ntfs.c failure in udev128

ntfs.c failure in udev128

[Scott Tester]

I am having problems with udev recognizing an ntfs filesystem under udev128.
Strangly, another ntfs partition is being recognized. (One formatted
using mkfs.ntfs.)

(I have added extra info() lines to ntfs.c and util.c as my syslog
config is not picking up debug statements.)
Under udev-127:
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: ntfs.c:134
sectorsize  0x200
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: ntfs.c:135
clustersize 0x1000
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: ntfs.c:136
mftcluster  786432
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: ntfs.c:137
mftoffset  0xc0000000
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: ntfs.c:138
cluster per mft_record  -10
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: ntfs.c:139
mft record size  1024
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: util.c:356
disk: 5
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: util.c:357
get buffer off 0xc0000c00(3221228544), len 0x400
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: util.c:406
read seekbuf off:0xc0000c00 len:0x400
Sep 18 11:06:54 coota hald-probe-volume: 5310: 11:06:54.657: util.c:407
lseek(5, 0xc0000c00, 0)
Sep 18 11:06:54 coota hald-probe-volume: [5310]: 11:06:54.658 [I]
probe-volume.c:623: volume_id_probe_all returned 0
Sep 18 11:06:54 coota hald-probe-volume: [5310]: 11:06:54.658 [D]
probe-volume.c:107: volume.fsusage = 'filesystem'
Sep 18 11:06:54 coota hald-probe-volume: [5310]: 11:06:54.658 [D]
probe-volume.c:113: volume.fstype = 'ntfs'
Sep 18 11:06:54 coota hald-probe-volume: [5310]: 11:06:54.658 [D]
probe-volume.c:118: volume.fsversion = '3.1'
Sep 18 11:06:54 coota hald-probe-volume: [5310]: 11:06:54.658 [D]
probe-volume.c:123: volume.uuid = '5A20835C20833E4F'
Sep 18 11:06:54 coota hald-probe-volume: [5310]: 11:06:54.658 [D]
probe-volume.c:135: volume.label = 'mytest'
Sep 18 11:06:54 coota hald-probe-volume: [5310]: 11:06:54.658 [I]
probe-volume.c:695: Loading part table

Under udev-128:
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: ntfs.c:138
sectorsize  0x200
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: ntfs.c:139
clustersize 0x1000
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: ntfs.c:140
mftcluster  786432
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: ntfs.c:141
mftoffset  0xc0000000
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: ntfs.c:142
cluster per mft_record  -10
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: ntfs.c:143
mft record size  1024
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: util.c:354
disk: 5
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: util.c:355
get buffer off 0xc0000c00(3221228544), len 0x400
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: util.c:404
read seekbuf off:0xc0000c00 len:0x400
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: util.c:405
lseek(5, 0xc0000c00, 0)
Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: util.c:407
lseek failed (Invalid argument)

I can run with udev127 till future ones work.

Is there a bug manager for udev?
-- 

Scott Tester

IT Department

SMBC  www.smbc.com.au <http://www.smbc.com.au>

Sydney Missionary and Bible College

43 Badminton Road, Croydon NSW 2132

Phone: 02 9747 4780  ext 147

Mobile: 0411 145 584

Re: ntfs.c failure in udev128

[Robby Workman]

Resending to list - sorry for the bad CC to -owner :/

On Thu, 18 Sep 2008 11:34:59 +1000
Scott Tester <scott@smbc.com.au> wrote:

> I am having problems with udev recognizing an ntfs filesystem under
> udev128. Strangly, another ntfs partition is being recognized. (One
> formatted using mkfs.ntfs.)
> 
> (I have added extra info() lines to ntfs.c and util.c as my syslog
> config is not picking up debug statements.)
> Under udev-127:


...snip...


Have you rebuilt hal since you upgraded udev?  If not, you
probably/usr/libexec/hald-probe-{storage,volume} should do so.
libvolume_id version in udev was bumped, and the 
/usr/libexec/hald-probe-{storage,volume} files are linked to the old
one, so they don't work anymore.  There's also a patch in hal's git
repo to use udevadm instead of depending on the udevinfo symlink, so
you'll want that as well.

-RW

Re: ntfs.c failure in udev128

[Sergey Vlasov]

[-- Attachment #1: Type: text/plain, Size: 857 bytes --]

On Thu, 18 Sep 2008 11:34:59 +1000 Scott Tester wrote:

> I am having problems with udev recognizing an ntfs filesystem under udev128.
> Strangly, another ntfs partition is being recognized. (One formatted
> using mkfs.ntfs.)
[...]
> Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: util.c:405
> lseek(5, 0xc0000c00, 0)
> Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: util.c:407
> lseek failed (Invalid argument)

EINVAL from lseek() might mean that the code was compiled without
large file support (-D_FILE_OFFSET_BITS=64).  AC_SYS_LARGEFILE is used
in configure.ac - it should add this option, but maybe this is not
working for some reason, which could be found by analyzing the build
output and maybe config.log.

udev <= 127 did not use autotools, and the -D_FILE_OFFSET_BITS=64
option was hardcoded in the toplevel Makefile.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: ntfs.c failure in udev128

[Kay Sievers]

On Fri, Sep 19, 2008 at 02:52, Sergey Vlasov <vsu@altlinux.ru> wrote:
> On Thu, 18 Sep 2008 11:34:59 +1000 Scott Tester wrote:
>
>> I am having problems with udev recognizing an ntfs filesystem under udev128.
>> Strangly, another ntfs partition is being recognized. (One formatted
>> using mkfs.ntfs.)
> [...]
>> Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: util.c:405
>> lseek(5, 0xc0000c00, 0)
>> Sep 18 10:24:06 coota hald-probe-volume: 4728: 10:24:06.793: util.c:407
>> lseek failed (Invalid argument)
>
> EINVAL from lseek() might mean that the code was compiled without
> large file support (-D_FILE_OFFSET_BITSd).  AC_SYS_LARGEFILE is used
> in configure.ac - it should add this option, but maybe this is not
> working for some reason, which could be found by analyzing the build
> output and maybe config.log.
>
> udev <= 127 did not use autotools, and the -D_FILE_OFFSET_BITSd
> option was hardcoded in the toplevel Makefile.

Oh, we might need to include config.h?
Is there something set in your config.h regarding LARGE_FILE, FILE_OFFSET?

Kay