Hi,
Unfortunately I'm currently without the usbmount using box, so I can't
test anything.
In git.katzien.de/usbmount.git (only available via git, not via
webinterface) are my latest attempts to fix some bugs in and add some
features to usbmount. This three commits may help with that problem, as
the mountpoint strategy is changed:
* [ac0c22cc] Add mountpoints based on vendor+model names. Patch
adapted from Andreas Pakulat, thanks! (Closes: #321694)
* [030096bf] new strategy for mountpoints: handle it completely in
usbmount, no preinstalled mountpoints nessesary. Oldstyle /media/usbX
is still the default.
* [4117d02b] Don't install the default mountpoints and remove them in
postinst, when upgrading from an older version
I've still locking enabaled, but the mountpoints are unique
(vender_model_deviceidXXX), so adding multiple partitions (with
different devideIDs) should not result in races even with locking
removed. But that interpretation is from looking at the code after
almost a year...
Kind regards,
Jan
On 12.08.2011 09:26, Hagen Fuchs wrote:
Package: usbmount
Severity: normal
Hello Jan,
I have just looked at #403209 and the patch proposed there seems to me
to lead straight into race conditions upon finding a free mount point
(see there for further reference).
The problem you described sounds almost as if you had the #403209-patch
applied, but I assume there's a different problem at work. I will try
to find and reactivate an old USB-IDE and partition it heavily in order
to try and reproduce your report.
Your patch, by the way, looks good to me! usbmount only needs to lock right
before search a free mount point and putting logically related code in a
separate function is almost always a good idea. So, yeah, I think we
should use that.
Best Regards,
Hagen Fuchs
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
