Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Thu, Oct 26, 2006 at 03:25:33AM +0200, Frans Pop wrote: On Wednesday 25 October 2006 23:12, Sven Luther wrote: It seems that nobootloader uses still devfs paths for some reason. The following line : That is not so strange as that line is using the exact same variable $bootfs_devfs as its base that the old code did... Should it be using something different instead? What is the value of $bootfs_disk_syspath and $bootfs_disk if you run the code with 'set -x'? For now I've added a comment that that should probably be changed at some point. (To be very honest, I don't see the point of the code added in 1.10 at all as the devfs path is still used as the base for the whole piece of code; I guess it is in preparation of a further future transition.) Don't be misled by the variable name. It's called $bootfs_devfs because it's the path before calling mapdevfs; if you aren't using devfs paths then $bootfs_devfs and $bootfs are identical. What $bootfs_devfs really means is the version of this path that can be used within d-i. I've just committed an IMHO better fix which will work for both devfs and non-devfs paths. Sven, thanks for the report. -- Colin Watson [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Wed, Nov 22, 2006 at 12:05:45PM +0100, Sven Luther wrote: On Wed, Nov 22, 2006 at 09:32:27AM +, Colin Watson wrote: Don't be misled by the variable name. It's called $bootfs_devfs because it's the path before calling mapdevfs; if you aren't using devfs paths then $bootfs_devfs and $bootfs are identical. What $bootfs_devfs really means is the version of this path that can be used within d-i. I've just committed an IMHO better fix which will work for both devfs and non-devfs paths. Sven, thanks for the report. Hehe, i suppose i will now need to build and test your new version, or will you be able to test it on your pegasos machine ? I need to buy a KVM before I can really get that machine up and running again (which is on the plan once I've recovered my finances a bit from buying a house). I already unit-tested the change with various example paths, of course. -- Colin Watson [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Fri, Oct 27, 2006 at 02:44:22PM -0700, Steve Langasek wrote: On Fri, Oct 27, 2006 at 07:23:43PM +0200, Sven Luther wrote: Ok, Frans, please step back into reality. I filled a correct patch back then, which i had tested and everything. This patch did not get applied, but modified slightly by Colin or you or whoever, which resulted in d-i being utterly broken on pegasos. And then you speak like this to me ? And to add to this the other patches which completed the hypervisor serial ports in finish-install, which i loast because of the conflict caused by not having my patches commited, but merged with a slight difference and with a untimely delay, causing a big mess in my checked out version, and you then later reproaching me that i did not send it. Furthermore, my attitude has been more than correct with you and the d-i team since at least the starting of september, and you have yourself indulged at least four-five times, in practices against me which are the exact echo of what was reproached against me back then. Bla, bla, bla and fuck you too. Sure, exact the same behaviour on your part, ad-hominem attacks and bashings, while i did nothing but be constructive. No, Sven, there is nothing constructive at all about self-righteously Steve, there is a lot whole of constructive thing, in me submitting patches, especially patches for RC-breakage caused by him or Colin applying untested modifications of my patches. And i want you to comment on the fact that Frans continuously pick on me, like making comments like the line which preceeded the 'step back into reality'line above, and which you chose to conveniently cut. How hypocrit of you, to side with Frans so, and to remove the line which would be damaging to him. And then i was told manipulative. telling Frans to step back into reality, and proclaiming your indignance that he would speak like this to you. There is nothing correct about the attitude that you show here. Ah, yeah. And so, there is nothing i ever can do, and i will always be an outcast of d-i, and every effort i could make will always be greated by the same amount of random bashing from Frans. I mean, look at my mail above, it was most reasonable, not aggressive, much more tame that what is usually common on debian mailing lists and irc channels, so what is reproached to me this time ? Come on Steve, you are not being fair to me, and you perfectly know it. Please stop this now, and reinstate my commit right, and stop this insidious and unprovoked bashing. You can be self-righteous, or you can have people be willing to cooperate with you. You can't have both. I have been cooperating, but the only result is that Frans is continuing in his insidious and unprovoked bashing. This in my opinion, makes him lose all justification to be judge and to still keep me as an outcast. Its the fifth time i count where he has been unprovocatedly bashing against me in small ways, and this is not acceptable. So, please, before posting such one-sided unfair stuff, look at what really is happening, which is something quite different from what happened in spring, consider it has been over 6 months now, and think a bit before making such posts. Again, Frans, i ask you to stop this insidious bashing, and if you cannot do it alone, Anthony, as DPL, i take you to your word when you issued that inperfect settlement back then, and revisit the issue today. Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Fri, Oct 27, 2006 at 03:03:29AM +0200, Frans Pop wrote: On Thursday 26 October 2006 19:39, you wrote: Did you see the new version i attached : No, initially I did not see this. Reason is simple: I had already applied your original patch and closed the BR, so I looked in the original bug report instead of in your new mail. Is that my mistake, maybe. But it is also your mistake as you did not check if the original BR was already closed or not. And it _was_ closed at the time you received my mail... Yeah, right, you do that a lot, going in the offensive, and trying to assign blame. I wonder how you can work with folk in that way. Don't you think taking the advice in the constructive way it was given, and replying in a constructive way instead of assigning blame and going into ad-hominem attacks will be more forward-looking ? I know i used to do the same, and i know that you willingly provoked me into that also, but it has been months since i last did so, and a few events have made me reflect on this, but apparently you are still in the old scheme of things. - elif [ $fv1 -ge 2 ]; then + elif [ $fv1 -ge 3 ]; then partition_offset=0 This change still looks completely broken to me. Your reply does not show that you looked into the error I explained _at all_. My request stands: please file this as a new BR and fix the error in it. Totally unrelated. I have looked into the bug, i accept the fact that it may be broken (altough 'completely' broken is not warranted here), but we are discussing something else here, and you are still in your old conflictive behaviour, and think yourself better than me in some way. Please stop this now, i have reflected upon my errors, and what happened with dato made think also, since i believe the various flamewars we had almost all this year (well, not only you and me, but in general), may have affected him some. So, it is past time you put aside your childish pride, stop this continuous bashing, and at least make some effort for us to work together again in a sane and constructive way. The ball is in your camp. Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Fri, Oct 27, 2006 at 03:40:51AM +0200, Frans Pop wrote: On Thursday 26 October 2006 21:04, Sven Luther wrote: Ok, Frans, please step back into reality. I filled a correct patch back then, which i had tested and everything. This patch did not get applied, but modified slightly by Colin or you or whoever, which resulted in d-i being utterly broken on pegasos. And then you speak like this to me ? And to add to this the other patches which completed the hypervisor serial ports in finish-install, which i loast because of the conflict caused by not having my patches commited, but merged with a slight difference and with a untimely delay, causing a big mess in my checked out version, and you then later reproaching me that i did not send it. Furthermore, my attitude has been more than correct with you and the d-i team since at least the starting of september, and you have yourself indulged at least four-five times, in practices against me which are the exact echo of what was reproached against me back then. Bla, bla, bla and fuck you too. This still does not answer why your current patch is broken. When I replied I only looked at this current one BR, and not at the entire history. In this current BR you have made three basic mistakes: - you have combined unrelated patches in one BR - you have changed the patch _after_ it was already applied and the BR closed - both patches had fairly basic errors in them (and you don't even admit to the second error) Sorry if I'm not impressed. Sure, exact the same behaviour on your part, ad-hominem attacks and bashings, while i did nothing but be constructive. Please stop this now, and reinstate my commit right, and stop this insidious and unprovoked bashing. Then we can speak about patches, since it is clear that me playing by the rules set by the DPL is never going to be enough. Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Fri, Oct 27, 2006 at 07:23:43PM +0200, Sven Luther wrote: Ok, Frans, please step back into reality. I filled a correct patch back then, which i had tested and everything. This patch did not get applied, but modified slightly by Colin or you or whoever, which resulted in d-i being utterly broken on pegasos. And then you speak like this to me ? And to add to this the other patches which completed the hypervisor serial ports in finish-install, which i loast because of the conflict caused by not having my patches commited, but merged with a slight difference and with a untimely delay, causing a big mess in my checked out version, and you then later reproaching me that i did not send it. Furthermore, my attitude has been more than correct with you and the d-i team since at least the starting of september, and you have yourself indulged at least four-five times, in practices against me which are the exact echo of what was reproached against me back then. Bla, bla, bla and fuck you too. Sure, exact the same behaviour on your part, ad-hominem attacks and bashings, while i did nothing but be constructive. No, Sven, there is nothing constructive at all about self-righteously telling Frans to step back into reality, and proclaiming your indignance that he would speak like this to you. There is nothing correct about the attitude that you show here. Please stop this now, and reinstate my commit right, and stop this insidious and unprovoked bashing. You can be self-righteous, or you can have people be willing to cooperate with you. You can't have both. -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Thu, Oct 26, 2006 at 03:25:33AM +0200, Frans Pop wrote:
On Wednesday 25 October 2006 23:12, Sven Luther wrote:
It seems that nobootloader uses still devfs paths for some reason. The
following line :
That is not so strange as that line is using the exact same variable
$bootfs_devfs as its base that the old code did...
Yeah, i suppose this should have gone, but the patch that got applied back
then is not the one i wrote and tested, but the one you or more probably
Colin Watson adapted.
Should it be using something different instead? What is the value of
$bootfs_disk_syspath and $bootfs_disk if you run the code with 'set -x'?
I suppose that the code should be rewritten to remove all devfs information, i
am not sure how to do this because i didn't look at the devfs-removal code,
and Colin was handling that.
For now I've added a comment that that should probably be changed at some
point.
(To be very honest, I don't see the point of the code added in 1.10 at all
as the devfs path is still used as the base for the whole piece of code;
I guess it is in preparation of a further future transition.)
Well, my old patch was supposed to check for the version of the pegasos
firmware, and then substract 1 for older firmwares for the disk number.
That said, please also apply the first hunk of the attached patch, which bumps
the check to 1.4, as my colegues released a 1.3 firmware from an older tree,
and i had to bump the version of the version with the correct disk numbering.
P.S. Your patch left the old line in the code instead of replacing it...
Yeah, sorry. Would still have worked though :)
Friendly,
Sven Luther
Index: postinst
===
--- postinst(revision 42226)
+++ postinst(working copy)
@@ -39,12 +39,12 @@
rest=${rest#*.}
fv3=${rest%%.*}
if [ $fv1 -eq 1 ]; then
- if [ $fv2 -eq 2 ] [ $fv3 -ge 99 ]; then
+ if [ $fv2 -eq 3 ] [ $fv3 -ge 99 ]; then
partition_offset=0
- elif [ $fv2 -ge 3 ]; then
+ elif [ $fv2 -ge 4 ]; then
partition_offset=0
fi
- elif [ $fv1 -ge 2 ]; then
+ elif [ $fv1 -ge 3 ]; then
partition_offset=0
fi
fi
@@ -74,14 +74,14 @@
lun=$(echo $bootfs_disk | cut -d:
-f4)
;;
esac
- part=$(($(echo $bootfs_devfs | sed 's/[^0-9]*//') -
$partition_offset))
+ part=$(($(echo $bootfs_devfs | sed 's%^.*part%%') -
$partition_offset))
else
kind=`echo $bootfs_devfs | sed -e 's%/dev/%%' -e
's%/host.*$%%'`
host=`echo $bootfs_devfs | sed -e 's%^.*host%%' -e
's%/bus.*$%%'`
bus=`echo $bootfs_devfs | sed -e 's%^.*bus%%' -e
's%/target.*$%%'`
target=`echo $bootfs_devfs | sed -e 's%^.*target%%' -e
's%/lun.*$%%'`
lun=`echo $bootfs_devfs | sed -e 's%^.*lun%%' -e
's%/part.*$%%'`
- part=$(($(echo $bootfs_devfs | sed -e 's%^.*part%%') -
$partition_offset))
+ part=$(($(echo $bootfs_devfs | sed -e 's%^.*part%%')
- $partition_offset))
fi
# We don't know how to map non ide or scsi disks
Index: changelog
===
--- changelog (revision 42226)
+++ changelog (working copy)
@@ -1,3 +1,10 @@
+nobootloader (1.13) UNRELEASED; urgency=low
+
+ [ Sven Luther ]
+ * Fixed bad sed invocation, which failed on devfs-style paths.
+
+ -- Sven Luther [EMAIL PROTECTED] Wed, 25 Oct 2006 22:59:00 +0200
+
nobootloader (1.12) unstable; urgency=low
[ Updated translations ]
@@ -31,7 +38,7 @@
partitions at 0.
[ Sven Luther ]
- * Update template for Genisi systems. Closes #388591.
+ * Update template for Genesi systems. Closes #388591.
[ Christian Perrier ]
* Avoid splitting a sentence in two parts which can make translations
Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Thursday 26 October 2006 09:10, Sven Luther wrote: That said, please also apply the first hunk of the attached patch, which bumps the check to 1.4, as my colegues released a 1.3 firmware from an older tree, and i had to bump the version of the version with the correct disk numbering. No idea what you're talking about here. I see nothing regarding that in your patch. pgpb7kOIeEFZL.pgp Description: PGP signature
Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Thursday 26 October 2006 09:10, Sven Luther wrote: That said, please also apply the first hunk of the attached patch, which bumps the check to 1.4, as my colegues released a 1.3 firmware from an older tree, and i had to bump the version of the version with the correct disk numbering. Please file a new BR for this as the current one is already closed. Also, the patch looks broken to me as the last change means that the offset will not be set to zero if the firmware major number is 2. So you get: 1.4: 1 1.4 - 1.9: 0 2.0 - 2.9: 1 3.0 -: 0 It really would be nice if you were more careful about your patches. pgpF8ESL9MB8y.pgp Description: PGP signature
Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Thu, Oct 26, 2006 at 01:33:55PM +0200, Frans Pop wrote:
On Thursday 26 October 2006 09:10, Sven Luther wrote:
That said, please also apply the first hunk of the attached patch,
which bumps the check to 1.4, as my colegues released a 1.3 firmware
from an older tree, and i had to bump the version of the version with
the correct disk numbering.
No idea what you're talking about here. I see nothing regarding that in
your patch.
Did you see the new version i attached :
Index: postinst
===
--- postinst(revision 42226)
+++ postinst(working copy)
@@ -39,12 +39,12 @@
rest=${rest#*.}
fv3=${rest%%.*}
if [ $fv1 -eq 1 ]; then
- if [ $fv2 -eq 2 ] [ $fv3 -ge 99 ]; then
+ if [ $fv2 -eq 3 ] [ $fv3 -ge 99 ]; then
partition_offset=0
- elif [ $fv2 -ge 3 ]; then
+ elif [ $fv2 -ge 4 ]; then
partition_offset=0
fi
- elif [ $fv1 -ge 2 ]; then
+ elif [ $fv1 -ge 3 ]; then
partition_offset=0
fi
fi
Friendly,
Sven Luther
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Thu, Oct 26, 2006 at 02:05:58PM +0200, Frans Pop wrote: On Thursday 26 October 2006 09:10, Sven Luther wrote: That said, please also apply the first hunk of the attached patch, which bumps the check to 1.4, as my colegues released a 1.3 firmware from an older tree, and i had to bump the version of the version with the correct disk numbering. Please file a new BR for this as the current one is already closed. Also, the patch looks broken to me as the last change means that the offset will not be set to zero if the firmware major number is 2. So you get: 1.4: 1 1.4 - 1.9: 0 2.0 - 2.9: 1 3.0 -: 0 It really would be nice if you were more careful about your patches. Ok, Frans, please step back into reality. I filled a correct patch back then, which i had tested and everything. This patch did not get applied, but modified slightly by Colin or you or whoever, which resulted in d-i being utterly broken on pegasos. And then you speak like this to me ? And to add to this the other patches which completed the hypervisor serial ports in finish-install, which i loast because of the conflict caused by not having my patches commited, but merged with a slight difference and with a untimely delay, causing a big mess in my checked out version, and you then later reproaching me that i did not send it. Furthermore, my attitude has been more than correct with you and the d-i team since at least the starting of september, and you have yourself indulged at least four-five times, in practices against me which are the exact echo of what was reproached against me back then. I thus now call to our DPL, as well as the remainer of the d-i team. The mediation procedure which was set upon me doesn't seem to be ever enough to satisfy Frans, and i have suffered enough and we are now in a a phase where there is no reason to keep this on, and it is time that the situation is now re-evaluated, and that this childish nonsense is left aside, and my commit rights to the d-i project is re-instated. Anthony, you promised to have a look at this again at some later time, i think now it is this time. Fellow d-i team members, please also speak with Frans, his current attitude is not helping, and i don't believe things will progress in any way from now on in a better direction as long as he keps this continuous and unprovoked bashing up. Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Thursday 26 October 2006 19:39, you wrote: Did you see the new version i attached : No, initially I did not see this. Reason is simple: I had already applied your original patch and closed the BR, so I looked in the original bug report instead of in your new mail. Is that my mistake, maybe. But it is also your mistake as you did not check if the original BR was already closed or not. And it _was_ closed at the time you received my mail... - elif [ $fv1 -ge 2 ]; then + elif [ $fv1 -ge 3 ]; then partition_offset=0 This change still looks completely broken to me. Your reply does not show that you looked into the error I explained _at all_. My request stands: please file this as a new BR and fix the error in it. pgp7EdAYZMEe8.pgp Description: PGP signature
Re: Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Thursday 26 October 2006 21:04, Sven Luther wrote: Ok, Frans, please step back into reality. I filled a correct patch back then, which i had tested and everything. This patch did not get applied, but modified slightly by Colin or you or whoever, which resulted in d-i being utterly broken on pegasos. And then you speak like this to me ? And to add to this the other patches which completed the hypervisor serial ports in finish-install, which i loast because of the conflict caused by not having my patches commited, but merged with a slight difference and with a untimely delay, causing a big mess in my checked out version, and you then later reproaching me that i did not send it. Furthermore, my attitude has been more than correct with you and the d-i team since at least the starting of september, and you have yourself indulged at least four-five times, in practices against me which are the exact echo of what was reproached against me back then. Bla, bla, bla and fuck you too. This still does not answer why your current patch is broken. When I replied I only looked at this current one BR, and not at the entire history. In this current BR you have made three basic mistakes: - you have combined unrelated patches in one BR - you have changed the patch _after_ it was already applied and the BR closed - both patches had fairly basic errors in them (and you don't even admit to the second error) Sorry if I'm not impressed. pgpY1Ebv91x6L.pgp Description: PGP signature
Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
Package: nobootloader Version: 1.12 Severity: grave Tags: patch Justification: renders package unusable It seems that nobootloader uses still devfs paths for some reason. The following line : part=$(($(echo $bootfs_devfs | sed 's/[^0-9]*//') - $partition_offset)) will fail on paths like this one : /dev/ide/host0/bus0/target0/lun0/part4 Since : $ echo /dev/ide/host0/bus0/target0/lun0/part4 | sed 's/[^0-9]*//' 0/bus0/target0/lun0/part4 which causes the calculation to result in a division by zero, thus making it impossible to create a bootable pegasos system. Please apply the below patch to fix this problem. Friendly, Sven Luther -- System Information: Debian Release: testing/unstable APT prefers testing APT policy: (500, 'testing') Architecture: powerpc (ppc) Shell: /bin/sh linked to /bin/bash Kernel: Linux 2.6.15-1-powerpc Locale: LANG=fr_FR.UTF-8, LC_CTYPE=fr_FR.UTF-8 (charmap=UTF-8) Index: nobootloader/debian/postinst === --- nobootloader/debian/postinst (revision 42226) +++ nobootloader/debian/postinst (working copy) @@ -75,13 +75,14 @@ ;; esac part=$(($(echo $bootfs_devfs | sed 's/[^0-9]*//') - $partition_offset)) + part=$(($(echo $bootfs_devfs | sed 's%^.*part%%') - $partition_offset)) else kind=`echo $bootfs_devfs | sed -e 's%/dev/%%' -e 's%/host.*$%%'` host=`echo $bootfs_devfs | sed -e 's%^.*host%%' -e 's%/bus.*$%%'` bus=`echo $bootfs_devfs | sed -e 's%^.*bus%%' -e 's%/target.*$%%'` target=`echo $bootfs_devfs | sed -e 's%^.*target%%' -e 's%/lun.*$%%'` lun=`echo $bootfs_devfs | sed -e 's%^.*lun%%' -e 's%/part.*$%%'` - part=$(($(echo $bootfs_devfs | sed -e 's%^.*part%%') - $partition_offset)) + part=$(($(echo $bootfs_devfs | sed -e 's%^.*part%%') - $partition_offset)) fi # We don't know how to map non ide or scsi disks Index: nobootloader/debian/changelog === --- nobootloader/debian/changelog (revision 42226) +++ nobootloader/debian/changelog (working copy) @@ -1,3 +1,10 @@ +nobootloader (1.13) UNRELEASED; urgency=low + + [ Sven Luther ] + * Fixed bad sed invocation, which failed on devfs-style paths. + + -- Sven Luther [EMAIL PROTECTED] Wed, 25 Oct 2006 22:59:00 +0200 + nobootloader (1.12) unstable; urgency=low [ Updated translations ] @@ -31,7 +38,7 @@ partitions at 0. [ Sven Luther ] - * Update template for Genisi systems. Closes #388591. + * Update template for Genesi systems. Closes #388591. [ Christian Perrier ] * Avoid splitting a sentence in two parts which can make translations
Bug#395259: nobootloader: [powerpc/pegasos] bad sed invocation breaks devfs style paths (division by zero)
On Wednesday 25 October 2006 23:12, Sven Luther wrote: It seems that nobootloader uses still devfs paths for some reason. The following line : That is not so strange as that line is using the exact same variable $bootfs_devfs as its base that the old code did... Should it be using something different instead? What is the value of $bootfs_disk_syspath and $bootfs_disk if you run the code with 'set -x'? For now I've added a comment that that should probably be changed at some point. (To be very honest, I don't see the point of the code added in 1.10 at all as the devfs path is still used as the base for the whole piece of code; I guess it is in preparation of a further future transition.) P.S. Your patch left the old line in the code instead of replacing it... pgpr3UgV3gg1r.pgp Description: PGP signature
