Dell Optiplex GX620 hangs after firmware update... help

[Michael E Brown]

On Tue, Oct 23, 2007 at 04:03:56PM +0200, davide rossetti wrote:
> On 10/23/07, Matt Domsch <Matt_Domsch at dell.com> wrote:
> > On Mon, Oct 22, 2007 at 11:55:36PM +0200, davide rossetti wrote:
> > > dear all,
> > > I'm so sad :( I ruined my GX620 using formware tools.... I know it's
> > > beta but please help me to get out of it...
> > >
> > > I followed the instructions on
> > > http://linux.dell.com/wiki/index.php/Oss/Firmware_Tools/announce
> > >
> > > I did:
> > > 1) yum install $(inventory_firmware -b)
> > > it found a brand new firmware (A11, old was A03)
> > > 2) apply_updates --yes
> > > 3) apply_updates --yes
> > > 4) apply_updates --yes

This is safe. I do it all the time during my testing... :)

> > >
> > > yes, I did it 3 times as the apply was so quick that I thought there
> > > was some error somewhere...
> > > Then I did a power-down
> > > Then a powen on
> > > and now I'm in the situation that the PC stops during BIOS:
> > > - LCD stay in power-save
> > > - a bunch of beeps are played, then it goes silent
> >
> > Listen to those beeps - they're a code that can be deciphered using this:
> >
> > http://support.dell.com/support/edocs/systems/opgx620/en/ug/A02/tools00.htm#wp1120374
> 
> as reported in a previous email, it is 1-3-2, indicating memory
> problems. So I tried the following:
> - swapped my 2x512MB DIMMs with another fully-functional Dell GX620 2x1GB,
> - but still had the SAME problems (beeping 1-3-2), while my DIMMs on
> the other GX620 was behaving properly.
> 
> > You might also try clearing out the CMOS settings described here:
> > http://support.dell.com/support/edocs/systems/opgx620/en/ug/A02/advfeat0.htm#wp1132466
> 
> bingo! clearing CMOS was the right winning thing. now it's ok.
> 
> - It booted back in A03 BIOS so in the end my first upgrade was not successful

Right. I think I went over it in private email, but for those following
along at home:

The BIOS update uses the "RBU" feature of the BIOS. We basically leave
the BIOS image in RAM, set a bit in CMOS, then warm reboot. The bios
notices the CMOS bit on reboot, finds the image and flashes itself.

If you power-cycle the box rather than warm-booting, the image is lost
from RAM. This normally wont cause a problem. (I do this *all* the time
on my laptop when I am testing.) The machine simply boots the next time
and says the bios update failed.

The RBU method of flashing BIOS is *quite* safe. Since the BIOS does all
teh work, there isnt much *I* can screw up in my utilities that will
brick a box. (I've not yet done it, and I have upgraded dozens of
machines in my lab.) For other things like BMC, it is a different
story...

> - did 'update_firmware --yes'
> - did warm restart
> - A03 BIOS found and upgraded to A11
> - auto reset
> - reboot A11 BIOS
> 
> now I'm happy again ;)
> 
> as to what was wrong with my original procedure I can only guess:
> 1) the fact that I did 'update_firmware --yes' three times in a raw
> 2) the fact that I did poweroff, i.e. cold restart.
> one or both of these may have garbaged some CMOS setting...

It's quite odd. I have not ever seen this on new machines. The older
machines all had crap BIOS with a ton of bugs (misreported checksum
algorithms, byte offsets wrong, etc). It is possible that there is a bug
there, but since you can't reproduce after update I'm not sure I will be
able to figure it out (especially now that the CMOS is clear.)

> thank you so much for your support
> 
> If you what me to do some more testing, please tell me. I'm really
> interested in supporting your work. confronted with the time I spent
> trying to do the BIOS upgrade the traditional way without the floppy
> disk (I tried with a USB floppy drive with no luck, tried to create a
> MSDOS USB flash disk no luck).

Thanks.
--
Michael