From 4b32b00759fc23aa5f397f4f1c563b431f096602 Mon Sep 17 00:00:00 2001 From: Luc Verhaegen Date: Fri, 25 Aug 2006 16:12:46 +0200 Subject: Prepare for release: update patchlevel and create Changelog. --- Changelog | 4974 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ configure.ac | 2 +- src/via_driver.h | 2 +- 3 files changed, 4976 insertions(+), 2 deletions(-) create mode 100644 Changelog diff --git a/Changelog b/Changelog new file mode 100644 index 0000000..a0668c6 --- /dev/null +++ b/Changelog @@ -0,0 +1,4974 @@ +commit ea05867d5e0ef3761ec6a61f4713a8efb23bd86e +Author: Luc Verhaegen +Date: Fri Aug 25 08:59:24 2006 +0200 + + Update README, manpage and remove VT3108 limited support warning. + + Looks like there will be a release today! + +commit b1dad94ef4e34c4badeac89e7fbb05df6da2d62f +Author: Luc Verhaegen +Date: Thu Aug 24 16:48:54 2006 +0200 + + XFree86 4.6.0 Build: better version seperation. + +commit f9202a85a6db6e651531816c2c8e0bcc63876d65 +Author: Luc Verhaegen +Date: Thu Aug 24 16:45:03 2006 +0200 + + Update build for XFree86 4.6.0 (Barry Scott) + +commit e32696ecc057a032155b756134c549512302525d +Author: Luc Verhaegen +Date: Thu Aug 24 14:08:01 2006 +0200 + + Imake: add install and any targets to git_version.h depend. + +commit a5ebf0e2914bf7f07955f834ad467bb5b3109fc8 +Author: Luc Verhaegen +Date: Thu Aug 24 12:58:55 2006 +0200 + + VT restoration: forego VGA clocking when using both CRTCs. + + Mitac 8899 textmode doesn't restore when booted without vgafb. Culprit is the + fact that textmode had VGA clock 0 set in Misc. This wreaks havoc on the + simultaneous mode and scaling used by this laptop, leaving us in a bad state + upon restoration. Fix used here is keeping VGA clock set to external, the + same route VBE takes. + + This means that console might still break when something else sets the VGA + clocks again. + +commit 75cccc5add455809fcf7606dd25ec00472e07ac2 +Author: Luc Verhaegen +Date: Wed Aug 23 21:01:21 2006 +0200 + + VT163x/SiI16xInit: logical OR instead of binary. + + This broke VT163x and SiI16x detection. + +commit 4770e52eb070bc213c2efa17e9d79dd602c32eab +Author: Luc Verhaegen +Date: Wed Aug 23 20:56:10 2006 +0200 + + Fix: Wrap vgaHWSaveScreen again. + + Yesterdays commit had us using vgaHWSaveScreen directly again, instead of + VIASaveScreen. pScreen hooks on xorg 7.x require wrapped vgaHWSaveScreen + again. + +commit 46edb4dc2773ac07088ebf00c653f5418f7f8887 +Author: Luc Verhaegen +Date: Tue Aug 22 22:46:57 2006 +0200 + + Implement FBEnable. + + Clean up MapMMIO/MapFB/Save handling in ScreenInit. This leads to a more + correct Save/Restore cycle. + FBEnable now enables everything needed to be able to reliably read/write to the + FB. FB is now nulled at the end of MapFB, right after FBEnable. + +commit 754560ede562304e224a7bba8486f3523006ce81 +Author: Luc Verhaegen +Date: Sat Aug 19 07:24:19 2006 +0200 + + IDs: time for an update. + + VT3118: Twinhead E12BL (not attributed by openchrome.org) + VT3118: Fujitsu/Siemens Amilo L73x0 (Andris Jankevics - to unichrome-devel@) + VT3108: Fix Packard Bell Bali. + VT3108: Fujitsu/Siemens Amilo K7610 (Botos Pavel - to openchrome.org) + +commit e6a2a21daf46ad162242dd2301b1f67efd4d4c38 +Author: Luc Verhaegen +Date: Sat Aug 19 06:44:33 2006 +0200 + + CRT: Add CRTBandWidth Option. + + Stop modes with too high a dotclock in CRTModeValid. This value is + usually retrieved from EDID but a fallback option is provided. + + The fact that X doesn't stop modes on bandwidth was discovered by + Adam Jackson. + +commit 038ab4a78b281623fb25ea7783ccc21b407c3462 +Author: Luc Verhaegen +Date: Tue Aug 15 04:24:04 2006 +0200 + + Add git based versioning. + + Add git_version.sh, which gathers the following information: + * wether git is used. + * what SHA-ID the latest commit has. + * the name of the branch we're in. + * whether there are local changes. + This information is piped to git_version.h, which is included in + src/via_driver.c to print extra version information in VIAProbe. + + This is not a foolproof solution as the file doesn't get rebuilt all the time, + only when it's not present, like after make clean. + +commit 8c17a22472a6b2183ce284f6cba34a0d452ad81a +Author: Luc Verhaegen +Date: Sun Aug 13 12:33:05 2006 +0200 + + Xv: Fix VT3122Ax HQV initialisation. + + We need to fire the HQV once before we enable the overlay, otherwise the old + memory contents will pass by for a bit. Now the only flashing that remains is + the painting of the colour key. + +commit b3b97f9abe8a11b7830c33da6b385e2c8f05c55b +Author: Luc Verhaegen +Date: Sat Aug 12 13:22:00 2006 +0200 + + Xv: Fix firing of Video3 after StopVideo. + + Video3 now fires when a source flip has happened, but StopVideo needs to fire + to disable the engine, yet the HQV will never get flipped. The next PutImage's + IdleLoop will then see Video3 still waiting. ViaStopVideo now sets fire to next + VBlank while ViaVideo3Fire now resets the firing to source flip. + + Two fires when doing enabling/disabling the colorkey during Alpha were + superfluous. + +commit d04d8cc031e70087ea28e04e099da5d936aa0244 +Author: Luc Verhaegen +Date: Wed Aug 9 20:09:40 2006 +0200 + + CRT: Add CRTSense option. + + This has the driver always ignore the CRT connection sensing, which means that, + as far as the driver is concerned, the CRT is always attached. + At the same time, CRT output code is moved to via_crt.c. + +commit 2ef86673944d1bc59b048778cc0e29b14ba8df39 +Author: Luc Verhaegen +Date: Sun Aug 6 13:05:47 2006 +0200 + + README: update modular build instructions + +commit b3ed8267dbcd32c333c5724db148c38a3a98d0f3 +Author: Luc Verhaegen +Date: Wed Aug 2 09:48:35 2006 +0200 + + Xv: Increase V3 FIFO thresholds for VT3122Ax. + +commit 5bf1aabcc91e83df55a556f1e5ffe5f07b1bb891 +Author: Luc Verhaegen +Date: Wed Aug 2 09:24:47 2006 +0200 + + Xv: Split HQV Idle and V3 Fire loop. + + We hold off touching V3 a bit longer now, so we don't need to check it that + early any more. + +commit 8e0cb4f68bb4f3b23d4655beecf5993debea0971 +Author: Luc Verhaegen +Date: Tue Aug 1 14:55:46 2006 +0200 + + Xv: Cleaner version of last commit (92e5c3a1..) + +commit 92e5c3a1c0fc067aa2725a1b40fd9cbd5913277d +Author: Luc Verhaegen +Date: Tue Aug 1 13:20:30 2006 +0200 + + Xv: Fix the uncovering issue that was blamed on X all along. + + https://bugs.freedesktop.org/show_bug.cgi?id=4653 needs to be reviewed, but + then, it should've been fixed/removed/closed ages ago. + +commit b912474a7eb5590d64ecabf2ac9c8e8cf33d028b +Author: Luc Verhaegen +Date: Sun Jul 30 14:20:41 2006 +0200 + + Xv: Fix colorkey and alpha for VT3122Cx. + + Out of all pre-VT3118 devices, the VT3122Cxen do have these registers/bits + implemented. + + Funded by ONELAN: Digital signage systems (http://www.onelan.co.uk/) + +commit 5ee0a82d77f30b374b50ea33e30ef7ebaaa6e85f +Author: Luc Verhaegen +Date: Sun Jul 30 14:14:01 2006 +0200 + + Xv: Fix broken bottom line when zooming out. + + Funded by ONELAN: Digital signage systems (http://www.onelan.co.uk/) + +commit dc47154c229f9be0a13ce2209bbf00ec5862c174 +Author: Luc Verhaegen +Date: Sun Jul 30 12:19:43 2006 +0200 + + Xv: Fire V3 only on HQV flip. + + Have V3 wait on HQV flip instead of nearest (sometimes wrong) VBlank. + + Funded by ONELAN: Digital signage systems (http://www.onelan.co.uk/) + +commit 8ef66fa01816db8286a6c4e6d1932255c870cd4c +Author: Luc Verhaegen +Date: Sun Jul 30 11:07:09 2006 +0200 + + Xv: Split up buffer allocation. + + Back buffer now gets reallocated right before the HQV is fired, further + reducing the risk of flashing. + + Funded by ONELAN: Digital signage systems (http://www.onelan.co.uk/) + +commit 531a596e554fa2d128671d021b43990779fd2397 +Author: Luc Verhaegen +Date: Sat Jul 29 21:43:56 2006 +0200 + + Xv: Use shadow registers for filling out the Video3 overlay engine. + + This writes out all commonly altered Video3 engine registers right before the + engine is fired. It's as transparent as before, clean and easy, and will get us + a free, but unnoticable, speed-up. + + Funded by ONELAN: Digital signage systems (http://www.onelan.co.uk/) + +commit e3c8cf7b1d9a03b4409fd7274239ec7f757841bd +Author: Luc Verhaegen +Date: Sat Jul 29 07:37:08 2006 +0200 + + Xv: Fix Video3 overlay firing for VT3122Cx. + + Funded by ONELAN: Digital signage systems (http://www.onelan.co.uk/) + +commit 9a21ab89ef66e8b778a8d5d5ebe398c99b014520 +Author: Luc Verhaegen +Date: Mon Jul 17 15:18:59 2006 +0200 + + XVideo: fix startaddress when adjusting frame when zooming out. + +commit b9a9a003132c6be561999bf4786e32b22b1f60de +Author: Luc Verhaegen +Date: Sun Jul 16 01:56:19 2006 +0200 + + Nasty Panel Code: Full register write for each mode. + +commit 7c882e1790143039f4ebfec1d8b9567627dfadf4 +Author: Luc Verhaegen +Date: Sat Jul 15 05:05:09 2006 +0200 + + Nasty Panel Code: fix some glaring issues. + + This while trying to run this on the M10k. + * Expand hsync/vrefresh of panel to all acceptable modes. + * Was incrementing wrong counter in ValidMode. + * Properly break from counter into lcdTable in Mode. + +commit 7bc2c9ba277140ff63cd4bda5d53c85987a6b23d +Author: Luc Verhaegen +Date: Sat Jul 15 00:56:06 2006 +0200 + + Nasty Panel Code: Adjust Mode Validation. + +commit 17400392e7398658376cb8df2cf784fc7b22537c +Author: Luc Verhaegen +Date: Sat Jul 15 00:13:59 2006 +0200 + + Nasty Panel Code: strip trailing nulls. + +commit 066d94abbb4359246e5ef0e1d3948d6b4a52358b +Author: Luc Verhaegen +Date: Fri Jul 14 17:51:43 2006 +0200 + + Nasty Panel Code: Add Scaling. + +commit 0720b4fed341b0805b3a3cc85cba9cd1c0caf5d7 +Author: Luc Verhaegen +Date: Fri Jul 14 16:57:13 2006 +0200 + + Nasty Panel Code: Clean up dotclocks. + +commit 46263d30992951d76ea0439e14fc5bf96d2d34df +Author: Luc Verhaegen +Date: Thu Jul 13 00:46:57 2006 +0200 + + Nasty Panel Code: Remove Ports. + +commit 04e178110f8533fbfe4ff52b1c914e0a129ff50d +Author: Luc Verhaegen +Date: Wed Jul 12 23:18:18 2006 +0200 + + Nasty Panel Code: Remove unused modes. + + * Replace VGA mode numbers by H/VDisplay. + * Remove modes with H/VDisplay 0. + * Remove supported Modes, match on H/VDisplay directly. + * Add MCtr/MExp modes to Patch2 directly. + * Rename table names. + +commit 76c1ce44315a62fcbc6965edbac495a4a1a4d7a5 +Author: Luc Verhaegen +Date: Wed Jul 12 04:58:26 2006 +0200 + + Nasty Panel Code: Remove resIdx 3: 1152x864 + +commit 7fe29ade16386a289e6dad1b62278028bbf658ea +Author: Luc Verhaegen +Date: Wed Jul 12 04:24:20 2006 +0200 + + Nasty Panel Code: Clean up ResolutionIndex/resIdx. + +commit 2aa76a443b2c58781855f36767846fed5d79e90a +Author: Luc Verhaegen +Date: Wed Jul 12 03:34:43 2006 +0200 + + Nasty Panel Code: remove vague general vga register writes. + + * Only in one case can we encounter a Sequencer access, others are Crtc. + * Strip most of via_vgahw.* thanks to this. + +commit 1e265d0dc33fdef01998644f065ea8205bd65f4f +Author: Luc Verhaegen +Date: Tue Jul 11 02:24:14 2006 +0200 + + Enable Gamma for DVP and DFP. + +commit 70084ad22b35f9d663271925e5a5e3adcadf9b54 +Author: Luc Verhaegen +Date: Tue Jul 11 01:25:24 2006 +0200 + + lcdTable: Clean up InitTB. + + * only CR registers being written. + * Scaling is disabled. + +commit cb617a6ff4389fde2f791f87aed6527a30d7e046 +Author: Luc Verhaegen +Date: Sun Jul 9 23:46:48 2006 +0200 + + Fix old panel code issues. + + * fix the primary versus secondary relationship. + * fix vesa modenumber usage. + +commit 329ea92cfa6c689dcc01a44a9c1b25352faedee4 +Author: Luc Verhaegen +Date: Sun Jul 9 21:37:59 2006 +0200 + + Fix stray paste from last commit (43cf4f73...). + +commit 43cf4f73e3a83877e77badcf591a5201ff82837f +Author: Luc Verhaegen +Date: Sun Jul 9 19:17:36 2006 +0200 + + Clean up output bus handling. + + * DI0/1 get renamed to DVP0/1 + * SR12 (scratch register for board strap) is used to detect where the TV + encoder sits instead of the I2CBus (although there is a strong relationship + between I2C Bus and output bus). + * This adds basic bus power management as well. + * VT1623 attached to VT3108 through DVP1 works with this patch. + +commit 749ca87de22065cdfdec8cd09a77f74124e448ad +Author: Luc Verhaegen +Date: Sat Jul 8 02:20:59 2006 +0200 + + VT3108: Add some quick (but wrongly named) DVP0 support for TV encoder. + + This is a quick hack giving me a working mode on the CRT, with VT1623 defined + timing and dotclock. + +commit f8c9ab6478636f73f78b5d0afd2d09a48491a5e4 +Author: Luc Verhaegen +Date: Fri Jul 7 03:53:29 2006 +0200 + + VT3108: Fix UMA latency issues for CRT and Video3 + + * Enable CRT prefetch + * Enable V3 prefetch and double the size of the FIFO. + +commit 16e6ad978495b57d982b0429963fbafd1d52dc02 +Author: Luc Verhaegen +Date: Thu Jul 6 19:15:01 2006 +0200 + + Enable VT3108 CRT gamma correction. + +commit d941a76a9f57ddc0b02f0be74d2064207815e28d +Author: Luc Verhaegen +Date: Thu Jul 6 14:26:17 2006 +0200 + + Clean up Panel Sizing. + + * Stop bothering with parsing EDID ourselves (aspire 135x will be handled + properly when the hardware becomes available). + * Rework the handling of the config option "PanelSize" + * Rework the handling of the Scratch register and move initial retrieval to + ViaScratchGet + * Remove Private->Size completely. From now on we use Private->X/Y everywhere + * Adjust lcdTable to use X/Y instead of VIA_PANEL defines. + +commit 084831922d51503f06492d8e23c7c12ae4e100be +Author: Luc Verhaegen +Date: Thu Jul 6 00:54:49 2006 +0200 + + Add basic VT3108 Panel support. + + * This code circumvenes all VT3122 and VT7205 panel code, by attaching + different callbacks at init-time. This doesn't change a thing to that 300K bit + of dead weight code, sadly. + * Native and centered modes are now possible. This code adds common modes, + but, when adhering to native resolution timing (check your log for examples), + all smaller than native resolutions are possible. + * Has working DPMS and correct VT restauration. + * Only tested with a directly connected 1024x768panel. + +commit 6eeecb44846ec56a80fdf00b4a04dd8250ef7142 +Author: Luc Verhaegen +Date: Wed Jul 5 23:53:26 2006 +0200 + + SwitchMode also requires Virtual Queue reinitialisation. + +commit e358deda5a9303cbd7a02f87e467e6a01d216270 +Author: Luc Verhaegen +Date: Sat Jul 1 13:36:22 2006 +0200 + + Fix pixmap cache issues after VT restoration. + + - Break up VIAInitialize2DEngine. + - Re-init Virtual Queue upon VT restoration. + - Depend solely on pVia->VQEnable instead of having a biased approach + where also VQStart is used. + +commit 1fe1fd7e1c150d0fc487181a5a368ba8b1ad249f +Author: Luc Verhaegen +Date: Thu Jun 29 21:17:29 2006 +0200 + + Remove interlaced support properly. + + Very low probability that this is even implemented in the unichromes. + +commit 3d7fa0f60c69979069ca0610903244b73746ab16 +Author: Luc Verhaegen +Date: Thu Jun 29 21:05:10 2006 +0200 + + Fix VT restoration on VT7205. + + Don't force CR47 bit 0 to 1 when returning to VT. Original value got restored + anyway. + +commit 9a183bd9eac31d3fa3ad26835c37155212641fff +Author: Luc Verhaegen +Date: Fri Jun 23 09:31:36 2006 +0200 + + VT3108 dotclock PLL generator: very nice compared to VT3122. + +commit 017f6fd0f8b581fbb5ac26b7a302267500c1e40e +Author: Luc Verhaegen +Date: Thu Jun 22 20:43:06 2006 +0200 + + Don't try to enable the panel except on VT3122 or VT7205. + +commit 97081ab932ca992119436fbe88c289c74158a353 +Author: Luc Verhaegen +Date: Tue Jun 20 11:52:19 2006 +0200 + + Start work on VT3108 support. + + * Stop PreInit from bailing on VT3108; warn about early support instead. + * Fix Restore; CR47 bit 1 CRTC protection works on VT3108. + * Add preliminary DI1 support to stop the driver from complaining. + +commit 30120c6fc6ea78b67693039e02df4c382cfda97b +Author: Luc Verhaegen +Date: Sun Jun 18 19:36:49 2006 +0200 + + Expand ViaVideoRamGetFromRamController hardware support. + +commit f4019b6fe347e42aa60540da5f0264d81d7d2ae3 +Author: Luc Verhaegen +Date: Sat Jun 17 02:15:06 2006 +0200 + + CRTSense will probably work on VT3122Cx too. + +commit 6cb56bf5e271cf990162ac1cea43efca52bcde7b +Author: Luc Verhaegen +Date: Thu Jun 15 10:09:34 2006 +0200 + + panel: Remove LCDPowerSeq tables for sane code. + +commit c4feaf0cd977e95e5efcf6717d097a26466b9d07 +Author: Luc Verhaegen +Date: Wed Jun 14 14:52:50 2006 +0200 + + Remove LCDModeFix table: handled by the vesa mode table. + +commit 84325ea25581d76cd0a611694566eb49e4dddabc +Author: Luc Verhaegen +Date: Wed Jun 14 14:45:49 2006 +0200 + + LCDModeTable: remove unused members. + +commit 015321a7f7a02ba5172035e5a49f6f84ed04802a +Author: Luc Verhaegen +Date: Wed Jun 14 11:19:32 2006 +0200 + + Bring structure in panel tables. + +commit df83f909bcf06982920a50a52165ef1a87483ce1 +Author: Luc Verhaegen +Date: Wed Jun 14 10:53:41 2006 +0200 + + Add *.html to man/.gitignore and *.so to src/.gitignore + +commit c9ae591e9d1cce3bfd0f89d2bd41d69352e1c92c +Author: Luc Verhaegen +Date: Wed Jun 14 10:52:05 2006 +0200 + + Move static cursor allocation to right after the Pixmap Cache. + +commit 299de5b350e30c8143d186ebc33b7888f114cd7d +Author: Luc Verhaegen +Date: Wed Jun 14 09:12:20 2006 +0200 + + Fix static FB allocation: + - Make sure DRM handles lack of memory. + - Fix XAA pixmap cache sizing. + +commit 462b399fde1b77613c6309020f1978fbff2a311f +Author: Luc Verhaegen +Date: Wed Jun 14 07:20:42 2006 +0200 + + Add ids for K8M890 and P4M890. Of course non-functional until hardware is + acquired. + +commit 766d6ffe2bd7ee901e5cb450ecbec0ba2c628887 +Author: Luc Verhaegen +Date: Mon May 15 03:36:36 2006 +0200 + + Add CRT detection; only useful on VT7205 though. + +commit fd7ded12a234bc5830fe9825a12a9112cef1ff08 +Author: Luc Verhaegen +Date: Wed May 3 10:31:07 2006 +0200 + + ID: VT3108: Clevo L570W (Jeremie Bortier) + +commit 748c9d4492ed9f8b4532355d05473236910d5a5e +Author: Luc Verhaegen +Date: Tue May 2 08:59:32 2006 +0200 + + Adjust ViaSwovBandwidth for DDR400. + +commit 8ebb157f91e3b1870bb19ec49338e8bca61ee10e +Author: Luc Verhaegen +Date: Fri Apr 28 02:11:17 2006 +0200 + + Honour configured gamma. (Reported by Hindrik Hettema) + +commit 887e96b900a68c973dae0e28204fbe9bf977469c +Author: Luc Verhaegen +Date: Tue Apr 25 17:36:28 2006 +0200 + + Remove ChangeLog. Will be replaced in packaged releases by git-log output. + +commit 782ba78a11334ed3e01d62a34ef1cd71871f5a4c +Author: Luc Verhaegen +Date: Tue Apr 25 16:48:19 2006 +0200 + + IDS: VT3108: Packard Bell Bali (Jean-Damien Durand) + +commit f83a5309c865252ba4267a1aa4a9d3c7023cf610 +Author: Luc Verhaegen +Date: Thu Apr 13 22:16:07 2006 +0200 + + Adjust to recent (but contested) Xv changes. + +commit 0e821a6503f0e8aa286acae5c6abea45db62dc1b +Author: Luc Verhaegen +Date: Thu Apr 13 15:22:05 2006 +0200 + + Fix typo: bouncebuffer versus bounce_buffer (Reported by Jake Thompson). + +commit 5a6ff82b1ead19f954fd8ca413fd5144b564fa54 +Author: Luc Verhaegen +Date: Wed Apr 12 20:07:21 2006 +0200 + + Introduce ModesExclusive. This way TV encoders and panel can impose their own + modes, and only their own modes, before validation. + +commit 705866aeaefdce9eb8be105837bd2c6fc8a92d74 +Author: Luc Verhaegen +Date: Tue Apr 11 22:33:12 2006 +0200 + + Split TV encoder modetables in NTSC and PAL. + +commit ef0700d4ad5660ca73fa752d62fe6233140f057e +Author: Luc Verhaegen +Date: Tue Apr 11 18:55:08 2006 +0200 + + Massively advance autoconfiguration by introducing MonRecs for all Outputs: + - Implement Monitor for CRT based on my EDID parsing code from #5386. Add + code to combine this information with configured or default information. + - Implement Monitors for CH7011 and VT162x, with PAL or NTSC vrefresh. + - Implement Monitor for Panel based on the timing of the native panel + resolution, as we have no further functionality with respect to panels + currently. + - Add code to combine the various outputs MonRecs and set pScrn->monitor to + the result, so that this can be used for mode validation. We are now no + longer naively assuming that the CRT will follow everything, we properly + disable the CRT when incompatible. + +commit 1e1868983782ec21cf663824d37b0707283f30a7 +Author: Luc Verhaegen +Date: Tue Apr 11 15:40:26 2006 +0200 + + Add *~ (emacs backups) to gitignores. + Fix stray dot in xf86DrvMsg. + +commit 020715fd12c4b18e8ae0c492861c16c61f994ed1 +Author: Luc Verhaegen +Date: Tue Apr 11 13:21:17 2006 +0200 + + Add gitignores, fix via_id.c build after yesterdays commit. + +commit 10a823be3f26e827de07e10bfeb08dbb613849d7 +Author: Luc Verhaegen +Date: Mon Apr 10 23:30:48 2006 +0200 + + ID: VT3108: ECS K8M800-M2 (2.0) (Vasile Rotaru) + VT3118: Sotec Winbook WA2330S5 (Jeffrey Gu) + +commit 995452a545cf13a3c97d5ec82a8cf0f8fbf47b11 +Author: Luc Verhaegen +Date: Sat Apr 8 12:32:32 2006 +0200 + + VT162x: Update dotclocks for NTSC modes to match actual timing. + +commit 0e1579396ed18ef9b88c04c81dba277a6347a7a1 +Author: Luc Verhaegen +Date: Mon Mar 27 23:39:59 2006 +0000 + + ID: VT3108: Clevo D400J (Peter Lord). + +commit 5809290a7ff38898d39e26f938fd708cf097a660 +Author: Luc Verhaegen +Date: Mon Mar 27 18:57:59 2006 +0000 + + Move the CRT into struct ViaOutput as well. + +commit 21c3b50f6e65e3cad87fba4f5f4e1140e7f67f79 +Author: Luc Verhaegen +Date: Mon Mar 27 14:22:29 2006 +0000 + + Autoconf magic failed to catch earlier modular builds properly. Better + solution is to #ifndef _XF86_ANSIC_H after xf86.h is included in code + itself. + +commit c1a305aa1c9f4f4679965d402c7ec36313b51cef +Author: Luc Verhaegen +Date: Sun Mar 26 19:04:42 2006 +0000 + + Move Changelog to ChangeLog. Bleh. + +commit fc07843872ae09220f13accee7389a916a1b87ad +Author: Luc Verhaegen +Date: Sun Mar 26 11:59:37 2006 +0000 + + Add Changelog and manually add entries from 0.2.3 on. + +commit 967ad53adfa03419fbcf3ea04008cb14c7015998 +Author: Luc Verhaegen +Date: Sat Mar 25 04:00:05 2006 +0000 + + HW video overlay and alpha updates: + Blit Size, not width*height. Require at least via DRM 2.7.5. + Use a local buffer for alpha. Use SwovCopy for that too. + +commit ee72ad2c9ff4b5d590a93711f979039f3f3cc061 +Author: Luc Verhaegen +Date: Fri Mar 24 21:27:52 2006 +0000 + + missed via_shadow.c in previous commit (libc unwrapping) + +commit a0748798e8eb7a313689e4f2776fcab2dcde7c6a +Author: Luc Verhaegen +Date: Fri Mar 24 20:45:30 2006 +0000 + + Add autoconf check for server side libc-unwrapping. Unwrap driver + accordingly. + +commit bb6ecf8c676f2804ff17834913b0a4585d26e363 +Author: Luc Verhaegen +Date: Thu Mar 23 02:29:23 2006 +0000 + + Properly encapsulte the gccisms -Wall and -pedantic. + +commit 7c5ade372a41bdcd24d888e4755d84a3e5a56e59 +Author: Luc Verhaegen +Date: Wed Mar 22 22:25:30 2006 +0000 + + [devel-swov_handle_fourcc_change_while_active] PutImage didn't alter the + HQV and Video3 Format fields when the FourCC changed. All it did was + allocate different buffers and set the dimensions correctly. + +commit 337694e157a974deb93f9c874adaac2023e940d0 +Author: Luc Verhaegen +Date: Tue Mar 21 10:43:36 2006 +0000 + + [devel-ids_20060321] VT3344: ECS P4M800PRO-M (Amy Overmyer to openchrome) + +commit 9dc33d012f73e1aca8a0c45937da61cf166550a9 +Author: Luc Verhaegen +Date: Wed Mar 8 21:08:22 2006 +0000 + + [devel-ids_20060308] VT3344: MSI PM8M3-V (Douglas Bell) + +commit 3afe7cc52cf2954446e3264cb93fc9de6ac6ad12 +Author: Luc Verhaegen +Date: Wed Mar 8 01:03:34 2006 +0000 + + [devel-revert_mode_VBlankEnd_limit] VBlankEnd _does_ use the undefined VGA + high bit of CR16. This is not VGA spec material but it still gets used. + +commit 71fc2b5f5d8aa72d6a5cb78395fa527735a7eefb +Author: Luc Verhaegen +Date: Mon Mar 6 21:53:36 2006 +0000 + + [devel-ids_20060306] VT3344: MSI PM8M-V (akshaykk) + +commit 10636c9f1d094861b52a63d2d86da89033ff7e6f +Author: Luc Verhaegen +Date: Mon Mar 6 15:32:39 2006 +0000 + + [devel-drm_versioning_in_via_drm_h] Adjust to DRM driver version defines + existing in the shared header. + +commit e9b8c4909a83afc10ccc8f80e1abb4ca2a39a8de +Author: Luc Verhaegen +Date: Mon Mar 6 10:49:15 2006 +0000 + + [devel-fix_vt3122a_secondary_fifo_depth] When secondary fails, the primary + still thinks secondary is present. And the bandwidth code goes for + different fifosettings. Wrong fifo settings. + +commit 299f2dd769a9254c191ce57600328a2b3ecc952c +Author: Luc Verhaegen +Date: Thu Mar 2 12:27:16 2006 +0000 + + [devel-cleanly_exit_when_unsupported] We don't have a dotclock generator + for VT3108, VT3118, VT3344, VT3157. So we can't produce a workable + mode. So exit cleanly. + +commit f50bc12e595731e69155d02c20fca5d0724a0f47 +Author: Luc Verhaegen +Date: Thu Mar 2 11:44:26 2006 +0000 + + [devel-output_device_whining] Add placeholder init functions that verbosely + detect output devices but which then don't go on implementing them. + +commit 407f42362c9b28403256ccda9bbb462b6e56fa51 +Author: Luc Verhaegen +Date: Wed Mar 1 08:02:05 2006 +0000 + + [devel-ids_20060301] VT3118: Packard Bell Easynote R1100 (Sven Giersig) + +commit 3a98fb4048e1fc2594e7c01b2d093646beca066c +Author: Luc Verhaegen +Date: Sun Feb 12 17:12:46 2006 +0000 + + [devel-ids_20060212] VT3344: Gateway MX3210 (Brendan LaMarche). + +commit e5be7c9cab52ce9358d3956943b47f38fa6be4a1 +Author: Luc Verhaegen +Date: Fri Feb 10 15:11:30 2006 +0000 + + [devel-dma_blit_flags] Fix build against newish via_drm.h: unused + blit.bouncebuffer changed to blit.flags. (reported by Thomas + Hellstrom). + +commit 8c7ab5483ba1f601152f28cc1a0317288b3b2e95 +Author: Luc Verhaegen +Date: Tue Feb 7 14:19:14 2006 +0000 + + [devel-ids_20060207] VT3344: Giga-byte GA-8VM800M (Wayne Gemmell - to + openchrome). + +commit 609adcfdec1eababd745a88097dbbec59ff9661f +Author: Luc Verhaegen +Date: Sun Feb 5 18:22:12 2006 +0000 + + [devel-shield_modes_by_power] + - Move up memset from PrimaryMode to ScreenInit. + - Power off or unsync all outputs before WriteMode. Power on again + afterwards. + +commit 44d140ed0db6d22e53f3e345aef4f33b055fe2a2 +Author: Luc Verhaegen +Date: Fri Feb 3 12:46:49 2006 +0000 + + [devel-ViaVideoRamGetFromRamController_P4M800_exception_fix] Fix warnings + in P4M800 messages; forgot __func__. + +commit 87402ed3e2eea4a48103abb8088b5581df3e4d59 +Author: Luc Verhaegen +Date: Fri Feb 3 12:32:08 2006 +0000 + + [devel-ids_20060203] + - VT7205: ECS KM400-M2 (Yuri Pakhomov) + +commit a141b4906b9057827878c149bed308828b6f263e +Author: Luc Verhaegen +Date: Fri Jan 27 18:11:28 2006 +0000 + + [devel-ViaVideoRamGetFromRamController_P4M800_exception] The P4M800 comes + with a VT7205, but it has a different setup entirely compared to KM400. + We need to handle this like the VT3118, VT3344 and VT3157 should one + day be handled. + +commit bc8836a56e6affcdf95f662352629d4a82b12c48 +Author: Luc Verhaegen +Date: Thu Jan 26 08:55:12 2006 +0000 + + [devel-ids_20060126] VT7205: Packard Bell Quasar2 ("Foo Bar"). VT3108: + Asustek A8V-MX (Martin Kaiser). VT3344: MSI PM8PM (Jeremy Shaw). + VT3108: Shuttle FX21 (Merten Michling). + +commit 8b5ac15ccb914ccc459daf6f4e5706ee05fb996b +Author: Luc Verhaegen +Date: Sun Jan 22 07:30:09 2006 +0000 + + [devel-fix_CRT_ownership] Some bioses, when set to using the LCD as well, + do the whole primary / secondary mess-up. When this happens, make sure + that the CRT gets switched to primary in our still-primary only driver. + +commit 122aeaa83251440aad26f13fb52a67d9eab5711f +Author: Luc Verhaegen +Date: Sat Jan 14 00:18:29 2006 +0000 + + [devel-fix_16bpp_gamma] + - Even in 16bpp, the VIA hw requires a full lut. X doesn't deliver this, it + instead delivers 565bit values, causing the actual gamma to be way off. + Code is a quick and ugly hack, but does work. + - Remove CrtGamma option as this was the same issue. It is terrible how i + managed to miss 16bpp twice before. + - Fix 8bpp too, this is palette, not gamma. + +commit d06e08595f001d2ded3121aa7912a06df0494286 +Author: Luc Verhaegen +Date: Tue Jan 10 07:35:29 2006 +0000 + + [devel-videoram_from_ramcontroller] VIAs latest codedrop stopped using the + inconsistently used scratch registers. They are now poking the ram + controller directly. Sadly they use faulty pci config handling code, + they poke the wrong registers, and they don't interprete the register + content. I of course do things properly, at least for those chips i do + claim to actively support. + +commit 4931e8f7f20f0ae402290df3657856c7fa9c9acb +Author: Luc Verhaegen +Date: Tue Jan 10 06:14:23 2006 +0000 + + devel-output_bus] + - Add OUTPUT_BUS_DI0 and DFP for Output->Position. Should get more when i + can test the relevant hardware. + - Move Bus and Modesetting into a seperate function instead of calling it + from ViaModePrimary. + +commit a5c98258a5bb2e57d14f991d13efe55ede52f344 +Author: Luc Verhaegen +Date: Sun Jan 1 03:24:18 2006 +0000 + + [devel-ViaScratch] + - Create struct ViaScratch, where all highly stable and trustworthy VIA + scratch area info is held for the driver to ignore. + - Move last remaining members of ModeInfo into pVia. Most of the info sits + in ViaOutput. + - Adjust to new year. Wee. + +commit bfbd72e9ac5018b73cb9b7373207be14e74aa954 +Author: Luc Verhaegen +Date: Fri Dec 30 23:01:40 2005 +0000 + + [devel-PLL_generation] This has been the week of one-thousand-and-one modes + for me. Although these crts of mine have seen way more than that. But + it does somewhat situate what work it took to get this one finally + done. But here it is: + Completely Free Modes! Now with added non-borking dotclock PLL generator! + Happy New Years! + +commit 8f4442a8fa88d5a233e3620d179908c5716bc731 +Author: Luc Verhaegen +Date: Fri Dec 30 22:26:40 2005 +0000 + + [devel-territorial_pissings] Review all copyright statements. + - No VIA code remains in via_mode.c and via_video.h. + - There are many files which are unmistakably written by only a single + person, as such, the rather wide "The Unichrome Project" copyright + statement gets replaced. Files are: via_ch7xxx.c/h, via_id.c/h, + via_memory.c/h, via_mode.h, via_vgahw.c/h. + - Then there are the files where VIA code (and thus VIA copyright) remains, + but all code written when and since the "The Unichrome Project" + copyright was added, was by a single person: via_bandwidth.c, + via_cursor.c, via_i2c.c, via_panel.c/h, via_videoregs.h + - Then there are files like via_video.c, via_vt162x.c and via_vt162x.h, + where, under the "The Unichrome Project" banner, changes were made by + other authors too. Here the years are adjusted accordingly, and a new + statement added. + - The files via_driver.c and via_driver.h will keep the current copyright + statement and gain "2004-2006 Luc Verhaegen" ones when i make my first + 2006 commits to these files. + +commit aa5f3bcad160020f4d586ad06da371b624161057 +Author: Luc Verhaegen +Date: Wed Dec 28 16:00:54 2005 +0000 + + [devel-Option_CRTGamma] This hopefully gives users like erc a workaround + for weird colours. I'm unable to reproduce this problem though. + +commit a1e18cfd64ec2f70f28fd2635084112a126255c2 +Author: Luc Verhaegen +Date: Tue Dec 27 01:15:37 2005 +0000 + + [devel-extend_I2CScan] Extend I2CScan to also include I2CBus1, so that the + monitor gets poked too. + +commit f05a33d1ac6bb46405aedb39d99befa4212ffe2a +Author: Luc Verhaegen +Date: Sat Dec 24 01:11:59 2005 +0000 + + [devel-pre_0_2_1_leftovers] + - Add UseDMACopy to manpage. + - Remove LCDDualEdge option, no code backing it anyway. + +commit 1472c8f242adf5da236b3e401c7afcdbc94b58a7 +Author: Luc Verhaegen +Date: Thu Dec 22 21:11:44 2005 +0000 + + [devel-Add_DMACopy] + - Move DRMVersion to a CARD32. Makes a world of difference when comparing + versions. CARD32 will suffice for at least the next 2 years. + - Add a routine to SwovCopy implementing Thomas Hellstroms DMA blit. This + very dramatically reduces Xv overhead. + +commit b6eb2568a1ed12472acd17fbc8a90fcf5a11dae4 +Author: Luc Verhaegen +Date: Mon Dec 19 13:49:15 2005 +0000 + + [devel-fix_glammal_collection] + - Fix glammal collection. + - Provide an overscan function to the cmap code directly. + +commit 1a695ccb88445d75d48c3795dabd8241a5b93537 +Author: Luc Verhaegen +Date: Mon Dec 19 10:33:43 2005 +0000 + + [devel-SetOverScan] Set overscan colour. + +commit ab8672209c8597859e33ceb4d5bbc4ad4fd103d3 +Author: Luc Verhaegen +Date: Sun Dec 18 18:54:11 2005 +0000 + + [devel-ids_20051218] + - VT3108: PCChips M861G (Thomasq Giezak) + - VT3118: Clevo M5xxV (Anton Farygin, originally reported to openchrome) + - VT3118: Mitac 8965 (reported to openchrome by unknown) + - VT3108: Twinhead 12KF (reported to openchrome by unknown) + - debunked averatecs and the element computers. + +commit 3c0b77c8ba43239bd5f0c531d2ed1e823c217b31 +Author: Luc Verhaegen +Date: Wed Dec 14 11:37:04 2005 +0000 + + [devel-I_dont_believe_in_PROBE_DETECT] I'm having a dog of a time running X + so that it calls PreInit with PROBE_DETECT. Everyone seems to be using + getconfig.pl anyway. Ditch the VBE DDC gathering, as X modegathering + will be changed anyway. + +commit 5f66c6c7bae2b890ec6af5f0be2eb1f9f8ff13e9 +Author: Luc Verhaegen +Date: Wed Dec 14 01:05:35 2005 +0000 + + [devel-review_virtual_for_VT7205] + - Offset restricts the virtual resolution too. + - VT7205 does have bits 24 and 25 of Primary StartAddress. + - H values are all in 8 pixels. Why do VIA monkeys think it is 16? + +commit f0720502ccd0b1f734bf89fcf25d068606e93b0c +Author: Luc Verhaegen +Date: Tue Dec 13 19:44:51 2005 +0000 + + [devel-cursor_failure_fix] pVia->CursorInfoRec was not-NULL even when the + Cursor init failed. + +commit c90b9ca5e12e5199a8a97059b5b86b03762fd34d +Author: Luc Verhaegen +Date: Tue Dec 13 13:52:26 2005 +0000 + + [devel-remove_unused_symbols] Complain as soon as you get unresolved + symbols warnings. These should all be down to the loaded modules + themselves, but this can still happen. For instance, the vbe module is + known not to do decent checking itself. + +commit 4ab096fceb43277e57358c5b7713b51162f4e12b +Author: Luc Verhaegen +Date: Tue Dec 13 11:56:36 2005 +0000 + + [devel-mode_h_remove_copyrights] Remove VIA and S3 copyrights from + via_mode.h. No code remains. + +commit d251ba6f3106ac59090c734288c59c56ad9704e8 +Author: Luc Verhaegen +Date: Tue Dec 13 11:44:19 2005 +0000 + + [devel-DotClocks_prune] Prune the dotclocks list to have only those modes + that are known to X. There used to be panel modes in there too, but the + panel code doesn't care one bit. + +commit e4d09d981ff1c7d20d10210d90c025a9ba253b46 +Author: Luc Verhaegen +Date: Mon Dec 12 15:02:17 2005 +0000 + + [devel-ViaPanel_to_ViaOutput] + - Move ViaPanel into ViaOutput code too. + - We have an ugly workaround for the primary DotClock value and still have + to keep some dotclock setting code global. + - Move TV bus handling closer to the modesetting of the TV, panel does this + inside PanelMode still. + - Have a special run through the outputs to disable all other devices when + a panel is found. + - Fix daft || versus && issue when warning that the panel is unsupported. + - Remove VIA/S3 copyright from via_mode.h, not a single line of the + original via_bios.h remains. + This abstraction is quite artificial and will not survive a panel clean-up, + if that will ever happen. + +commit d6ea81daf1cd5c7c507897a7454cf5810ea52cb3 +Author: Luc Verhaegen +Date: Sun Dec 11 13:52:59 2005 +0000 + + [devel-ViaOutput] + - Move TV encoders over to struct ViaOutput, making TV encoders 95% + independent from the rest of the driver. + - Create the doubly linked list from pVia. + - Add functions and further code to handle this linked list. + +commit 7790406ed9157820370c2570f2afb6f28a17bb80 +Author: Luc Verhaegen +Date: Wed Dec 7 00:23:27 2005 +0000 + + [devel-ids_20051206] VT3344: Asus P5V800-MX (Jordan Thompson) + +commit 6769172d9253dc373390ea088fabb1a56c63ed81 +Author: Luc Verhaegen +Date: Tue Dec 6 23:36:27 2005 +0000 + + [devel-bandwidth_VT3344] + - Add VT3344 and VT3157 to the bandwidth table. + - Update panel and limited support warnings. + +commit 64ac0909c1858841664b6b1619937f3cc04b47e8 +Author: Luc Verhaegen +Date: Tue Dec 6 23:27:21 2005 +0000 + + [devel-driverman_PRE] Fix the "unused variable" warnings spawned by + automake by renaming driverman_SOURCES to driverman_PRE in + man/Makefile.am. (Kevin E. Martin) + +commit 053c411a45e1bf15147642ce2a6615b9616f4fed +Author: Luc Verhaegen +Date: Mon Dec 5 14:57:45 2005 +0000 + + [devel-move_option_handling_to_output] + - Rename TVType to TVStandard. + - Move TV encoder specific option handling to via_vt162x.c and + via_ch7xxx.c. + - Move Panel options to via_panel.c. + - Update manpage. + +commit b603424bc88c0d580c0093f133ae164d5caed74e +Author: Luc Verhaegen +Date: Sun Dec 4 13:34:54 2005 +0000 + + [devel-remove_HAVE_DEBUG_entirely] + - Remove HAVE_DEBUG dependance. + - Have all PrintReg code present at all times. + - Pass __func__ to PrintReg functions directly. + +commit 4407ed551196f7c13ed1be0ed8df8c7f3f65f2b7 +Author: Luc Verhaegen +Date: Sat Dec 3 13:02:49 2005 +0000 + + [devel-add_VT3157] Add pci id for VT3157. This requires another dotclock + calculation and has internal LVDS and TMDS encoders, so this will be + quite an exciting device whenever it actually materialises. + +commit 6b76640abec3f28461849f32a98bfeb0c61ca4d3 +Author: Luc Verhaegen +Date: Sat Dec 3 12:30:34 2005 +0000 + + [devel-remove_modular_XvExtension] Remove AH_BOTTOM #definition of + XvExtension. XvExtension was added to xorg-server.h directly and the + superflous XV will be removed before 7.1. + +commit e75b4e369b1eeadb7b211961ba33dc849a6662bd +Author: Luc Verhaegen +Date: Fri Dec 2 15:08:38 2005 +0000 + + [devel-via_panel_c] + - Move panel code to via_panel_c + - Make ViaSetUseExternalClock and ViaSetSecondaryDotclock global. + +commit 520948f9e08eb6bf0431dfaddf1302e6280cc6f5 +Author: Luc Verhaegen +Date: Fri Dec 2 11:07:52 2005 +0000 + + [devel-panel_callbacks] + - Move to callbacks for the panel code. This allows for clean separation + like for TV encoders (but the Crtc must still be aware of a panel being + in use). + - Rename all panel code to ViaPanel*. + - Move ModeInfo->Clock to ModeInfo->PanelClock. + - Move ModeInfo->ResolutionIndex to ModeInfo->PanelResolutionIndex. + +commit 9a092d7f50870192f919bc3e339c5fa07b8362b6 +Author: Luc Verhaegen +Date: Thu Dec 1 21:27:13 2005 +0000 + + [devel-via_mode_h_cleanup] + - Remove pBIOSInfo dependance of via_video.c: modes are no longer lying. + - Remove pBIOSInfo dependance of via_cursor.c: waiting for fallout from + laptop users. + - Rename VIABIOSInfoRec to struct ViaModeInfo. + - Move VIA_MEM_ and VIA_DEVICE_ defines to via_driver.h. + - Move via_bandwidth prototypes to via_driver.h. + - Remove via_mode.h inclusion from via_driver.h. + +commit e5cadac21d04214b72a95d99ee44da221941f010 +Author: Luc Verhaegen +Date: Thu Dec 1 19:57:02 2005 +0000 + + [devel-via_panel_h] Move via_mode.h to via_panel.h, and move via_bios.h to + via_mode.h. This plays merry hell with CVS history, but this is a very + neccessary change. + +commit 9c0c875e908a561227d38010666b594cccec75ea +Author: Luc Verhaegen +Date: Thu Dec 1 19:43:57 2005 +0000 + + [devel-move_ViaDotClocks] + - Move ViaDotClocks from via_mode.h to via_mode.c + - Bring all dotclock related code next to eachother. + +commit 06a11b1f96f497bbed4fc62587180015d5272cc9 +Author: Luc Verhaegen +Date: Thu Dec 1 19:33:04 2005 +0000 + + [devel-id_HasPanel] + - Rename struct ViaCardIdStruct to plainly struct ViaCardId. Rename table + to ViaCardIds. + - Remove CARD8 Output to Bool HasPanel, and reduce ViaCardIds. + - Simplify ViaOutputsDetect + +commit cb2a98ac51755036259f5d45e0d2347a9f189c78 +Author: Luc Verhaegen +Date: Wed Nov 30 10:51:14 2005 +0000 + + [devel-remove_vgaHWInit] + - My modesetting doesn't depend on vgaHW initialised hwp->ModeRegs. It does + everything from scratch itself. Remove vgaHWInit from WriteMode. + - This removes our single point of failure within WriteMode, which was a + vgaHWInit call to vgaHWGetHWRec, which I already did in PreInit. So + WriteMode becomes a static void. + - This cleans up return handling in EnterVT, ScreenInit and SwitchMode. + - Remove leftover debug entry in pVia (for dumping rom). + +commit 113d9b38159ca7a375576cb6682febca7fdaa0b2 +Author: Luc Verhaegen +Date: Wed Nov 30 10:24:52 2005 +0000 + + [devel-update_vendor] + - Update vendor name to no longer claim that Xorg is responsible for this + driver. + - Sanitise Chipset listing prelude. + - Sanitise support notice. + +commit 1c6c43fde31e6d00b2f06628d0653528afb67ca2 +Author: Luc Verhaegen +Date: Wed Nov 30 10:11:04 2005 +0000 + + [devel-update_modular_XvExtension] + - Cleaner XvExtension handling: + - add an extra #define to config depending on XV + - check for xf86xv.h and xf86xvpriv.h for inclusion of via_video.o + - Import recent Xorg changes. + - configure.ac: don't build static (ajax) + - configure.ac: pkgcheck dependencies (kem) + - man/Makefile.am: update drivermandir (alanc) + +commit b6b8c303acdaf9979a3e42bc0905a324964c9f4f +Author: Luc Verhaegen +Date: Tue Nov 29 14:37:10 2005 +0000 + + [devel-drmCommandWriteRead_to_drmSymbols] Add missing drmCommandWriteRead + to drmSymbols to silence loader warnings. + +commit c00d6200b319b158586be8f70742942e461a9d3d +Author: Luc Verhaegen +Date: Tue Nov 29 14:30:42 2005 +0000 + + [devel-ids_20051129] VT3108: Abit KV-81 (Tobias Blom) + +commit 9eb3cf2da2956ba0fc65d7a5391caba5c1109727 +Author: Luc Verhaegen +Date: Fri Nov 25 14:26:05 2005 +0000 + + [devel-__func__] + - Move away from __FUNCTION__ to the more portable __func__. __FUNCTION__ + was a gccism. __func__ is, on gcc, equal to __FUNCTION__ but is a C99 + standard. + - #define __func__ to unknown for other compilers (like solaris cc) + +commit 4deb467f21ac11731c3143cd1c10d87506a310e5 +Author: Luc Verhaegen +Date: Wed Nov 16 16:36:44 2005 +0000 + + [devel-ids_20051116] VT7205: Acorp 7KM400QP (Thomas R.) VT7205: Biostar + P4M800-M7 (Alex Labram) + +commit 2c98dd28f448ef0b2dfd877c11982d5b1ef51ea6 +Author: Luc Verhaegen +Date: Wed Nov 16 16:35:24 2005 +0000 + + [devel-drmVersion] + - Fix build to build against old and new via_drm.h. + - Get pVia->drmVersion early on, so it can be used for more than the + ringbuffer. Relay DRM version in log. + - Include drm/via_drm.h instead of plain via_drm.h when using modular. + +commit e2ce1961008bf80ea789b5f8d068526310bed083 +Author: Luc Verhaegen +Date: Wed Nov 16 16:32:24 2005 +0000 + + [devel-drmVersion] Committed test code from dereferenced directory. + Reverting. + +commit 379d8e4125262668f1678795c59ea78c02dffb1a +Author: Luc Verhaegen +Date: Wed Nov 16 16:00:28 2005 +0000 + + [devel-drmVersion] + - Fix build to build against old and new via_drm.h. + - Get pVia->drmVersion early on, so it can be used for more than the + ringbuffer. Relay DRM version in log. + - Include drm/via_drm.h instead of plain via_drm.h when using modular. + +commit b0b628e4b35791a04f56fa5509589d6da72d65fe +Author: Luc Verhaegen +Date: Wed Nov 9 08:41:14 2005 +0000 + + [devel-ids_20051108] VT3108: Twinhead N14RA (Steve Webb) VT3108: DFI + K8M800-MLVF (Barvaz Zan) VT7205: MSI PM8M2-V (Jeremy Shaw) + +commit cbc427767e9fbc54cccb628437ee8530b1260699 +Author: Luc Verhaegen +Date: Mon Nov 7 15:17:31 2005 +0000 + + [devel-cursor_cleanup] + - Fix cursor memory allocation, both by timing, size and alignment. + - Add support for 64*64 cursor. + +commit 816598b6c3d4e5350f7f19fc2a5296f85c35f0b0 +Author: Luc Verhaegen +Date: Fri Nov 4 11:05:38 2005 +0000 + + [devel-p4m800_bandwidth] + - P4M800 turns out to be a VT7205A but on an intel northbridge. VT7205 + revision says that it's a plain VT7205. Adjust ViaGetMemoryBandwidth to + check for host bridge device id and get the appropriate bandwidth. + - Move bandwith checking to via_driver.c, right after the memory clock + check. + +commit cc9a4a0533dfc5abc8459b4bb08b1bd73162b90a +Author: Luc Verhaegen +Date: Fri Nov 4 10:13:17 2005 +0000 + + [devel-naming_VTpciid] Rework naming scheme once again. It is clear that + there is no distinct VT naming structure for the unichromes themselves. + Sometimes the northbridge VT name was used, sometimes the number after + VT is the plain pci-id. Now we start using pci-id all the way, so that + the amount of confusion is kept at the absolute minimum. + +commit 7cf9c45b8ad08bf2460403258ff6e1f136368572 +Author: Luc Verhaegen +Date: Mon Oct 31 12:33:12 2005 +0000 + + [devel-remove_ViaDoubleCheckVT3122Revision] Remove + ViaDoubleCheckVT3122Revision. This uncovered nothing in more than a + year, and was thus completely useless. + +commit e6ee9ee394f28d9acad16dba800f7945acf2bf7d +Author: Luc Verhaegen +Date: Mon Oct 31 12:01:18 2005 +0000 + + [devel-README_modular_build] Update the README with instructions for + building the driver against modular. + +commit 517ff1503ff48f30d60070467ea98702aa670df2 +Author: Luc Verhaegen +Date: Mon Oct 31 10:58:52 2005 +0000 + + [devel-ids_20051031] VT3205: Asustek A7V8X-LA (Bryan Donlan) VT3314: + Biostar P4M80-M4 (Joe U) VT3204: MSI K8MM3-V (Joerg Woelke) + +commit 45b5ae537d17057719df2e305e6b23aa91a9d196 +Author: Luc Verhaegen +Date: Thu Oct 27 20:17:49 2005 +0000 + + [devel-ids_20051026_fix] Fix wrong Chip for FSC Amilo Pro V2030 (Bernhard + M. Wiedemann). + +commit 89aa25c7244dd705b84db57e47fedbeacbbe3c40 +Author: Luc Verhaegen +Date: Thu Oct 27 20:08:24 2005 +0000 + + [devel-fix_bsd_drm_allocmem] Import X.org fix that makes via_memory.c work + on the BSDs too. Closes + 1206267. (Jake Burkholder, Eric Anholt) + +commit eae87ddf37554b95ca112d3c7a37a8c0869e3147 +Author: Luc Verhaegen +Date: Thu Oct 27 00:21:56 2005 +0000 + + [devel-modular_build_initial] + - Steal autogen.sh, configure.ac and Makefile.ams from Xorg xf86-video-via + - Adjust to match xf86-video-unichrome peculiarities and enable XV. + - Import X.org changes with respect to the Modular build: + - include config.h. + - adjust header locations to match sparse -Is. + +commit 0e2136a0aa28c91446f8daedc9c7d21c29c921c6 +Author: Luc Verhaegen +Date: Wed Oct 26 10:32:52 2005 +0000 + + [devel-ids_20051026] + - Add comment for 1106:3122 moronic id sharing. + - VT3314: Fujitsu-Siemens Amilo Pro V2030 (Jan Palus) + - VT3204: ECS 536 (Mike Kirsanow) + +commit 139a5e90e6d98e467efade3a808a52df3905e447 +Author: Luc Verhaegen +Date: Wed Oct 26 08:23:38 2005 +0000 + + [devel-update_via_dri_h_5_0_0] Update via_dri.h to Xorg version. This, + sadly, requires the latest Mesa to work. And i don't see a clean way to + work around this, or to build things conditionally. + +commit 16839ccba820e1cb92668cccd5fdcdf8ad19e842 +Author: Luc Verhaegen +Date: Fri Oct 21 13:22:26 2005 +0000 + + [devel-fix_swov_striping] Vertical Filtering was disabled when zooming in. + Can be traced back to via_swov.c:Upd_Video's hqvFilterCtl &= + 0xfffdffff; + +commit a26c55ba16ce9e4180c9e9017161e904115bd3e9 +Author: Luc Verhaegen +Date: Fri Oct 21 13:18:43 2005 +0000 + + [devel-remove_SwovCopy_idiocy] Previous commit was from wrong and untested + tree. + +commit 23c5d0f55f2e30a8860c6fa0fbefc1380c90e81d +Author: Luc Verhaegen +Date: Fri Oct 21 13:14:19 2005 +0000 + + [devel-remove_SwovCopy_idiocy] Playing with cache prefetching in asm leads + to fuzzy thinking. Remove clueless stuff. + +commit 3d6ea3d0d713f1129613acdf3a8185cb56d5ddbf +Author: Luc Verhaegen +Date: Fri Oct 21 12:20:05 2005 +0000 + + [devel-fix_nonXv_includes] Fix includes for when Xv is disabled. Instead of + depending on XvExtension define, depend on _VIA_VIDEO_H_. That way, + when something else causes via_video.h not to be included, this is + handled just as cleanly. + +commit 4ed2721166df9d22b65eaa8e74cf59270e474bb2 +Author: Luc Verhaegen +Date: Fri Oct 21 10:44:00 2005 +0000 + + [devel-xorg_version_current] Fix drivers XF86ModuleVersionInfo to report + the Xorg version instead of the xfree86 one, when built against Xorg. + +commit 96de2cc67fe2b436b2d5e0258c6d37f350eae4ea +Author: Luc Verhaegen +Date: Tue Oct 18 14:51:10 2005 +0000 + + [devel-tv_more_flexible_i2c_detection] + - Use a table so that different addresses can be matched to the same + devices. + - Combine TVDetect and TVInit. Move basic I2CDev handling up. + - Remove general TVEncoder ids and go for (tv)driver local enums. + +commit 9d5dcbc18211486a8cef2768f42b11cf6011d799 +Author: Luc Verhaegen +Date: Fri Oct 14 22:24:51 2005 +0000 + + [devel-mode_VBlankEnd_limit] + - Halve primary VBlankEnd. + - Remove SubCarrier value from VT1623 720x576Noscale table. + +commit 2a8ab929ecbac42c624d2a5dac3ce4f3b439ff9a +Author: Luc Verhaegen +Date: Fri Oct 14 22:20:19 2005 +0000 + + [devel-man_add_CH7011] + - Add CH7011 information to the manpage. + - Clean up man page a bit. + +commit 463e3ff9a48d21fda50842828e17f40f517d597c +Author: Luc Verhaegen +Date: Fri Oct 14 22:09:37 2005 +0000 + + [devel-fix_Xvless_build] Fix build for the case where BuildXvExt is defined + as NO. + +commit bab2b22ce2e11b9832b70ef8a958fd7a3c153109 +Author: Luc Verhaegen +Date: Fri Oct 14 16:57:21 2005 +0000 + + [devel-xfree86_4_5_dga] Fix build against XFree86 4.5.0. DGAOpenFrameBuffer + arguments changed. + +commit b33b3757e38d28f05f72b56cea86ea8afc7cc410 +Author: Luc Verhaegen +Date: Fri Oct 14 09:17:29 2005 +0000 + + [devel-swov_idles] + - FourCC is passed down from the DIX as an int, not a CARD32. + - Reintroduce HQVDoubleInit, so that the brief green flash of YUV 0x0000 + can be avoided on CLE266Ax. + - Wait for HQV to go idle before disabling the HQV. This hopefully fixes + CLE266Cx (ported from Barry Scotts code - still needs testing) + +commit 250ff89aebc1750b3d547b0930a6c2323b5fe669 +Author: Luc Verhaegen +Date: Thu Oct 13 18:21:26 2005 +0000 + + [devel-add_ch7011] Add support for the Chrontel CH7011, as found on the + shuttle FX43. + +commit e0a847e05582bca43e36ece718fe9ef94f1bb59b +Author: Luc Verhaegen +Date: Wed Oct 12 13:40:50 2005 +0000 + + [devel-ids_20051012] VT3122: LT21 VA28 (Anes Lihovac) + +commit b1abf37ca6809ab77435e3ca51b7e541d2757b1c +Author: Luc Verhaegen +Date: Wed Oct 12 10:47:43 2005 +0000 + + [devel-ids_20051011] VT3314: Asrock P4VM800 (Fede Heinz) VT3204: + Packard-Bell Easynote E6300 (Leandro Nini) VT3204: Asus K8V-MX (Jasper + Van Der Neut) VT3205: Abit VA-20 (Ofir Avni) VT3205: FIC K7M-400A (Jiri + Pinkava) VT3259: Asrock P4VM8 ("Tango") VT3122: AsusTek Terminator C3V + (Brad Davis) VT3204: Averatec 3270 (Miguel Marte) VT3205: Soltek + SL-B7C-FGR ("conteb") VT3204: Packard Bell B3 800 (MikeH) VT3204: MSI + K8MM-V (Philip Prindeville) VT3204: ECS K8M800-M2 (Daniel Syddall) + VT3204: Asrock K8Upgrade-VM800 ("Strat") VT3204: FIC K8M-800M (Vijay + Shamdas) VT3205: Asus Terminator A7VT (Ben C) VT3122: Element Computers + Helium 2100 (Brian Cornell) VT3259: Mitac 8666 (Gert Lavsen) VT3204: + Mitac 8889 (Rico Rommel) VT3259: Chaintech MPM800 (Kevin Kuphal) + +commit eb74d5cb108ecca8ddd88dcffd115813245afe27 +Author: Luc Verhaegen +Date: Mon Oct 10 04:47:46 2005 +0000 + + [devel-MemorySize_fallback] When CR34 is not set on VT3122, try using CR39 + instead. (Brad Davis). + +commit b2743c68db9f1b30ec7f121b14eb058612e07faf +Author: Luc Verhaegen +Date: Sun Oct 9 14:31:15 2005 +0000 + + [devel-fix_DRIINFO_VERSION_bump] + - Fix build warning concerning frameBufferPhysicalAddress + - Add build dependant DRIQueryVersion checking instead of idiotically + setting the latest version manually. The extra work required by + properly fixing version differences and keeping things building against + many versions is negligable. + +commit bf564bff34ce851b84d864348a91ec69c1b45cec +Author: Luc Verhaegen +Date: Sat Oct 8 11:26:10 2005 +0000 + + [devel-rename_to_VT3xxx] + - Rename CLE266, KM400, K8M800 and PM800 to VT3122, VT3205, VT3204, VT3259 + respectively. + - Add pci ids for the VT3314. + - Rework VIAIdentify to decently outline which is which. + - Adjust manpage to follow naming structure. + - Add missing defines to XFree86 case for manpage. + +commit d71ae26478dca0976b4934d7e0f5e0bbdefcd60c +Author: Luc Verhaegen +Date: Fri Oct 7 13:05:55 2005 +0000 + + [devel-adjust_build_to_restructure] + - Add toplevel Imakefile and man/Imakefile. + - Add preliminary README explaining how to build the driver (new structure + requires make Makefiles on top of xmkmf) + - Bump version number to 0.2.0. This is to avoid the ambiguity introduced + by the forkers when they decided to bump their patchlevel to 0.1.31, + while r31 never appeared on unichrome.sf.net. xf86-video-unichrome will + have 0.2.0 for CVS, and 0.2.1 for the first release. + +commit b7775ba1e8ea506f231bc18937420453e9586158 +Author: Luc Verhaegen +Date: Wed Oct 5 07:59:47 2005 +0000 + + [devel-swov_restructure_cleanup] + - Update some comments. Some are no longer applicable. + - Review Xv adaptor flags. + - Review Reset. + - Remove ViaVideoStore, ViaVideoRestore and ViaVideoExit hooks. All cases + are handled by StopVideo and on Restore, PutImage. Remaining + awkwardness is due to server side Xv awkwardness. + +commit 4e33b3848da771678a68d826fbe1483f70dec15f +Author: Luc Verhaegen +Date: Tue Oct 4 06:26:04 2005 +0000 + + [devel-swov_restructure_via_video] Sanitize ordering of via_video.c. + +commit a65a7ef1b7d2674e249486b166a71577070fc80b +Author: Luc Verhaegen +Date: Tue Oct 4 05:03:40 2005 +0000 + + [devel-swov_fourcc_rgb] + - Add RV15, RV16 and RV32. + - Check alignment for all fourccs. + - Move ImageRec definitions to the start of via_video.c. + +commit 4b1eb56e8b1ed4518bc7f295a569e7c3c34a869d +Author: Luc Verhaegen +Date: Fri Sep 30 04:11:31 2005 +0000 + + [devel-swov_SurfaceCreate_return] + - Make the FALSE returns of ViaSwovSurfaceCreate more verbose. + +commit 9de82d2dda1e15ccd89975f9ef3a518133ba9ee8 +Author: Luc Verhaegen +Date: Fri Sep 30 02:34:42 2005 +0000 + + [devel-swov_I420] Add support for the I420 FourCC. + +commit ad5e204ecff8e3259906a8fc2a65a5ab7e279ff6 +Author: Luc Verhaegen +Date: Tue Sep 27 12:28:55 2005 +0000 + + [devel-swov_opacity_limit] Clued people tend to properly check the value of + parameters passed. + +commit 92e53cd460e8aa3722e953b40420a3ba8e4b31d9 +Author: Luc Verhaegen +Date: Tue Sep 27 12:18:07 2005 +0000 + + [devel-swov_yuy2] Fix HQV Stride for YUY2. + +commit 794daa0211f27a546fc5cde223b82804f5827a48 +Author: Luc Verhaegen +Date: Mon Sep 26 04:00:57 2005 +0000 + + [devel-swov_alpha_size] Include H/VDisplay limitation in AlphaWidth/Height + calculation. Seperate this, and the alignment to a seperate function. + +commit 1d60cbef27099ec260b06a59f89554400529e593 +Author: Luc Verhaegen +Date: Mon Sep 26 02:57:12 2005 +0000 + + [devel-memory_alignment_fix] Fix oversight in DRI memory allocation + alignment. + +commit 128ad2bda7841c71a2307e231b293b7e60b306a1 +Author: Luc Verhaegen +Date: Sun Sep 25 17:52:40 2005 +0000 + + [devel-swov_front_doublebuffer] + - ReputImage incurs some delay in a subsequent PutImages ViaHQVSwovIdle. To + reduce this the SwovCopy should happen before the idle wait, requiring + double front buffering. + - The start of PutImage now handles each of the three cases distinctively: + Swov not active; size change; normal update. This increases codesize + slightly, but it vastly increases transparency. + +commit 0aa3ec5b05f8a32d487beba3388b77a43fae5542 +Author: Luc Verhaegen +Date: Sun Sep 25 17:29:14 2005 +0000 + + [devel-swov_alpha_doublebuffer] + - Use 2 alpha planes. This allows us to write to an alpha plane without the + overlay seeing any of it, up until the point where we make the switch. + - clean up AlphaSurface handling, together with a solidification of the + alpha/colorkey code in PutImage. + +commit a0b5d83d3a9377a6f5fdb2cff1a4a7d8e82a1a6b +Author: Luc Verhaegen +Date: Sun Sep 25 06:40:02 2005 +0000 + + [devel-memory_alignment] + - Introduce a third variable to ViaMemAlloc: alignment + - It turns out that the HQV requires 16byte aligned addressen, both for the + front and back buffers. Video3 doesn't care, luckily. + +commit 574d018c1b84176b56fb45324da49b4f7a18b029 +Author: Luc Verhaegen +Date: Sun Sep 25 05:41:25 2005 +0000 + + [devel-swov_Alpha_inactive] Fix handling of the XV_OPACITY attribute when + SWOV is not active. + +commit c79ff45da0113aaec632bbec6197ca3001a44c98 +Author: Luc Verhaegen +Date: Fri Sep 23 23:26:58 2005 +0000 + + [devel-swov_ReputImage_fixing] Handle some stuff that went broken by moving + the old AdjustFrame hooking to proper ReputImage: + - set Port->Drw_X/Y in ReputImage (not only called on FrameAdjust) + - disable V3 as part of StopVideo (StopVideo calls are now rarer than they + used to be) + - rework Alpha Plane clipping position. + +commit 824c4cb698a05ed7ad953f9808a424d89865ffc1 +Author: Luc Verhaegen +Date: Tue Sep 20 05:04:11 2005 +0000 + + [devel-swov_Opacity] + - Add support for the Overlay Alpha Plane: a 4bit surface describing the + opacity of the overlay versus the graphics. + - Work around the various issues of the unichromes alpha plane + implementation. + - Add XV_OPACITY attribute: 0 disables completely (colorkey, no alpha), + 15 makes higher level windows completely invisible. + - Use colorkey clipping for selection. Clipped area (video) is full + opacity. Rest (normally graphics) is at XV_OPACITY. This means that + whatever is normally put on top of the overlay now becomes transparent. + A limited effect, but about as much use as this code will ever be + without proper compositing. + +commit b4dcdb30a2c286c473c13a42bf22f1e222d379db +Author: Luc Verhaegen +Date: Sun Sep 18 23:52:22 2005 +0000 + + [devel-swov_ColorKey] + - Fix ColorKey, it was eternally locked at the initial value. + - Remove SupportTwoColorKey, we only use a single overlay anyway. + +commit e3abb082d68693cc820b828f858329578398a0a6 +Author: Luc Verhaegen +Date: Thu Sep 15 18:20:42 2005 +0000 + + [devel-swov_ReputImage] + - Use pScrn->frameX0/Y0 instead of tracking the frame position ourselves. + Retain some Swov entries to reduce the time AdjustFrame takes. + - Use ReputImage (undocumented) instead of an AdjustFrame call. StopVideo + now doesn't get called anymore when adjusting the frame. + +commit e4a7be15247e0f8eaca7ad6b45bae49330cebb82 +Author: Luc Verhaegen +Date: Thu Sep 15 00:02:08 2005 +0000 + + [devel-swov_REGION_EQUAL] Fix RegionEqual/RegionsEqual oversight. + +commit a1a73f2b60b0d7ffa5408ff26bf61f795a516f09 +Author: Luc Verhaegen +Date: Wed Sep 14 05:49:55 2005 +0000 + + [devel-swov_REGION_EQUAL] Fix build on newer X versions by ditching the + daft define i instituted just recently. #ifdef around REGION_EQUAL, and + if ! then include RegionEqual and #define it to REGION_EQUAL. + +commit c75f7bc735f05c9e1879ec6a13a7f836c41ee4d2 +Author: Luc Verhaegen +Date: Wed Sep 14 05:19:29 2005 +0000 + + [devel-memory_size] + - check for size, return NULL immediately when 0. + +commit 6811aa742fcb2ba47a36b8abb7e4593ed5374a90 +Author: Luc Verhaegen +Date: Thu Sep 8 12:43:40 2005 +0000 + + [devel-swov_lowmemory] + - Move HQVSurface and SwovSurface to Back and Front respectively. + - Only depend on ThreeHQVBuffer in SwovSurfaceCreate. Check for Back[2] + everywhere else. Move Back[2] allocation last. X_WARNING only on + failure. This allows us to continue functioning without this allocated. + - The SW_FLIP loop barely ever delays, dual buffering gains us barely + anything. Maybe in a kernel based fb memory manager, where the Xv + client can get handed a surface, but not here. So drop the second + buffer. + - With Front[1] gone, HQV SrcAddress can be set from SwovSurfaceCreate, and + Cr/CbOffset can be dropped from Port. memcpy from HQVFlip directly. + Drop SwovCopy. Drop FrameNum. + - reduce noisyness of memory manager, X_WARNING suffices. + +commit d65ee9804958779f5beb0c0c07ddc8a336973198 +Author: Luc Verhaegen +Date: Wed Sep 7 23:02:51 2005 +0000 + + [devel-swov_enableflipfire] + - Move SwovInitControl to HQVInit, move setting Video3Control up. + - Clean up waiting for the HQV to get idle and V3 fire to be completed. + Move out to its own function. + - Properly wait for flipping to end at HQVFlip. + - Depend on Swov->Active for the Status wait. + - Move X_ERRORs to X_WARNINGs and stop throwing in the towel when something + fails to happen (HQVFlip mainly). + - Set up FIFO at the start of PutImage when !Swov->Active instead of at + VideoInit. + +commit 0a915a18f9938d0d5028d1ff7ee8e9ef2c410b2d +Author: Luc Verhaegen +Date: Sun Sep 4 14:12:46 2005 +0000 + + [devel-swov_ViaSwovColorKey] + - Rename SetColor/ChromaKey to ViaSwovColor/ChromaKey. + - Have Xv attributes use ViaSwovColorKey. + +commit 1e6c4e403dd81eeb92d6a530a37f27b96a146e07 +Author: Luc Verhaegen +Date: Sun Sep 4 13:42:11 2005 +0000 + + [devel-swov_hide] + - Move ViaSwovHide to ViaSwovStop. + - Have the position calculations hide/display through disabling/enabling V3 + +commit 1e7974f661ee5d868389cf7a69fb3d29c6bc1e96 +Author: Luc Verhaegen +Date: Fri Sep 2 20:48:01 2005 +0000 + + [devel-swov_rework_flip] + - Move the HQV Flip and HQV stuff from SwovFire into HQVFlip. + - Rework actual flipping. + - Blacking of surface no longer is necessary. Drop ViaSwovFillBlack. + - HQVInitPatch no longer seems necessary either. + - Go for a doubleword copy routine. Exa upload would be real welcome. + - Fire V3, when necessary manually. + +commit 5156527b22e60a4e925d1a491210507d8e4e169a +Author: Luc Verhaegen +Date: Wed Aug 31 21:48:37 2005 +0000 + + [devel-swov_remove_DBG_DD] + - Remove the last remnants of DBG_DD. + - Quick onceover of DecideOverlaySupport. Rename to ViaSwovBandwidth. + - Remove pVia->OverlaySupported, add pVia->Swov->Allowed instead. + - #ifdef out RegionsEqual when using a "recent" X version. + +commit eb421189aeaeafe90cbb4d8f8015d68d1f30799b +Author: Luc Verhaegen +Date: Tue Aug 30 22:35:19 2005 +0000 + + [devel-swov_videoregs] + - rename via_swov.h to via_videoregs.h + - add what's useful from via.h to via_videoregs.h, next to the memory + mapped registers. remove via.h + - make a few cosmetic changes in regard to register and define naming. + +commit 3f07f9b833662f38ee66200da355889575e4cc7f +Author: Luc Verhaegen +Date: Tue Aug 30 20:08:04 2005 +0000 + + [devel-swov_Adaptors] + - clean up Adaptor handling. Only a single adaptor is used. + - move StructViaSwov to via_video.h + +commit 408a572c47cdc7cf3ae425de35791274d681fdc0 +Author: Luc Verhaegen +Date: Tue Aug 30 19:11:38 2005 +0000 + + [devel-swov_Attributes] + - clean up Attribute handling. + - introduce some superficial sanity to viaCalculateVideoColor + +commit ad02479d946564c071ee24f81997e9f1afbc3d2c +Author: Luc Verhaegen +Date: Tue Aug 30 12:29:34 2005 +0000 + + [devel-swov_QueryImageAttributes] + - clean up QueryImageAttributes. + - enforce pitch alignment in QueryImageAttributes. + - the copy to FB now becomes only a single memcpy. + +commit 1112376091ba9da9158df9d491dc96b95ba25cf7 +Author: Luc Verhaegen +Date: Sun Aug 28 17:27:04 2005 +0000 + + [devel-Vsync_restore_fix] Wrong. This is linux kernel display blanking at + work. + +commit 572b5270c193afb25063452e177b2b06e3b16d2b +Author: Luc Verhaegen +Date: Sun Aug 28 16:03:55 2005 +0000 + + [devel-Vsync_restore_fix] Sometimes bit 9 of vsync isn't properly restored. + Mostly when ambient temperature is high. This has been happening on + both CLE266A3 and KM400. In VIARestore, after vgaHWRestore, we check + the contents of CR07 and, if not matching the contents of + hwp->SavedReg.Crtc[0x07], we write it out again. + +commit 55f20a9de437c170d09f4eede3786a0b2f080d02 +Author: Luc Verhaegen +Date: Fri Aug 19 23:12:50 2005 +0000 + + [devel-swov_sanitise_PutImage] + - Sanitise the order in which PutImage does things. Add 3 bools to track + what needs to be done. + - Fix AdjustFrame to fire V3 and to check for X/Y adjustment, instead of + just recalculating both. + - Fix rounding error in VisibleV when calculating Offset_Y, which caused a + green band when full screened here. + +commit 4e2ca019c16ed458a0b5b07873f6da6d4f8a0bbe +Author: Luc Verhaegen +Date: Fri Aug 19 14:21:25 2005 +0000 + + [devel-swov_move_to_via_video_2] + - move the remainder of functions to via_video.c + - drop the daft switch from PutImage. + - drop SwovUpdate into PutImage. + - call VisibleH/V and Startaddress from AdjustFrame. + +commit 900b23b2828c1f978608e8350242b3569b0b70f2 +Author: Luc Verhaegen +Date: Fri Aug 19 11:09:12 2005 +0000 + + [devel-swov_move_to_via_video_1] + - Start moving some stuff from via_swov.c to via_video.c. There's only 1 + thing left. Move ViaSwovPrintRegs, ViaSwovFillBlack, ViaHQVFetch, + ViaSwovStride, ViaSwovSurfaceCreate, ViaSwovSurfaceDestroy, + ViaSwovFIFO, ViaSwovInit. + - Call ViaHQVFetch and ViaSwovStride from ViaSwovSurfaceCreate, since both + only depend on the size of the buffers. + - Call ViaSwovFIFO from ViaSwovInit since we currently don't alter the FIFO + at all later on. + +commit 9f6a9510a964c0894c8bda47bfc53a9316d1361f +Author: Luc Verhaegen +Date: Fri Aug 19 09:26:12 2005 +0000 + + [devel-swov_shift_Swov_entries] + - move "stream" specific stuff out from Swov to Port: HQVSurfaces, + SwovSurfaces, Width, Height, Pitch, CrOffset, CbOffset, FrameNum. + - remove Swov->Flags to Port->Deinterlace and use saner defines. + +commit 150fae6fd9cd27519ec64b55b6e47e67fb009e06 +Author: Luc Verhaegen +Date: Fri Aug 19 07:15:09 2005 +0000 + + [devel-swov_VisibleHV] + - Split SwovRectangle in SwovVisibleH/V. + - Set Horizontal Window info, Video3 Fetch and SrcWidth at the end of + SwovVisibleH directly. + - Set Vertical Window info at the end of VisibleV. + - The VisibleH/V calculations each only need a single value each outside to + calculate the startaddress offset. Make this to Port->Offset_X/Y. + +commit 1bc33f01744ff4910a869f402d09d9df9aca93df +Author: Luc Verhaegen +Date: Thu Aug 18 09:19:05 2005 +0000 + + [devel-swov_ViaXvPort] + - move viaPortPriv to struct ViaXvPort + - since we use only a single port; add Swov->Port and use that one. + +commit 3e0927e869a99bdcf31fac691fa665f8d7aa9d86 +Author: Luc Verhaegen +Date: Wed Aug 17 12:05:23 2005 +0000 + + [devel-swov_SurfaceCreate_consolidate] + - Move all surface creation to a single function. + - Use SurfaceDestroy for cleanup when allocation fails. + - Use width and height to check if the surfaces need to be altered. + +commit 5a8f860d313212180284bb3418c15487bbca6fe0 +Author: Luc Verhaegen +Date: Wed Aug 17 02:53:15 2005 +0000 + + [devel-swov_split_zoom] + - split up zoom code in both a Horizontal and Vertical function. + - pass Src_W/H and Drw_W/H directly for calculating zoom factor. + - clean up minify filter setting. + +commit db7f1756e36f9f670788f96f185d399fb0f0444e +Author: Luc Verhaegen +Date: Tue Aug 16 03:24:32 2005 +0000 + + [devel-swov_Video3_takes_YUV422] Fix up loads of hacks/assumptions around + Video3. In the current situation Video3 takes in YUV422 from the HQV, + no matter what FourCC the HQV takes in. + - the buffer is 2*width*height + - the FIFO code suddenly becomes really simple. + - there's also a single offset calculation for Video3 StartAddress. Ditch + all the alignment stuff for a single one to 32bit. + +commit 5a54e2684b723f4845561b56c4b5bdab47178e9e +Author: Luc Verhaegen +Date: Sun Aug 14 07:48:35 2005 +0000 + + [devel-swov_FillBlack] libv is an idiot. + +commit b38edb25857ef9a03c00580313439c8291cf28c1 +Author: Luc Verhaegen +Date: Sun Aug 14 07:25:45 2005 +0000 + + [devel-swov_FillBlack] + - Fix ViaSwovFillBlack to do YV12 properly. + - Alter old YUVFillBlack code to fit in with the YV12 case. (this remains + untested until i get hold of YUY2 media). + +commit 8ce7febd40840495418667cc17f5ddce72080d29 +Author: Luc Verhaegen +Date: Sun Aug 14 03:24:54 2005 +0000 + + [devel-swov_remove_ExtendedFIFO] Due to massive cleanups, ExtendedFIFO has + become pretty much useless. Remove it. A useful bandwidth checking + system should be created some day. + +commit b81f1fd329155dc6ab457e7842bfdb7da1d6b422 +Author: Luc Verhaegen +Date: Sat Aug 13 02:50:34 2005 +0000 + + [devel-swov_pass_pPriv] + - rename coordinates in pPriv + - rename VIAVidUpdateOverlay to ViaVidUpdate + - rename ViaOverlayHide to ViaSwovHide + - pass pPriv down to ViaSwovUpdate and ViaSwovRectangle. + - remove pUpdate->ColorSpaceLowValue and Orig_* in favour of pPriv entries. + +commit 1dbfab84e5dac953dd00c066b87f42aeeb9c17d3 +Author: Luc Verhaegen +Date: Sat Aug 13 02:34:03 2005 +0000 + + [devel-swov_width_and_height] + - fix casting issue with previous commit. + +commit 947c648c737ac707c872c37fad1052b3e624b753 +Author: Luc Verhaegen +Date: Fri Aug 12 19:38:03 2005 +0000 + + [devel-swov_width_and_height] + - Move to Width and Height tracking instead of endpoint coordinates. Only + SwovWindow needs X + W and Y + H. + - Adjust SwovRectangle and pUpdate entry naming accordingly. + +commit 9bdcf182a25ccd8f7c952de129d0304cf1ee937c +Author: Luc Verhaegen +Date: Fri Aug 12 08:22:13 2005 +0000 + + [devel-swov_frame] + - introduce Swov->Frame_X/Y and track them properly. + - use Crtc StartAddr to figure out initial Frame coordinates. + - fix FrameAdjust/panning by cleaning up ViaSwovRectangle + +commit f107d3bc9b7e1682368c827f865ccd5d8f10b15b +Author: Luc Verhaegen +Date: Wed Aug 10 23:45:37 2005 +0000 + + [devel-swov_Active] + - remove Swov->Status for Bool Swov-Active. + - remove leftover ViaSetPrimaryFIFO call, no primary fifo settings are + altered any more. + - shift setting up UpdateOverlay until right before the VIAVidUpdateOverlay + call. + +commit c929168f338dc8f7402260a9a694a94a209b9c9a +Author: Luc Verhaegen +Date: Wed Aug 10 00:24:01 2005 +0000 + + [devel-swov_cleanup_allocation] + - Rename HQVMem and SwovBuffer to HQVSurface and SwovSurface. + - Clean up properly when allocation fails. + - Have PutImage depend on SwovSurface[0] existing instead of using flags. + - Move OverlayDecide to the front of PutImage + - Move memcpy to right before the Flip in PutImage. + +commit b326b64fa736a9bb9e40d43af159e0dd8a2c0fc1 +Author: Luc Verhaegen +Date: Tue Aug 9 17:50:33 2005 +0000 + + [devel-ViaSwov] + - use struct ViaSwov instead of swovrec. + - rename a few global functions. + +commit 317caeb33b21db60cf6c01523ef736eff94765ed +Author: Luc Verhaegen +Date: Tue Aug 9 03:37:21 2005 +0000 + + [devel-swov_remove_hwdiff] + - Fix HQVFetchByteUnit usage. CLE266A is the only device that needs >> 3. + - move all HWDiff entries to swovrec and use proper Bools. + +commit 0a5db41323eb9a4cc1f7273f9b2dab69568ef377 +Author: Luc Verhaegen +Date: Tue Aug 9 01:48:36 2005 +0000 + + [devel-ViaSwovPrintRegs] + - Add function that dumps all HQV and Video regs to the log. + - Disable HQV and Video1/3 at ViaVideoInit. This stops renegades scribbling + over the fb. + +commit 511f8accaa0c3f959d0cd825b4e500c34fbb8a02 +Author: Luc Verhaegen +Date: Fri Aug 5 04:13:40 2005 +0000 + + [devel-memclk_option_fix] Fix a missing argument to xf86DrvMsg. + +commit b516c068832555911f46eb62726f4306a77ffb8b +Author: Luc Verhaegen +Date: Fri Aug 5 03:52:58 2005 +0000 + + [devel-memory_management_cleanup] + - clean up the ugly macro mess that is via_memory.h. + - move to a more transparent mechanism. + - split up HQV and SWOV buffers into seperate entities. This makes handling + them more straightforward. + +commit 4631b05263110940209c858a7671d9dda6e90448 +Author: Luc Verhaegen +Date: Wed Jul 27 20:20:23 2005 +0000 + + [devel-swov_address_cleanup] + - pass HQV Dst Addresses directly. + - split up addresses to HQVAddr0, HQVAddr1, FBSize, FBPitch, CrOffset, + CbOffset. + - move via_priv.h VIAMem to via_memory.h + - move the remainder of via_priv.h to via_swov.h, removing SWDevice and + Overlay in the process. + - move via_xvpriv.h contents to via_video.h + +commit 4653f32b7aeba7d8862b3165f117e6cb5f972dcd +Author: Luc Verhaegen +Date: Wed Jul 27 12:45:27 2005 +0000 + + [devel-swov_zoom_cleanup] + - remove minifyH. + - move DisplayCount to end of SwovFetch. + - clean up setting HQVMinify, HQVFilter, Video3Zoom and Video3Minify. + +commit 2643f47c80d9e6b8036b9967aab4739e39db9e63 +Author: Luc Verhaegen +Date: Mon Jul 25 15:21:18 2005 +0000 + + [devel-remove_TVModeCrtc] + - Move setting of the output bus to ViaModePrimary entirely. + - Ditch tv encoder specific ModeCrtc routines, making tv encoders totally + independent of the unichrome. + +commit 65dc08afc67c90d850f29acf7bd36ef2e32effb3 +Author: Luc Verhaegen +Date: Mon Jul 25 13:45:47 2005 +0000 + + [devel-memclk_option] + - Add option ForceMemClock. Available values: 200,266,333,400 + - Use saner MemClk values. Remove SDR. + +commit f0ba088b475a30bc4bb34e75fe6505b905b7912e +Author: Luc Verhaegen +Date: Sun Jul 24 22:07:09 2005 +0000 + + [devel-mode_rework_typos] Fix 2 typos in the huge mode setting rework. + Wrong KM400a version for bandwidth checking. Fix primary sync polarity. + Already present in CVS and xorg. + +commit fc7c225d6e44f5658f52fba6ccd3c3fcca2317bc +Author: Luc Verhaegen +Date: Tue Jul 12 14:22:10 2005 +0000 + + [devel-swov_v3_only] + - move to V3 code entirely. + - ditch swov.VideoFlag. + +commit 7a116af78fbdae126f67ae7873e624f5706a3c4d +Author: Luc Verhaegen +Date: Sun Jul 10 21:03:56 2005 +0000 + + [devel-swov_hqv_only] End hqvless hypocrisy, scaling code is hqv only, flip + is too. Ditch all HQV less code. + +commit 66e7e28a89da544b9a4ef39e1f9321dca7232bc6 +Author: Luc Verhaegen +Date: Sun Jul 10 17:23:03 2005 +0000 + + [devel-swov_video3] + - move to video3 entirely: there is no manager for v1/v3 placing, we are + always using HQV, and video3 is capable of alpha blending (later). + - fix fifo settings for v3 + - fix colorkey for v3 + +commit 222cb3ccfe86b47e17cd81416fa15e1d068290a0 +Author: Luc Verhaegen +Date: Fri Jul 8 22:52:20 2005 +0000 + + [devel-swov_Zoom] + - integrate SetMiniAndZoom, viaOverlayHQVCalcZoomHeight/Width into + Upd_Video (or what remains of it). + - rename Upd_Video to ViaSwovZoom. + +commit 6e7eea9aca16f3a51e8eefcdb16b62d584259286 +Author: Luc Verhaegen +Date: Fri Jul 8 20:34:26 2005 +0000 + + [devel-swov_Compose] + - Dynamically set Video->Compose. + - Move ExpireNumber setting from SwovInitControl to SwovFIFO + +commit 6555d3e9fc5148e1ed6f2df78c56743999acee42 +Author: Luc Verhaegen +Date: Tue Jul 5 16:28:41 2005 +0000 + + [devel-swov_hqvCtl] + - Dynamically set hqvCtl. + - Integrate all viaWaitHQV/VBI code in Upd_Video. + +commit 70f4214b01241d207cc508e994ba5abf1457071e +Author: Luc Verhaegen +Date: Tue Jul 5 09:17:21 2005 +0000 + + [devel-swov_vidCtl] + - Dynamically set vidCtl. + - Lump together viaOverlayGetV1V3Format and ViaSetVidCtl + +commit 46a3cc6d9af9d3dae701bd3efc93c15052d57ff7 +Author: Luc Verhaegen +Date: Mon Jul 4 21:58:49 2005 +0000 + + devel-swov_remove_SWVideo_ON] + - swovRec->SWVideo_ON + - OVERLAYRECORD.UVoffset + +commit 3484cabbe88bf55c9ed7a01f11ca227d7ac922ec +Author: Luc Verhaegen +Date: Mon Jul 4 19:55:25 2005 +0000 + + [devel-swov_remove_VidRegBuffer] + - move all remaining stuff to direct MMIO calls. + - fix up wrong HQV assignments (SrcStride). + - remove all VidRegBuffer stuff. + +commit f9ec888943cf64e744a5e77585c5dd42fb226336 +Author: Luc Verhaegen +Date: Sun Jul 3 20:59:31 2005 +0000 + + [devel-swov_memory_mapped] + - Create swov.Video and swov.HQV, memory mapped to Video and HQV registers + - Get rid of VidMapBase, MpegMapBase and related macros VIDOut, MPGOut + - Remove old memory mapped thingiemabob for the new implementation. + +commit 7c583e9d3505b3dba9cb2f7bb6ce7aec6b851dc3 +Author: Luc Verhaegen +Date: Fri Jul 1 10:02:26 2005 +0000 + + [devel-swov_fetch_stride] Move out Fetch and Stride to ViaSwovFetch and + ViaSwovStride. + +commit cd617052cacac537525800d5828d1c9d4960c1e1 +Author: Luc Verhaegen +Date: Thu Jun 30 15:44:43 2005 +0000 + + [devel-swov_fifo] + - undo clueless abstractions. + - top level conditional is now Chipset, followed by V1/V2 + - move from Upd_Video to VIAVidUpdateOverlay + +commit 5e1ca6727571bafd48c7af9e93841880d8e154e8 +Author: Luc Verhaegen +Date: Thu Jun 30 11:53:30 2005 +0000 + + [devel-swov_flags] + - stop using intermediate values for passing flags to Upd_Video. use + pUpdate->Flags directly. + - move ViaSwovStartAddress to VIAVidUpdateOverlay. + +commit df275bb38d9669bb169223b55f376e65390dec47 +Author: Luc Verhaegen +Date: Thu Jun 30 09:23:21 2005 +0000 + + [devel-ViaSwovRectangle] Move out the calculation of the viewable rectangle + to ViaSwovRectangle. Move pUpdate entries to short to match PutImage + (these are most certainly not unsigned values). + +commit a7818b465f30ca9b6873c140d7e63dca54aa9668 +Author: Luc Verhaegen +Date: Mon Jun 27 14:29:43 2005 +0000 + + [devel-swov_startaddr] move out StartAddress specific code from Upd_Vid to + ViaSwovStartAddress. + +commit 403cc9a524aeb76f25a19f68dc5aaf5a2e0258bb +Author: Luc Verhaegen +Date: Mon Jun 27 14:11:44 2005 +0000 + + [devel-cleanup_pScreen_fix] Fix up overzealous pScreen cleanup. + pScrn->pScreen is only initialised after ScreenInit. + +commit 5ae65581fe851614cb69bd1499264c2ae0ed3465 +Author: Luc Verhaegen +Date: Sun Jun 26 14:18:55 2005 +0000 + + [devel-cleanup_priv] + - Clean up usage of SWDEVICE entries. + - Clean up usage of OVERLAYRECORD entries. + - remove FOURCC_HQVSW + - remove overlayRecordV3 + +commit c259c981711913f9740336a95d34f552d55639f4 +Author: Luc Verhaegen +Date: Fri Jun 24 20:30:03 2005 +0000 + + [devel-stop_passing_videoFlag] Use pVia->swov.VideoFlag directly. + +commit 8fdf469c0d36685a6df9f23ebbd636f542c26757 +Author: Luc Verhaegen +Date: Fri Jun 24 19:03:09 2005 +0000 + + [devel-ViaDebug] Create VIAFUNC and ViaDebug, replace DEBUG macro. + +commit f9cbc7770dda1d4e35351861b6f83176e3835aae +Author: Luc Verhaegen +Date: Thu Jun 23 13:19:51 2005 +0000 + + Variable passing sanitation. Added scrnIndex to pVia (for future Msg + cleanup and reduced pScrn passing). Remove daft pScreen passing for + either pScrn or pVia. + +commit 03be94ad49532ca57c13eeb330d240c166ade631 +Author: Luc Verhaegen +Date: Thu Jun 9 19:58:28 2005 +0000 + + [devel-remove_xvmc] Remove XvMC support. The x-wide XvMC thing should be + properly worked into Xv anyway. + +commit 562bed8f37df6dcc7cbacf3af45415d048a29894 +Author: Luc Verhaegen +Date: Thu Jun 9 18:19:34 2005 +0000 + + [devel-remove_memcpy] Remove a big blob of nonportable nonsense. + +commit 237add9247871824134a4181f80f8258c157c77d +Author: Luc Verhaegen +Date: Thu Jun 9 18:19:34 2005 +0000 + + [devel-remove_memcpy] Remove the caching game. + +commit 1560989fa84790af0bcc821ff4dd78fe42036677 +Author: Luc Verhaegen +Date: Thu Jun 9 17:27:04 2005 +0000 + + [devel-remove_vbe] Toss out VBE crap. + +commit cd9659bb0beb32cfff199d798c7859c964460ce8 +Author: Luc Verhaegen +Date: Thu Jun 9 10:48:17 2005 +0000 + + [devel-memory_pScrn] Clean up some via_memory functions. Pass pScrn around + properly. + +commit 6b7014e0fea684c096eb226e3362be34172fe067 +Author: Terry Barnaby +Date: Sun Apr 24 18:56:27 2005 +0000 + + Added 720x576Noscale mode for VT1622a TV chip. + +commit 325e1cc2d78c8ba4dea9700dea690b5eb68297f8 +Author: Luc Verhaegen +Date: Wed Apr 13 13:33:16 2005 +0000 + + [devel-vt1622_noscale_ntsc] + - Bring new mode in line with other via_vt162x.h entries. + - small VT162x manpage changes. + +commit bb2c940e6a52ff82ab253cfb1b147d7ca84ce1ee +Author: Terry Barnaby +Date: Wed Apr 13 05:57:19 2005 +0000 + + Added the "720x480Noscale" mode for NTSC TV's. + +commit 79101bf8a991f553e24886a6e3c19ef6b6f16126 +Author: Luc Verhaegen +Date: Sun Apr 10 11:48:42 2005 +0000 + + [devel-fix_disclaimers] Move to OSI standard MIT disclaimer: + - all VIA copyrighted files: with permission of Joseph Chan of VIA. + - via_memory.c: with permission of Alan Cox. + - via_memcpy.c/.h: with permission of Thomas Hellstrom. + - via_vgahw.c/.h: with permission of myself :p + +commit aff66bc010f3cad1a21eb1a4a54d1fb2d44a616a +Author: Luc Verhaegen +Date: Sun Mar 20 18:38:00 2005 +0000 + + [devel-symbol_weak_xf86ExecX86int10] More suitable fix for the + xf86ExecX86int10 problem: load int10 right before vbe. Add this symbol + to the vbeSymbols list. + +commit ed30bdb1f59d5cc60f16fe1089fa6b3d2688c61f +Author: Luc Verhaegen +Date: Sat Mar 19 01:08:51 2005 +0000 + + [devel-symbol_weak_xf86ExecX86int10] Call xf86LoaderRef on xf86ExecX86int10 + to reassure driver that this doesn't need to be resolved unless we say + so. + +commit 40eb487202cf9964ec829e5aa964819253af349b +Author: Luc Verhaegen +Date: Fri Mar 18 03:32:05 2005 +0000 + + [devel-vt162x_subcarrier] + - Move 4byte subcarrier value out of the table. Adjust DotCrawlSubcarrier + to full 4byte so that writing this comes in line with standard + dotcrawl. + - stop writing 0x1B (version) and 0x1D (software reset) + +commit 3c5b6386a980b4d8959c2e34d376331f05adb5c8 +Author: Luc Verhaegen +Date: Thu Mar 17 19:37:50 2005 +0000 + + [devel-vt162x_RGB_YCbCr_table] + - The positions in the RGB and YCbCr tables in VT1622 and VT1623 are always + the same for the same TV register. So simplify this. + - move hex values in 720x576Noscale to caps, move 0x00 to 0 for + readability. + +commit 5a3d50984a3c42436112830032b5f212de7778e1 +Author: Luc Verhaegen +Date: Thu Mar 17 16:44:37 2005 +0000 + + [devel-Xv_SetVideoWindow] pass pUpdate directly to SetVideoWindow. + +commit 8770371783777a6bd2db6607e806d42c3fb28e14 +Author: Luc Verhaegen +Date: Thu Mar 17 16:16:32 2005 +0000 + + [devel-Xv_remove_RECTL] + - Move pUpdate->rSrc and pUpdate->rDest to SrcLeft, SrcTop, SrcRight, + SrcBottom, DstLeft, DstTop, DstRight, DstLeft. Pass pUpdate directly to + viaOverlayGetSrcStartAddress, Upd_Video. + - remove RECTL struct + - remove useless pUpdate->dwColorSpaceHighValue and dwFourcc + +commit 2794a9d9386121df065a2b1468559f41a14a6e5b +Author: Luc Verhaegen +Date: Thu Mar 17 16:00:22 2005 +0000 + + [devel-fixup_DBG_DD] + - remove 2 superfluous definitions of XV_DEBUG and DBG_DD + - move all long arguments of DBG_DD to long + +commit 472ea7be06451e73abac7f12b3d6acc8c59f81aa +Author: Luc Verhaegen +Date: Thu Mar 17 15:02:40 2005 +0000 + + [devel-add_via_drm_h_fix] + - define drm_hw_lock_t to drmLock on < xfree86-4.4.99.7 and on xorg-6.7.0. + - Realphabetize SDK list. + +commit 651daab381c0496922a7311c1331b8cec8624503 +Author: Luc Verhaegen +Date: Thu Mar 17 02:44:16 2005 +0000 + + [devel-mode_vga_restrictions_fix] Alter dotclock formatting to handle long + instead of short. + +commit e3b6b7daa5b81e81b97550f2025f785542fce492 +Author: Thomas Hellstrom +Date: Mon Mar 14 23:37:22 2005 +0000 + + Remove via_common.h in favour of via_drm.h and via_drmclient.h. Now + via_drm.h is identical to the kernel include file, and they should be + kept in sync. Adjust other files to new type names. Note that this + severely brakes libviaXvMC and when this commit is transferred to Xorg, + libviaXvMC build must be disabled until an update is made of that + library. + +commit f03b172a193417f947bd7ad5b4ec2daedfac2033 +Author: Luc Verhaegen +Date: Mon Mar 14 20:10:59 2005 +0000 + + [devel-xv_SetDisplayCount] Concat viaOverlayGetDisplayCount and + SetDisplayCount. + +commit 645cbdbf079c6dc7df396a8f3a03c487602dcf40 +Author: Luc Verhaegen +Date: Mon Mar 14 19:46:22 2005 +0000 + + [devel-support_notice] Have VIAProbe X_NOTICE that VIA Technologies doesn't + care and that help can be found here. + +commit 6b4995466def1a94e68bcdc19d3de543a9802bdc +Author: Luc Verhaegen +Date: Mon Mar 14 17:13:22 2005 +0000 + + [devel-tv_clock_nonprimary_vga_hack] Set up clock before setting up the tv + encoder. In nonprimary vga case, the clock is completely unset + beforehand resulting in a displaced/unstable image on the tv. A + modeswitch fixes this, a vt restore reintroduces this. This should be + kept in mind when removing TVCrtcMode. + +commit 3274dc6293f001817c629c3ff1aa9a636edd15da +Author: Luc Verhaegen +Date: Sun Mar 13 19:53:47 2005 +0000 + + [devel-vbe_request_symbols] + - add missing symbols to vbeSymbols. + - request xf86ExecX86int10 from int10 module too. + +commit 6e659e1a9aaec9dc9cff54b476b1b4951a24bf09 +Author: Thomas Hellstrom +Date: Sun Mar 13 18:23:41 2005 +0000 + + Hopefully fix debian build failure by including the "drm.h" file + explicitly. + Remove drm memory type references in via_priv.h unless XF86DRI is defined. + +commit fbff491f7034d4a7c32dc3c074c4f254269debfd +Author: Thomas Hellstrom +Date: Sun Mar 13 10:15:58 2005 +0000 + + Moved dri client common dri defines to via_common.h from via_dri.h + Implemented video lock / unlock macros in via_common.h so that both the X + server and the XvMC dri clients share the same implementation. The idea + is to protect the HQV with one of these locks so that a client wanting + to write to the HQV can just take one of these locks instead of the + heavyweight hardware lock which stops _all_ X server activities. This + means the X server also has to take the HQV lock when writing to HQV. + In the long run, similar locks might even enable us to make a clean + interface to a unichrome_v4l_drv for video capture. + Adjusted the unichrome XvMC internal protocol to carry the ChipID to be + able to differentiate different hardware in libviaXvMC. + +commit d0e952eb64aadf6936dfe53ccf1caeda9e8ae9eb +Author: Thomas Hellstrom +Date: Sat Mar 12 16:13:24 2005 +0000 + + Removed unnused pointer in XvMC common struct. Bumped internal XvMC + protocol minor. + +commit e6f77ff491ec646988175a6830c5aff1d86df141 +Author: Thomas Hellstrom +Date: Sat Mar 12 14:47:32 2005 +0000 + + FOURCC_VIA becomes FOURCC_XVMC. It was bogus coded 'WLKN' anyway. + Fix annoying first XvMC flip of uninitialized image by setting the HQV + source address in libviaXvMC before the X server is called. The X + server now never sets the HQV source address for XvMC images. + This code needs updated libviaXvMC. + +commit 31ebea3c8ce54ce1c931a5433944ade578744b6a +Author: Luc Verhaegen +Date: Wed Mar 9 12:40:48 2005 +0000 + + [devel-mode_vga_restrictions] + - restrict the timing on the basis of the available registers. + - implement crtc2 VGA (untested, needs higher level work first) + +commit 24cd110f9d00cdeac6dcfda98a148f6b4f90c3ef +Author: Luc Verhaegen +Date: Tue Mar 8 12:36:41 2005 +0000 + + [devel-InstallVideoObjectModule] Bring in Roland Mainz his + InstallObjectModule change. + +commit 857ab8e9a1c818d67944d9eaba0bbcfd736aa420 +Author: Luc Verhaegen +Date: Sun Mar 6 12:28:04 2005 +0000 + + [devel-pro_dotclock_skeleton] Adjust all functions and the ViaDotClocks + struct to take/handle the 24bit dotclock value needed for Unichrome Pro + devices. This does not fix the unichrome pro dotclock yet, values need + to be tested and inserted in the table first. + +commit 8a5ae0d2cc14c12b469d7bbda4310e59064adb47 +Author: Luc Verhaegen +Date: Thu Mar 3 18:33:53 2005 +0000 + + [devel-ViaTVPrintRegs] + - add another callback, TVPrintRegs. This moves out the decision to print + registers to ViaTVInit and ScreenInit and allows full register printing + even when using VBE. + - move up initial saving of TV registers to ViaTVInit + - properly initialise Crt/PanelPresent in ViaOutputsDetect. + +commit 247b9fd61ebb36b0d041447da78dc7e3cdfffe75 +Author: Luc Verhaegen +Date: Thu Mar 3 17:16:12 2005 +0000 + + [devel-remove_DumpVGARom] + - remove ViaDumpVGARom (will be replaced by bit of python in + unichrome/utils/) + - move via_bios.c to via_vbe.c, remove Via/S3 copyright (no code left), fix + up MIT license disclaimer. + +commit c325be5a544398e60ba8ceb29678c423b1a458d0 +Author: Luc Verhaegen +Date: Thu Mar 3 16:07:44 2005 +0000 + + [devel-vbeModes_cleanup] + - VBEModes option should be limited to HAVE_DEBUG (todo: reg dumping) + - only enable vbeModes on unichrome pro panel. + - remove old (and unneccessary; DDC and info) vbe uses, only use vbe now + for vbeModes. + - depend entirely on pVia->pVbe existing instead of pVia->vbeModes. (and + fix the nonPrimaryPci case) + - remove unneccessary substructing from ViaVbeModeInfo + - ModesAttach and ViaPanelModes are for standard modes only. + +commit e19545b4384e1da877a6007f6efeabacccedd06b +Author: Thomas Hellstrom +Date: Wed Mar 2 08:41:30 2005 +0000 + + Work around Linux 2.6 mtrr removal problem by doing it twice. Ugly, but it + works. + +commit 1e3cba2d353bcecb4577cb7468d9863ad6f42a47 +Author: Luc Verhaegen +Date: Wed Mar 2 01:30:58 2005 +0000 + + [devel-accel_undefined_pDRIInfo] + - stop referencing pVia->pDRIInfo->devPrivate when + !pVia->directRenderingEnabled in dispatchCBufferAGP + - properly NULL pVia->pDRIInfo in VIADRIScreenInit + +commit 04823a6ccccf3bd30d5ef5ef57c1022c2c893341 +Author: Luc Verhaegen +Date: Tue Mar 1 13:34:09 2005 +0000 + + [devel-dotclock_25200_fix] + 0x5958 was found to be unstable. Use 0x513C instead. + +commit aeb6b11d17ec2610b78a8b1e7fe7f7f4f9076d3d +Author: Thomas Hellstrom +Date: Mon Feb 28 21:39:17 2005 +0000 + + Apparently fix the ol' 3D-hang-or-misrendering-on-vt-swich bug. + +commit 06c6ecd79213f2a9b1368cae3143599482312c33 +Author: Luc Verhaegen +Date: Mon Feb 28 10:41:38 2005 +0000 + + [devel-devel-64bitcleanup-1_fix] Fix tiny oversight (gviaPortPriv instead + of viaPortPriv) inside !X_USE_REGION_NULL macro. + +commit e266cbc0326ec9ee8c98a0c4989cbd5ac69f988d +Author: Luc Verhaegen +Date: Mon Feb 28 10:35:29 2005 +0000 + + [devel-ids_20050228] K8M800: Abit KV-80 (Jyrki Muukkonen) K8M800: MSI K8M + Neo-V (David Harder) PM800: Fujitsu/Siemens Amilo Pro V2010 (Alois + Nespor) + +commit d9359fd8edf5b236e4193bd0fb1580355d577c25 +Author: Thomas Hellstrom +Date: Sun Feb 27 23:06:35 2005 +0000 + + Fix strange bug in last vbe commit. Moving tag devel-vbe-3 + +commit 8d77be6439bd7a6911919a6f74f73bbaa5e16fd8 +Author: Thomas Hellstrom +Date: Sun Feb 27 12:11:01 2005 +0000 + + More VBE cleanups: + More understandable printout whe VBE customized modesetting fails. + Fix 3D hang after logout / X server kill in the same what as is done in + VIA's binary drivers. + Clear whole frame-buffer on X start also with VBE modes. + Cleanup Save / Restore. + +commit 17da9937ae675133b86339f1821427e0fe069bab +Author: Thomas Hellstrom +Date: Fri Feb 25 16:22:23 2005 +0000 + + Fix up VBE DPMS, screen blanking and VBE panel power switch. + +commit 2ce5e888260cd79b70e4f1b5c195a6b714f81324 +Author: Thomas Hellstrom +Date: Thu Feb 24 22:23:33 2005 +0000 + + Obvious 64-bit incompatibility in via_video.h + +commit c5b9101dcf58ce744b792b921bc02b3d0a4ae969 +Author: Thomas Hellstrom +Date: Thu Feb 24 20:55:14 2005 +0000 + + Fix up XvMC after previous commits. One cannot touch Xv adaptor info + records after Xv has been initialized. + Removed the "XV_DUMMY" adaptor. It had no use. + Distinguish between Xv ports and adaptors. Via had mixed the two concepts. + Some defines renamed. Fix up XvMC to initialize every port on + XvMC-capable adaptors. + Make Xv init code take other, previously initialized, Xv adaptors into + account. + Fix up memory leaks in Xv and XvMC initialization. + +commit 0f1da0c284c850f5cb0973cb3f2b4fa7b69579ac +Author: Luc Verhaegen +Date: Thu Feb 24 12:05:48 2005 +0000 + + [devel-xvmc_fix_init_order] ViaInitXvMC should not be called before + xf86XVScreenInit. Pass viaAdaptPtr and XV_PORT_NUM to ViaInitXvMC and + init viaAdaptPtrs for XvMC there. + +commit 2995197b729eea152f469cd46b920f6ffd6753f4 +Author: Luc Verhaegen +Date: Wed Feb 23 23:06:48 2005 +0000 + + [devel-symbols_clear_remaining] data: ViaCardId, libc_cpuflags, + kernel_cpuflags, sse_cpuflags, mmx_cpuflags, now_cpuflags, + mmx2_cpuflags, ViaBandwidthTable, ViaDotClock, ViaPanelPrivate, + ViaResolutionTable, ViaVesaModes text: VIASwitchMode, VIAAdjustFrame + +commit e166ff6639460d9387cebc3c816483607e09dcc7 +Author: Luc Verhaegen +Date: Wed Feb 23 22:41:42 2005 +0000 + + [devel-XvMC_symbols] + - wrap around PutImage/SetPortAttribute/GetPortAttribute by saving the old + XvAdapt fields inside pPriv->xvmc_priv and only replacing these in + viaXvMCInitXv instead of earlier. This no longer requires the wrapped + around functions and the wrappers to be global. + - Move ViaInitXVMC and ViaCleanupXVMC from via_driver.c to viaInitVideo and + viaExitVideo. This removes the need to check XvMCEnabled each time a + wrapper is called. + - remove the Gs appended to via_video functions assuming they stand for + Global. + +commit fdf1c721f83e5f032ae6d758f8df2da9d9ac99d7 +Author: Thomas Hellstrom +Date: Wed Feb 23 14:42:38 2005 +0000 + + First version of VBE modesetting code. Some things still needs to be fixed + / checked: + Should we use a new file for the VBE functions? + Bandwidth check in via_video.c is very rough. + Save / Restore is broken w respect to 3D, but it behaves differently than + ViaSave / ViaRestore. Seems like both versions can be used, but i'm not + sure yet which one is the best. + Enabled HAVE_KM800 and HAVE_PM800. + Maybe this should've been put on hold 'til we have real modes for those. + Added a 1200x800@70 modeline to the built-in panel modelines. + +commit a9d0240d17ca99f7c1e7565039a657f0d27ed93f +Author: Terry Barnaby +Date: Wed Feb 23 13:47:40 2005 +0000 + + Added new TV mode "720x576Noscale" for improved quality TV display on PAL + TV's. + +commit d6de7d1d632bc8c51338df86ef058d040960f236 +Author: Luc Verhaegen +Date: Wed Feb 23 12:31:59 2005 +0000 + + [devel-vt1625_detection] Add detection of vt1625 (id reg holds 0x50). Not + supported yet though - no hardware confirmed with this tv encoder. + +commit 81804c74f9ce3e6285af5c76a02e29b3ea994c0b +Author: Luc Verhaegen +Date: Wed Feb 23 10:45:18 2005 +0000 + + [devel-remove_via_overlay] All of via_overlay.c is completely via_swov + specific, there is no reason to keep this outside of via_swov. Move + code over, make it all static, remove via_overlay.c and via_overlay.h. + via_swov now measuring more than + 60kB will shrink severely with cleanups. Saves 8 global symbols. + +commit 8e755657382ab860a52d900cda45e6b799bc9bc0 +Author: Luc Verhaegen +Date: Wed Feb 23 10:26:49 2005 +0000 + + [devel-remove_regrec] All of via_regrec.c is completely via_swov specific, + there is no reason to keep this outside of via_swov. Move code over, + make it all static, remove via_regrec.c and via_regrec.h. Saves 9 + global symbols, of which 3 are not prepended by VIA/Via/via. + +commit 67901148c6e1d7ec154a4e353a1590df99b738af +Author: Luc Verhaegen +Date: Tue Feb 22 23:04:38 2005 +0000 + + [devel-ViaShadowFBInit] Introduce ViaShadowFBInit, this moves shadow + specific code from ScreenInit to via_shadow.c, reducing 6 global + symbols to only one. + +commit c843db9dac8089193f95cbaa56f1fa06f184fbea +Author: Luc Verhaegen +Date: Tue Feb 22 17:23:32 2005 +0000 + + [devel-vt162x_fix_PrintRegs_call] + - PrintRegs depended on pBIOSInfo->TVI2CDev, it was called at Detect, + before TVI2CDev was assigned. PrintRegs now takes I2CDevPtr directly. + - remove "Via" from all local vt162x symbols. Make VT162xModePrivates + static. + +commit 1d97fb2eeda48b8582a101e84c6b47cbb76a5455 +Author: Luc Verhaegen +Date: Tue Feb 22 12:29:54 2005 +0000 + + [devel-vt162x_fix_poweron] Bad transfer of power on for vt162xx on + devel-fix_i2cbus3, when moving from I2CReadWrite to I2CWriteByte. + (Reported by Hans-Jürgen Ficker, Rene Mayrhofer, Robert Winder) + +commit 062d5dba53befb4358f871a7623a755461d81e51 +Author: Luc Verhaegen +Date: Sun Feb 20 15:29:45 2005 +0000 + + [devel-Xv_ViaOverlayHide] + - split code depending on DDOVER_HIDE to ViaOverlayHide. Remove + viaStopSWOVerlay, use ViaOverlayHide directly. Remove VIDEO_SHOW/HIDE + from VideoFlags + - use Bool as Upd_Video and VIAVidUpdateOverlay return values. Remove all + PI_ defines. + +commit 9e20fd06b2903c0c110b653eea3868022d0dcb7c +Author: Luc Verhaegen +Date: Sun Feb 20 14:30:19 2005 +0000 + + [devel-Xv_remove_Save_3C4] Leftover from previous commit. + +commit 7d1c657929c908117b496f9e17adb45613cab578 +Author: Luc Verhaegen +Date: Sun Feb 20 14:27:05 2005 +0000 + + [devel-Xv_remove_Save_3C4] Use ViaSetPrimaryFIFO directly for FIFO + restoration. Panel code doesn't touch FIFO. Remove Save_3C4_16/17/18 + from pVia->swov. + +commit db196133e8873c78f7d9bae16ec9e86892e156ce +Author: Thomas Hellstrom +Date: Sat Feb 19 12:18:50 2005 +0000 + + Silence some compilation warnings. + +commit 38d1b278bf07d9b6b14d271a5345107f3c69c7d1 +Author: Thomas Hellstrom +Date: Sat Feb 19 12:15:05 2005 +0000 + + Assume that we have the same bandwidth available on the Unichrome PRO as on + the KM400A. This will allow us to do more interesting modes on those. + +commit 1cc80ea875344e3bffb95877365bfdaad8dc49ca +Author: Thomas Hellstrom +Date: Sat Feb 19 12:09:27 2005 +0000 + + Remove the ÁllowInsecureDRI option and update the manpage. + +commit 7444024462675b8f04b6eae00fe0296619c3faf2 +Author: Thomas Hellstrom +Date: Sat Feb 19 11:56:02 2005 +0000 + + Remove putvideo advertizing. We cannot do it anyway. + +commit 24552ab3fa35c5bca4b626987c43d016b3dd73e7 +Author: Thomas Hellstrom +Date: Thu Feb 17 20:21:18 2005 +0000 + + revert The unichrome pro dotclock hack commit. + +commit 0170938d16013603637d3449dc613f3830836e60 +Author: Thomas Hellstrom +Date: Thu Feb 17 20:15:22 2005 +0000 + + Xv and XvMC (mpeg2 vld) support for K8M800 and friends. + +commit 81fcd209f2e08163f73b1039df941460014f2e65 +Author: Thomas Hellstrom +Date: Wed Feb 16 21:45:14 2005 +0000 + + Bring dxperimental Unichrome Pro mode hack to HEAD since we now seem to + have enough hardware to support it. (Luc, Thomas) + +commit cdfbb69127b662340dc1e30e4f74283561c01d10 +Author: Luc Verhaegen +Date: Tue Feb 15 05:11:13 2005 +0000 + + [devel-Xv_remove_SurfaceDesc] + - move CreateSWOVSurface from via_video.c to via_swov.c, rename to + ViaSwovSurfaceCreate, merge with VIAVidCreateSurface + - move DestroySWOVSurface from via_video.c to via_swov.c, rename to + ViaSwovSurfaceDestroy, merge with VIAVidDestroySurface + - replace pPriv->SurfaceDesc with CARD32 FourCC, emove DDSURFACEDESC struct + +commit 5bc7b4d99a0d15d32dbea9edaf214280edca96b4 +Author: Luc Verhaegen +Date: Mon Feb 14 21:30:22 2005 +0000 + + [devel-panelindex_add_linefeed] forgot \n in message in ViaPanelGetIndex + message + +commit 6c10a7ebddb2869f0c71658113abd7fdb263dd40 +Author: Luc Verhaegen +Date: Mon Feb 14 14:39:34 2005 +0000 + + [devel-ids_20050214] K8M800: Averatec 54xx (Cobe Higginbottom) + +commit 61b081d8be4a45df56a74af2f645c4264a38c06e +Author: Luc Verhaegen +Date: Mon Feb 14 13:22:17 2005 +0000 + + [devel-vt162x_table_i2ccleanup] clean up what remains of the vt162x tables. + - split up main table, allowing for more transparent writing. + - remove mask. + - align all bytes for readability. + - alter usage of RGB and YCbCr fields + +commit 6e9b6610f14a3e17e7853ef429e1a856c84b7211 +Author: Luc Verhaegen +Date: Mon Feb 14 12:34:04 2005 +0000 + + [devel-add_common_dotclocks] + - add dotclocks for common and acceptable modes + - move ViaModes to ViaPanelModes, remove those with VRefresh != 60Hz attach + only when panel is active. + - remove panel checking for 60Hz, depend entirely on mode->Private + - remove ViaModeInit, all checking is done as either part of ValidMode or + ViaPanelGetIndex (gets called from ValidMode and ModePrimary). + +commit 18b81123bca632e49c5a82ff26466984c789d66d +Author: Luc Verhaegen +Date: Sun Feb 13 21:39:35 2005 +0000 + + [devel-vt162x_table_DisplayModeRec] + 2 more things with this tag: + - only add TV modes when tv encoder is actually active + - re-add SetUseExternalClock, this is needed to reset the Primary clock + when not depending on the tv encoders clock. + +commit 616189a002a507d544396a4411d3821571b8ee12 +Author: Luc Verhaegen +Date: Fri Feb 11 19:46:30 2005 +0000 + + [devel-vt162x_table_DisplayModeRec] Forgot to save a tiny mode validation + change. + +commit 278dba089999369640098c78a1d64ce01a5181ae +Author: Luc Verhaegen +Date: Fri Feb 11 19:44:33 2005 +0000 + + [devel-vt162x_table_DisplayModeRec] + - remove CRT1/CRT2 and the crt2 "patch" fields from tables. + - add proper modelines, attach VT162x specific Privates. + - match up modelines with table entries on Private, standard, width, height + and name. + - update manpage to explain new mode based behaviour. + +commit a767054759d4e60232ebb47796ffcad559374b6b +Author: Luc Verhaegen +Date: Thu Feb 10 18:14:01 2005 +0000 + + [devel-old_VGA_blanking_limit] Fix X working around old VGA blanking limit. + This limitation doesn't apply to modern cards. + +commit 7e109215645fad9836d2736167d934dae660e3b0 +Author: Luc Verhaegen +Date: Thu Feb 10 17:23:15 2005 +0000 + + [devel-stop_using_PrivFlags] mode->PrivFlags gets altered by the server as + it is currently used to tell the Tseng driver which clockrange is being + used. Use mode->Private instead. + +commit 63515dfffd2eaf84699a4ec96b9361f6a4c76dd1 +Author: Luc Verhaegen +Date: Wed Feb 9 17:29:26 2005 +0000 + + [devel-panel_remove_nonnative_modes] This is what happens when you rush + things: VIA_PANEL12X7 exists too, even if it is unhandled. + +commit 48d2482e39e9106cda32f24e37704605f59d3ac5 +Author: Luc Verhaegen +Date: Wed Feb 9 15:23:49 2005 +0000 + + [devel-panel_remove_nonnative_modes] + - print out which panel resolution has been detected. + - have the panel code invalidate nonnative modes. This has been broken for + ages and will probably not be fixed soon. + +commit 0c5dfde2d60076ad791f083989ac707110c5853c +Author: Luc Verhaegen +Date: Wed Feb 9 14:46:49 2005 +0000 + + [devel-ViaModes_becomes_DisplayModeRec] + - move ViaModes to full and proper DisplayModeRec. + - attach modes to pScrn->monitor before mode validation. + - have ModePrimaryVGA use the DisplayModeRec entirely. + - remove refresh checking, check for dotclock in ValidMode. + +commit 22943a6994ec93ad6affdfdfdeea072e45ade7df +Author: Luc Verhaegen +Date: Tue Feb 8 08:16:26 2005 +0000 + + [devel-Xv_remove_pixelformat] DDPIXELFORMAT only had 3 entries still in + use; dwFlags, dwFourCC, dwGBitMask dwflags was always either + DDPF_FOURCC or 0, so DDPF_RGB support was removed, making dwGBitMask + redundant. Moving everything to FOURCC, and changing + swov.DPFsrc.dwFourCC to swov.SrcFourCC removes DDPIXELFORMAT + completely. + +commit 14d89836fb6b1618a164741fd513b13abfcd6d18 +Author: Luc Verhaegen +Date: Mon Feb 7 19:10:17 2005 +0000 + + [devel-ViaBandwidthTable] + - Replace ViaRefreshAllowedTable with a table for Chip/ChipRev and MemClk, + holding the maximum allowed value (gathered from the old + ViaRefreshTable). At ValidMode, the actual bandwidth needed is compared + against this maximum value. + - move ValidMode to via_mode.c + - remove duplicate MemClk values in via_video.h + +commit f62897e8e66c3d228ee94b7043419a0a36bd7d4e +Author: Luc Verhaegen +Date: Thu Feb 3 11:39:34 2005 +0000 + + [devel-Xv_VidAdjustFrame_cleanup] + - Move Xv related content of AdjustFrame into VidAdjustFrame. + - Track down VidLockSurface which leads to the removal of VidLockSurface + and pPriv->ddLock + +commit d8fd2fc7f86c096f4e5419153019c80d26cbb16f +Author: Luc Verhaegen +Date: Thu Feb 3 10:46:38 2005 +0000 + + [devel-vt162x_table_dotcrawl] Handle dotcrawl in a more elegant manner. + +commit 677f5126f3812f87895712c0ce335d436be68402 +Author: Luc Verhaegen +Date: Thu Feb 3 01:02:02 2005 +0000 + + [devel-set_CR33_low_bit_zero] Keep the low bit of CR33 0. Setting it 1 as + part of line compare is problematic on VT1622/NTSC/Overscan/720x480. + +commit 587581acff2e492a1eb413ed1ea3f70016c2bc46 +Author: Luc Verhaegen +Date: Thu Feb 3 00:31:52 2005 +0000 + + [devel-VgahwPrint_sequencer_range] The sequencer registers only go from + 0x00-0x7F. High bit is ignored: + 0x80 equals 0x00, etc. + +commit 5a5b7408983d354e4c515669e633611537800838 +Author: Luc Verhaegen +Date: Sun Jan 30 22:43:43 2005 +0000 + + [devel-vt162x_table_crtc_regs] + - Remove bitdepth/fetchcount/offset from CRTC2 tables. + - drop 8/16/24, keep only a single array for registers 0x50 - 0x5F. + - reduce CRTC1 array according to mask. + +commit 9e42d38316206d804093b49de14787e578552642 +Author: Luc Verhaegen +Date: Sun Jan 30 00:41:57 2005 +0000 + + [devel-vt162x_table_remove_output] Remove output control register (CR6A, + CR6B, CR6C) specifics from vt162x tables. + +commit 00a37df879932f0a3bbe48d2bc696e3b2658bb90 +Author: Luc Verhaegen +Date: Wed Jan 26 01:45:26 2005 +0000 + + [devel-reinstate_directRenderingEnabled] directRenderingEnabled is the + established standard, even though DRIEnabled is much easier on the + fingers :) + +commit d7c9bfc686b84f8b0c3d5983e9144981131f60d4 +Author: Luc Verhaegen +Date: Tue Jan 25 19:50:17 2005 +0000 + + [devel-improve_power_debug] + - Fix LCDPower to no longer log "ViaPower" in debug. + - Make LCDPower and TVPower debug messages reflect the value of On. + +commit fe79ca6a53ddbfe06be0270a622db70ce51db350 +Author: Thomas Hellstrom +Date: Mon Jan 24 22:12:18 2005 +0000 + + Remove stray debug message from previous commit. + +commit 47c6d671bdf627bf37d2f7bcac81719e2e6b680d +Author: Thomas Hellstrom +Date: Mon Jan 24 22:04:08 2005 +0000 + + Import Alan H's DRI visual fixes from Xorg. (Alan H) Fix minor bugs in the + above (Thomas H). These bugfixes will be backported to Xorg before next + unichrome merge. + +commit f7cb3dcaf5d0bdae2d8df5011fafef893b815837 +Author: Thomas Hellstrom +Date: Sun Jan 23 09:45:16 2005 +0000 + + Xv: Remove a debug line causing compilation error when XV_DEBUG is turned + on. + +commit f33e3f7fb325bc979f5e622e0f028e927e385386 +Author: Thomas Hellstrom +Date: Sun Jan 23 09:43:16 2005 +0000 + + Disable ScreenToScreenColorExpand Acceleration. Source address setup is + buggy. + +commit 7fa0b7ff6faf54addf6584642226f490ea50ca74 +Author: Luc Verhaegen +Date: Sat Jan 22 20:11:09 2005 +0000 + + [devel-vt162x_table_consolidate] + - Reduce all tables to one table per type. + - Seperate NTSC and PAL to their own entries. + - Add fields describing width, height, tvstandard and normal/overscan. + - Remove unused HDTV and SDTV fields. + +commit d0c3bbc251831111032e2f337ec20713149cb5ef +Author: Luc Verhaegen +Date: Thu Jan 20 21:55:38 2005 +0000 + + [devel-remove_Modes_table] Remove the last remnants of the Modes table, + using table with dotclock and VGA timing instead, called ViaModes. + +commit c2de5c18f026e9f283743cbea42c62494e6281f4 +Author: Luc Verhaegen +Date: Thu Jan 20 17:12:22 2005 +0000 + + [devel-Fix_dotclock_vt_hardlock_2] Properly fix the hardlock experienced + with certain high values of the dotclock divider by resetting the + dotclock after new values are written in Restore. + +commit fd968776eb242e746936bfca1985cd922a5e280d +Author: Luc Verhaegen +Date: Thu Jan 20 15:26:36 2005 +0000 + + [devel-UnmapMem_stop_touching_SR1A_bit3] Disabling 0x08 means that the + driver fails to restore CH7019/panel properly upon Xserver exit. Bit + tends to be enabled by the BIOS in textmode anyway, so Restore handles + it. (Terry Lewis) + +commit a2eec0ad7a226aa24647713e9532dd668cd66aa7 +Author: Luc Verhaegen +Date: Thu Jan 20 11:30:44 2005 +0000 + + [devel-tv_callbacks_enforce] Start enforcing the callbacks. When there is a + problem, fail gracefully and dealloc cleanly. + +commit ce17a5ffdea8e872451a61d6c5f3407bc5b5e901 +Author: Thomas Hellstrom +Date: Thu Jan 20 09:14:02 2005 +0000 + + Increased max Xv source Y resolution to 1200. Fixed bug where YUY2 overlays + did not free the allocated image buffers when a surface was destroyed. + +commit bc6b0a3365ee0ccf7c1537b0c161846383434ad3 +Author: Thomas Hellstrom +Date: Wed Jan 19 16:35:44 2005 +0000 + + XvMC update: + Adaptation to DRM which now cleans up stale decoder locks. No longer + necessary to do that in the X server. + Bumped required DRM version to 2.4.x + Started preparation for IDCT / MOCOMP surfaces. + +commit 7bab0fbc560c01bf1aa7e8ec257386f4756b45b6 +Author: Luc Verhaegen +Date: Tue Jan 18 15:02:09 2005 +0000 + + [devel-ids_20050118] + - K8M800: MSI K8MM-ILSR (Karl J. Runge) + - K8M800: Biostar K8VGA-M (Brendan Burns) + - Added commented list of common devices that haven't been reported yet. + - replace disclaimer with proper MIT on via_id.c and via_id.h + - update copyright date. + +commit 8c03d8abe8ecf22aa7112a92d418bc458f68d2a3 +Author: Andreas Robinson +Date: Fri Jan 14 09:31:56 2005 +0000 + + Bugfix: YUY2 overlays were broken, because VIAVidCreateSurface() did not + allocate any HQV surfaces. + +commit f1b7f877211bf94cf91ae94889e9c27b199e5bd3 +Author: Luc Verhaegen +Date: Thu Jan 13 22:31:20 2005 +0000 + + [devel-modetable_remove_fifo] Remove setting of fifo from the Modes table. + Set it from ViaPrimaryFIFO instead. + +commit 976d8417c146bbe43d3f0819a1f9b63df9b23173 +Author: Luc Verhaegen +Date: Sat Jan 8 22:44:04 2005 +0000 + + [devel-modetable_bpp_independent] Remove all bitsPerPixel dependencies from + the modetables: namely: setting SR15 (bpp?) and CR13/35 (offset). + +commit c01a704e8c5b0b6b8bcc3c7c6d8d316a949bb7af +Author: Luc Verhaegen +Date: Sat Jan 8 14:05:24 2005 +0000 + + [devel-add_vesamodetable] Add a table which maps known vesa modes to vesa + mode numbers. These mode numbers are necessary for the panel support + (still) and moving this to a seperate table removes a bpp specific + entry from the Modes table. + +commit ce2bf2ac891de9751439671115b0440dc0497532 +Author: Luc Verhaegen +Date: Sat Jan 8 13:42:51 2005 +0000 + + [devel-fix_panel_numPatch] Properly propagate numPatch when setting + Centered or Expanded panel. This is a possible fix for the non-native + resolution problem. + +commit 3a74ba9d3a8b385454bbad88460d1a41fa7c30ca +Author: Luc Verhaegen +Date: Sat Jan 8 13:31:40 2005 +0000 + + [devel-ViaModePrimary] + - Rename VIAFindModeUseBIOSTable, VIASetModeUseBIOSTable and + VIASetModeForMHS to ViaModeInit, ViaModePrimary and ViaModeSecondary + respectively. + - Move vga mode specific stuff to ViaMode[Primary/Secondary]VGA. + +commit e0008449d63b90aded5e374d0ea9b900f2e4f015 +Author: Luc Verhaegen +Date: Mon Jan 3 11:12:57 2005 +0000 + + [devel-cleanup_Modes] + - remove unused Modes entries (unacceptable resolutions, 4bpp, etc), move + to null ended list instead of numModes. + - remove unused Mode struct entries. + - embed remaining stdVgaTable entries in Mode struct directly. + - reduce Ext to 8 entries (null ended list now) + +commit b4f9e28a8778d49b4e59e38e448a1957b9b50f18 +Author: Luc Verhaegen +Date: Mon Jan 3 10:29:46 2005 +0000 + + [devel-remove_via_refresh_h] + - create ViaRefreshAllowedTable, replacing via_refresh.h entirely. This + table now holds the minimum memory clock required, per bitdepth, + resolution and refreshrate, instead of a table per memory clock and a + boolean. + - recreate refreshTable, which includes resolution and refreshrate, + reducing a 17x5 table to 17. + - alter ViaGetNearestRefresh, ViaRefreshAllowed and ViaPrimaryRefreshPatch + to use the new tables. + - Limit TV and Panel to 60Hz modes. + +commit b5bd456cfdfa17a59d73db2eab4c37eee70d7b45 +Author: Luc Verhaegen +Date: Mon Jan 3 07:34:05 2005 +0000 + + [devel-remove_commExtTable] + - add ViaGrMask (graphics register) to via_vgahw.c + - remove commExtTable, replace by small function which sets the same + registers. + - remove unused stdModeExtTable. + - embed VIABIOSRegTable struct in VIAModeEntry, rename to Ext. + - alter Mode.Ext handling to switch on port and using the right via_vgahw + call. + +commit 136bf2953d9151e7a7c8f80639a51a6052730044 +Author: Luc Verhaegen +Date: Fri Dec 31 16:33:25 2004 +0000 + + [devel-ResolutionTable_to_mode_h] + - move VIA_RES_, VIA_TVRES_, VIA_NUM_REFRESH_RATE defines to via_mode.h + - remove unused defines in via_mode.h + - move ViaResolutionTable to via_mode.h + - move structs in via_mode.h close to the actual variables + +commit 87163cfc4681a78ceb4575f97f3386854720d618 +Author: Luc Verhaegen +Date: Fri Dec 31 04:58:21 2004 +0000 + + [devel-TVModeValid] + - move TVResolutionValid to TVModeValid, pass DisplayModePtr to it. + - have vt162[12]ModeValid check on Crtc[HV]Display instead of + pBIOSInfo->ResolutionIndex. + +commit 7ab908989bef914337ab2ebfb44e6da96b51032e +Author: Luc Verhaegen +Date: Fri Dec 31 02:30:23 2004 +0000 + + [devel-fix_dotclock_vt_hardlock] VT hardlock is dotclock related. When the + divider multiplier is too high, the CLE266A hardlocks on returning to + VT. Reduced multiplier for + 1280x1024@60Hz, 1280x960@60Hz, 1024x576@60Hz, 640x480@100Hz, 640x480@120Hz. + Proper fix will probably be resetting the dotclock on LeaveVT. + +commit 34bbf38ded61747fa0c0cdacdb0cd57a03b10cfe +Author: Luc Verhaegen +Date: Mon Dec 27 19:35:29 2004 +0000 + + [devel-tv_callbacks] + - move everything vt162x related to callbacks, except detect and init. + - split up vt1621/vt1622 shared functions to vt1621 and vt1622. + +commit 196fb984b88cb723d4df9490e6e585e2c890c86f +Author: Luc Verhaegen +Date: Sun Dec 26 23:35:49 2004 +0000 + + [devel-dump_actual_rom_size] Stop dumping the full 64k while we know the + actual size of the video rom. + +commit 4bebf7a39761fadae4cddb8ab18975d950d1d8b5 +Author: Luc Verhaegen +Date: Sun Dec 26 23:23:30 2004 +0000 + + [devel-ids_20041226] + - KM400: ?another? Abit VA-10 (Chris Hessing) + - KM400: rename Maxdata M-Book/Gericom Hummer to Uniwill 755CI (examined + .infs for pciids in uniwill windows drivers) + - KM400: Epox EP-8KMM3I (Yury Tarasievich) + - K8M800: Giga-byte GA-K8VM800M has its own ids (Michael Patra) Suspect + possible pciinfo misalignment with AMD64. + - K8M800: Sharp Actius AL27 (Jeremy Moles) + - PM800: ECS PM800-V2 (thomas) + +commit 2081c4473ad8cb9e9721d4cfb2672a16ed03db62 +Author: Thomas Hellstrom +Date: Sun Dec 26 21:31:20 2004 +0000 + + Apparent fix for the problems people have been having with small monochrome + bitmaps that sometimes are not drawn correctly. The lowest 3 bits of + the bitmap source address were lost. + This is a trial-and-error fix that might not work in all cases. The meaning + of the register bits is undocumented for this operation and the bug is + present also in VIA's own drivers. + +commit 9101defc712a96edb8c46d0088c20fa93bf0cddc +Author: Luc Verhaegen +Date: Sun Dec 19 14:34:03 2004 +0000 + + [devel-seperate_vt162x] + - move via_vt2.h, via_tv2.h, via_vt1622a.h and the relevant structs from + via_mode.h into via_vt162x.h + - remove VIA_VT1622A define, use VIA_VT1623 instead (devices are + equivalent) + - move all vt162x specific functions/code into via_vt162x.c + - move TVRegs array from VIARegPtr to pBIOSInfo + - replace calls to vt162x functions by ViaTVxxxx awaiting the next step: + callbacks in pBIOSInfo + +commit 1fbe44d571a06acf4ab2e591345e36a3eb7245f6 +Author: Thomas Hellstrom +Date: Sat Dec 18 12:35:50 2004 +0000 + + Fixed long standing bug where pVia->Overlaysupported could be used + uninitialized, hence blocking the overlay for applications not setting + up any Xv attributes before starting to play (mplayer for example). + (YT, cybervision) + +commit 5ead308b799fc53c0d5bd64919f7eaebc1ec6fbf +Author: Luc Verhaegen +Date: Fri Dec 17 23:40:55 2004 +0000 + + [devel-pBIOSInfo_Clock_panel] Write Secondary clock and set + SetUseExternalclock when setting panel. CR6A, CR6B and CR93 need some + serious poking. (Terry Lewis) + +commit 0f9a00d82aa2b46c48fe2c542003c1ae0ef2cb6e +Author: Luc Verhaegen +Date: Fri Dec 17 18:50:05 2004 +0000 + + [devel-rework_output_activation_2] + - remove VIA_DEVICE_CRT1, VIA_DEVICE_CRT2 + - remove leftover function prototypes + - remove ViaCardIdStruct member Force + +commit 30313e309e1572659b3c9adeae909d9141a2e612 +Author: Thomas Hellstrom +Date: Fri Dec 17 15:51:18 2004 +0000 + + Transfer copyright to the Unichrome Project. + +commit ccc53101c4c69100b4f87f2ec5617519e471bdf2 +Author: Luc Verhaegen +Date: Fri Dec 17 15:26:30 2004 +0000 + + [devel-remove_vbegetdisplaydeviceinfo] remove leftover vbe function. + +commit 2d93e23b685cf53902c3abd312375c3059d305b2 +Author: Luc Verhaegen +Date: Fri Dec 17 12:22:42 2004 +0000 + + [devel-remove_cvs_id] + - remove old xfree86 id strings left over from import from xfree86 + - removes the carriage returns in via_overlay.c + +commit d999ab711caebcb96448477b5b6f3138a48aa895 +Author: Luc Verhaegen +Date: Wed Dec 15 15:38:35 2004 +0000 + + [devel-DRIEnabled] + - remove directRenderingEnabled for DRIEnabled. + - fix segfault while attempting to switch modes with DRI not enabled. + - remove DRIEnter/LeaveVT; DRILock/Unlock and RingBufferInit/Cleanup + directly in EnterVT, LeaveVT, SwitchMode + +commit a0086434925b903d890f293d025b614f3a2ee7b5 +Author: Luc Verhaegen +Date: Wed Dec 15 15:09:11 2004 +0000 + + [devel-remove_ViaWaitIdle] VIAAccelSync simply calls this; move ViaWaitIdle + to VIAAccelSync and into via_accel.c + +commit 7139fd0820009f8bb9479d30542179b1f2b635b3 +Author: Luc Verhaegen +Date: Sun Dec 12 13:24:24 2004 +0000 + + [devel-pBIOSInfo_Clock] Introduce pBIOSInfo->Clock. This CARD16 gets set to + the value that needs to be written to the dotclock registers later on, + and these only get written once now. + +commit ff85584be00936efdb5784a4b6c8ad103c6dc90c +Author: Luc Verhaegen +Date: Thu Dec 9 07:32:15 2004 +0000 + + [devel-remove_EnableMMIO] + - Work seperate Enable and DisableMMIO into the relevant functions. + - Remove pVia->FBStart, a duplicate of FBBase. + +commit d1c9ddf283e1107653aefd782ba946fea7bae2ba +Author: Luc Verhaegen +Date: Thu Dec 9 04:51:53 2004 +0000 + + [devel-rework_output_activation] + - remove tvhscale option entirely. + - Alter manpage entry for ActiveDevice and go for saner ActiveDevice option + parsing. + - remove the Active/Connected/DefaultActiveDevice mess. Go for a more + transparent configuration: Crt/Panel/TVActive, based on detected + outputs and possibly the ActiveDevice option. + - All output detection and activation code is now contained in + OutputsDetect and Select. + +commit eb9a051986075a949eb1035e5c0e1507e34febdc +Author: Luc Verhaegen +Date: Thu Dec 9 02:45:47 2004 +0000 + + [devel-LoadPalette_move] Move VIALoadPalette back to via_driver. + +commit 03f200daf383669d1bf5578d134a05da76e1ff93 +Author: Luc Verhaegen +Date: Thu Dec 9 02:34:43 2004 +0000 + + [devel-ovl_cleanup_5_static] Leftover prototype warning. + +commit cbcafbd1539cf05b25ef2164c5e53620d906ce2e +Author: Thomas Hellstrom +Date: Wed Dec 8 10:46:34 2004 +0000 + + Fix 8x8 pattern drawing bug. (Felipe Rodriguez) + +commit 666363a995e8ae46a4178da58b065a91d02e9b8b +Author: Thomas Hellstrom +Date: Tue Dec 7 13:53:59 2004 +0000 + + Blitter fix that should make the HW blitter cover a larger area of the + frame buffer. If this works well it might be possible to loosen the + restrictions on the pixmap cache size. + Foreground color fix. The 2D engine may loose its setting of this value + between two subsequent calls if the blitter is used by DRM to track + completion of command buffer submissions. (That is to blit a unique + number to a reserved region of the frame- buffer when a certain point + has been reached in the DMA command buffer). + +commit fb5e1655e587104e5d0b552e6a1938f3c4b7ca4c +Author: Bernhard Rosenkraenzer +Date: Thu Dec 2 12:00:57 2004 +0000 + + Fix up make install.sdk after via_i2c.h removal + +commit a605571015112fe6dbf7dc1458a60159f743cece +Author: Luc Verhaegen +Date: Tue Nov 30 21:56:33 2004 +0000 + + [devel-ids_20041130] KM400: Mitac 8375X (Jan Sarenik) + +commit 3007ea74818eae41fff3e0d5a50abc99c9c02b71 +Author: Luc Verhaegen +Date: Tue Nov 30 03:11:35 2004 +0000 + + [devel-I2CScan] Add HAVE_DEBUG specific option "I2CScan". This has the + driver probe all possible addresses on the second and third bus at + I2CInit. + +commit e4bfbbad1b4dfcc514bb895b2f8074defa558b2d +Author: Andreas Robinson +Date: Mon Nov 29 08:18:15 2004 +0000 + + [devel-ovl_cleanup_5] + - Added back (16,16,8) V3 FIFO setting in two (out of four) places. + +commit e8e2aa3c51f46890ebff913469bed41e3bfdad47 +Author: Andreas Robinson +Date: Mon Nov 29 08:05:26 2004 +0000 + + [devel-ovl_cleanup_5] Bugfixes: + - Reinstated out-commented u/v pitch in the V1-without-HQV case. YV12 + picture was black and white because the u/v pitch was zero. + - Set V3 colorspace for CLE266. + - (16,16,8) is the wrong V3 FIFO setting for my CLE266 rev3 board; bad + picture. VIA uses (32,16,16) and that works much better. + - Corrected V3 FIFO setting (0, 0, 0) => (1, 0, 0) + - Removed unneeded call to viaOverlayGetV1V3Format() (V1 case) + Cleanups: + - Partial HQV-idle cleanup Upd_Video() TODO: Separate and move HQV-wait + functions to via_regrec.c + - Readability in VIAVidUpdateOverlay(). + - Modified VIAVidAdjustFrame() parameters. ADJUSTFRAME struct now obsolete. + TODO: Remove definition + +commit a9b11ae44478bd9bd1627aa9009384dd1bf67178 +Author: Thomas Hellstrom +Date: Sun Nov 28 19:00:23 2004 +0000 + + Fixed uninitialized video engine colorkey value. On KM400's this caused + mplayer not to view the videos properly until the colorkey value was + reinitialized by another application. + +commit 4b1e43bef40cfb7a90175a23f6e2e668cbd3d977 +Author: Luc Verhaegen +Date: Fri Nov 26 10:42:23 2004 +0000 + + [devel-fix_i2cbus3_cleanup] + - Raise SDA before reading, to avoid false ACKs when bus is not + implemented. + - fix I2C3Stop + - Remove via_i2c.h, rename VIA to Via in via_i2c.c, and add Unichrome + Project copyright. + - fix unknown VT162x message address + - check for Bus existence in TVDetect + - reshuffle TVEncoder handling in GetDeviceDetect + +commit 564acaf364c744a5d17cedc7221f3b5a706c0356 +Author: Luc Verhaegen +Date: Tue Nov 23 23:56:15 2004 +0000 + + [devel-tone_down_PanelGetIndex_message] X_ERROR, together with + FindModeUseBIOSTable being called from ValidMode, is a nuissance. Moved + to X_INFO. + +commit e6d6a0cc3a8855a5770c1c686fd4f398877a659b +Author: Luc Verhaegen +Date: Tue Nov 23 23:45:38 2004 +0000 + + [devel-ids_20041123] + - Warn about manufacturers using the main pci ids as subsystem ids. BAD + form. + - drop "pogolinux konabook 3100", use only "Mitac 8399" as i doubt + pogolinux still intends to pursue this. + - rename "Gericom Hummer Advance" to "Maxdata M-Book 1200X", should be + "Uniwill something" though, but uniwill site is not very definitive on + this. + - KM400: add "Chaintech 7VIF4/Biostar iDEQ 200V", these are 2 different + devices using the KM400 id (Damjan Jovanovic/Stefan Foerster) + - KM400: "Abit VA-10" (Gustavo Broos) + - K8M800: "DFI K8M800-MLVF", shares K8M800 id with "Giga-byte GA-K8VM800M" + (Livio Soares) + +commit 2549c033bc7ba07273aac02fc0d88da7328f93f5 +Author: Luc Verhaegen +Date: Tue Nov 23 22:49:02 2004 +0000 + + [devel-ovl_cleanup_4] Set SetupFIFOs static to silence missing prototype + warning. + +commit 67a3c4f5f58dca6fc232e010de8854dc9cdf6653 +Author: Andreas Robinson +Date: Tue Nov 23 17:13:03 2004 +0000 + + [devel-ovl_cleanup_4] More Upd_Video() cleanups: + - Removed hungarian notation. + - Moved many V1/V3 register writes into separate functions. + +commit ee0cf9ab807a89db1021d12294fa9fdc601bad0c +Author: Thomas Hellstrom +Date: Sun Nov 21 11:08:50 2004 +0000 + + Added a fix to the asm memcpy timing routines that makes it possible to + compile it with PIC. One asm instruction clobbered the PIC register. + This fix is already in xorg. + +commit 36fa7215d605cf579a6b7fd3b54b649ebb4bcd69 +Author: Thomas Hellstrom +Date: Sun Nov 21 10:48:05 2004 +0000 + + Added cropping ability to the XvPutImage function (Kenneth Aafløy) Also + loosened the restriction of XvImages having to have a witdth that is a + multiple of 8, since the above fixes this. + +commit 19ecef92d5a7a2ac98c83faf34f02f664143613a +Author: Luc Verhaegen +Date: Sat Nov 20 22:51:24 2004 +0000 + + [devel-fix_i2cbus3] + - Add backend to pI2CBus3 (high level). Tested on an acer aspire 135x with + ch7019b. (made available by Terry Lewis) + - remove everything related to gpioi2c. + - remove useless PostDVI, SensorDVI and related pBIOSInfo entries. + - move to xf86I2CRead/WriteByte where possible, replacing clunky + xf86I2CWriteRead + - move TV Post and Preset next to eachother, call the encoder specific + functions from ViaTVSetMode. + - move Enable/DisableLCD to VIALCDPower + - create ViaTVPower, call this from DPMS. + +commit 5ee0084c4ac1f78bda77ebe828bb2fd145f4db00 +Author: Luc Verhaegen +Date: Sun Nov 14 17:41:28 2004 +0000 + + [devel-virtual_extend] Extend Virtual ranges to 3344x2508. This is the + highest resolution (4:3) possible while not breaching the 32MB + frameadjust limit. This 32MB limit might be breached on chipsets other + than CLE266Ax. Current X can only be limited with max pitch and max + height, apertureSize has no effect. + +commit 6629e3b22165d1120940b388bb5a7b54abe91b8f +Author: Thomas Hellstrom +Date: Sat Nov 13 16:15:21 2004 +0000 + + Fix compilation warning with older XvMC protocol revision. + +commit 92eecc8df678625c63335acfc0814bdfd847175c +Author: Andreas Robinson +Date: Sat Nov 13 15:19:57 2004 +0000 + + [devel-ovl_cleanup_3] via_swov.c: + - Refactored VIAVidCreateSurface() (surface allocator). + - Cleaned up the display fifo (V_FIFO_CONTROL) setting code. + - Cleaned up the color/chroma-keying setting code. + +commit 2cb4def724c59b46a90a4246064af38171d86933 +Author: Luc Verhaegen +Date: Sun Nov 7 21:17:48 2004 +0000 + + [devel-Drop_tuner] + - Drop horribly broken tuner support. Whoever reinstates this will have to + fix it, clean it up severely, and have drivers/I2C as an end goal. + - remove pVia->Video for pVia->VideoStatus, rename its values. + - various unused cruft removal + +commit c5c8d67deefe7fa03f1b40a2502bc72cb88b6e3c +Author: Thomas Hellstrom +Date: Sun Nov 7 16:22:10 2004 +0000 + + Add fix segfault in for enter / leave VT when DRI was not enabled. + +commit cff1d726cb53bb5ff0c8dd58d4311c7ce39ccbb8 +Author: Thomas Hellstrom +Date: Sun Nov 7 16:18:09 2004 +0000 + + Add support for telling the XvMC lib the driver name to load, and to tell + libviaXvMC what busID to use and the internal XvMC protocol version. + +commit 7fd62bc2b0fd7fd611633b28abd4a5c7e820b382 +Author: Luc Verhaegen +Date: Sun Nov 7 02:46:03 2004 +0000 + + [devel-reinstate_memset] Upon relentless request, reinstate blacking of + memory. + +commit a1de55851ef7687705221a91f01918b891812a9e +Author: Andreas Robinson +Date: Fri Nov 5 20:06:16 2004 +0000 + + Total makeover of via_overlay.c: General cleanup (variable renaming etc) of + of all functions. Merged viaOverlayGetV1Format() and + viaOverlayGetV3Format(). Rewrote functions calculating overlay scaling + register settings. Rewrote functions calculating overlay 'fetch' and + 'displaycount' variables. [devel-ovl_cleanup_2] CVS: + ---------------------------------------------------------------------- CVS: + Enter Log. Lines beginning with `CVS:' are removed automatically CVS: + CVS: Committing in . CVS: CVS: Modified Files: CVS: via_driver.h + via_overlay.c via_swov.c CVS: + ---------------------------------------------------------------------- + +commit e24a417f38b183f941da115054141b7f363b9567 +Author: Andreas Robinson +Date: Thu Nov 4 19:27:21 2004 +0000 + + Whitespace cleanup in via_swov and via_overlay.c (indentation, hard tabs, + trailing spaces etc.) Replaced ALIGN_TO_nn macros in via_priv.h with + single ALIGN_TO macro. Replaced viaMacro_* functions with + SaveVideoRegister(), ResetVidRegBuffer() etc. + +commit c089d8ef1746d85e04d6cf59365307524097c46f +Author: Thomas Hellstrom +Date: Wed Nov 3 16:11:01 2004 +0000 + + Moved AGP ring-buffer support to an option which makes it easier to + configure. Also added another option that makes it possible to avoid 2D + acceleration commands to use AGP. + Updated the man-page. + +commit 79850f4bb1ecf5161f3a20044a6f1984fba1aa40 +Author: Thomas Hellstrom +Date: Wed Nov 3 14:22:50 2004 +0000 + + Fixed rare bug where a failing XvMC init would remove the DRI mmio map. + +commit 30d29ed3cdebbe70fe5b3403df6483f3a3bccf98 +Author: Thomas Hellstrom +Date: Sat Oct 30 12:29:39 2004 +0000 + + Experimental stop overlay from flickering when a window is moved over it. + (among other thing this makes watching unscaled shape overlays in xine + much more enjoyable). Removed compilation warning in via_swov.c + +commit 0e8e6a9ad380e932f870a876134fff246a63e7a3 +Author: Thomas Hellstrom +Date: Sat Oct 30 10:14:34 2004 +0000 + + altered BusID transmission and authentication success info to XvMC client. + This is for faster startup of XvMC and correct handling of multiple + video cards. + +commit 2f94aad0763f7471b69d712ad465cb410ca86798 +Author: Luc Verhaegen +Date: Sat Oct 30 00:31:41 2004 +0000 + + [devel-fix_tvres_invalid] Using FindModeUseBIOSTable exposed that + ViaTVGetIndex returns TRUE even when the TVIndex value in the table was + TVRES_INVALID. + +commit e873ebf79b431b3626743fb837ebaf520a668f04 +Author: Luc Verhaegen +Date: Fri Oct 29 04:25:48 2004 +0000 + + [devel-fix_handling_of_bad_mode] + - Fix the horrible bug introduced in clean_Save_and_Restore: An assignment + of a variable is a copy, not a reference. Or how an utter idiot could + still break something this trivial. This fixes the hardlocking upon + Restore. + - Move FindModeUseBIOSTable to the front of WriteMode. Remove FreeRec from + WriteMode (Pieter Peers) + - Call FindModeUseBIOSTable from ValidMode. Unacceptable modes are now + removed beforehand. + - Adjust messaging of FindModeUseBIOSTable by passing a boolean to it. + +commit e7eed47426702c50b7f5c887da8baaa568e0cea6 +Author: Thomas Hellstrom +Date: Thu Oct 28 19:55:19 2004 +0000 + + Fixed chroma bug for YV12 when the image width was not a multiple of 8. + +commit e0ecf40f4fef435adaf7994f53605c009b894ced +Author: Luc Verhaegen +Date: Wed Oct 27 06:39:37 2004 +0000 + + [devel-ids_20041027] + - KM400: Asustek A7V8X-MX SE (Koos Pol) + +commit 4f66d1fc414c9658830b3bd7a6be4d79e0aabd3a +Author: Luc Verhaegen +Date: Wed Oct 27 05:42:44 2004 +0000 + + [devel-remove_s3utility_leftover_2] + - fix XVPORTNAME + - remove dual define for XV portnames. go for XV_PORT_something + +commit a25f83461ec9f04e0e4bbdaf2ae6af019f48d81d +Author: Luc Verhaegen +Date: Sun Oct 24 19:15:34 2004 +0000 + + [devel-clean_Save_and_Restore] + - remove useless pVia entries: ModeReg, ModeStructInit + - VIARestoreRegs(pscrn, vgaReg, VIAReg) -> VIARestore(pScrn) + - clean up VIAInternalScreenInit + +commit 1032bbffd4729d856cf7451245631ecd2cde9b46 +Author: Luc Verhaegen +Date: Fri Oct 22 09:10:46 2004 +0000 + + [devel-ids_20041022] + - km400: Soltek SL-75MIV2 (Hendrik Muhs) + - km400: DFI KM400-MLV (Richard Biddle) + - fixed Giga-byte GA-K8VM800M (my bad, mixed up ids) + - ordered entries on chipset, vendor, device. + +commit 4a2814c36aaf8e035c1e64bad7561dd3c6aa36c2 +Author: Luc Verhaegen +Date: Fri Oct 22 07:29:32 2004 +0000 + + [devel-fix_viaYUVFillBlack] + - fix compile issue on BSD (missing stdint.h) by moving to CARD[8/16/32] + - the 32bit version only filled half of what the 8bit version did. + (4B*(num/4) != (B+B)*num), buffer is num*16bits, so use CARD16. + - check endianness. + - move to via_swov.c, remove via_lib.h + +commit 013a5532d25960aa986b97b1e22ae06afe524853 +Author: Ivor Hewitt +Date: Thu Oct 21 19:51:30 2004 +0000 + + Updated flicker filter with Terrys correction for the VT1622 chip. + +commit bcf0c76f388c3bde50199a7b7da194f16333a4c4 +Author: Ivor Hewitt +Date: Wed Oct 20 13:23:19 2004 +0000 + + Workaround for CLE266 Rev 18 chipset hanging. + +commit bb907af36ec7332ba2fd4cf5286ead0d366a55fe +Author: Luc Verhaegen +Date: Wed Oct 13 07:12:27 2004 +0000 + + [devel-add_option_ForcePanel] Add an option to force the panel, making life + easier for people who are currently told to alter the return value of + GetDeviceDetect. Should only be used as a last resort, users should + depend on pci-ids, and therefor it should not feature in the manpage. + +commit f51837722ffb120eecd38f0ec304cbb73bfa130c +Author: Luc Verhaegen +Date: Wed Oct 13 02:19:36 2004 +0000 + + [devel-remove_s3utility_leftover] + - Remove leftover UTCTRL portid. + - Replace direct numeric index to viaPortPriv[] with XV_SWOV_PORTID. + +commit fdab1b57e8fa6f78dc6e2c711cff6dbe187c235d +Author: Luc Verhaegen +Date: Tue Oct 12 02:55:24 2004 +0000 + + [devel-ids_20041012] Move "Giga-byte 7VM400M-RZ" to "Giga-byte 7VM400" + (reported by Jan Tesacek) + +commit 08c0d0e75ebe63b966d93d57ed41754af63937e4 +Author: Luc Verhaegen +Date: Mon Oct 11 23:51:51 2004 +0000 + + [devel-create_HAVE_DEBUG] + - replace DEBUG_PRINT macro with HAVE_DEBUG, remove REGISTER_PRINT in + favour of HAVE_DEBUG. + - Add ViaPrintTVRegs to via_mode.c + - Add options DumpVGARom, PrintVGARegs, PrintTVRegs and subsequent booleans + in pVia. These options, when TRUE, will run ViaDumpVGAROM, + ViaVgahwPrint, and ViaPrintTVRegs respectively at key points. These + functions (and options) are available when HAVE_DEBUG is defined (as + currently in HEAD). + +commit fb478eae78033ead048991e04ab161e539d807dc +Author: Thomas Hellstrom +Date: Mon Oct 11 21:06:15 2004 +0000 + + Fix AGP Ring-buffer related crashes at mode-switch. Reported by and fix + suggestion by Pieter Peers. + +commit 81c8c87e82b0484089018d0ff100f9a632736312 +Author: Ivor Hewitt +Date: Sun Oct 10 19:16:29 2004 +0000 + + Fake change to test commit mail. + +commit a307190fa5254c7a3ac6db97a9eb7c1e2b51938a +Author: Ivor Hewitt +Date: Sun Oct 10 18:22:21 2004 +0000 + + Tiny commit to see if I've unbroken the cvsmail. :) + +commit 4e4c1518c8e200b76658dddedcb51e051a6d8757 +Author: Ivor Hewitt +Date: Sun Oct 10 18:20:03 2004 +0000 + + Finally got around to adding deflicker option. + +commit 0dedc81f7c50eb15b2b5bf0c30c940ff11700255 +Author: Thomas Hellstrom +Date: Sun Oct 10 13:02:36 2004 +0000 + + Update XvMC server side to new major DRM version. + +commit f0e5edafa70ed6ca076f81a00eb5831988ec3373 +Author: Thomas Hellstrom +Date: Sun Oct 10 13:00:35 2004 +0000 + + AllowInsecureDRI option and manpage update for it and DisableIRQ. + +commit dffa30012d0e4b423fbfbec395b38f030a38d238 +Author: Luc Verhaegen +Date: Sun Oct 10 11:39:39 2004 +0000 + + [devel-remove_via_vbe] + - remove unused via_vbe functions + - remove long time broken ViaVBEGetVideoMemSize and adjust memory size + detection accordingly. + - Remove BIOS version/date entries from pBIOSInfo, make VBE functions that + acquired those only print this information. (was unused anyway) + - move remaining via_vbe functions (4) to via_bios.c + +commit d1c47b2d7b96b1a2f674a60bc4f12550d3498d57 +Author: Luc Verhaegen +Date: Sun Oct 10 10:40:32 2004 +0000 + + [devel-drop_pViaModeTable] + - move modetable specific defines from via_bios.h to via_mode.h + - drop pViaModeTable and use included modetables directly. + +commit 240db04c8d4a06e5aed902b5e6c31b7ed046b686 +Author: Thomas Hellstrom +Date: Fri Oct 8 21:21:54 2004 +0000 + + Updated manual page with new DRI security policy. + +commit 2d21fc36a76450e1739bb8aa6d4e70cb12ab0e14 +Author: Thomas Hellstrom +Date: Fri Oct 8 21:17:56 2004 +0000 + + Updated to drm 2.0. DRI mmio changed to be exported read-only, which fixes + the dri security issue. 3D OpenGL is now only available as root. + +commit ac6fcb48cf212c77b10d90d2641d334493c35f7e +Author: Luc Verhaegen +Date: Fri Oct 8 04:32:37 2004 +0000 + + [devel-tv_single_I2Cdev] + - replace I2C_PORT[1/2] with pI2CBus1/2 + - remove TVRegs from pBIOSInfo + - add TVUseGpioI2C, fix VT1623 == GPIO idiocy. + - add TVI2CDev, only xf86CreateI2CDevRec/xf86I2CDevInit once for TV. + - alter TV GpioI2c to use pBIOSInfo->TVI2CAddr instead of "0x40" + +commit 5636b8bc12f09bf2b1cef5a7baa05aba2aaa70cf +Author: Luc Verhaegen +Date: Thu Oct 7 14:22:51 2004 +0000 + + [devel-remove_vgabios_modetables_retrieval] + - remove writing out modetables. + - remove acquiring modetables. + - remove Option "UseBIOS" + - add code to dump VGA BIOS image (Option "DumpVGAROM"). Only available + when DEBUG_PRINT is defined. No-one is (hopefully) using this code, and + the modetables will go soon anyway. + +commit d4cca339f3cf95772cb4d050e6b546d8570e5407 +Author: Luc Verhaegen +Date: Wed Oct 6 22:58:52 2004 +0000 + + [devel-remove_unused_palette] + - Remove unused VIASavePalette, VIARestorePalette + - pedantic whitespace cleanups discovered while syncing + devel-branch-bios_cleanup + +commit e577b22c82efd6e6133fcf4bbac4a4a2272fd466 +Author: Luc Verhaegen +Date: Tue Oct 5 22:16:53 2004 +0000 + + [devel-expand_ViaVgahwPrint] + - expand ViaVgahwPrint to also print Graphics, attribute and misc + registers. + - Rename ViaVgahwChange, ViaSeqChange, ViaCrtcChange to ViaVgahwMask, + ViaSeqMask, ViaCrtcMask. I never was entirely happy with "Change". + +commit 76c44db3bfce27183733fd642f2f033bbf4dc74f +Author: Luc Verhaegen +Date: Tue Oct 5 19:39:03 2004 +0000 + + [devel-ids_20041005] + - K8M800: Giga-byte GA-K8VM800M (Michael Jedlick) + - K8M800: Acer Aspire 1362WLC (Timothy Pearce) + - KM400: Soyo K7VME (Jason Parekh) + - PM800: Biostar P4VMA-M (Ted Taylor) + +commit d288a5aaab7d711719d4e2d637f3a2962dbed2d2 +Author: Luc Verhaegen +Date: Tue Oct 5 18:44:38 2004 +0000 + + [devel-VT162xDACDetect] + - Move VIASensorTV2/TV3 code into VT162xDACDetect. + - Move actual TVOutput decision to VT162xDACDetect, cleaning up + VIAGetDeviceDetect considerably. + - remove dozen unused defines from via_bios.h + - move VIA_TV2PLUS, VIA_TV3 to VIA_VT1621, VIA_VT1622 respectively. + +commit 44a069a64da788b401d1ecc95506c9ce8fe7494a +Author: Thomas Hellstrom +Date: Sat Oct 2 10:13:21 2004 +0000 + + Updated man-page with the virtual terminal switch issue. + +commit 6acfe10c687ff551b8ee67f805a2038598c3e22b +Author: Thomas Hellstrom +Date: Sat Oct 2 10:05:52 2004 +0000 + + Fix part of the DRI VT switch issue + +commit 7900e0b2962f831c180d77b14a7d1f33508f0a44 +Author: Bernhard Rosenkraenzer +Date: Thu Sep 23 23:12:05 2004 +0000 + + (trivial) fix for make install.sdk + +commit 991050ed893ddea619ffb070c57012991d7affd1 +Author: Luc Verhaegen +Date: Tue Sep 21 21:35:56 2004 +0000 + + [devel-man_adjust_option_values] + - clearly mark "over" and "under" as the values for TVVScan (suggested by + Anders Ellenshøj Andersen) + - put values for TVType, Cap0Deinterlace, and Cap1Deinterlace in double + quotes. + +commit 48c7c435835ff254302b34f7e3495b816156a485 +Author: Luc Verhaegen +Date: Tue Sep 21 19:09:30 2004 +0000 + + [devel-alter_driver_version] + - move VIA_VERSION_MAJOR/MINOR to VIA_DRI_VERSION_... + - change VERSION_MAJOR, VERSION_MINOR, PATCHLEVEL to 0,0,0 + +commit 0b2f3bcaf0b04fb15a16b35cfb8e5c231b76b1c9 +Author: Thomas Hellstrom +Date: Fri Sep 17 11:08:23 2004 +0000 + + Fixed a race condition in the XvMC image displaying code. Current XvMC libs + will not work with this. + +commit 07e360c5ed30ce73549f6a40417091e252800864 +Author: Thomas Hellstrom +Date: Thu Sep 16 20:01:33 2004 +0000 + + Modified the way unsupported chipsets were handled in via_video.c + +commit a57552363aa8e5cc50378e680b744c77e813c30b +Author: Thomas Hellstrom +Date: Thu Sep 16 17:01:39 2004 +0000 + + Added XvMC and Bug info to the manpage. + +commit 06a77c3142dd0c23ff9ecd417edd6267c79c5d83 +Author: Luc Verhaegen +Date: Thu Sep 16 11:53:59 2004 +0000 + + [devel-prepare_manpage_for_xorg] + - change nondescript "VIA video" to "VIA unichrome graphics" + - remove information for UseBIOS option. Should not be used regulary. + - adjust Xfree86 specific fields, usually with macros provided by x.org + - add k8m800 under development + - add list of possible modes + +commit 9b691b40b0c53d11694e30afb703c50f16bd5506 +Author: Luc Verhaegen +Date: Wed Sep 15 21:37:41 2004 +0000 + + [devel-remove_tvencoder_option] Was pretty useless. If we can't detect it, + then we can't talk to it through I2C and then we can't set up a new + mode. So why bother? + +commit da3f1d93b5f0c4c319021fab4f38f4c6c0293388 +Author: Thomas Hellstrom +Date: Tue Sep 14 18:09:17 2004 +0000 + + Minor Xv / XvMC attribute fixup. + 1. Changed back Hue to loop between -180 and 180 since this seems to be + what most applications expect. Default is still 0. + 2. Fixed sign and number of digits for the CLE266AX color settings. The + formulas are still not according to docs but the ones in the docs + simply gives totally wrong colors. + 3. Checked CLE266CX and KM400 color settings against docs. The number of + digits in the docs are wrong and inconsistent. Kept current formulas. + Happy with the reesult. + +commit d23e5a9adf46406b5c39f3a0a39814088fd4617d +Author: Thomas Hellstrom +Date: Tue Sep 14 17:44:08 2004 +0000 + + Made XAA call sync after a system-to-framebuffer video blit. Disabled + system-to-memory hardware video copies on CLE266 since they are + probably much faster in hardware. Made MMIO commands to the 2D engine + wait also for VQ. + +commit fb94089340122e38cd7d3cee28581f71928756c5 +Author: Thomas Hellstrom +Date: Tue Sep 14 14:22:53 2004 +0000 + + Fixed XF86Config forced chipset bug. + +commit b5f0e3f1a01b3653117e4f5339fa291dc953004f +Author: Thomas Hellstrom +Date: Tue Sep 14 14:19:55 2004 +0000 + + Verbously disabled Xv for non-supported chipsets. Milder warnings when XvMC + is not working. Don't add a duplicate MMIO map for XvMC. + +commit 96845743b6ef407665eeaf916add3d8dfbcec470 +Author: Thomas Hellstrom +Date: Sat Sep 11 10:29:26 2004 +0000 + + removed ddmpeg.h removed unsupported ddmpeg stuff from structs removed some + unsupported pixel formats from via_overlay.c Moved all dependencies to + via_priv.h + +commit 3f9b9b4dd9f3008024c9519b34a49d0991787980 +Author: Thomas Hellstrom +Date: Sat Sep 11 09:32:54 2004 +0000 + + Removed via_v4l_drv support + +commit cf134d7586f58fc6cf46331e2d4a0fae3d1990f7 +Author: Thomas Hellstrom +Date: Sat Sep 11 08:01:41 2004 +0000 + + Fixed a bug where fifo was not restored after video was shut down. + +commit a2411fdf7720d96523ded1ca442e15b1c62c338e +Author: Luc Verhaegen +Date: Fri Sep 10 11:58:06 2004 +0000 + + [devel-ids_giga-byte_and_asrock_km400] Fix mitac/pogolinux string, print + ids in hex instead of decimal in unknown message. Add new ids: + - Giga-byte 7VM400M-RZ (Mark C.) + - ASRock K7VM4 (Miki L.) + +commit 3bed8deab1565219e9db01516a064362d5f98a4a +Author: Thomas Hellstrom +Date: Wed Sep 8 16:48:43 2004 +0000 + + Adding support for immediate update of XvMC attributes, so that they take + effect immediately and not when the next image is displayed. Good for + color adjustments on still pictures. + +commit 405656eb3fc1363cb169604cdc7ee71e4b2be8e2 +Author: Thomas Hellstrom +Date: Wed Sep 8 13:12:47 2004 +0000 + + Updated XvMC protocol with a flag that tells wether AGP is available. + libXvMC needs to be recompiled to be compatible with the new protocol. + Bumped XvMC protocol version minor. + +commit 74a54f0f76cc95dd103e6d07b33b635be7d9bf52 +Author: Luc Verhaegen +Date: Fri Sep 3 10:49:43 2004 +0000 + + [devel-add_first_k8m800_cardids] + - Add pci card-ids for: Shuttle FX83 (Christoph Lampert) Mitac + 8399/Pogolinux KonaBook 3100 (Kosmo King) + - Clean up unknown message. Turn this into an error as i'm not getting much + responses to a warning. + +commit 62a90a62db7b3bbdc5853f413d5422a8ed6e1fb5 +Author: Thomas Hellstrom +Date: Thu Sep 2 22:34:58 2004 +0000 + + Fixed problem with access to unitialized XvMC data when XvMC was not + enabled. + +commit 222c023790a1669bb25bbaf2ed3fe0a82a98944a +Author: Thomas Hellstrom +Date: Thu Sep 2 08:30:43 2004 +0000 + + 1. Temporarily disable the AGP ringbuffer init since it seems to be able to + wreak havoc among current dri OpenGL Clients. Use #define + ENABLE_AGP_RINGBUF to enable + 2. Make XAA save certain small parts of the 2D engine state in software to + automatically restore it when it is used. DRI clients may alter it. + Removed the old hacky context save used by XvMC. + 3. Moved XAA to use command buffers. These are dispatched with AGP DMA if + available, otherwise using PCI. See the define above. + 4. Made XAA wait for engine idle before issuing commands over PCI. This + will otherwise stall the processor until the video engine becomes idle, + which is BAD for interrupt latency. + +commit 363a6257f79ce7c85586be4347a0d3c5524f2821 +Author: Luc Verhaegen +Date: Mon Aug 30 23:21:17 2004 +0000 + + [devel-fix_binary_AND_when_setting_dotclock] Fix feeding + VIASetPrimary/SecondaryClock 0x0000, caused by a & instead of |. This + effectively resolves the VT1622A/VT1623 bug, which was actually + CLE266Cx based. Bug was introduced in release-12-clock_abstractions. + (Hans-Jürgen Ficker) + +commit a5237ffcd4f735cbd81a2479a8e354ec7ae5f083 +Author: Luc Verhaegen +Date: Sun Aug 29 20:32:00 2004 +0000 + + [devel-remove_di_enable_from_panelsize] GPIOI2C panelsize detection is + removed, so we don't really need to enable DI1 or LCD anymore. DI0 + usually is enabled (probably not in non-primary VGA). The only hardware + actually getting DDC information currently is the aspire, which + responded to DDC without enabling DI0/DI1/LCD seperately (in + GetDeviceDetect). + +commit 2adfeea4fbfe9f72e3b32225828c5df7e01246bc +Author: Thomas Hellstrom +Date: Sat Aug 28 13:23:55 2004 +0000 + + Added support for initializing the DRM DMA ring buffer in DRM 1.4.0. + Hardware specific information from Erdi Chen. (Pause register). + +commit e5fd92e1bfb2ea063b9e27cdcc215c4f951a0389 +Author: Luc Verhaegen +Date: Thu Aug 26 01:07:24 2004 +0000 + + [devel-remove_VIAdcc1] This has been #ifdeffed out for quite a while, + because it was pretty broken to begin with. The (complete) removal of + vgaHWddc1SetSpeed eases the move to xorg dlopen support. + +commit 25c8d77fe3b890a5412615e1a359d82056af5f13 +Author: Luc Verhaegen +Date: Thu Aug 26 00:53:28 2004 +0000 + + [devel-remove_memset] Remove the memset which nulls the FB memory when the + primary mode is first written. It is of limited use and will cause + problems on KM400 and K8M800 when used before the sequencer is first + initialised. (Found by Christoph Lampe, further testing by Phil + DeBecker) + +commit 5ef2175dac8563dda2c5bef9f51897464d1e352f +Author: Luc Verhaegen +Date: Wed Aug 25 23:59:55 2004 +0000 + + [devel-backport_xorg_dlopen_XAAGetxxxROP] Backport X.org changes to the via + driver for dlopen. These are limited to the move of XAACopyROP, + XAACopyROP_PM, XAAPatternROP arrays to the functions XAAGetCopyROP, + XAAGetCopyROP_PM, XAAGetPatternROP. These changes are not necessary, + but I backported them anyway since we can hook them off a nice #define. + (x.org changes by Adam Jackson) + +commit dfce5c6c76518e7ac5083634a1c7c7efbfb1842b +Author: Thomas Hellstrom +Date: Tue Aug 24 17:25:55 2004 +0000 + + Added format of IA44 and AI44 surfaces used for subpicture blending to + via_video.c. Advertised the Backend Subpicture capability of the XvMC + implementation. + +commit 9cb61842ba13e180c32080c07f0930fe71342550 +Author: Luc Verhaegen +Date: Wed Aug 18 23:45:44 2004 +0000 + + [devel-fix_id_message] + - fix the message in via_id.c which states the device detected. + - fix the Chipset not implemented messages in via_bandwidth.c + - add unichrome project copyright notice to via_mode.c + +commit ec4fbad0397c1cfd458a0686ae683feb0ea9141f +Author: Luc Verhaegen +Date: Thu Aug 12 19:31:26 2004 +0000 + + [devel-fix_PanelSize_detection] + - remove GPIOI2C support for DDC panelsize detection (broken + duplicate of + normal i2c) + - remove manual detection of DDCv1, use xf86DoEDID_DDC2 instead. Add + ViaGetPanelSizeFromEDID to parse pMon for maximum allowed size. + - Move VIAGetPanelInfo to VIAGetPanelSize. + - Add IO PanelSize detection (preset by BIOS) + default to 1024x768. + - also detect PanelSize when using LCD (not only DFP) + +commit 21e4e5be9f9247e4e28a95723cdb6416d696aa5a +Author: Luc Verhaegen +Date: Thu Aug 12 12:44:48 2004 +0000 + + [devel-add_option_DRIIrqEnable] Allow for a quick workaround to keep the + NIC from giving up on lost interrupts on certain unichrome laptops. + This problem should get a proper seeing to though. (Phil Debecker) + +commit fefc31c78ac92db4ec3b05582bbf36ef693c37de +Author: Luc Verhaegen +Date: Thu Aug 12 12:14:46 2004 +0000 + + [devel-fix_CursorImage_initialisation] Properly set CursorImage to NULL in + VIAGetRec. (reported by Phil Debecker) + +commit 48fa9006987cc87fcb24b84ce5ba49a65c47e25a +Author: Thomas Hellstrom +Date: Wed Aug 11 19:41:17 2004 +0000 + + Fixed Xv / XvMC related building problems when building DRI-less driver. + Reported by Olaf Müller-Michaels, Luc Verhaegen. + +commit 880ea6622d82267694ef367f5bc95c939f4809ce +Author: Luc Verhaegen +Date: Sun Aug 8 21:09:47 2004 +0000 + + [devel-fix_GetNearestRefresh] + - correct routine for checking nearest refresh + - use mode->VRefresh directly instead of clock/(HTotal*VTotal) + - remove pBIOSInfo->Refresh + +commit 99aac78804a866b66f444dc66109dda31be91354 +Author: Luc Verhaegen +Date: Sun Aug 8 10:20:47 2004 +0000 + + [devel-fix_driless_build_of_v4l] + - add check for XF86DRI or linux at the top of via_privioctl.h + - encapsulate via_privioctl.h contents and V4L/ddmpeg stuff in VIAHAVEV4L + - remove strange uses of graphicInfo for pScrn, currentMode, pVia, + pBIOSInfo + +commit b459f91b0bf1cc2e7ba141361e8011994e3708ab +Author: Luc Verhaegen +Date: Thu Aug 5 14:38:09 2004 +0000 + + [devel-via_id] + - Add files via_id.c and via_id.h + - move pci ids and revision stuff to via_id.h + - move double checking of CLE266 revision to via_id.c + - introduce a ViaCardId table. + - set return value of ViaGetDeviceDetect acording to ViaCardId->force + +commit 7b4053122db174dea6535b7fdbcec69bac371890 +Author: Luc Verhaegen +Date: Wed Aug 4 17:01:15 2004 +0000 + + [devel-pBIOSInfo_remove_pVia_entries] + - Remove following entries from pBIOSInfo which are duplicated from pVia: + Chipset, ChipRev, FBBase, IsSecondary, HasSecondary, I2C_Port1, + I2C_Port2, DDC1, DDC2. + - Move MemClk to pVia, move retrieval to PreInit. + - Move FirstInit to pVia, move clearing memory to WriteMode. + - Move SAMM and GpioI2c to pVIA, remove I2CDevPtr dev. + - Stop DecideOverlaySupport from checking memory bandwidth itself, move to + pVia->MemClk instead. + - Remove handful of unused pVia entries. + +commit f3dd7f7586e4a25d8f162b697b5a9edc93c6e070 +Author: Luc Verhaegen +Date: Wed Aug 4 10:29:12 2004 +0000 + + [devel-pBIOSInfo_remove_pMode_entries] + - Remove the following entries of pBIOSInfo which are replicated from + DisplayModePtr: Clock, HDisplay, VDisplay, CrtcHDisplay, CrtcVDisplay, + SaveHDisplay, SaveVDisplay, SaveCrtcHDisplay, SaveCrtcVDisplay. + - Split WriteMode to WriteMode and RestoreRegs + - Move Contents of ModeInit to WriteMode + - remove mode, refresh, resMode, Clock from VIARegRec + +commit 39e5886a653e6305c8820362943bc2651e3f0a7f +Author: Luc Verhaegen +Date: Mon Aug 2 15:55:40 2004 +0000 + + [devel-pBIOSInfo_remove_pScrn_entries] + - remove entries from pBIOSInfo which are present in ScrnInfoPtr: hwp, + bitsPerPixel, displayWidth. + - Remove unused functions which depend on those entries + - Move xf86DrvMsg to pScrn->scrnIndex where pScrn is available. + - Group/Reshuffle entries in pBIOSInfo + +commit 9ed22161eb387fa9c849fe1ea3e9d7c77647f5d3 +Author: Luc Verhaegen +Date: Sat Jul 31 12:26:59 2004 +0000 + + [devel-ViaResolutionTable] + - Create table containing resolutions, even TV and PANEL. + - Rework contents of FindModeUseBIOSTable and the functions it depends on. + - Remove option Refresh. + - Remove or rename related pBIOSInfo entries. + +commit b834d75dc2b918593fb9a6660163a782c20519ec +Author: Luc Verhaegen +Date: Mon Jul 19 00:20:41 2004 +0000 + + [devel-move_version_macros_to_Imakefile] Move all X version specific macros + to the Imakefile. This also takes into account the X.org which + removal/renaming of the XF86_VERSION macros. + +commit 4e1eaa9c2e21761410440bdc5e0fde566242ac7a +Author: Luc Verhaegen +Date: Thu Jul 15 22:50:27 2004 +0000 + + [devel-FreeBSD_build_fix_for_via_privioctl_h] Remove superfluous inclusion + of asm/ioctl.h, this include breaks FreeBSD build. (Cory Jones) + +commit ad65383721fbbd90c0af04ab4b9e5b4a806cf079 +Author: Thomas Hellstrom +Date: Thu Jul 15 22:19:25 2004 +0000 + + Fix probable bug in Mono8x8PatternFill acceleration that causes the lower + part of the screen to become black in the KDM logout screen. + +commit 4a0ce1bd9042cdc21e1a13e94b09b00916799344 +Author: Luc Verhaegen +Date: Thu Jul 15 17:44:35 2004 +0000 + + [devel-pciids_k8m800_and_pm800] Clean up comments surrounding K8m800 and + pm800 pci-ids. The ones currently present are reproduced in the windows + drivers too and there has been a submission fo pciids.sf.net for the + pm800. + +commit b92cb0f356f33a3e5ccc5002da54180dec5f0319 +Author: Luc Verhaegen +Date: Wed Jul 14 15:40:35 2004 +0000 + + [devel-xfree86_cvs_build_fix_for_drm_h] Fix build against xfree86 CVS. + - Adjust ObjectModuleTarget (bero) when using version 4.4.99.2 or higher + - include correct directory for drm.h when using version 4.3.99.901 or + higher + - fix 2 xf86DrvMsg warnings. + +commit 30e3c2aed497e666a9f866d54da64753e7076643 +Author: Thomas Hellstrom +Date: Wed Jul 14 09:35:40 2004 +0000 + + Improved SMP cpu feature detection in via_memcpy.c + +commit 5fd1a8c1317151e015647a260f78cfe99375214c +Author: Luc Verhaegen +Date: Tue Jul 13 21:06:16 2004 +0000 + + [devel-debian_build_fix_for_xvpriv] Fixes build against debian + xfree86-4.3.0-dfsg1-6, which includes commit r1502. This requires + inclusion of xf86xvpriv.h for xvpriv. + +commit f14272a70a1e896eb20b38f54688b017bdb90277 +Author: Thomas Hellstrom +Date: Tue Jul 13 19:42:41 2004 +0000 + + Modified for dri.sf.net and X.org recent xf86drm.h changes. Thanks to Nick + Hall and Ian Romanick. + Fixed version check for inclusion of xf86xvpriv.h in via_xvmc.c + +commit c6e74d6bf333e552ef1126dce60888f25b54cab6 +Author: Luc Verhaegen +Date: Tue Jul 13 08:09:32 2004 +0000 + + [devel-fix_for_fixes_for_vgaHW_remaining] Guess what, i only had to alter 2 + lines, and i still managed to break it. (luc) + - Fix disabling HQV (again - thomas) + - got confused over remaining outb/inb, which were commented out, moved to + vgaHW to avoid further confusion (luc) + +commit 5e45b9bf455c90f090f9488d98dd67d0820e92fd +Author: Luc Verhaegen +Date: Mon Jul 12 20:10:43 2004 +0000 + + [devel-fixes_for_vgaHW_remaining] This fixes 2 bugs i introduced in + vgaHW_remaining: + - fix disabling HQV (thomas) + - fix setting primary fifo threshold (me) + +commit b592e107f53d2ac3e31ec2f96c9e905c9e4152e1 +Author: Thomas Hellstrom +Date: Fri Jul 9 07:43:21 2004 +0000 + + Small fixes to via_overlay.c. + +commit 1e7252ebc18d7ee49a90a451cdb33233b8615368 +Author: Luc Verhaegen +Date: Thu Jul 8 20:47:19 2004 +0000 + + [devel-vt1622_force_DI0] Fix tv-out when not primary VGA; enable DI0 at the + end of PostSetTV3. + +commit eade4fa56e61d0e98234581aa3c72ea2a1a3ce5c +Author: Luc Verhaegen +Date: Thu Jul 8 20:29:37 2004 +0000 + + [devel-VgahwPrint] Implement ViaVgahwPrint which prints all "known" + Sequence and CRT registers. Encapsulate in REGISTER_PRINT + (via_vgahw.h). Add to PreInit and ScreenInit. + +commit 10b5d92f46ed9881c6668b4a4a7ffd4aaa832d2f +Author: Thomas Hellstrom +Date: Thu Jul 8 20:27:51 2004 +0000 + + Changed name of the 3d client driver from via_dri.so to unichrome_dri.so to + be compatible with the development at dri.sourceforge.net + +commit f2640da9dea17c856b53ad26f39de243d3ee42cc +Author: Thomas Hellstrom +Date: Thu Jul 8 20:20:34 2004 +0000 + + Implemented ad-hoc formulas for Xv hue, saturation, brightness and + contrast. Enabled corresponding Xv and XvMC attributes. + +commit 859ab8901fc3ea076018414750a51711d1657344 +Author: Luc Verhaegen +Date: Thu Jul 8 19:59:55 2004 +0000 + + [devel-secondary_vga_cleanups] Some cleanups that turned up when fixing + usage as secondary VGA device. + - Clean up WaitIdle. Only one routine is used for all unichrome + incarnations. + - remove comments with register descriptions introduced in the bandwidth + cleanup. + - abstract Cursor(Re)Store, VQDisable + - move LoadPalette to join other (now unused) palette functions + (via_mode.c). + - some minor cleanups and message improvements + +commit f8e71f34b855a570f4de3a6ac07cc31501d3babe +Author: Luc Verhaegen +Date: Thu Jul 8 15:30:23 2004 +0000 + + [devel-vgaHW_remaining] + - Moved all remaining VGA register accesses to vgaHW + - Always Use MMIO. Move MMIO mapping to sit before any vgaHW calls. Move + vgaHWUnlock after MMIO mapping. + +commit 892600765ce20e548ecbda55b1cb81c16807baab +Author: Luc Verhaegen +Date: Mon Jun 28 02:48:22 2004 +0000 + + [devel-via_gpioi2c_into_via_i2c] + - move i2c bus 1 & 2 to vgaHW (used to be PIO directly) + - Remove unused code from via_i2c.h + - Move code from via_gpioi2c.c to via_i2c.c + - rework gpioi2c to be able to use vgaHW, rework gpioi2c struct + - remove I2CUDelay, pass I2CBus1->Delay to gpioi2c struct instead + - move struct and function prototypes to via_i2c.h + - Split up I2C_RW_Control + - Add ViaGpioI2c_Probe + - Remove 12bit subaddress support + - Fix gpioi2c device detection: VIAGPIOI2C_Read return no longer ignored + when "probing". (former results on gpioi2c probably happened due to + uninitialised memory only) + +commit 9abacaa1738db05c9ed3021ccdbc9f3f1ec2e5cf +Author: Luc Verhaegen +Date: Thu Jun 10 21:16:41 2004 +0000 + + [devel-bandwidth_cleanup_fix] Add missing ViaSetPrimaryFetchCount to KM400 + case in ViaSetPrimaryFIFO. + +commit 10322a1c976d9390549bedc10b3a6805a77910d6 +Author: Thomas Hellstrom +Date: Tue Jun 8 21:59:44 2004 +0000 + + Merged devel-branch-km400-xv providing KM400 Xv support + XV_AUTOPAINT_COLORKEY attribute added to Xv and XvMC. XV_HUE attribute + with new scale in Xv and added in XvMC. XV_CONTRAST attribute with new + scale in Xv and added in XvMC XV_SATURATION attribute with new scale in + Xv and added in XvMC XV_BRIGHTNESS attribute with new scale in Xv and + added in XvMC Note that the four last attributes are temporarily + disabled until I've corrected VIA's completely broken formulas. + +commit bb006a6536d1c4a3cabb01408d884a4fd92ed9d8 +Author: Luc Verhaegen +Date: Sat May 29 09:58:19 2004 +0000 + + [devel-Imakefile_SDK_cleanup] Found fixes lost in the merge between the + devel-arklinux-merge branch and HEAD. These add files to the SDK + filelist in the Imakefile. Realphabetized. (bero) + +commit f7ec012c12ebeae8b880a89208813ee754a281c7 +Author: Luc Verhaegen +Date: Sat May 29 08:38:41 2004 +0000 + + [devel-fix_static_build_viaFillGraphicInfo] Move viaFillGraphicInfo outside + of #ifdef XFree86LOADER. + +commit 37012fafc30984eb4253188dbadca3e9422dff7a +Author: Luc Verhaegen +Date: Thu May 27 20:21:43 2004 +0000 + + [devel-remove_utility] + - Remove pBIOSInfo->UserSetting, struct and code. + - Remove VIAXVUtilityProc from viaPutImageG. + - Handle COMMAND_FOR_UTCTRL by default: in PutImageG and PutVideo. + - Remove via_utility.c/.h + +commit 5ad30b91d1c5f32572ad58531bb95d4de5424888 +Author: Luc Verhaegen +Date: Thu May 27 14:51:59 2004 +0000 + + [devel-revise_pciids_and_revisions] + - Aggregated all information i could find in comments in via_reg.h. + - (Re)Defined pci-ids. Altered symtab strings. Added commented + HAVE_K8M800/PM800. Created warnings for when either is used. + - Reviewed CLE266 revisions. Created CLE266_REV_IS_[A/C]X to handle this + cleanly. Removed pBIOSInfo->A2 and the related option. Added + doublecheck for Ax/Cx found in fb code. Replaced DispatchVGARevisionID + with new define. Added comments where only the revision is checked and + not the chipset. + - removed VIAvfInitHWDiff to VIAVidHWDiffInit and moved into via_swov.c, + removed via_hwdiff.c. Renamed VIAHW to HWDiff, removed unused entries, + added PM800. + - Removed K8M800/K8N800 from manpage, only detection is really implemented. + +commit 775d4dcc186f753d7d050f865faf7472a86ac2a9 +Author: Luc Verhaegen +Date: Tue May 25 15:46:26 2004 +0000 + + [devel-bandwidth_cleanup] + - Move the tables in via_bandwidth.h into via_bandwidth.c, they are not + needed anywhere else. Remove via_bandwidth.h. + - Cleanup via_bandwidth.c, move to vgahw. Rename + VIAEnabled[Primary,Secondary]ExtendedFIFO to ViaSet[]FIFO. + - Move VIAFillExpireNumber and Fetch Count setting (ViaSet[]Pitch) to + via_bandwidth.c + - Replace the copyright on via_vgahw with "The Unichrome Project". + - Create comments with crude register descriptions for fifo related + registers. + - Move blanking end overflow patch out of VIASetModeUseBIOSTable + - Rename ViaSet[]Pitch to ViaSet[]Offset. + - Add K8M800/PM800 code to ViaSet[]FIFO, enclosed in HAVE_K8M800/PM800, for + future implementation. + +commit b25a6e6a0630670f4da1d1ab707641c594f2b48a +Author: Luc Verhaegen +Date: Thu May 20 07:40:14 2004 +0000 + + [devel-remove_via_modeheaders] Move the code in via_modeheaders.c back into + via_bios.c, as the latter has become managable (somewhat). + +commit 8f557eb903a930bf9a7f94ff0e9e666e73aa1005 +Author: Luc Verhaegen +Date: Thu May 20 07:16:23 2004 +0000 + + [devel-remove_saa71xx] Remove support for philips saa71xx. No hardware + currently available/known. + +commit 5e09c1683ad70429d62af9a2bbdd81632547965f +Author: Luc Verhaegen +Date: Thu May 20 06:41:22 2004 +0000 + + [devel-remove_fs454] Remove support for Focus Enhancements FS454 Tv + encoder: no hardware (except on nvidia on newer xbox) and no + information currently available. + +commit e227a96aa3d91333648486c75561999e60c8ecb8 +Author: Luc Verhaegen +Date: Thu May 20 06:22:20 2004 +0000 + + [devel-remove_ch7xxx] Remove support for CH7017, CH7019, CH7304, CH7305: no + hardware currently available. CH7011 support will be added seperately + later (shuttle box has this and register level documentation is + available) + +commit 2b60c6619172063b2c442ff58e21334680ee8501 +Author: Luc Verhaegen +Date: Thu Apr 29 18:55:05 2004 +0000 + + [devel-failsafe_xvmc_attributes] Remove unused variable. + +commit a629c78c5f140191b2557e0342d7dd435d19caa3 +Author: Thomas Hellstrom +Date: Thu Apr 29 08:20:03 2004 +0000 + + Correct failsafe handling of possible future non-gettable XvMC Attributes. + +commit 1365f84ba4d657a22ac72efb14e1b583b00da915 +Author: Luc Verhaegen +Date: Sun Apr 25 23:26:20 2004 +0000 + + [devel-ViaSetColorspace_and_ViaYUVFillBlack] Move duplicated code to + ViaYUVFillBlack, reimplement. Move duplicated code to ViaSetColorspace. + Experiment with 32bit alignment on ViaYUVFillBlack. Various cleanups + and fixes. (bero) + +commit 0d5c3b0beed4a2be6c9ac746f9e70305943b3c9b +Author: Bernhard Rosenkraenzer +Date: Sun Apr 25 14:00:11 2004 +0000 + + file via_lib.h was initially added on branch devel-arklinux-merge. + +commit ea0ddf936bfb9a67f812bbb34e5b529d87b331c8 +Author: Luc Verhaegen +Date: Thu Apr 22 17:22:24 2004 +0000 + + [devel-ViaModeTablesInit2] Revert old and slighty broken ViaModeTablesInit + for a superior and fixed implementation. This does try to get the image + but should handle failure properly. + +commit 4dbbbd24e52fc6e50f59ea7f577b165c6acac226 +Author: Luc Verhaegen +Date: Thu Apr 22 11:28:40 2004 +0000 + + [devel-remove_wrong_pciids] Removed 3204/3205 pciids for k8m800 and km400. + These are the pci ids of the host bridge, not the agp device. Commented + out 3022 for CLE266, this device does likely not exist in the wild, if + it does, please contact -devel. + +commit f20a16f4fb6ca16669dc0811f2781c41fce4b9a0 +Author: Bernhard Rosenkraenzer +Date: Tue Apr 20 23:11:58 2004 +0000 + + FOURCC debug messages in hex are more useful than in decimal... + +commit da7f0febe45a21960aa75758a26625c79b51d3c2 +Author: Bernhard Rosenkraenzer +Date: Tue Apr 20 22:48:08 2004 +0000 + + VT7204 and VT7205 support + +commit f16aa032fa9b0cfc296b75c6293585068bab679b +Author: Bernhard Rosenkraenzer +Date: Tue Apr 20 22:12:01 2004 +0000 + + Remove comment about 3D being available only for XFree86 4.2 -- it's no + longer true + +commit 9cd9a62e90535100cdce60453794302d9b73b75a +Author: Thomas Hellstrom +Date: Mon Apr 19 13:46:34 2004 +0000 + + Xfree86 4.3 compatibility issue: Removed inclusion of xf86xvpriv.h if + XFREE86_44 is not defined. + +commit 198ed1e24e9d2a8f6b6acb13ac270b42dc71b566 +Author: Thomas Hellstrom +Date: Mon Apr 12 18:51:28 2004 +0000 + + XvMC updates: Added workaround for possible XvMC client crashes while doing + hardware blitting thus corrupting the via 2d engine. Added support for + future multiple XvMC ports if number of overlays increases. Added + support for XvMC Attributes. XvMC support is now complete, but probably + not bugfree ;-) + +commit fe45860716172384070b9b2ff48a98d2087ee88b +Author: Luc Verhaegen +Date: Fri Apr 9 17:33:51 2004 +0000 + + [devel-ViaModeTablesInit] introduce ViaModeTablesInit to replace the + overgrown VIAGetBIOSTable. Stop getting the bios image and computing + its checksum when the option UseBIOS is not specified. + +commit 648cf586042078770cf25f5244b2fb9365b54a77 +Author: Luc Verhaegen +Date: Fri Apr 9 17:13:28 2004 +0000 + + [devel-add_I2CBus3_stub] Add a stub for the third I2CBus (which i still + should get working). Add pI2CBusX hooks in pVia, next to the two + existing I2C_PortX, these will be weaned off as part of + output-abstraction. + +commit 394ce1f6f1fd06dcd01bfc5c0a61eb5ae0e8645e +Author: Thomas Hellstrom +Date: Fri Apr 9 13:12:30 2004 +0000 + + Defines for max xvimage size. Set max xvimage size to 1920x1080. Works fine + on CLE 3122. Made a symbol in via_video.c static. + +commit 899c9243e5fb690e5bf589fe09fd8a6a981a191a +Author: Luc Verhaegen +Date: Thu Apr 8 20:36:36 2004 +0000 + + [devel-pitchpatchfix] Fix a bad mask being used when setting the actual + primary pitch. + +commit 09c09d2dee544824821eb7572d2abcd2a463f385 +Author: Luc Verhaegen +Date: Thu Apr 8 19:50:25 2004 +0000 + + [devel-shuffle_blanking_bug_fix] Fix the daft and badly tested bug + introduced in devel-via_mode_shuffle_4. + +commit f4aeb622e88990bb642e60250ddddd7f114a7e73 +Author: Luc Verhaegen +Date: Fri Apr 2 18:27:13 2004 +0000 + + [devel-via_mode_shuffle_5] Part5: VIAEnableLCD/VIADisableLCD + +commit f3e8ff3d93dab3a7250fcc3836f53bedfdf3cbd2 +Author: Luc Verhaegen +Date: Fri Apr 2 18:25:33 2004 +0000 + + [devel-via_mode_shuffle_4] Part4: VIASetModeUseBIOSTable and + VIASetModeForMHS + +commit 009ca5dfe4759e551d5c806b217d7596083d3f64 +Author: Luc Verhaegen +Date: Fri Apr 2 18:23:20 2004 +0000 + + [devel-via_mode_shuffle_3] Part3: VIA[Pre|Post]set + +commit d9b404ce70cba2422cdb755a9e6dbf22df1aa75a +Author: Luc Verhaegen +Date: Fri Apr 2 18:19:52 2004 +0000 + + [devel-via_mode_shuffle_2] Part2: VIASetLCDMode + +commit d74595b2ab8b2d7eacfbe9496ea179f3a602d074 +Author: Luc Verhaegen +Date: Fri Apr 2 18:15:54 2004 +0000 + + [devel-via_mode_shuffle_1] This is what a few different choices in + variables and shuffling a few bits of code off to their own function + can do. All without actually changing the flow or functionality. Part1: + VIAFindModeUseBIOSTable + +commit 77d9cd374753205baeec7f7bf15ed9d4fa3eb09f +Author: Luc Verhaegen +Date: Fri Apr 2 18:03:43 2004 +0000 + + [devel-split_via_bios_to_via_mode] Move everything but retrieval of mode + tables from the bios to via_mode.c + +commit 895d3f84762ed888efd119fbcdb9a192d40c02ad +Author: Luc Verhaegen +Date: Fri Apr 2 17:58:07 2004 +0000 + + [devel-small_bios_cleanups] A handful of daft and pedantic cleanups. + unsigned char to CARD8, add a few returns to GetBiosTable, make + BIOSDate static. + +commit 0a79a863ffd66dde66015e62030cfc80bebeee80 +Author: Luc Verhaegen +Date: Fri Apr 2 17:47:52 2004 +0000 + + [devel-pitchpatch] Clean up setting line pitch, remove pitchalignmentpatch + because we are aligning ourselves already. + +commit afd938397448970acfd369d5f0180b0b72281799 +Author: Luc Verhaegen +Date: Fri Mar 26 19:29:50 2004 +0000 + + [devel-clock_abstractions] Implement the pretty common and straightforward + abstractions: SetUseExternalClock, Set[Primary|Secondary]Clock. + +commit 6cc017cb6382de80ffc60e1ea3b1a32ab5bc7544 +Author: Luc Verhaegen +Date: Fri Mar 26 19:24:04 2004 +0000 + + [devel-vgahw] Replace MMIOIN/OUT8 with vgaHW function calls. Add a few + convenience functions in via_vgahw.c/h. + +commit ce9bc2f7d1cbd3d19edb76645fce5b49f5c891fe +Author: Luc Verhaegen +Date: Fri Mar 26 01:49:54 2004 +0000 + + file via_mode.c was initially added on branch devel-branch-bios_cleanup. + +commit 12d23e26cf13afe182a57d521a4cf0fbbf696939 +Author: Luc Verhaegen +Date: Wed Mar 24 14:43:08 2004 +0000 + + file via_vgahw.c was initially added on branch devel-branch-bios_cleanup. + +commit 8194bd3b5530c1e4fdc4e418c1c4ead9b4a708e0 +Author: Luc Verhaegen +Date: Wed Mar 24 14:43:08 2004 +0000 + + file via_vgahw.h was initially added on branch devel-branch-bios_cleanup. + +commit 4c1a81caabfe3874fa06c9d5f2f50680636b6c06 +Author: Thomas Hellstrom +Date: Wed Mar 24 10:55:21 2004 +0000 + + Workaround for problem where the old drm module would return Success even + if memory allocation failed. (A zero size was returned instead). Also + changed indentation to preffered X "standards". + +commit 920d4f73e515fcd2132c01d7f39623314ac54630 +Author: Thomas Hellstrom +Date: Wed Mar 24 10:13:30 2004 +0000 + + Disabled XvMC for All chipsets except for CLE266. + +commit 6fdbf2070bb6ad91f467fc8b0f9ccf4c0f2a0b08 +Author: Luc Verhaegen +Date: Wed Mar 24 04:46:08 2004 +0000 + + [devel-resync_20030315] Resync with xfree86.org: fix driver name (tsi) + +commit 1e0597209487403b85c6e521f1d57821658482bf +Author: Thomas Hellstrom +Date: Sun Mar 21 18:49:04 2004 +0000 + + Final (?) fix of bug reported by Huynh Tran. The server probably called + drmRmMap on non-existing maps when it exited. This probably caused a + total machine lockup. :-( + +commit 7299f048ee48f5a287c3129713f93665276e583b +Author: Thomas Hellstrom +Date: Sun Mar 21 08:46:50 2004 +0000 + + XvMC Fix in case of failed init. Also Disabled XvMC for KM400 for now. + +commit 1b658bb04a605649fa26bf3bebc171595a9d17e9 +Author: Thomas Hellstrom +Date: Sat Mar 20 16:16:20 2004 +0000 + + XvMC lock fixes and drm irq cleanup + +commit 45c9b45f1a84f38dcfb53ee69cdc989d7af1173a +Author: Luc Verhaegen +Date: Fri Mar 19 19:25:51 2004 +0000 + + [devel-add_via_modeheaders] Move modeheader creation out to + via_modeheaders.c. extend #ifdef UNUSED across via_bios.c + +commit 99aa0ba91785a33f801378fa650a7b04ee483d6a +Author: Luc Verhaegen +Date: Fri Mar 19 19:11:02 2004 +0000 + + file via_modeheaders.c was initially added on branch + devel-branch-bios_cleanup. + +commit b2231e1b9caf207c7a832ebeb0ba30f25bffdab1 +Author: Thomas Hellstrom +Date: Thu Mar 18 18:39:14 2004 +0000 + + Added DRM Irq support (Terry Barnaby) + +commit 7b76e68c183adb3684916a7089b0e82985562360 +Author: Luc Verhaegen +Date: Tue Mar 16 18:14:05 2004 +0000 + + [devel-marginally_cleaner_v4l] Clean up V4L, a bit, still a mess though. + +commit c37148d206a435daa1916114f802b2b8b5e81b7b +Author: Luc Verhaegen +Date: Tue Mar 16 14:59:33 2004 +0000 + + [devel-manpage_remove_noDDCValue] Remove leftover from ddc patch. + +commit 60e2b138906abbbb593dc1a08e578ade4647e6e6 +Author: Thomas Hellstrom +Date: Sun Mar 14 17:30:02 2004 +0000 + + Add missing file from previous commit. + +commit 45590210792bd57498fb5d91c2e03f275723b1fa +Author: Thomas Hellstrom +Date: Sun Mar 14 17:26:09 2004 +0000 + + Further XvMC Updates. Multiple decoding, Subpictures and Bugfixes. + +commit 9726d449e61b9845a14e3655cd218e773256963e +Author: Thomas Hellstrom +Date: Sat Mar 13 19:07:05 2004 +0000 + + Prepared XvMC for subpictures and multiple decoding. Bugfixes. + 1. Updated the driver's view of the private part of the drm shared memory + area (SAREA) to be the same as the drm module's view. + 2. Extended that area to include mpeg2 decoder locking for multiple + decoding. + 3. Adjusted the maximal number of XvMC contexts, surfaces and buffers. + 4. Allowed non-root XvMC clients to write to the frame-buffer. Previously + the frame-buffer was exported read-only. This is for subpictures. + 5. Fixed the XvMC code to not leak drmContexts. + 6. Fixed the XvMC code to not attempt to free already freed surface private + info on X server exit or reset. (May cause segfault!) + 7. Fixed a test for malloc success in subpicture creation that was testing + on an unitialized pointer -> subpicture creation failed with BadAlloc. + +commit 81edc630b5c390c96906b55e242ab240134cf401 +Author: Luc Verhaegen +Date: Sat Mar 13 12:12:48 2004 +0000 + + Backport ifdeffing of ye olde CopyDataYUV420/422 + +commit a5090632dbef6a79841ab38fd9871438710dd50b +Author: Luc Verhaegen +Date: Fri Mar 12 20:06:28 2004 +0000 + + [devel-ddmpeg_import] Add Ivor's ddmpeg code (v4l and graphicInfo) + +commit f708600695abe0f91135736ce88152192a39f9a6 +Author: Luc Verhaegen +Date: Fri Mar 12 18:32:23 2004 +0000 + + file via_privioctl.h was initially added on branch + devel-branch-ddmpeg_import. + +commit beb1882b72309cd2ea9ebd3ddd4c3c262ca7f73e +Author: Thomas Hellstrom +Date: Thu Mar 11 21:25:46 2004 +0000 + + Added X server side of XvMC extension for HW mpeg2 decoding: New file + via_xvmc.c. Cleanup of Xv picture types. Added a dummy zero size Xv + picture type, FOURCC_VIA, that only updates the overlay when it is + displayed. An aid for the XvMC client lib. Redefined the FOURCC_VIA + code in via_swov.c and via_overlay.c to have the same meaning as + FOURCC_YV12, which is the HW mpeg2 format. + +commit 7382a56ddac6a9bd92b8fa30bd08492b9c97a2b1 +Author: Luc Verhaegen +Date: Mon Mar 8 19:29:08 2004 +0000 + + [devel-vbe_cleanup] Remove seperate int10. Clean up PreInit returns. Move + VBE OEM calls to via_vbe.c. Rename them from VIABIOS_ to ViaVBE. Add + VBE OEM calls added in CLEXF40037 to via_vbe.c (for possible future + use). (this is the vbe cleanup on xfree86 bugzilla + move to ViaVBE and + addition of new calls) + +commit 337d821863bf54892c78f85a3a74f4b85b7785e5 +Author: Thomas Hellstrom +Date: Fri Mar 5 08:52:34 2004 +0000 + + Aargh. Fixed another compilation bug if XFREE86_44 is not defined. /Thomas + +commit 74b0f4c073b13959cbbee1eec0ef0d539fc5f6a3 +Author: Thomas Hellstrom +Date: Thu Mar 4 23:14:12 2004 +0000 + + Fixed a compilation error introduced by previous commit. Fixed unresolved + drm symbols warning. + +commit 3db13481a99185b2e194edee6bd2065aa0d42222 +Author: Thomas Hellstrom +Date: Thu Mar 4 23:02:02 2004 +0000 + + Additional small cleanup that got lost in the previous commit /Thomas + +commit 954fe7066e722886a318f671777d8478710b6ddc +Author: Thomas Hellstrom +Date: Thu Mar 4 23:00:44 2004 +0000 + + Cleanup of offscreen memory allocation code. This was basically once + submitted to XFree86 but something went terribly wrong during + application of that patch. No new features. Just cleanup. /Thomas + Hellström + +commit e00bcebe64bab4a3dd9837b0fec715d76dd55e4f +Author: Thomas Hellstrom +Date: Thu Mar 4 22:42:05 2004 +0000 + + Added asm videoblit functions in via_memcpy.c and via_memcpy.h. Altered + file: via_video.c. This one times the fastest of a number of videoblit + imple- mentations and chooses and uses the fastest for Xv PutImage. + Currently the CPU capabilities detection is non-portable and therefore + this feature should not be submitted to XFree86 in it's current form. + The code was originally inspired by xine code which was very much + inspired from mplayer code which was very much inspired by linux kernel + code. These are released under GPL. However the present code is a + complete macro rewrite and shares no common codebase with xine / + mplayer / linux kernel implementations. /Thomas Hellström. + +commit 6d96dbc6db37c949ff2967c3e53e6363eb061da2 +Author: Luc Verhaegen +Date: Thu Mar 4 18:37:21 2004 +0000 + + Formalize xf86AllocateOffscreenLinear into offScreenLinear and add fallback + to this when drm alloc fails. (thomas) + +commit c72ee4bd4be981307b1dbe1066422969a985df5f +Author: Luc Verhaegen +Date: Thu Mar 4 18:16:34 2004 +0000 + + Increase PIXMAP_CACHE_SIZE (thomas) + +commit 477926d9a1314e118df3d153953af4a1e46f2f4d +Author: Thomas Hellstrom +Date: Tue Mar 2 22:06:17 2004 +0000 + + file via_xvmc.c was initially added on branch devel-branch-xvmc. + +commit 620a7a6e147e67ba525bd91767af3ee7e97967ca +Author: Luc Verhaegen +Date: Mon Mar 1 13:54:02 2004 +0000 + + Add (primary) DDC cleanup as already submitted to xfree86 bugzilla. + +commit 956f4482340e22d79f5421768d9ed806dd9b907d +Author: Luc Verhaegen +Date: Mon Mar 1 11:36:38 2004 +0000 + + Initial import of the xfree86 module, contains the via driver as found in + xfree86 4.4.0 + +commit 4e38918912a1fac9b8b0a55de227ddfc857dbee2 +Author: Luc Verhaegen +Date: Mon Mar 1 11:36:38 2004 +0000 + + Initial revision + +commit 1f0f0270e7a4385a04b66d4fb9f722b8ff3831ba +Author: Luc Verhaegen +Date: Mon Mar 1 11:36:09 2004 +0000 + + Initial import of the xfree86 module, contains the via driver as found in + xfree86 4.4.0 + +commit 60f73e4688d5125395d1ccff75005043bc6c8de0 +Author: Luc Verhaegen +Date: Mon Mar 1 11:36:09 2004 +0000 + + Initial revision + diff --git a/configure.ac b/configure.ac index 8516789..12503eb 100644 --- a/configure.ac +++ b/configure.ac @@ -22,7 +22,7 @@ AC_PREREQ(2.57) AC_INIT([xf86-video-unichrome], - 0.2.0.0, + 0.2.6.0, [https://unichrome.sf.net/], xf86-video-unichrome) diff --git a/src/via_driver.h b/src/via_driver.h index 9fd1144..11ca60c 100644 --- a/src/via_driver.h +++ b/src/via_driver.h @@ -69,7 +69,7 @@ #define DRIVER_NAME "via" #define VERSION_MAJOR 0 #define VERSION_MINOR 2 -#define PATCHLEVEL 0 +#define PATCHLEVEL 6 #define VIA_VERSION ((VERSION_MAJOR<<24) | (VERSION_MINOR<<16) | PATCHLEVEL) #define VIA_MAX_ACCEL_X (2047) -- cgit v1.2.3