Syslinux - Dec 2010 - Looking for debugging ideas and help with testing

Hi,

Please let me start a new thread for this.  During the past weeks Debian
accumulated three bug reports, which seem to be the manifestations of a
Syslinux regression between 3.86 and 4.00 (see #604245):

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=604245
  Sony vaio Z12C5E
  USB boot of
http://ftp.nl.debian.org/debian/dists/testing/main/installer-amd64/current/images/hd-media/boot.img.gz

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=604560
  Acer Aspire One 531h-0Bk
  USB boot of
http://people.debian.org/~joeyh/d-i/images/daily/hd-media/boot.img.gz

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=603960
  Toshiba U205-S5057
  CD boot of
http://cdimage.debian.org/cdimage/squeeze_di_beta1/i386/iso-cd/debian-squeeze-di-beta1-i386-netinst.iso
    (now available as beta2 only)

#604245, which is the only closely examined one, was crossposted here on
the Syslinux list as well.  The submitter of the second didn't answer
our requests for further info, unfortunately.  I talked with that of the
third on IRC, he seemed fairly eager to help, but I decided to come here
first to gather some debugging ideas and perhaps testing by the Syslinux
community, as the issue seems to affect a rather diverse class of
laptops -- but not mine, so I can't debug this myself.

The theory is that the boot.img.gz-s fail to boot correctly if written
to a full disk device (/dev/sda), but work fine if written to a
partition (/dev/sda1) and invoked by some MBR (see #604245).  No theory
about the ISO, but the reported symptoms are rather similar.  Anybody
inclined to help, please test whether the above images (or other similar
ones) fail to boot the Debian Installer on their systems.  Thanks.

Concerning the bug reports themselves: should I ask the reporter(s) to
try an image with isolinux-debug.bin?  I could create one without all
the distraction of the graphics and the menu.  Is there something like
ldlinux-debug.sys easily buildable?  Should I ask them to run meminfo.c32
(if possible by some other means)?  Or HDT?  Or something else?

These are some things that came to my mind, but I'm not particularly
experienced on this field, so comments and ideas most welcome!  Debian
uses Syslinux 4.02, btw.
-- 
Thanks,
Feri.

Most likely the problem is that the newer Syslinux misidentifies
something in memory as a partition handover structure.

Someone -- and I can't remember who right now -- built a debugging MBR
for testing this case.

On 12/13/2010 02:34 PM, Ferenc Wagner wrote:
> 
> The theory is that the boot.img.gz-s fail to boot correctly if written
> to a full disk device (/dev/sda), but work fine if written to a
> partition (/dev/sda1) and invoked by some MBR (see #604245).  No theory
> about the ISO, but the reported symptoms are rather similar.  Anybody
> inclined to help, please test whether the above images (or other similar
> ones) fail to boot the Debian Installer on their systems.  Thanks.
> 
	-hpa