summaryrefslogtreecommitdiff
path: root/src
AgeCommit message (Collapse)Author
2007-08-08Fixed major blending and compositing breakage.Tilman Sauerbeck
Do not adjust the destination blend factor if we need source alpha but our source doesn't have an alpha channel. This makes no sense at all, because the intermediate picture will always have alpha bits (either from the original source, or forced to 0xff).
2007-07-30Bug #4249: Further increase the minimum dot clock.Tilman Sauerbeck
2007-07-27Change minimal frequency guess from 12 to 16MHzBrice Goglin
G550 seems to lock up with frequencies below 16MHz. Reported by Benjamin Wong <hackerb9@u.washington.edu> in http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=430112 and https://bugs.freedesktop.org/show_bug.cgi?id=11339
2007-07-19Add Xv support for Millennium II chipsGabriele Gorla
Bug 11537
2007-04-15Added memcpy-based DownloadFromScreen implementation.Tilman Sauerbeck
2007-04-15On G550, do A8 adds in hardware.Tilman Sauerbeck
2007-04-15Make DRI initialization work when using EXA.Tilman Sauerbeck
Still need to implement InitBuffers and MoveBuffers for EXA.
2007-03-30Removed the debug messages for solid and copy ops.Tilman Sauerbeck
2007-03-30Sync after UploadToScreen.Tilman Sauerbeck
2007-03-28Default mode on G200SE of 1024x768 @ 16bpp.Adam Jackson
The poor thing has so little VRAM that you can't do any bigger than 640x480 at 32bpp, so in the interest of having a reasonable screen size...
2007-03-28Disable MTRR support for PCIE G200SE cards, as it tickles hardware bugs.Adam Jackson
2007-03-28Fix various timing bugs on the G200SE.Adam Jackson
2006-12-26Fixed a typo that caused MGAREG_C2OFFSET to be set to a bad value.Tilman Sauerbeck
Unfortunately, this doesn't fix any bugs because afterwards, the correct value would be stored in MGAREG_C2OFFSET anyway.
2006-12-26More boring janitoring work.Tilman Sauerbeck
2006-12-26Fixed a typo in a debug message.Tilman Sauerbeck
2006-12-26Bug #9448: Actually write the C2STARTADD0 register.Tilman Sauerbeck
2006-12-10Replaced some more magic values with register names.Tilman Sauerbeck
2006-12-02Removed a duplicated register definition.Tilman Sauerbeck
2006-12-01More janitoring work.Tilman Sauerbeck
Moved the XPWRCTRL, XDISPCTRL and C2CTL register definitions to mga_reg.h. Cleaned up the DAC2 routing.
2006-12-01Use the XORG_RELEASE_VERSION autoconf macro.Tilman Sauerbeck
2006-11-26Made some XAA function static.Tilman Sauerbeck
2006-11-26Removed a duplicated register definition.Tilman Sauerbeck
2006-11-26Replaced more magic values with register names.Tilman Sauerbeck
2006-11-26Use register names in MGASaveScreenMerged() rather than magic numbers.Tilman Sauerbeck
2006-11-26Register name tweak.Tilman Sauerbeck
The "DAC power down" register powers down the DAC when it's set to zero, and enables it when it's set to one. To avoid confusion, I renamed it.
2006-11-25Removed some unused defines from mga_reg.h.Tilman Sauerbeck
2006-11-25Removed some duplicated register definitions.Tilman Sauerbeck
2006-11-19Bug #2168: Fix graphics corruptions with Mystique rev 2.Tilman Sauerbeck
2006-11-17Enabled the Exa composite hooks on G550.Tilman Sauerbeck
2006-10-27Bump version to 1.4.4.xf86-video-mga-1.4.4Ian Romanick
2006-10-27Convert tests of pMga->Chipset to tests of single bit flags.Ian Romanick
Several places in the driver contain if-statements that test Chipset against several different values to determine what functionality to use. Some of these tests, HALCHIPSETS in particular, test as many as 6 different values. This bloats the code size unnecessarilly. Instead we cache the results of some of those tests in single bit flags stored in pMga. These changes are similar to some changes in the pci-rework branch.
2006-10-26Fix bug #8666. git-bisect for the win!Ian Romanick
Since && takes precedence over ||, put parenthesis around the || block to make all the operations associate in the intended manner.
2006-09-28Make the DUALSTAGE0 code slightly easier to read.Tilman Sauerbeck
2006-09-20Unbreak the build: MGAStormSync() is required by mga_dga.c.Tilman Sauerbeck
2006-09-20Removed the XAA atype tables.Tilman Sauerbeck
Added a FIXME on BLK support for GXcopy instead.
2006-09-20Removed the DFS stub.Tilman Sauerbeck
2006-09-20Fall back to software rendering for A8 add A8.Tilman Sauerbeck
2006-09-16Enable the BYPASS332 and NODITHER flags when writing to 8 bpp dests.Tilman Sauerbeck
Thanks to Ville Syrjälä, who noticed these were missing. Note that this messes up glyph compositing (A8 add A8) pretty much, but I don't know what's really going on there yet.
2006-09-15More DUALSTAGE0 fixes related to A8 writes.Tilman Sauerbeck
When writing from a texture without an alpha channel to an A8 texture, we now set the alpha channel to 0xff via FCOL.
2006-09-15Reworked DUALSTAGE0 setup.Tilman Sauerbeck
If we're copying from A8 to A8, we need to replicate the alpha to the color channels.
2006-09-13Only use the source texture's alpha channel if it actually has one.Tilman Sauerbeck
2006-09-11Support masks without alpha.Tilman Sauerbeck
2006-09-10Fixed the remaining MGAAllocateMemory() issues.Tilman Sauerbeck
2006-09-10Fixed the engine sync needed for TexturedVideo.Tilman Sauerbeck
2006-09-10The XAA path in MGA_SYNC() now only syncs the engine if NeedToSync is set.Tilman Sauerbeck
This makes both paths behave the same way.
2006-09-09Dont't multiply the retval of MGAAllocateMemory() by cpp again.Tilman Sauerbeck
2006-09-09Make use of MGA_MARK_SYNC.Tilman Sauerbeck
2006-09-09XV works with EXA now (dumb port of the Radeon code).Tilman Sauerbeck
2006-09-09Do A8->A8 ops in hardware.Tilman Sauerbeck
2006-09-09PrepareSourceTexture() cannot fail.Tilman Sauerbeck
generated by cgit v1.2.3 (git 2.25.1) at 2024-12-28 12:51:56 +0000