ChangeLog | 1400 ++++++++++++++++++ Makefile.am | 2 README | 14 configure.ac | 3 debian/changelog | 6 man/radeonhd.man | 124 - src/Makefile.am | 9 src/r5xx_accel.c | 9 src/r5xx_accel.h | 2 src/r5xx_exa.c | 12 src/r5xx_xaa.c | 2 src/r600_exa.c | 2150 ++++++++++++++++++++++++++++ src/r600_reg.h | 140 + src/r600_reg_auto_r6xx.h | 3087 +++++++++++++++++++++++++++++++++++++++++ src/r600_reg_r6xx.h | 504 ++++++ src/r600_reg_r7xx.h | 149 + src/r600_shader.c | 1848 ++++++++++++++++++++++++ src/r600_shader.h | 359 ++++ src/r600_state.h | 293 +++ src/r600_textured_videofuncs.c | 531 +++++++ src/r6xx_accel.c | 1233 ++++++++++++++++ src/r6xx_accel.h | 89 + src/radeon_exa_render.c | 2 src/rhd.h | 49 src/rhd_atombios.c | 4 src/rhd_atombios.h | 2 src/rhd_atomcrtc.c | 15 src/rhd_atomout.c | 152 +- src/rhd_atomout.h | 2 src/rhd_atompll.c | 2 src/rhd_atomwrapper.c | 2 src/rhd_atomwrapper.h | 2 src/rhd_audio.c | 4 src/rhd_audio.h | 2 src/rhd_biosscratch.c | 2 src/rhd_biosscratch.h | 2 src/rhd_card.h | 2 src/rhd_connector.c | 2 src/rhd_connector.h | 2 src/rhd_crtc.c | 47 src/rhd_crtc.h | 4 src/rhd_cs.c | 27 src/rhd_cs.h | 2 src/rhd_cursor.c | 10 src/rhd_cursor.h | 2 src/rhd_dac.c | 2 src/rhd_ddia.c | 2 src/rhd_dig.c | 208 +- src/rhd_dri.c | 171 +- src/rhd_dri.h | 2 src/rhd_driver.c | 215 +- src/rhd_edid.c | 2 src/rhd_hdmi.c | 2 src/rhd_hdmi.h | 2 src/rhd_helper.c | 32 src/rhd_i2c.c | 16 src/rhd_i2c.h | 2 src/rhd_id.c | 68 src/rhd_lut.c | 77 - src/rhd_lut.h | 2 src/rhd_lvtma.c | 75 src/rhd_mc.c | 948 +++++------- src/rhd_mc.h | 16 src/rhd_modes.c | 2 src/rhd_modes.h | 2 src/rhd_monitor.c | 2 src/rhd_monitor.h | 2 src/rhd_output.c | 58 src/rhd_output.h | 6 src/rhd_pll.c | 2 src/rhd_pll.h | 2 src/rhd_randr.c | 293 ++- src/rhd_randr.h | 2 src/rhd_regs.h | 6 src/rhd_shadow.c | 2 src/rhd_shadow.h | 2 src/rhd_tmds.c | 57 src/rhd_vga.c | 133 + src/rhd_vga.h | 21 src/rhd_video.c | 223 ++ src/rhd_video.h | 12 utils/conntest/rhd_conntest.c | 2 utils/conntest/rhd_dump.c | 38 83 files changed, 13675 insertions(+), 1340 deletions(-)
New commits: commit f6fcadb86a6664e4288462025943b308cbcb9061 Author: Brice Goglin <[email protected]> Date: Fri Apr 10 08:09:19 2009 +0200 Prepare changelog for upload diff --git a/debian/changelog b/debian/changelog index 8501e9c..540e5a9 100644 --- a/debian/changelog +++ b/debian/changelog @@ -1,8 +1,8 @@ -xserver-xorg-video-radeonhd (1.2.5-1) UNRELEASED; urgency=low +xserver-xorg-video-radeonhd (1.2.5-1) unstable; urgency=low * New upstream release. - -- Brice Goglin <[email protected]> Fri, 10 Apr 2009 08:07:39 +0200 + -- Brice Goglin <[email protected]> Fri, 10 Apr 2009 08:09:08 +0200 xserver-xorg-video-radeonhd (1.2.4-3) unstable; urgency=low commit 177fc11b5f05c3fcb7a8d3cc98a2a6537825efdc Author: Brice Goglin <[email protected]> Date: Fri Apr 10 08:09:04 2009 +0200 New upstream release diff --git a/ChangeLog b/ChangeLog index ba39083..7e64580 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,1393 @@ +commit cb54f48b212d5ae54e13bbdf24575b6163798c0d +Author: Matthias Hopf <[email protected]> +Date: Thu Apr 9 15:11:37 2009 +0200 + + Bump to 1.2.5. Updated README. + +commit 9d2a508892f16572127d192497bcf505556538b8 +Author: Yang Zhao <[email protected]> +Date: Thu Apr 9 15:01:05 2009 +0200 + + Fix resume from suspend for r6xx/r7xx + + Move DRM calls to DRM_RADEON_CP_RESUME and RADEON_SETPARAM_VBLANK_CRTC + to after doing GART backup. Also move call to DRIUnlock() to end of + RHDEnterVT() from RHDDRIEnterVT(). + +commit 1137e0963e322dcb41816bdc76c57d3e3d407166 +Author: Matthias Hopf <[email protected]> +Date: Thu Apr 9 11:42:35 2009 +0200 + + RHDDRIGetIntGARTLocation is called too often to be logged. Nuked RHDFUNC there. + +commit 76d6973a5c18b5aff7e7027a2f87d3616179e2e9 +Author: Matthias Hopf <[email protected]> +Date: Wed Apr 8 17:01:49 2009 +0200 + + man: R6xx and R7xx 2D possible now. + +commit 0065697647401fd4b0df13df94bd9ebedcac96dc +Author: Yang Zhao <[email protected]> +Date: Wed Mar 25 19:26:02 2009 -0700 + + LUT: reduce number of writes to DC_LUT_RW_INDEX + + The value of DC_LUT_RW_INDEX is increased everytime DC_LUT_30_COLOR is + accessed. It is pointless to set this explicitly for each row of the + input colour table. + + Signed-off-by: Yang Zhao <[email protected]> + +commit 6d7f2486ea5f7794dee0c8d8d2655ddad378eff5 +Author: Matthias Hopf <[email protected]> +Date: Fri Apr 3 11:54:14 2009 +0200 + + Use () in macro arguments of new RHDRegMask. + +commit d6c3727751a2de2c333e654aea1bd9b1d9c2ef44 +Author: Yang Zhao <[email protected]> +Date: Fri Mar 27 21:21:25 2009 -0700 + + CS: Replace register read/write macros with ones from compiler.h + + Signed-off-by: Yang Zhao <[email protected]> + +commit c9d1af91acc579d7930ecf221090ff558cbc2f99 +Author: Yang Zhao <[email protected]> +Date: Fri Mar 27 19:46:11 2009 -0700 + + Use X MMIO macros instead of own register read/write functions + + The old functions were not portable. + + Signed-off-by: Yang Zhao <[email protected]> + +commit f65014d8dd50eb0d72a1d9da5371db0840c16d7e +Author: Alex Deucher <[email protected]> +Date: Thu Apr 2 12:11:03 2009 -0400 + + Add RV790 (HD 4890) Support + +commit 79efc6093f26daea654d3f748ebe3d4aa1229fe4 +Author: Egbert Eich <[email protected]> +Date: Wed Apr 1 09:38:41 2009 +0200 + + AtomBIOS: Fixed wrong logic: switch -> if. + +commit ecd616670e73d57397cc3e01017c84b9076c9f4f +Author: Yang Zhao <[email protected]> +Date: Sat Mar 21 15:21:50 2009 -0700 + + rhd_dump: fix error introduced by LUT dump addition + + 6e71f01f225a0cb13d10c7f7960db9d20c89a7db introduced a bug in rhd_dump + where register address is scanned as %d instead of %x. + + Thanks to Rafal for pointing out the regression. + +commit 7e4948a3ad80ae0885df89ac7320457cb90f0b55 +Author: Alex Deucher <[email protected]> +Date: Thu Mar 26 14:22:35 2009 -0400 + + R6xx/R7xx EXA: rework composite pixel shader + + - move to vram storage + - move swizzle logic to tex setup + +commit 461701c6abf76eef296a33fbdbda77a3bde2aa62 +Author: Alex Deucher <[email protected]> +Date: Thu Mar 26 14:08:14 2009 -0400 + + R6xx/R7xx: clean up bool const code + + 3 regs: 1 bit per bool, 32 bools per ps/vs/gs + +commit cc6e6fe4e8b4f4557dd43b97b2b864f6dda76087 +Author: Alex Deucher <[email protected]> +Date: Sun Mar 22 01:31:57 2009 -0400 + + add new chip ids + +commit f0f640f105da161e7dbb08bdd85f7ff065696e72 +Author: Alex Deucher <[email protected]> +Date: Thu Mar 19 20:49:42 2009 -0400 + + R6xx/R7xx EXA: fix maxPitchBytes + + should now allow accel up the hw max of 8192x8192 + +commit 37da5e5abf7907f9341354732775d45ba88319f0 +Author: Alex Deucher <[email protected]> +Date: Thu Mar 19 20:18:50 2009 -0400 + + Add new pci ids + +commit 4fa650628d7688045cdda22683d49131f858ed4e +Author: Alex Deucher <[email protected]> +Date: Thu Mar 19 10:56:04 2009 -0400 + + Really disable UTS/DFS on r6xx/r7xx AGP + + typo in the last patch + +commit b075ec9cee108061a1abb7c3f786a8b76da99997 +Author: Alex Deucher <[email protected]> +Date: Wed Mar 18 14:37:51 2009 -0400 + + R6xx/R7xx AGP: disable gart data transfers + + Doesn't seem to be reliable on AGP. + +commit d9c8f9ce4aabc7c1a85e4f015c6b19bde4f96866 +Author: Dave Airlie <[email protected]> +Date: Fri Mar 13 10:36:52 2009 -0400 + + r600: reload shaders into VRAM on resume + + As VRAM gets zeroed out over s/r, we need to reload the + shaders. + +commit 12611c8f1b3f6e6b2ed7f6cfb1b7271d9a313bdf +Author: Yang Zhao <[email protected]> +Date: Wed Mar 11 19:29:41 2009 -0700 + + R6xx/R7xx shader: Fix OFFSET_[XYZ] macro for TEX_DWORD2 to accept floats + + Values for OFFSET_[XYZ] are 5-bits two's-complement fixed-point with + one-bit after decimal point. Values in [-8.0, 7.5] are valid. Inputs + that do not exactly land on 0.5 increments are rounded towards 0 to the + nearest increment. + +commit 70490504ef74069e657daefff4f28a9bd9cb8778 +Author: Alex Deucher <[email protected]> +Date: Thu Mar 12 03:45:24 2009 -0400 + + RS600: fix page table size for rs600 as well + +commit 10ebf657755b993a09046a664311a8309a0950b9 +Author: Dave Airlie <[email protected]> +Date: Thu Mar 12 01:03:43 2009 -0400 + + r600: fix sizing of PCI GART table for r600 + +commit f27383df4a5f3a9349ead2d833142f82ee6acd92 +Author: Yang Zhao <[email protected]> +Date: Wed Mar 11 16:20:42 2009 +0100 + + R6xx/R7xx: Fix OFFSET_[XYZ] macros for negative values. + +commit 827fb141f5be6bf02acb8ab650602c56136433fa +Author: Matthias Hopf <[email protected]> +Date: Tue Mar 10 18:31:17 2009 +0100 + + randr: Set use_screen_monitor correctly using helper function. + + Fixes novell bug #370656. + +commit 76490dd8d052586f8737babdd8f383dbe39f4291 +Author: Matthias Hopf <[email protected]> +Date: Tue Mar 10 17:14:26 2009 +0100 + + Enable DRI by default on R5xx chips. + + R[67]xx still need an explicit 'Option "DRI"'. + +commit 093e5934a26f784f70b5febf8ae6d50a3459552c +Author: Alex Deucher <[email protected]> +Date: Thu Mar 5 11:55:54 2009 -0500 + + R6xx/R7xx Xv: only stall on vline if the destination pixmap is the scanout buffer + + With composite we may be rendering to an offscreen pixmap. + +commit 70c6b680118097cddf72c0c82eabf7dc7e438f2f +Author: Rafał Miłecki <[email protected]> +Date: Wed Mar 4 20:43:40 2009 +0100 + + Fixes to the radeonhd manpage + + Signed-off-by: Christian König <[email protected]> + +commit ad21c9ccc56bfb0957e59eaad152f1c2fcdde73a +Author: Christian Koenig <[email protected]> +Date: Thu Dec 25 15:24:10 2008 +0100 + + Remove no interlaced modes limitation from TMDS outputs. + + This patch removes the limitation of displaying no interlaced modes from + the TMDS outputs. Since interlaced modes are just a matter of CRTC + programming this makes them work out of the box. + +commit 742a4b4e0e68e88bf91c28237675877831f605b0 +Author: Christian Koenig <[email protected]> +Date: Sat Dec 20 23:15:14 2008 +0100 + + Randr HDMI property + + Adds an HDMI property to randr and while we are at it rework the + property handling a bit. Also adds an RHDOutputAttachConnector function + which is called every time an output <-> connection relation is changed. + +commit 2013be5c86b11fd05cce25814213f9a2bf0bd704 +Author: Christian Koenig <[email protected]> +Date: Sat Dec 20 23:05:27 2008 +0100 + + Cleanup and include true/false as match in RhdParseBooleanOption + + Further improve RhdParseBooleanOption to include true/false as match, + since people seems to be used to it because of the behaviour of Boolean + options. + +commit 54600950219ddb83b73c4ac17546093c36402e36 +Author: Christian Koenig <[email protected]> +Date: Sat Dec 20 23:03:22 2008 +0100 + + Update RHDAudioInit to include RS600, RS690 and RS740 + + RHDAudioInit decides from the chipset type if there is an audio engine. + This commit changes this test to include more chipsets. It should be + removed entirely when i figure out how to get this information from + atombios. + +commit 9c8ab2dfbe61120298c4b46a2b49245c6779dbc2 +Author: Alex Deucher <[email protected]> +Date: Mon Mar 2 16:26:14 2009 -0500 + + R6xx/R7xx EXA: cleanup composite texture setup + +commit 3826cd76b04664f66b19892b9d937b27a1c69534 +Author: Alex Deucher <[email protected]> +Date: Mon Mar 2 16:16:36 2009 -0500 + + R6xx/R7xx: combine Xv packed/planar PS and EXA composite VS + + use bool consts to select the fetch routines + +commit 2a06a508dfe64212887192a3126b668415507d34 +Author: Evgeni Golov <[email protected]> +Date: Mon Mar 2 14:52:27 2009 -0500 + + Fix compile with EXA but no DRI + +commit b02af166bed7706d6909acd4037af2df646a18a6 +Author: Alex Deucher <[email protected]> +Date: Mon Mar 2 11:23:39 2009 -0500 + + R6xx/R7xx: fix build errors in some cases + +commit 7338dc0d9d2bff677e9f10bdd0f59d8694fe274d +Author: Alex Deucher <[email protected]> +Date: Mon Mar 2 11:22:05 2009 -0500 + + R6xx/R7xx: fix regression in 347b23d419c877de0ac95b1f2f69d6953dddfebd + + Should fix bug 20426 + +commit 1935b2aae5cf53fe2777eb31ca39c4ce551b3292 +Author: Luc Verhaegen <[email protected]> +Date: Mon Mar 2 15:07:43 2009 +0100 + + Byebye novell. + +commit b35c26fc4a318ec30b76bae6d420a1b73fa14fa6 +Author: Alex Deucher <[email protected]> +Date: Mon Mar 2 00:21:12 2009 -0500 + + R6xx/R7xx: write vertexes directly to the IB + + Reduces the vertex buffer setup overhead + +commit 347b23d419c877de0ac95b1f2f69d6953dddfebd +Author: Alex Deucher <[email protected]> +Date: Mon Mar 2 00:07:23 2009 -0500 + + R6xx/R7xx: switch emit functions to macros + + This improves performance due to decreased function call + overhead. + +commit 89a835f03993da087c361de7ee3f6bc13bcb33f8 +Author: Christian Koenig <[email protected]> +Date: Sun Mar 1 23:03:24 2009 -0500 + + R6xx/R7xx cleanup: Replace Xv shaders with generated ones + + Replace Xv shaders with generated one fixing some minor typos in the + pixel shaders. + +commit 2b725913b1a91f2d178e77cd542c35067fc16d7c +Author: Christian Koenig <[email protected]> +Date: Sun Mar 1 23:01:32 2009 -0500 + + R6xx/R7xx cleanup: Move shaders to r600_shader.c + + Move all shaders from r600_exa.c to the new file r600_shader.c + +commit 6f9373384d3ebe58333e1ee0c64ddcc11d235a06 +Author: Christian Koenig <[email protected]> +Date: Sun Mar 1 22:38:06 2009 -0500 + + R6xx/R7xx cleanup: Fix comments in r600_exa.c + + Replace all C++ comments in r600_exa.c with C ones. + +commit 812ff17853d4d8a07418d0b28b2f3f80f6db74ed +Author: Alex Deucher <[email protected]> +Date: Sun Mar 1 12:22:14 2009 -0500 + + Fix R5xx Xv after r6xx/r7xx accel merge + +commit df71658688e47614a3f660c28ac1dc77970eb491 +Author: Matthias Hopf <[email protected]> +Date: Thu Feb 26 12:24:44 2009 +0100 + + EXA: fix for (now compatible) PREPARE_AUX buffers. + + Partially reverts commit 047bd7059bcfec55e06a35db4a2b16d52fe8dbf6. + +commit 10c211913de10bb4c7ec0747671f1b3e61231674 +Author: Alex Deucher <[email protected]> +Date: Fri Feb 27 11:34:20 2009 -0500 + + Only pad the IB on r6xx+ + +commit f10c8c797f3a0b53625cf8850b949fb304bcc4da +Author: Alex Deucher <[email protected]> +Date: Wed Feb 25 22:44:32 2009 -0500 + + CloseScreen(): Don't mess with the engine if the xserver isn't active + + This fixes problems with killing X when accel is active and the server + is switched away. + +commit a900f07b64bebe26d66b02decb146391937ea18d +Author: Alex Deucher <[email protected]> +Date: Wed Feb 25 09:20:03 2009 -0500 + + R6xx/R7xx EXA: same surface and same coords equals nop + +commit fdf6150ad9988b2e6e391c54b2f5fb9bc1cda837 +Author: Christian Koenig <[email protected]> +Date: Tue Feb 24 16:57:48 2009 -0500 + + R6xx/R7xx: Bugfix for wait_vline_range + + Check if CurrentiMode is set, not if Active is true. + +commit c16e74a2d36781fee495a72005a2decfec0d1b91 +Author: Alex Deucher <[email protected]> +Date: Tue Feb 24 11:06:32 2009 -0500 + + R6xx/R7xx EXA: optimize overlapping copy using temp surface + + - allocate temp surface in PrepareCopy() + - fallback to old OverlapCopy() path is we're not able to + allocate a temp surface + +commit 435e159d5c15ff3f6631d038e02860681c541eb5 +Author: Mark van Doesburg <[email protected]> +Date: Tue Feb 24 11:01:56 2009 -0500 + + R6xx/R7xx EXA: use a temp surface for overlapping copies + +commit 0de4db723c4d9bd40dcaa7fd79b507f7fe9270aa +Author: Alex Deucher <[email protected]> +Date: Tue Feb 24 10:10:08 2009 -0500 + + R6xx/R7xx EXA: switch to surface sync packet + +commit db97dd5db4f95c0c4e056a4a7c3e9055dce337c1 +Author: Alex Deucher <[email protected]> +Date: Tue Feb 24 03:11:36 2009 -0500 + + r7xx: Fix typo in cmd fifo mask + +commit 33a84c9b4fddc5bb396ec90b93e174e0a1a4f91b +Author: Alex Deucher <[email protected]> +Date: Mon Feb 23 19:06:22 2009 -0500 + + R6xx/R7xx: add wait for idle MMIO path + +commit c2b2155fcc499a253bef99e2462369162640a37c +Author: Alex Deucher <[email protected]> +Date: Mon Feb 23 18:35:04 2009 -0500 + + R6xx/R7xx: reset 3D state after VT switch + +commit ca198357d3f3f9906d35585f3e22bfd00e92d2ec +Author: Christian Koenig <[email protected]> +Date: Mon Feb 23 10:14:39 2009 -0500 + + R6xx/R7xx: Improve wait_vline_range + + Make shure the given crtc is active and the range we are waiting for is + in the visible area of this crtc. + +commit 793cebe3dccc176da1c326b6be979d63d09b3ba5 +Author: Alex Deucher <[email protected]> +Date: Wed Feb 18 20:23:55 2009 -0500 + + R6xx/R7xx EXA/Xv: properly deal with running out of vertex buffer space + + As noted by mhopf, if VGT_MAX/MIN_INDX, etc. regs change, you need to re-emit + CB blocks to avoid a hang. So, just set the VGT_MAX_INDX to a reasonably large value + in the default state and don't touch them when drawing. When we run out of VB space, + just draw the current buffer, grab a new one, and continue. + +commit ef9225f75d8376d1f648b1f673ee5f25fe40564c +Author: Alex Deucher <[email protected]> +Date: Thu Feb 12 17:47:44 2009 -0500 + + R6xx/R7xx: switch to drm for wait for idle + + THIS REQUIRES AN UPDATED DRM + +commit 1442be75b3a8f72dbbd4c33b4c92d7754999e4b2 +Author: Alex Deucher <[email protected]> +Date: Thu Feb 12 15:16:27 2009 -0500 + + R6xx/R7xx EXA: handle running out of vertex space in the copy path + +commit e16691af2a4e35b14fbbc39aed1d527d21c80ad6 +Author: Alex Deucher <[email protected]> +Date: Thu Feb 12 15:09:15 2009 -0500 + + R6xx/R7xx EXA: properly handle non repeat cases in the texture setup + +commit 976d651b7f472ef44eecb8ef07b1ab8675418c19 +Author: Yang Zhao <[email protected]> +Date: Thu Feb 12 14:52:22 2009 -0500 + + R6xx/R7xx EXA: Further optimizations to overlapping copy + + Diagonal overlapping copies can be reduced to either horizontal- or + vertical-only offset, and the one with fewer copies is picked. + +commit f02dd7cbc523160aedaa21599234c58c53447dc1 +Author: Christian Koenig <[email protected]> +Date: Wed Feb 11 18:13:47 2009 -0500 + + R6xx/R7xx: Cleanup register definitions + + Comment every definition in r600_reg_r6xx.h and r600_reg_r7xx.h which + is also in rhd_regs.h, making it possible to include both files at the same + time. + +commit b9e0ef98ba846d8051e1ee7ad685e98394579493 +Author: Christian Koenig <[email protected]> +Date: Wed Feb 11 18:10:22 2009 -0500 + + R6xx/R7xx: Implements wait_vline_range + + Implements wait_vline_range function and use this in + R600DisplayTexturedVideo. Also fixes a nasty bug in "ereg". + + Based on radeon patch from Pierre Ossman. + +commit 47f375d7dee8a5291b87edc7d14238134be6ae15 +Author: Christian Koenig <[email protected]> +Date: Wed Feb 11 18:09:33 2009 -0500 + + R6xx/R7xx: Define WAIT_x constants for IT_WAIT_REG_MEM + + And also fix some compiler warnings. + +commit 936838143d5431f6a0b33c6a396ad0ff44cf9c35 +Author: Alex Deucher <[email protected]> +Date: Wed Feb 11 15:25:17 2009 -0500 + + R6xx/R7xx Xv: switch packed over to Yang's new shader code + +commit e315314c98ae7750fe5107625f35976697e4fa79 +Author: Yang Zhao <[email protected]> +Date: Mon Feb 9 22:28:03 2009 -0800 + + R6xx/R7xx Xv: Planar - Properly scale Y'CbCr values before converting to RGB + + According to MPEG-2 spec, Y' and Cb/Cr values are scaled to [16, 235] + and [16, 240], respectively, when packed into bytes. Properly take care + of the reverse scaling before translating to RGB. + + Conversion matrix has been simplified to remove 3rd column, as the fitting + to [-0.5, 0.5] can be done with scaling. + + Redundant MOV instructions were also removed, and now only 3 GPRs are + required. + +commit 8dc7161ce426fb3968bba02d62db0a9980b26197 +Author: Alex Deucher <[email protected]> +Date: Wed Feb 11 13:14:01 2009 -0500 + + R6xx/R7xx: be more verbose about what function ran out of VB space + +commit 78c0f342282a1105ec30305dc2a44324702b151e +Author: Alex Deucher <[email protected]> +Date: Wed Feb 11 12:52:33 2009 -0500 + + R6xx/R7xx: move engine idle to sync function + +commit a04e5b212523d877cd6591c367125745c27049c7 +Author: Alex Deucher <[email protected]> +Date: Wed Feb 11 11:45:18 2009 -0500 + + R6xx/R7xx Xv: add support for packed uploads + +commit 4205f30ae97c4771078988c6494f82a0d7e1f8de +Author: Alex Deucher <[email protected]> +Date: Wed Feb 11 11:40:23 2009 -0500 + + R6xx/R7xx Xv: Add native support for packed formats + +commit d4079f7a68392fe0486c0ea0ddd6f2ca6d376149 +Author: Alex Deucher <[email protected]> +Date: Tue Feb 10 19:38:12 2009 -0500 + + M82: Missed one in the previous commit + +commit 7ea3e14f4a3904ab2c5d858d26d23c793f393ade +Author: Alex Deucher <[email protected]> +Date: Tue Feb 10 17:21:02 2009 -0500 + + R6xx/R7xx: fix M82 + + M82 is an RV620 + +commit 47c57d1d994a9427b3e6068326eeb5971d2664cf +Author: Alex Deucher <[email protected]> +Date: Mon Feb 9 19:57:30 2009 -0500 + + R6xx/R7xx: Add checks to make sure we don't overrun VB space + +commit 531de3518857e585c869c1d4c004ff2174984389 +Author: Alex Deucher <[email protected]> +Date: Mon Feb 9 14:09:24 2009 -0500 + + R6xx/R7xx Xv: switch to native planar shader and add accelerated planar uploads + +commit 4e574120feec720dc1086596bf1b2a315a1f1e0e +Author: Alex Deucher <[email protected]> +Date: Mon Feb 9 13:33:13 2009 -0500 + + R6xx/R7xx UTS: move actual upload to separate function + + So it can be shared with Xv + +commit 916a3a08255aeda24987ae32cc1369c0c10b7bcd +Author: Yang Zhao <[email protected]> +Date: Sat Feb 7 13:42:22 2009 -0500 + + R6xx/R7xx EXA: Optimize overlapping copy + + Overlapping copy is now done in chunks proportional to the + non-overlapping area. + + Diagonal overlaps are also handled properly. + +commit 295648454253f98e30e3680e8c9dd94c91bde6bb +Author: Alex Deucher <[email protected]> +Date: Sat Feb 7 10:34:12 2009 -0500 + + R6xx/R7xx Xv: fix typos in cache flushing + +commit fab5c673418a44aafb0314766ce0edc25740c51d +Author: Alex Deucher <[email protected]> +Date: Fri Feb 6 18:31:14 2009 -0500 + + R6xx/R7xx EXA: Fix typo in DFS + + noticed by pzad in IRC + +commit 8f7fca4bbe8e3d29026f8cf603602e09f00af9e3 +Author: Alex Deucher <[email protected]> +Date: Fri Feb 6 18:04:43 2009 -0500 + + R6xx/R7xx EXA: fallback on overlapping blits for now + + Leave this disabled until we get a proper solution. + +commit 4120d13a4e1f84a97903c839193b39acbab0b5c9 +Author: Alex Deucher <[email protected]> +Date: Fri Feb 6 17:56:42 2009 -0500 + + Revert "r6xx/r7xx EXA: Optimize overlapping copy" + + This reverts commit 262126ef7292a851bcb757462615c015639e50d7. + + This is also not adequate for all cases. + +commit f1b9c19683090272a29a3e4c9e2c054b4a32b47c +Author: Alex Deucher <[email protected]> +Date: Fri Feb 6 11:17:02 2009 -0500 + + R6xx/R7xx EXA: add accelerated UTS/DFS hooks + + I'm not sure how much of a win these are. I need to + do some benchmarking. + +commit 8f90a3432e0f234b4c0f778aed31367b3dc4110d +Author: Yang Zhao <[email protected]> +Date: Fri Feb 6 10:28:39 2009 -0500 + + r6xx/r7xx EXA: Optimize overlapping copy + + When source and destination blocks are only offset horizontally, it + appears to be unnecessary to perform careful, segment-by-segment copy. + The code path that does this is taken out completely. + + For the case where offset is only vertical, copying is now done by + height of the non-overlapping area each time, instead of always + line-by-line. + +commit c725d146e333930e410733073b88849e8fb91bf2 +Author: Alex Deucher <[email protected]> +Date: Thu Feb 5 19:26:12 2009 -0500 + + Revert "R6xx/R7xx EXA: improve overlapping copy performance" + + This reverts commit bd28b5817ed96e86a2cb275f72a7233d8fc5eccc. + + THis seems to cause corruption in some cases. + +commit 7ef8c0b06522477cc2abcc2c9f1fa65efd2fab02 +Author: Alex Deucher <[email protected]> +Date: Thu Feb 5 17:33:36 2009 -0500 + + r6xx/r7xx EXA: fix corruption when doing sw access + + need to wait until the engine is idle. Ideally we wait + on a timestamp shadowed in memory, but polling the + GRBM_STATUS reg will do for now. + +commit f0ac61a002debc29442492229c1465fe55d2a791 +Author: Alex Deucher <[email protected]> +Date: Thu Feb 5 16:25:06 2009 -0500 + + R6xx/R7xx EXA: improve overlapping copy performance + + send vertices for each line of the copy, but only draw once + +commit 9f160eaadb944bc80a62cc03195a3245cae2df2a +Author: Alex Deucher <[email protected]> +Date: Thu Feb 5 16:12:28 2009 -0500 + + r6xx/r7xx EXA: cleanup overlapping copy + +commit 4d20f2ce1b3a248ebf7093b7b8c35f87dd1ec673 +Author: Christian Koenig <[email protected]> +Date: Wed Feb 4 19:24:44 2009 -0500 + + Fix RHDDRIGARTBaseGet to work with AGP based card + + Remove a check in RHDDRIGARTBaseGet makes AGP based r6xx card working, + and also fix a minor debugging output typo. + +commit a5c129abb9e9ff9484c468056766f7281d8bc56e +Author: Alex Deucher <[email protected]> +Date: Tue Feb 3 18:16:06 2009 -0500 + + EXA: fix and re-enable Solid() on R7xx + + last bit was set in final alu instruction. + +commit 59fa04881f598602a3163070c62349a3ec65f242 +Author: Alex Deucher <[email protected]> +Date: Mon Feb 2 16:38:05 2009 -0500 + + Updates the comments about SPI setup for Solid() + +commit ded1312e6478ca205d5b5f0c470e24cc77405436 +Author: Alex Deucher <[email protected]> +Date: Mon Feb 2 16:20:36 2009 -0500 + + R7xx: temporarily disable Solid() to prevent hangs + + Something about using a PS constant for color seems to + disagree with some r7xx cards. + +commit f723ad4a591c3940373566c56f8d03a01ddee67f +Author: Alex Deucher <[email protected]> +Date: Mon Feb 2 16:08:00 2009 -0500 + + EXA solid fixes + + - fix fetch count in solid VS + - fix SPI PS input setup (we aren't using any VS outputs) + - always have at least one VS export besides position, even if + it's not used by the PS (SPI expects at least one). + +commit 53c1b328bda7103c06fa4545f72976ab71abce10 +Author: Alex Deucher <[email protected]> +Date: Mon Feb 2 11:34:09 2009 -0500 + + R6xx/R7xx: Avoid emitting default state when possible + +commit b32618d0c1799ed12ca4c2eaf96dde2700785ef5 +Author: Alex Deucher <[email protected]> +Date: Mon Feb 2 11:08:24 2009 -0500 + + Add R6xx/R7xx AGP support + + - Needs latest drm bits + - doesn't seem to be as reliable as PCIE + +commit 8ae410b1d9a7a19477c00794047eadf5c5400ff6 +Author: Alex Deucher <[email protected]> +Date: Sat Jan 31 15:37:06 2009 -0500 + + Fix and enable composite mask + + bad offset into the vertex buffer for the mask coordinates + +commit 1a3fe25e9def5fa8f371d3a2c96566318ed4a195 +Author: Alex Deucher <[email protected]> +Date: Fri Jan 30 20:22:07 2009 -0500 + + EXA: Remove repeat tile code + + As far as I can tell from documentation and experimentation, + all the clamp modes work fine with NPOT or POT textures. + +commit 84a7741e21c0293a4ac1c8693fc6010c2ff5c869 +Author: Alex Deucher <[email protected]> +Date: Fri Jan 30 14:30:58 2009 -0500 + + EXA: Re-enable a8 solid and copy. + + it's just masks with repeat that are broken. + +commit efbd11d434f4b9327effa5792b219c3ba3505abe +Author: Alex Deucher <[email protected]> +Date: Thu Jan 29 13:50:42 2009 -0500 + + R6xx/R7xx: move some VGT setup to drm + + This fixes accel on rs780/rv610/rv620, etc. + VGT_OUT_DEALLOC_CNTL and VGT_VERTEX_REUSE_BLOCK_CNTL + are set up based on the backend config so set them up + in the drm and forget them. + + YOU MUST UPDATE YOUR DRM! + +commit 8f4ab199975a76597dcf354996822af8816433d5 +Author: Alex Deucher <[email protected]> +Date: Thu Jan 29 13:06:41 2009 -0500 + + Discard command buffers when nothing is drawn + + This fixes hangs on r7xx. + + Changing the same SQ constant (ALU, VTX/TEX, etc.) + more than once between draw calls can result in a hang. + +commit cea356d9b0b89b34659c729e4781497fd81bee88 +Author: Alex Deucher <[email protected]> +Date: Wed Jan 28 19:33:04 2009 -0500 + + switch to R600PitchMatches() due to changes in pitch alignment + + render repeat modes seem the be the last thing left to getting + this working properly + +commit 381ca1a2a1b9b05d41e07f7d0548da753b66753e +Author: Alex Deucher <[email protected]> +Date: Wed Jan 28 17:26:09 2009 -0500 + + EXA Composite: add a bit of missing state setup + +commit 68311d8865152bee1ffe1f1233ccde51a18bf4d3 +Author: Alex Deucher <[email protected]> +Date: Wed Jan 28 02:34:25 2009 -0500 + + minor code re-org + +commit e6c753f409fbddfbd8f706061743d1d486de672d +Author: Alex Deucher <[email protected]> +Date: Wed Jan 28 02:31:01 2009 -0500 + + removed leftover vertex color param export + +commit 16b6e8d30db237acd354fc717d15939c45a6e095 +Author: Alex Deucher <[email protected]> +Date: Wed Jan 28 02:16:44 2009 -0500 + + EXA: fix mask coordinate export + +commit f808fe852596c9a6d6742bd19d2bb972b2268dec +Author: Ludovic Aubry <[email protected]> +Date: Wed Jan 28 00:37:12 2009 -0500 + + use correct mask resource and sampler id + +commit c1db8638979e5e47eb580f0ebcf1b2db7043888b +Author: Alex Deucher <[email protected]> +Date: Tue Jan 27 11:58:59 2009 -0500 + + Remove Xv vertex debugging output + +commit a4bfc7f0695db02dcea3bc07bf042b4b08930eb3 +Author: Alex Deucher <[email protected]> +Date: Mon Jan 26 20:46:04 2009 -0500 + + R6xx/R7xx: workaround text rendering issues + + - disable a8 acceleration until I figure out why it's not working properly + - disable DFS/UTS hooks. they are still SW at the moment anyway. + - fix pitch alignment (width has to be a multiple of 8, + bytes have to be a multiple of 256) + - fix height (height has to be a multiple of 8) + +commit 3b692f8ac29562d0b9b73037cd2b4dcf7a715adf +Author: Alex Deucher <[email protected]> +Date: Fri Jan 23 17:43:57 2009 -0500 + + R6xx/R7xx: add missing TEX instruction padding + +commit 43f54aa6c5014fcb81871d681b877f20673bccdb +Author: Alex Deucher <[email protected]> +Date: Thu Jan 22 20:20:45 2009 -0500 + + R6xx/R7xx: flush TC on chips without VC + + On chips without a dedicated vertex cache, + the texture cache is used for vertexes and textures. + +commit f6315d535c0084771030884bad8894008902d318 +Author: Alex Deucher <[email protected]> +Date: Thu Jan 22 19:44:47 2009 -0500 + + R6xx/R7xx: get the SQ setup right + + based on info from the hw guys + +commit b6351dc4079cc343e865ce6dab8e7dbde552274c +Author: Alex Deucher <[email protected]> +Date: Thu Jan 22 16:55:31 2009 -0500 + + Add missing return in PrepareAccess() + +commit a73ed442270077d9b148993ed1061224161c7e2f +Author: Alex Deucher <[email protected]> +Date: Thu Jan 22 13:04:06 2009 -0500 + + R6xx:R7xx: add fine grained cache flushing support + + - also flush HDP caches when doing sw access + +commit 5c4b2c38f23076d5f829212c165c207f5140e153 +Author: Alex Deucher <[email protected]> +Date: Sat Jan 17 15:51:34 2009 -0500 + + R6xx/R7xx: adjust SQ setup + -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
