On 09/29/14 09:22, Gerd Hoffmann wrote:
> On Sa, 2014-09-27 at 18:06 -0400, Gabriel L. Somlo wrote:
>> Factor out logic to distinguish between Q35 and PIIX4 platforms
>> into a separate header (OvmfPkg/Include/Library/PciHostBridge.h),
>> then refer to these macros from all relevant source locations
>> throughouth OvmfPkg.
>>
>> This patch also adds a Q35-specific PIC IRQ routing initialization
>> function to OvmfPkg/Library/PlatformBdsLib/BdsPlatform.c
>>
>> Contributed-under: TianoCore Contribution Agreement 1.0
>> Signed-off-by: Gabriel Somlo <so...@cmu.edu>
>> ---
>>
>> As discussed with Laszlo and Paolo in an earlier thread, this stuff should
>> hopefully be mostly uncontroversial and "the right thing to do" to support
>> Q35. Doesn't fix my weird OS X / Q35 / UHCI issues, but I guess that's a
>> whole different fight to be fought another day :)
> 
> It is definitively a step into the right direction, even if the puzzle
> isn't completely solved yet.
> 
>> The one thing I'm a bit unsure about is PciInitializationQ35 (BdsPlatform.c);
>> In PciInitializationPIIX (which used to be plain PciInitialization before
>> this patch), there's a whole bunch of other devices (ide, video, network,
>> etc.) being initialized by writing to registers
> 
> That is wrong.  The function should only initialize the chipset, i.e
> devices 0 (northbridge) and 1 (southbridge).  Everything else is just a
> shot into the dark, hoping the devices are there.  Which is the case in
> a default qemu configuration, i.e. if you boot the system with just
> "qemu -hda $image" and let qemu add default vga+nic.  It is possible to
> configure qemu completely different though.

Correct.

If you check SVN r13626 and r13627 (from more than two years ago),
you'll see that I didn't introduce that code, I only sought to preserve
it and to correct it / adapt it to qemu defaults. Background:

  [edk2] [PATCH v2 00/19] OvmfPkg: FADT, DSDT, MADT fixes
  http://thread.gmane.org/gmane.comp.bios.tianocore.devel/187/focus=205

If the devices-related register-massaging code proves removable, then it
should be removed.

Thanks,
Laszlo


------------------------------------------------------------------------------
Slashdot TV.  Videos for Nerds.  Stuff that Matters.
http://pubads.g.doubleclick.net/gampad/clk?id=160591471&iu=/4140/ostg.clktrk
_______________________________________________
edk2-devel mailing list
edk2-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/edk2-devel

Reply via email to