pmount fails in 12.2

Joe 6 Pack · 12-12-2008, 08:20 AM

Hi forum, I'm a first poster but been using Slackware since 12.1 came out. It cured me from distrohopping! Installed it on day one and never had a question.

But now I do. I use fvwm and I previously used pmount to mount my usb sticks and drives. With 12.2 I built and installed pmount from Slackbuilds.org. It installed okay but when I tried to mount any of 3 usb sticks or one usb/firefire drive, I get:

Error: device /dev/sdb1 is not removable

Thunar does work to mount all the devices but since I am in the command line a lot, I thought I'd ask anyway. Any ideas?

Joe

GazL · 12-12-2008, 10:04 AM

Ahh yes, I had a play with pmount the other week. I decided not to use it in the end, but if I remember rightly, pmount has a file called something like /etc/pmount.allow. You have to list your eligible devices in there.

You might want to check the pmount man page to check I've remembered the filename correctly.

p.s. Welcome to LQ, and the Slackware forum.

edit:

Just as an afterthought. If you add an entry to /etc/fstab for your usbsticks with a 'user' option, you may not need to use pmount at all. man pages for fstab and mount will explain it all.

shadowsnipes · 12-12-2008, 01:45 PM

GazL,

pmount.allow is a whitelist intended for internal devices. Yes, it is a workaround until a solution is found, but it is not ideal by any means (though certainly better than defeating pmount by adding an entry in fstab).

Joe_6_Pack,

I, too, am having the same issue as you. The problem seems to lie in something with the newer 2.6.27.7* kernel. My older custom 2.6.24.7 kernel on Slackware 12.2 (and I'm guessing the stock SW 12.1 kernels) does not have this problem, so clearly there is an issue with the kernel interaction. No other variable is changed.

Here is what the problem looks like

Code:

$> pmount-hal /dev/sda1 --debug
resolved /dev/sda1 to device /dev/sda1
Checking for device '/dev/sda1' in '/etc/fstab'
mount point to be used: /media/usbdisk
no iocharset given, current locale encoding is ISO-8859-1
Cleaning lock directory /var/lock/pmount_dev_sda1
Checking for device '/dev/sda1' in '/etc/mtab'
Checking for device '/dev/sda1' in '/proc/mounts'
device_whitelist: checking /etc/pmount.allow...
device_whitlisted(): nothing matched, returning 0
find_sysfs_device: looking for sysfs directory for device 8:1
find_sysfs_device: checking whether /dev/sda1 is on /sys/block/fd0 (2:0)
find_sysfs_device: checking whether /dev/sda1 is on /sys/block/ram0 (1:0)
[...]
find_sysfs_device: checking whether /dev/sda1 is on /sys/block/loop0 (7:0)
[...]
find_sysfs_device: checking whether /dev/sda1 is on /sys/block/hda (3:0)
find_sysfs_device: checking whether /dev/sda1 is on /sys/block/hdb (3:64)
find_sysfs_device: checking whether /dev/sda1 is on /sys/block/hdc (22:0)
find_sysfs_device: checking whether /dev/sda1 is on /sys/block/hdd (22:64)
find_sysfs_device: checking whether /dev/sda1 is on /sys/block/sda (8:0)
find_sysfs_device: major device numbers match
find_sysfs_device: minor device numbers do not match, checking partitions...
find_sysfs_device: checking whether device /dev/sda1 matches partition 8:0
find_sysfs_device: checking whether device /dev/sda1 matches partition 8:1
find_sysfs_device: -> partition matches, belongs to block device /sys/block/sda
device_removable: could not find a sysfs device for /dev/sda1
Error: device /dev/sda1 is not removable
policy check failed
Error: could not execute pmount

Yet...

Code:

$> cat /sys/block/sda/removable 
1

So, perhaps pmount needs a compatibility patch or perhaps a newer sysfsutils or kernel config will fix this problem.

I'm going to link to this thread in the SBo forum.

Joe 6 Pack · 12-12-2008, 11:38 PM

Thank you both for your replies. I tried whitelisting /dev/sdb1 in /etc/pmount.allow, and that certainly worked. shadowsnipes, you mentioned that is not ideal by any means. I'm just curious, why is that? Is there some kind of risk? As I said, it seems to work fine.

T3slider · 12-12-2008, 11:47 PM

Quote:

Originally Posted by Joe 6 Pack

Thank you both for your replies. I tried whitelisting /dev/sdb1 in /etc/pmount.allow, and that certainly worked. shadowsnipes, you mentioned that is not ideal by any means. I'm just curious, why is that? Is there some kind of risk? As I said, it seems to work fine.

I think he meant it's not ideal because pmount is supposed to mount any removeable media on-the-fly without the need for adding devices to configuration files. Using the pmount.allow file for removeable media is not really using pmount to its full capability. It works, and it shouldn't cause any problems. It's just a little annoying.

Joe 6 Pack · 12-13-2008, 12:04 AM

Okay, I see. For me it is more annoying to have to start Thunar!

I just happened to notice there is a newer version of pmount, 0.9.18, at

http://alioth.debian.org/frs/?group_id=31157

I changed the version number in the slackbuild script and made a new package and installed it, but unfortunately that didn't fix the problem.

shadowsnipes · 12-13-2008, 02:22 PM

Quote:

Originally Posted by T3slider

I think he meant it's not ideal because pmount is supposed to mount any removeable media on-the-fly without the need for adding devices to configuration files. Using the pmount.allow file for removeable media is not really using pmount to its full capability. It works, and it shouldn't cause any problems. It's just a little annoying.

Absolutely correct. You can make it a little easier on yourself, however, by using globs in pmount.allow such as

Code:

/dev/sd[a-z]?

Quote:

Originally Posted by Joe 6 Pack

Okay, I see. For me it is more annoying to have to start Thunar!

I just happened to notice there is a newer version of pmount, 0.9.18, at

http://alioth.debian.org/frs/?group_id=31157

I changed the version number in the slackbuild script and made a new package and installed it, but unfortunately that didn't fix the problem.

I don't think this problem can be fixed without a major rewrite for pmount (sysfs dependency) or a kernel reconfiguration. I believe I came across a page in my searching that had one of the authors (Vincent Fourmond) mentioning this (I don't remember where, but I found it through a quick Google search).