[Qemu-devel] Re: [PATCH 00/23] VGA cleanup

[Juan Quintela]

Pierre Riteau <address@hidden> wrote:
> On 31 août 2009, at 16:07, Juan Quintela wrote:
>
>> This patch:
>> - cleans the differences between VGAState/VGACommon state
>> - moves vga pci, isa, isa-mm out of vga.c (cirrus and blizzar uses
>>  VGA common state).
>> - Cleans all the cirrus_hook_* stuff is something that don't look
>> like a hook :)
>>
>> I would want this bits morged/reviewed before going the rest of the
>> cleanup:
>>
>> Todo:
>> - vga_ioport_read/write and cirrus_vga_ioport_read/write are almost
>> identical
>>  after this series of changes.  Creating read/write() callbacks for
>>  sr/gr/cr/palette.  Should put us there.  The cirrus ones were the
>> difficult ones,
>>  the vga ones are trivial.
>> - CONFIG_VBE_BOCHS code.  This code is only used by vga std (pci +
>> isa).  Neither
>>  cirrus/blizzard/isa-mm uses it.  It should be moved out of vga.c,
>> vga-std.c?, and
>>  then used by vga-isa and vga-pci?
>> - vmware_vga: it uses vga as embeded, but it don't save its state, I
>> am not sure
>>  if it wants/need VBE_BOCHS, ....  I haven't looked at vmware_vga
>> too much, just
>>  to vga and cirrus_vga.
>>
>> Comments?
>>
>> Later, Juan.
>
>
> This patchset introduced a bug. I noticed today that I can't boot an
> installation disk of ubuntu-9.04-server-i386.
> After entering the install program ("Install Ubuntu Server"), I just
> get a black background with gray vertical lines.
>
> More specifically, from oldest to newest commit:
> 86948bb104f419db9af6b621b85703e8f0d3234c works
> f705db9df04c6491f242a5a4585dfe72b708f197 throws a warning about an
> uninitialized variable (had to disable werror) but works

could you post the error message?  I didn't have that problem.

> 22286bc6468adac10b2eb7e603f1a8ba524bfb03 throws a warning about an
> uninitialized variable (had to disable werror) and doesn't work
> b863d51490b7c6e339c9565eda786cadc1218d48 compiles without warning and
> doesn't work

Weird, it did worked for me.  Could you post what command line did you
use?  I tested with fedora + win XP.

> So 22286bc6468adac10b2eb7e603f1a8ba524bfb03 must be the problematic
> commit.

Thanks, will take a look.