Subject: Re: OPTI bridge chipset bios bugs?
To: None <jonathan@DSG.Stanford.EDU>
From: Hayakawa Koichi <>
List: current-users
Date: 04/13/1999 10:12:40

From: Jonathan Stone <jonathan@DSG.Stanford.EDU>

 > The CardBus patches include a workaround for two OPTI PCI devices, the
 > (e PCI_PRODUCT_OPTI_82C558 and e PCI_PRODUCT_OPTI_82C700) which
 > according to the comment don't have interrupts enabled properly on at
 > least some BIOSes.

That's it.  Some laptops (especially old one) and deskside
computers don't route PCI INT to CardBus bridges.  This
workaround is really dirty hack and it must be fixed.

Anyway, CardBus bridge without PCI INT routing must be a big
problem.  I got a patch from Mr Uchimaya, which re-make
entire PCI device tree.  The patch re-numbers PCI device
number, PCI bus number and re-route PCI INT.  This treatment
must be required, because ordinary BIOSes on deskside
computers do neither route PCI INT nor set PCI subordinate
bus number.

 > If that's really what's going on, shouldn't we be doing it for 1.4
 > as well?

I think this is too dirty to incorporate into 1.4 tree.  I
suppose some better solution is required.