Timeout problems with "blkid" applet

[Denys Vlasenko]

On Tuesday 10 February 2009 10:58, Alexander Griesser wrote:
> I noticed a strange timeout issue with the "new" blkid applet.
> When running "blkid" on a busybox system (v1.13.1 and older versions
> too) where _NO_ floppy drive is connected (f.ex. Thinclients) or where
> the floppy drive needs some special kernel parameters, the command takes
> ages to complete.
> 
> In dmesg I can then see entries like:
> 
> end_request: I/O error, dev fd0, sector 0
> Buffer I/O error on device fd0, logical block 0
> ...
> 
> On the Thinkpad Notebooks I worked around that by adding
> "floppy=thinkpad" to the kernel command line, but that is not a proper
> solution because my desktop "blkid" does not have the problem, so there
> must be something wrong in the device enumeration or something like
> that.
> 
> Here's an example:
> 
> [usenet-test - ~ #] /system/blkid -v
> blkid 1.0.0 (12-Feb-2003)
> [usenet-test - ~ #] /system/blkid
> /dev/hda1: LABEL="LXTC_BOOT" UUID="9406c03b-e4dc-4cbe-867a-b18d3fceec15"
> TYPE="ext2"
> [usenet-test - ~ #]
> 
> This is blkid from my desktop system copied over to the thinclient.
> The command exits immediately printing the blkid for hda1, so, that
> works a treat.

Can you run "strace -o LOG /system/blkid" and post LOG file?

> Now, blkid from busybox v1.13.1:
> 
> [usenet-test - ~ #] blkid --help
> BusyBox v1.13.1 (2008-12-09 10:34:51 CET) multi-call binary
> 
> Usage: blkid
> 
> Print UUIDs of all filesystems.
> 
> [usenet-test - ~ #] blkid
> ---- now I have to wait for approx. 4m30s ----
> /dev/ram0: LABEL="Compressed" TYPE="cramfs"
> /dev/hda1: LABEL="LXTC_BOOT" UUID="9406c03b-e4dc-4cbe-867a-b18d3fceec15"
> SEC_TYPE="ext2" TYPE="ext3"
> [usenet-test - ~ #]
> 
> So, 4m30s is a bit tough in this case.

Can you produce a strace of this one too?

> Additionally, why does BB blkid show /dev/ram0 whereas the GNU(?) blkid
> doesn't?

busybox's blkid scans all block devices present in /dev.
Please provide more info, I might be able to add
code which probes for precense of removable media
and skips it if not present.

> And, SEC_TYPE is also only in the BB output by default.

What SEC_TYPE? I see no SEC_TYPE in busybox output.
--
vda