[syslinux] extlinux issue

[Gene Cumm]

On Wed, Apr 7, 2010 at 11:59 AM, Thomas Mieslinger
<Thomas.Mieslinger at 1und1.de> wrote:
> Maybe I missunderstand your problem. You want to do something like "localboot 0" to boot off the local harddrive?
>
> Please try
> Localboot 0x80
>
> Or go back to 3.36 which is the last known to me supporting localboot 0 syntax.
>
> BTW, the latest G4p BIOS is dated 07/16/2007
>
> Thomas
>
> P.S. I don't think those G4 Boxes are old... we still have 1850R in service in out Datacenter :-)
>
>> -----Original Message-----
>> From: syslinux-bounces at zytor.com
>> [mailto:syslinux-bounces at zytor.com] On Behalf Of BuraphaLinux Server
>> Sent: Wednesday, March 31, 2010 8:38 AM
>> To: syslinux at zytor.com
>> Subject: [syslinux] extlinux issue
>>
>> Trying again since the mailing list says I have a 'suspiscous header'.
>>
>>    I have been told to get linux running on two HP DL360G4p machines
>> (ancient rackmount xeon stuff).  Unfortunately, while isolinux boots
>> fine from CDROM, and extlinux can boot a USB stick, I cannot get
>> extlinux to boot from the CCISS hard disks (two disks in hardware raid
>> 1 configuration).  I searched on google a lot, and found (very old,
>> many years ago) messages saying these machines have bad BIOS and we
>> should upgrade.  We did find a 2007 BIOS upgrade CD iso image from HP,
>> but it claimed the machines were up-to-date.
>>
>>    1. Is this still a known issue?
>>    2. If so, is there a known workaround besides changing
>> bootloaders, or booting the machine from a cdrom every time?
>>

By "changing bootloaders", I would assume that you are wanting to
stick strictly with EXTLINUX and to not use SYSLINUX (with a FAT
filesystem instead of ext2/3)?

>> We were using extlinux from syslinux-3.85 compiled for 32 bit to boot
>> 64 bit linux-2.6.32.10.   Everything works for all our other
>> equipment, just not these specific machines.
>>
>>     We are scheduled to go back on site in two days to try again, and
>> I was hoping somebody on this list has experience with this hardware
>> and can provide some hints on what to try.  When we boot, we don't get
>> a single byte on the screen from syslinux, the machine just hangs.  We
>> did wait 5 minutes just in case it was slow, but that didn't help.  We
>> used vesamenu, but the exact same vesamenu works great with isolinux
>> booted from a CDROM, so I don't think it is a graphics card issue.
>>
>>    This is with the 'Smart Array 6i Controller'.  The information we
>> could get form the BIOS with F9 was this:
>>
>> Product Id: 380325-371
>> HP BIOS P54 08/16/2005
>> Backup Version 08/16/2005
>> Bootblock 03/16/2004
>>
>> The cpu is Xeon 3 ghz Family 15, Model ID 4, Stepping 3, and
>> hyperthreading and no-execute are enabled in the BIOS.
>>
>> Thank you,
>>
>> JGH

As Thomas mentioned, it looks like there _probably_ are a few
BIOS/firmware upgrades/updates that can be done.  Searching "DL360G4p"
didn't work but "DL360 G4p" did.  It does appear that there is a BIOS
and storage controller firmware update that may be relevant if you
feel it's a BIOS/storage issue.

Your scenario reminds me something I have seen so I feel the need to
ask the extremely obvious.  I've seen systems behave this exact way,
stared at a blank screen, sometimes with a blinking cursor and it
would frustrate me, only to find out that it had blank data in the MBR
boot code.  Have you verified that it does have valid boot code in the
MBR?  Using a hexeditor or something like hexdump can easily help to
quickly see if it's got anything but 0's.  Any time I've initialized
drives on any RAID controller, the resultant target disk seen from the
OS is completely blank.

Another thing that can help to ensure that it is doing something is to
temporarily not use vesamenu.c32 (or even menu.c32) and ensure you get
a prompt.  I agree that it shouldn't make a difference but it can't
hurt.

You could also try using chain.c32 from your CD to chainload to the
MBR of the hard drive or the desired partition of the hard drive as
another test.

I'd also assume you have the target disk partitioned in some way with
an ext2/3 filesystem.

-- 
-Gene