Re: [flashrom] My first question...

[Bernd Blaauw]

Op 30-9-2011 19:09, Marcos Felipe Rasia de Mello schreef:
> > I get differences with the previous one. I haven't enter in the bios or change
> > anything... I was expecting no differences between them. Am I missing
> > anything?
> >
> > Regards
> > Jorge
>
> http://flashrom.org/Random_notes#Bios_content_changes_between_reboots

It might be usefull if Flashrom gets a percentage counter:

'file savebios.bin is 99.3% identical to EEPROM contents, continue at 
your own risk. It's also 50.0 % identical to file asusbios.bin'

instead of something along the lines of 'verification failed, file is 
corrupt'.

Above example would also require a way to reference a 2nd file to verify 
against. Let's say that 2nd file is an internet-downloaded original 
motherboard vendor's BIOS instead of a flashrom dumped file.

After all, people might want to check if flashrom works correctly before 
writing anything. What better way than dump current bios, compare dumped 
file to bios but also compare to vendor bios file. All 3 should be 
(near-)identical.

My first question...

[Stefan Tauner]

On Fri, 30 Sep 2011 19:32:05 +0200
Bernd Blaauw  wrote:
> Op 30-9-2011 19:09, Marcos Felipe Rasia de Mello schreef:
> > > I get differences with the previous one. I haven't enter in the bios or change
> > > anything... I was expecting no differences between them. Am I missing
> > > anything?
> > >
> > > Regards
> > > Jorge
> >
> > http://flashrom.org/Random_notes#Bios_content_changes_between_reboots
>
> It might be usefull if Flashrom gets a percentage counter:
>
> 'file savebios.bin is 99.3% identical to EEPROM contents, continue at 
> your own risk. It's also 50.0 % identical to file asusbios.bin'
>
> instead of something along the lines of 'verification failed, file is 
> corrupt'.
>
> Above example would also require a way to reference a 2nd file to verify 
> against. Let's say that 2nd file is an internet-downloaded original 
> motherboard vendor's BIOS instead of a flashrom dumped file.
>
> After all, people might want to check if flashrom works correctly before 
> writing anything. What better way than dump current bios, compare dumped 
> file to bios but also compare to vendor bios file. All 3 should be 
> (near-)identical.

some additional output in verify mode might indeed be nice. OTOH bsdiff
etc might give similar insights - not that terse and
"interactively" though.

My first question...

[David Hendricks]

On Fri, Sep 30, 2011 at 10:32 AM, Bernd Blaauw  wrote:
> Op 30-9-2011 19:09, Marcos Felipe Rasia de Mello schreef:
>
> I get differences with the previous one. I haven't enter in the bios or
> > > change
> > > anything... I was expecting no differences between them. Am I missing
> > > anything?
> > >
> > > Regards
> > > Jorge
> >
> > http://flashrom.org/Random_**notes#Bios_content_changes_**between_rebootshttp://flashrom.org/Random_notes#Bios_content_changes_between_reboots
>
> It might be usefull if Flashrom gets a percentage counter:
>
> 'file savebios.bin is 99.3% identical to EEPROM contents, continue at your
> own risk. It's also 50.0 % identical to file asusbios.bin'
>
> instead of something along the lines of 'verification failed, file is
> corrupt'.

I disagree. If the ROM images do not match, then Flashrom can make no
guarantee or ascertain as to whether it is corrupt or not. This is entirely
dependent on the implementation of your BIOS.

On Fri, Sep 30, 2011 at 10:32 AM, Bernd Blaauw  wrote:
> Above example would also require a way to reference a 2nd file to verify
> against. Let's say that 2nd file is an internet-downloaded original
> motherboard vendor's BIOS instead of a flashrom dumped file.

Unfortunately, those are unlikely to match because many vendors place
board-specific information such as serial numbers, MAC addresses, etc. in
the firmware ROM.

For example, try running "dmidecode -t 1" on a machine which has never had
its firmware updated. Then change the firmware using Flashrom, reboot, and
run "dmidecode -t 1" again. You will likely see serial numbers, UUIDs, etc.
reset to a "default" value.