Failed detection of audio-CD fixed

January 27, 2013 — BarryK

01micko reported on this problem:
http://bkhome.org/archive/blog2/201301/xsane-remove-run-as-root-warning.html

I tracked down the cause of the problem. Utility 'probepart' was written in the 2.x kernel days, when /proc/partitions does not list optical media. That is, 'sr0' is not listed, regardless of whether any media is inserted or not.

Sometime, I presume with the 3.x series (I only tested 2.6.32.59, 3.2.29, 3.6.11), /proc/partitions does list sr0.
Initially, with no media in the optical drive, this is what is listed in /proc/partitions:

11 0 1048575 sr0

After inserting a media, the entry changes to reflect the amount of used space in the optical media, for example:

11 0 771392 sr0

...curiously, after removing the media, that size parameter stays the same in /proc/partitions.

Anyway, probepart obtains information about the optical media from /sys/block, and the inclusion of optical drive info in /proc/partitions causes trouble in the script -- the exact trouble reported was the desktop icon not appearing when an audio-CD was inserted.

The solution is to screen sr[0-9] out of /proc/partitions in script 'probepart'. I have done it, see the relevant code portion:

#devices that have partitions...

#([^k] is to eliminate mmcblk0 device -- allow mmcblk0p1 etc) v4.01 bugfix eliminate ram...

#130127 [^kr] screens out sr0-sr9. 2.x kernels do not have these in /proc/partitions, 3.x do, which causes desktop icon to not appear when audio-cd inserted.

PARTITIONS="`grep '^ .*[^kr][0-9]$' /proc/partitions | tr -s ' ' | cut -f 4-5 -d ' ' | grep -vE ' loop| ram'`" #each line ex: 16076800 sda5

PARTNAMES="`echo "$PARTITIONS" | cut -f 2 -d ' '`" #120602

This is quite an important bug fix, affecting all pups. I will need to bring out a Service Pack for Precise soon.

Woof commit:
http://bkhome.org/fossil/woof2.cgi/info/979470b8a8

Comments

/dev/sdr0-9
Username: BarryK
Of course, my fix is going to stuff things up if someone has drives /dev/sdr0+. Possible, but pretty unlikely. It would be an extreme system with that many drives.

3.2 series
Username: 01micko
"I notice the comment in /sbin/probepart script "(tested 2.6.32.29 and 3.2.29+, so not sure exact version this difference came in(tested 2.6.32.29 and 3.2.29+, so not sure exact version this difference came in(tested 2.6.32.29 and 3.2.29+, so not sure exact version this difference came in)" A lot changed in 3.2 series kernels and beyond, as old probepart worked fine in Slacko-5.3.3 with k3.1.10. On another note, we have found a serious bug, which seems to be kernel related (again 3.2 plus). follow slacko discussion from the post linked, (affects precise too apparently). It is a severe slow down of IO under certain conditions, several members are affected. Sorry, no link, forum is down for me, anyway, post is by forum member [b]SFR[/b] on about page 8 of slacko-5.4 bugs reports, and discussion is ongoing. I'll link when I can access the forum.

Forum down
Username: BarryK
"Yes, down for me too, last couple of hours.

Re udf
Username: BarryK
"Ted Dog, Yes, the udf problem has been around for a long time. I have finally decided to fix it properly. Working on it now.

udf support
Username: BarryK
"Done, see later blog post: http://bkhome.org/archive/blog2/201301/full-support-for-udf-filesystem.html Note, probepart has been modified again.

Tags: woof