Re: Rolling back Computer device in device manager (hal.dll) after installing SP2 gives BSOD after restarting?

Well, this is your opinion which I respect.

However, in my world, that is my REAL word, I cannot afford to create 15
different ghost images for 15 differents motherboards.

I create a single image for all mi IntelChipset PIV computers, I test, and
everything works fine. (I can assure this to you, 2 years since XP deployed
in 10's of workstations, and 0 incidences related to this problem)

The only problem I have had until now has been the problem I described in my
earlier postings...

The best solution is the solution that suits my needs.

"Walter Clayton" <w-claytonNO@xxxxxxxxxxxx> escribió en el mensaje
news:u44JAsOVFHA.4092@xxxxxxxxxxxxxxxxxxxxxxx
> <cuentanewslabogel@xxxxxxxxxxxxxx> wrote in message
> news:unKRTQJVFHA.3532@xxxxxxxxxxxxxxxxxxxxxxx
>> Yes, in part I agree with you, we've allways done that with Intel
>> Chipsets, in similar models.
>
> You're walking on thin ice. Similar models means what? Same chipset *and*
> rev *and* BIOS level, or just a chipset that happens to be from the same
> vendor and it's 'close enough' because it's of the same generation and has
> the same feature set, approximately?
>
>>
>> I'd never do that from VIA Chipsets to Intel, or viceversa without a
>> repair CD...
>
> And the reasons why you do repair installs in that instance is the same
> reason why you do a repair when swapping between Intel chipsets that
> support SMP/HT/APIC and those that don't. The fact that you're not
> changing vendor does not make it safe.
>
>>
>>>
>>> You want pre-SP2 kernel code to run with SP2 code. Why do you think that
>>> should be possible?
>>
>> Then why is it possible to rollback the driver in Windows XP SP2 GUI? I'm
>> doing something unsupported via the XP GUI? If it is unsupported, then
>> gray it out in device manager!!
>
>
> *Some* drivers are the same, especially peripherals and/or 3rd party. MS
> has to draw the line somewhere and assume at least a modicum of
> intelligence on the part of someone messing around with drivers. The sole
> intent of driver roll back is if something goes splat immediately after
> installing a driver (or shortly there after). If you've been running a
> driver for more than a few days, rolling it back, *might* become
> problematic depending on what else has been updated in the interim.
>
> There are also reason why it can be necessary to change the HAL sans
> setup. For example, remotely troubleshooting a suspected failing processor
> on an headless SMP server. In this instance there is no change in the
> chipset or enumeration; you're simply disabling one of the processors.
> However switching between an ACPI HAL and a standard HAL is dangerous.
>
> Playing with drivers is a crap shoot. Always has been and always will be.
> Playing with drivers at the abstraction layer is a real crap shoot
> especially if you're trying to cross service pack levels.
>
> --
> Walter Clayton
> Any technology distinguishable from magic is insufficiently advanced.
>
>


.

Relevant Pages

  • Re: Epox EP-8RDA3+ motherboard
    ... Nvidia motherboard chipsets just aren't Linux-friendly -- because ... Linux can deal with basically any ol' CPU and any ol' RAM circuitry; ... but doesn't list the driver. ...
    (comp.os.linux.hardware)
  • Re: [opensuse] KDE3-4 did teams change?
    ... KDE3 is still in KDE subversion repository and currently it will ... know if there will be a fix for the i965 and other Intel video chipsets ... now no driver update has occured. ... Intel video drivers and has been reported, ...
    (SuSE)
  • Re: Any known issues with MSI K8T Neo F* VIA K8T800 Socket 754 Motherboards?
    ... > driver in the drivers/ide collection has good enough support has been ... some motherboards come with extra drive controllers. ... I've only seen chipsets by NVidia, Via, and SiS. ... it appears that the NForce3 has the best I/O characteristics (lower CPU ...
    (comp.os.linux.hardware)
  • Re: Marvell chipsets on 8-CURRENT and XP x64 wont talk with one another
    ... The issue indeed is with the mskdriver in FreeBSD. ... If you use forced speed/duplex settings, ... because I think they come with mskcompatible chipsets onboard.. ... Bad checksum seems to be different issue to me. ...
    (freebsd-net)