:
1878 lstat64("/dev/fd0", {st_mode=S_IFBLK|0660, st_rdev=makedev(2,
0), ...}) = 0
1878 open("/dev/fd0", O_RDONLY) = 4
1878 ioctl(4, BLKGETSIZE64, 0xbf9bbdf8) = 0
1878 lseek(4, 0, SEEK_SET) = 0
1878 read(4, 0x94c41d0, 512) = -1 EIO (Input/output error)
1878 lseek(4, 0, SEEK_SET) = 0
1878 read(4, 0x94c41d0, 69632) = -1 EIO (Input/output error)
1878 lseek(4, 0, SEEK_SET) = 0
1878 read(4, 0x94c41d0, 1536) = -1 EIO (Input/output error)
1878 lseek(4, 0, SEEK_SET) = 0
1878 read(4, 0x94c41d0, 33280) = -1 EIO (Input/output error)
1878 lseek(4, 0, SEEK_SET) = 0
1878 read(4, 0x94c41d0, 512) = -1 EIO (Input/output error)
1878 close(4) = 0
I don't understand. Is new code running as slow as old? How is this possible?
I need to see the output of strace -tt to see
where does it spend most of the time.
--
vda
_______________________________________________
busybox mailing list
[email protected]
http://lists.busybox.net/mailman/listinfo/busybox
