[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#403556: xserver-xorg-video-mga: VT switch dims 16-bit-color screen 0 when screen 1 is 24-bit



Brice Goglin [2007-12-24 15:41]:

Hi,

sorry for the late response.

>>> - The DoubleScan flag seems to be ignored on the VGA2 output.
>>>   I am not entirely sure it worked in the previous version either.
>>>   However, the same mode works correctly on the VGA1 output.
>>>     
>>
>> I looked in xserver-xorg-video-mga_1.4.4.dfsg.1.orig.tar.gz
>> and it seems to ignore the V_DBLSCAN flag on the second CRTC
>> too.  Perhaps the hardware doesn't support that feature?
>> However, in xserver-xorg-video-mga-1.9.100.dfsg.1/mga_driver.c,
>> MGAPreInit does this:
>>
>>     clockRanges->interlaceAllowed = !pMga->SecondCrtc;
>>     clockRanges->doubleScanAllowed = TRUE;
>>
>> If doublescan were not possible on the second CRTC, presumably
>> this would check pMga->SecondCrtc on the second line too?
>>   
>
> Looks correct to me but Tilman will now more :)

This came up some months ago, too. Back then I promised to look into it,
but didn't :(

We need to find out whether the double scan failure is because of the
CRTC, or because of the output. Maybe CRTC2 doesn't support double scan,
but maybe it's output2.

>>> - I don't get a mouse cursor on the second monitor.  At first I
>>>   thought it might be some presentation feature, activated when
>>>   the monitors are displaying the same picture otherwise.
>>>   However it also happens when they are configured side by side.
>>>     
>>
>> In both versions, MGAGUseHWCursor in mga_dacG.c does:
>>
>>     if( pMga->SecondCrtc == TRUE )
>>         return FALSE;
>>
>> In 1.4.4.dfsg.1, MGAPreInitMergedFB used to set pMga->SecondCrtc
>> = TRUE.  This code has apparently been removed from
>> 1.9.100.dfsg.1; pMga->SecondCrtc is now always FALSE.
>> I presume MGAGUseHWCursor should now be somehow hooked to
>> crtc_mode_set in mga_g_crtc2.c, so that it begins returning
>> FALSE when the second CRTC is turned on.

You'll have to use Option "HWCursor" "Off" with xf86-video-mga 1.9.100.
Unfortunately, the cursor layer that comes with the randr 1.2 API
doesn't allow us to use the one hardware cursor that the chip has (tied
to crtc1) and a software cursor for crtc2 :/

HTH,
Tilman

-- 
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail?

Attachment: pgpMFOlUIpuQq.pgp
Description: PGP signature


Reply to: