[syslinux] syslinux 4.02 hangs at "boot:"

[Gene Cumm]

On Wed, Feb 2, 2011 at 09:36, Alex Zeffertt <alex.zeffertt at eu.citrix.com> wrote:
> On Wed, 2011-02-02 at 12:49 +0000, Gene Cumm wrote:
>> On Wed, Feb 2, 2011 at 06:25, Alex Zeffertt <alex.zeffertt at eu.citrix.com> wrote:
>> > Hi all,
>> >
>> > I have a machine that won't boot syslinux 4.02 but will boot syslinux
>> > 3.53.  We've installed syslinux 4.02 on a number of machines but only
>> > this one has a problem.
>> >
>> > The problem is that the machine hangs entirely after printing "boot:" on
>> > the tty.  The stanza being booted in /boot/extlinux.conf is:
>> >
>> >  label xe
>> >  # XenServer
>> >  kernel mboot.c32
>> >  append /boot/xen.gz dom0_mem=752M lowmem_emergency_pool=1M
>> > crashkernel=64M at 32M console= vga=mode-0x0311 --- /boot/vmlinuz-2.6-xen
>> > root=LABEL=root-ldndebvr ro xencons=hvc console=hvc0 console=tty0 quiet
>> > vga=785 splash --- /boot/initrd-2.6-xen.img
>>
>> 1) There's some global directives that have relevance.  I'd suggest
>> posting your entire config.
>>
>
> Hi Gene,
>
> Thanks for replying.  I've attached the full extlinux.conf below
>
>> 2) For testing, remove the command line option "quiet" as this will
>> change the behavior of any Syslinux variant.  It's possible that it's
>> not hanging at the "boot: " prompt but after it tries to run
>> mboot.c32.
>>
>
> I don't think that removing "quiet" would help as this is a *linux*
> command line parameter and we're not even seeing mboot.c32 output.  On

Yes, it is a Linux command line parameter however it affects the
behavior of Syslinux also.

> working machines we get the following on the tty immediately after the
> "boot:" prompt
>
>  Loading /boot/xen.gz... ok
>  Loading /boot/vmlinuz-2.6-xen... ok
>  Loading /boot/initrd-2.6-xen.img... ok
>
>
>> 3) Have you tried 4.03 or 4.04-pre5?  If not, I'd suggest trying them
>> in this order after removing "quiet" and testing 4.02 without "quiet".
>>
>
> Thanks, I'll have a go at that as soon as I am able.
>
>> 4) Do you have any machines that are identical to this machine that
>> have differing success?  If so, I'd compare BIOS versions.  I had one
>> machine that actually caused more issues with a BIOS upgrade as there
>> was a new memory hole around 32-64MiB.  I know that for Linux kernels,
>> linux.c32 has the new "loader" (I know there's a better word but can't
>> remember it; basically how it loads files into memory in order to work
>> around the memory holes) but I don't know what strategy mboot.c32
>> uses.
>>
>
> Yes we have two machines with
>
>  "CPU: Family - 15, Model - 65, Stepping - 2"
>
> one works and the other does not.  On the working one dmidecode shows
> the following BIOS:
>
>        Vendor: American Megatrends Inc.
>        Version: 080014
>        Release Date: 10/06/2006
>
> and the not-working one has the following BIOS:
>
>        Vendor: American Megatrends Inc.
>        Version: 080011
>        Release Date: 08/28/2006
>
> Could this explain the difference in behaviour?

Easily.  I'd recommend upgrading the non-working machine.

-- 
-Gene