IBM (and other) quirks mixed up?

[Richard Hughes]

On Wed, 2007-04-25 at 18:21 -0400, Zack Cerza wrote:
> The solution for ATI chips they recommended was the s3_bios quirk.
> Only the IBM quirks file [1] lists the quirks for my laptop (2525) as
> vbe_post and vbestate_restore. I don't know if there are somehow other
> 2525s that need those quirks and not s3_bios, but I do know the
> current fdi file is broken for mine.

Is the 2525 always the same video card? If so, then we should just
change the fdi to what works for you. Patches gleefully accepted! ;-)

> >From [2]:
>   516     // X41
>   517     { "IBM",            "1864*",        "",    "", VBE_POST|
> VBE_SAVE|UNSURE },
>   518     { "IBM",            "1865*",        "",    "", VBE_POST|
> VBE_SAVE|UNSURE },
>   519     { "IBM",            "2525*",        "",    "", VBE_POST|
> VBE_SAVE|UNSURE },
>   520     { "IBM",            "2526*",        "",    "", VBE_POST|
> VBE_SAVE|UNSURE },
>   521     { "IBM",            "2527*",        "",    "", VBE_POST|
> VBE_SAVE|UNSURE },
>   522     { "IBM",            "2528*",        "",    "", VBE_POST|
> VBE_SAVE|UNSURE },
> I'm guessing this is where hal's information comes from, but the lack
> of a confirmation note in that file, together with the 'UNSURE' note
> suggests to me that maybe we can change this safely. In fact, a *lot*
> of the entries in that file are 'UNSURE', and also coincidentally list
> the VBE_POST and VBE_SAVE quirks. I wonder if those are wrong, too.

Sure - they could be.

> As an aside, something's wrong with this picture (from [1]):
>   25       <!-- T42p -->
>   26       <match key="smbios.system.product" prefix="2371">
>   27         <match key="smbios.system.version" string="ThinkPad X40">
>   28           <merge key="power_management.quirk.s3_bios"
> type="bool">true</merge>
>   29           <merge key="power_management.quirk.s3_mode"
> type="bool">true</merge>
>   30         </match>
>   31       </match>
> 
> Which model is that again? ;)

The comment is wrong, the version check is correct, iirc.

Richard.