Mailing List Archive: isolinux regression? / problem? v3.50 and newer.

isolinux regression? / problem? v3.50 and newer.

Jan 16, 2008, 7:08 PM

Post #1 of 9 (5026 views)

Greetings,

I have been using isolinux with Microsoft's cdimage.exe for my Utility /
Preinstallation cds. I just upgraded from 3.36 to 3.55 and found it to
be non functional. I didn't use the -debug version, it just hangs at the
version banner. I went through several previous versions, and it appears
that 3.50-pre23 is the last working version. I didn't see anything
between that and 3.50-release?! 3.55 was tested as a burned cd on
several machines, all version testing after was with vmware and iso image.

Of course this could be some change that I didn't see listed, and I need
to modify my procedure for. I am using cdimage as I had issues with
mkisofs previously. I don't recall what / why. I can try mkisofs again
later.

I am calling cdimage like so:
cdimage.exe -Lmycd -t08/23/2001,09:00:00 -b\cd\isolinux\isolinux.bin -h
-n -o -m C:\cd C:\test.iso
cdimage is v 2.47
isolinux.bin and isolinux.cfg is in \isolinux. I do not have a \boot

-t sets the time stamp for all files/dirs
-h includes hidded files/dirs
-n allows long filenames
-o optimizes the image by encoding duplicate files only once

I will investigate playing with these options to see if I can get newer
versions working.

The non-working output (typed):
ISOLINUX 3.50 2007-06-09 Copyright (C) 1994-2007 H. Peter Anvin
isolinux: starting up DL=9F
isolinux: Loaded spec packet OK, drive = 9F
isolinux: No boot info table, assuming single session disk...
isolinux: Spec packet missing LBA information, trying to wing it...
isolinux: Loading main image from LBA = 0000011C
isolinux: Sectors to load = D5B7

and hangs there indefinitely.

The working 3.50-pre23 output (typed):
ISOLINUX 3.50 3.50-pre23 Copyright (c) 1994-2007 H. Peter Anvin
isolinux: Starting up, DL =9F
isolinux: Loaded spec packet OK, drive = 9F
isolinux: No boot info table, assuming single session disk...
isolinux: Spec packet missing LBA information, trying to wing it...
isolinux: Loading main image from LBA = 0000011C
isolinux: Sectors to load = 0005
isolinux: Loaded boot image, verifying...
isolinux: Main image read, jumping to main code...
isolinux: Root directory at LBA = 0000001C
isolinux: isolinux directory at LBA = 00000030
isolinux: About to load config file...
isolinux: Configuration file opened...
boot:

Any ideas? Currently my test image is quite large, I will try and create
a test image that is smaller tomorrow.

-Andrew

_______________________________________________
SYSLINUX mailing list
Submissions to SYSLINUX@zytor.com
Unsubscribe or set options at:
http://www.zytor.com/mailman/listinfo/syslinux
Please do not send private replies to mailing list traffic.

Re: isolinux regression? / problem? v3.50 and newer. [ In reply to ]

hpa at zytor

Jan 16, 2008, 7:57 PM

Post #2 of 9 (4864 views)

Permalink

Andrew Stuart wrote:
> Greetings,
>
> I have been using isolinux with Microsoft's cdimage.exe for my Utility /
> Preinstallation cds. I just upgraded from 3.36 to 3.55 and found it to
> be non functional. I didn't use the -debug version, it just hangs at the
> version banner. I went through several previous versions, and it appears
> that 3.50-pre23 is the last working version. I didn't see anything
> between that and 3.50-release?! 3.55 was tested as a burned cd on
> several machines, all version testing after was with vmware and iso image.
>
> Of course this could be some change that I didn't see listed, and I need
> to modify my procedure for. I am using cdimage as I had issues with
> mkisofs previously. I don't recall what / why. I can try mkisofs again
> later.

mkisofs might help, since the lack of boot info table seems to be biting
you.

> isolinux: No boot info table, assuming single session disk...
> isolinux: Spec packet missing LBA information, trying to wing it...

This means that neither the boot info table nor the spec packet has the
appropriate information.

Clearly something is still wrong in the "winging it" code. I don't get
around to testing it much.

-hpa

_______________________________________________
SYSLINUX mailing list
Submissions to SYSLINUX@zytor.com
Unsubscribe or set options at:
http://www.zytor.com/mailman/listinfo/syslinux
Please do not send private replies to mailing list traffic.

Re: isolinux regression? / problem? v3.50 and newer. [ In reply to ]

andrew at shopcusa

Jan 16, 2008, 9:29 PM

Post #3 of 9 (4868 views)

Permalink

H. Peter Anvin wrote:
> Andrew Stuart wrote:
>
>> isolinux: No boot info table, assuming single session disk...
>> isolinux: Spec packet missing LBA information, trying to wing it...
>>
>
> This means that neither the boot info table nor the spec packet has the
> appropriate information.
>
> Clearly something is still wrong in the "winging it" code. I don't get
> around to testing it much.
>
> -hpa
>
Hopefully me limiting it down to between 3.50-pre23 and 3.50-release
helps, otherwise I just wasted a good part of the day for nothing :)
(I doubt my boss would be pleased..)

You did catch in the initial message that everything was identical up to
Sectors to load = line? It seems both working/non-working versions
figured out to load from the same LBA.. or is the Sectors to load part
of 'winging it' as well?

-A

_______________________________________________
SYSLINUX mailing list
Submissions to SYSLINUX@zytor.com
Unsubscribe or set options at:
http://www.zytor.com/mailman/listinfo/syslinux
Please do not send private replies to mailing list traffic.

Re: isolinux regression? / problem? v3.50 and newer. [ In reply to ]

hpa at zytor

Jan 16, 2008, 9:47 PM

Post #4 of 9 (4862 views)

Permalink

Andrew Stuart wrote:
> Hopefully me limiting it down to between 3.50-pre23 and 3.50-release
> helps, otherwise I just wasted a good part of the day for nothing :)
> (I doubt my boss would be pleased..)

It certainly should make it much easier to get a clue about what might
be going on. I'm a bit confused, however, as there are no (zero) source
code changes between 3.50-pre23 and 3.50-release. This might indicate a
toolchain problem, so I'm going back to look at the archive binaries.

> You did catch in the initial message that everything was identical up to
> Sectors to load = line? It seems both working/non-working versions
> figured out to load from the same LBA.. or is the Sectors to load part
> of 'winging it' as well?

Certainly it ends up with a totally bogus sector count (whereas the LBA
looks sane)... that also narrows it down tremendously.

-hpa

_______________________________________________
SYSLINUX mailing list
Submissions to SYSLINUX@zytor.com
Unsubscribe or set options at:
http://www.zytor.com/mailman/listinfo/syslinux
Please do not send private replies to mailing list traffic.

Re: isolinux regression? / problem? v3.50 and newer. [ In reply to ]

hpa at zytor

Jan 16, 2008, 9:55 PM

Post #5 of 9 (4870 views)

Permalink

Andrew Stuart wrote:
> Hopefully me limiting it down to between 3.50-pre23 and 3.50-release
> helps, otherwise I just wasted a good part of the day for nothing :)
> (I doubt my boss would be pleased..)

Looking at the archive binaries I can see what happened.

The only change from 3.50-pre23 to 3.50 was a Makefile change. That
Makefile change apparently broke filling in the default values of the
isolinux boot info table, which are only used if mkisofs doesn't fill in
those values (with -boot-info-table).

So *thank you* for narrowing it down so closely, it made it very easy to
track this problem down. I will produce a fix and put it into 3.61-pre2.

-hpa

_______________________________________________
SYSLINUX mailing list
Submissions to SYSLINUX@zytor.com
Unsubscribe or set options at:
http://www.zytor.com/mailman/listinfo/syslinux
Please do not send private replies to mailing list traffic.

Re: isolinux regression? / problem? v3.50 and newer. [ In reply to ]

andrew at shopcusa

Jan 16, 2008, 10:02 PM

Post #6 of 9 (4865 views)

Permalink

I uploaded a zip with both testing isos, and the files needed to build
the isos. I removed all extra switches and confirmed it still didn't
work. Get it from www.shopcusa.com/files/testcd.zip I will give mkisofs
a go again. For some reason previously one of the programs on my cd
wouldn't work when created with mkisofs.

_______________________________________________
SYSLINUX mailing list
Submissions to SYSLINUX@zytor.com
Unsubscribe or set options at:
http://www.zytor.com/mailman/listinfo/syslinux
Please do not send private replies to mailing list traffic.

Re: isolinux regression? / problem? v3.50 and newer. [ In reply to ]

hpa at zytor

Jan 16, 2008, 10:33 PM

Post #7 of 9 (4854 views)

Permalink

Andrew Stuart wrote:
> I uploaded a zip with both testing isos, and the files needed to build
> the isos. I removed all extra switches and confirmed it still didn't
> work. Get it from www.shopcusa.com/files/testcd.zip I will give mkisofs
> a go again. For some reason previously one of the programs on my cd
> wouldn't work when created with mkisofs.

I just uploaded syslinux-3.61-pre2. Could you try it out? I believe it
should resolve your problem.

-hpa

_______________________________________________
SYSLINUX mailing list
Submissions to SYSLINUX@zytor.com
Unsubscribe or set options at:
http://www.zytor.com/mailman/listinfo/syslinux
Please do not send private replies to mailing list traffic.

Re: isolinux regression? / problem? v3.50 and newer. [ In reply to ]

andrew at shopcusa

Jan 16, 2008, 10:42 PM

Post #8 of 9 (4859 views)

Permalink

H. Peter Anvin wrote:
> Andrew Stuart wrote:
>
>> Hopefully me limiting it down to between 3.50-pre23 and 3.50-release
>> helps, otherwise I just wasted a good part of the day for nothing :)
>> (I doubt my boss would be pleased..)
>>
>
> Looking at the archive binaries I can see what happened.
>
> The only change from 3.50-pre23 to 3.50 was a Makefile change. That
> Makefile change apparently broke filling in the default values of the
> isolinux boot info table, which are only used if mkisofs doesn't fill in
> those values (with -boot-info-table).
>
> So *thank you* for narrowing it down so closely, it made it very easy to
> track this problem down. I will produce a fix and put it into 3.61-pre2.
>
> -hpa
>
Thank you for your efforts, and the quick fix.

As a side thought.. since I too diffed the two trees and came to the
same conclusion after you said there was no source changes (looking to
see if the packages had been tampered with somehow?). I noticed most of
the change in the Makefile seemed to do with nasm. Any chance that will
help / correct the problem with nasm 0.99.06 that was someone noticed
back in November?

-A

_______________________________________________
SYSLINUX mailing list
Submissions to SYSLINUX@zytor.com
Unsubscribe or set options at:
http://www.zytor.com/mailman/listinfo/syslinux
Please do not send private replies to mailing list traffic.

Re: isolinux regression? / problem? v3.50 and newer. [ In reply to ]

hpa at zytor

Jan 16, 2008, 10:56 PM

Post #9 of 9 (4861 views)

Permalink

Andrew Stuart wrote:
> Thank you for your efforts, and the quick fix.
>
> As a side thought.. since I too diffed the two trees and came to the
> same conclusion after you said there was no source changes (looking to
> see if the packages had been tampered with somehow?). I noticed most of
> the change in the Makefile seemed to do with nasm. Any chance that will
> help / correct the problem with nasm 0.99.06 that was someone noticed
> back in November?
>

No, that was a bug in NASM 0.99.x (which was an alpha development branch
of NASM and never meant for production use). Current NASM - 2.00
(release) or 2.01rc2 (beta) should work fine.

-hpa

_______________________________________________
SYSLINUX mailing list
Submissions to SYSLINUX@zytor.com
Unsubscribe or set options at:
http://www.zytor.com/mailman/listinfo/syslinux
Please do not send private replies to mailing list traffic.