Age | Commit message (Collapse) | Author | Files | Lines |
|
commit a245948 (Map FB during screen initialization rather than
pre-initialization) broke DRI1 initialization code. This was due to
frame buffer no longer being mapped prior to DRI1 initialization. Now,
frame buffer is mapped prior to DRI1 initialization.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
There are still some platforms where the cursor will not be displayed,
but it works on some, so it is better if the code get committed.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
In development code adding hardware cursor (bi-color hardware cursor)
support for older devices got committed accidentally.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
This eliminates duplicated second storage location for HI based cursor.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
The HI cursor is still broken in UMS mode for CLE266 and KM400
chipsets.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
All EXA related code should be inside via_exa.c.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
In UMS no acceleration and non-DRI1 EXA acceleration situations,
drm_bo_alloc() was not actually passing the alignment requirement to
function calls that allocate memory. They were blindly specifying
32 byte alignment.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
It appears that CN400 / PM800 chipsets have only 1 hardware icon rather
than 2 for newer graphics engines. Note that hardware icon is used as
cursor rather than the "real" hardware cursor. The fix also applies to
CN333 and PN800 chipsets as well. The fix was tested on VIA Technologies
C3 processor based Wyse Vx0 thin client that has CN333 chipset.
Suggested-by: Eric Kudzin <knoppix1337@yahoo.com>
Tested-by: Eric Kudzin <knoppix1337@yahoo.com>
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Use the actual X.Org X Server version for determining whether or not
ClockRange or ClockRanges should be used. This commit is effectively
redoing commit 8bc53d8 (Compilation hooks to support ClockRange and
ClockRanges). This commit and commit 8bc53d8 are ultimately related to
X Server commit 326429b (modes: Remove the ClockRanges type).
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Color depth and bpp (Bits Per Pixel) are not the same. VIA Technologies
Chrome IGP does not support 32 bit color depth. It does, however,
support 24 bit color depth and 32 bit bpp.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Commit 5622df16bebeffa50442c5e566d37d71b658f910 (Add viaUMSPreInitExit()
for unmapping MMIOs) should have declared viaUnmapMMIO() as a static
function.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
This is for releasing MMIOs when exiting VIAPreInit() in UMS mode.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
GCC 10 now activates -fno-common option by default. This results in
link time errors. Since GCC 10 is still not officially released, this
link time behavior was reproduced on older GCC releases by adding
-fno-common option manually to the generated code compilation script.
Suggested-by: Xavier Bachelot <xavier@bachelot.org>
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
EXA offscreen memory manager is used when DRM is not available.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
The HAVE_DRI macro should have been used when referring to DRI2.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
exaDriverAlloc() ensures that EXA driver struct is zero cleared.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
This code should have been implemented with commit
ac4b39b1c6aafe089e741f039cceb3a5684764ef (Make off screen back frame
buffer the last item to be allocated).
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|
|
Commit 9a6d8615a8ca999528cad288d7858edf5834ae04 (Minor code clean up
inside viaUMSCreate()) introduced a bug that causes a screen
initialization failure if the reserved VRAM is fairly small like 32 MB.
Suggested-by: Eric Kudzin <knoppix1337@yahoo.com>
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
|