At Mon, 12 Dec 2011 14:07:50 -0200, Gustavo Maciel Dias Vieira wrote:
Em Seg, 2011-12-12 às 15:04 +0100, Takashi Iwai escreveu:
At Mon, 12 Dec 2011 11:55:59 -0200, Gustavo Maciel Dias Vieira wrote:
Unfortunately no. Actually, it appears someone at HP is pulling a
prank
on us. In a Pavilion dm4 the BIOS string is "HP_Mute_LED_P_G", where
P
and G are not numbers, but the actual letters.
Doh, they forgot to fill the right numbers there. Obviously a BIOS bug.
For such machines, the IDs should be explicitly checked instead of marking all machines with this codec. That is, the influence must be as small as possible. Or, just check "HP_Mute_LED_P_G" string as an identifier of bad machines, and apply a special handling.
I can do both, to be double sure. Just tell me which ID to use. I assumed that codec->subsystem_id was unique enough (by inspecting hp_blike_system(u32))
Yes, subsytem_id can be used. OTOH, I guess HP_Mute_LED_P_G check would be easier and may cover more machines. So, I prefer the latter.
BTW, now IDT guys and I are discussing about dropping the fallback case for 92HD83x codecs because it can break potentially the EAPD. Your patch won't work any more if that change is applied. So, a more solid fix would be needed anyway.
It's a BIOS bug. Any fix is going to suck. :)
Yeah, but we try to keep less damage by the suck :)
thanks,
Takashi