ChangeLog revision d6c0b56e
1commit a04f4015d6afef20c2b79e2779f6555836ee2b07 2Author: Michel Dänzer <michel.daenzer@amd.com> 3Date: Thu Apr 7 16:47:25 2016 +0900 4 5 Bump version for 1.1.0 release 6 7commit aed1c17c43b2c0c983f6fc0973a5224d0faf32d9 8Author: Michel Dänzer <michel.daenzer@amd.com> 9Date: Mon Apr 4 18:28:02 2016 +0900 10 11 glamor: Force GPU rendering to/from pixmaps created via DRI3 12 13 Fixes crash when running DRI3 clients with ShadowPrimary enabled. 14 15 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94799 16 17 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 18 19commit faf9d720b7d650f5f1ea657a874d08eac3972e60 20Author: Michel Dänzer <michel.daenzer@amd.com> 21Date: Fri Apr 1 16:09:51 2016 +0900 22 23 Update manpage entry for Option "TearFree" 24 25 It's now effective for rotation as well. 26 27 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 28 29commit 5ba95c3abeb8df82aa8d33a47596eae6403ea7af 30Author: Michel Dänzer <michel.daenzer@amd.com> 31Date: Fri Apr 1 15:29:26 2016 +0900 32 33 Identify DRM event queue entries by sequence number instead of by pointer 34 35 If the memory for an entry was allocated at the same address as that for 36 a previously cancelled entry, the handler could theoretically be called 37 prematurely, triggered by the DRM event which was submitted for the 38 cancelled entry. 39 40 (Ported from radeon commit 4693b1bd5b5c381e8b7b68a6f7f0c6696d6a68df) 41 42 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 43 44commit 8ecfa69b5a833bd4c39e773a6acfd7eef9144d13 45Author: Michel Dänzer <michel.daenzer@amd.com> 46Date: Wed Mar 30 18:33:00 2016 +0900 47 48 DRI3: Refuse to open DRM file descriptor for ssh clients 49 50 Fixes hangs when attempting to use DRI3 on display connections forwarded 51 via SSH. 52 53 Don't do this for Xorg > 1.18.99.1 since the corresponding xserver 54 change has landed in Git master. 55 56 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93261 57 58 (Ported from radeon commit 0b3aac1de9db42bfca545fa331e4985836682ec7) 59 60 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 61 62commit b2a2e114eec0967f7b67f030fbab8983cf980489 63Author: Michel Dänzer <michel.daenzer@amd.com> 64Date: Fri Mar 25 11:55:34 2016 +0900 65 66 Revert "Use render node for DRI3 if available" 67 68 This reverts commit ea558e645786b08d75307716036045170e97b43e. 69 70 It broke VDPAU<->GL interop with DRI3 enabled, because the Gallium VDPAU 71 code doesn't support DRI3 yet. We can consider re-enabling this once 72 there is a Mesa release where the Gallium VDPAU code supports DRI3. 73 74 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94675 75 76 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 77 78commit e31a2d668a1b5ebaf75d423c8123cbc8e0dcbae9 79Author: Flora Cui <Flora.Cui@amd.com> 80Date: Wed Nov 18 16:44:13 2015 +0800 81 82 add polaris10 pci id 83 84 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 85 Signed-off-by: Flora Cui <Flora.Cui@amd.com> 86 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 87 88commit 6e09b8deb77f76b9bb7d393cc1ad924ebba62eff 89Author: Flora Cui <Flora.Cui@amd.com> 90Date: Thu Nov 5 14:16:39 2015 +0800 91 92 add polaris11 pci id 93 94 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 95 Signed-off-by: Flora Cui <Flora.Cui@amd.com> 96 Reviewed-By: Jammy Zhou <Jammy.Zhou@amd.com> 97 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 98 99commit 7d32c43fff4c8df32cce150223094f793e036cf3 100Author: Alex Deucher <alexander.deucher@amd.com> 101Date: Wed Oct 28 17:28:23 2015 -0400 102 103 add Polaris chip families 104 105 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 106 Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 107 108commit fbf9ae18cd241b8b78936aa30441e5fbfd9ba1c5 109Author: Michel Dänzer <michel.daenzer@amd.com> 110Date: Thu Mar 24 19:05:15 2016 +0900 111 112 Require xserver 1.9 or newer 113 114 1.9.0 was released in August 2010. 115 116 We were already unintentionally relying on things not available in 1.8 117 for at least a year, and nobody has complained. 118 119 (Ported from radeon commit e592f32f8b5f5873fcc18b10a69dd5e4ccf11073) 120 121 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 122 123commit 912db5fbbc6b9b1121c8a03168cb4bd870474376 124Author: Michel Dänzer <michel.daenzer@amd.com> 125Date: Thu Mar 24 18:59:05 2016 +0900 126 127 Fix build against older versions of xserver 128 129 Also slightly clean up the error handling in amdgpu_scanout_do_update. 130 131 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94614 132 133 (Ported from radeon commit bde466e5d44cad64b4e4eceaa5de80fdbf86356e) 134 135 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 136 137commit 3fb6280ab3b104b02841c7cab8ed68c1d463c834 138Author: Michel Dänzer <michel.daenzer@amd.com> 139Date: Thu Mar 24 18:56:44 2016 +0900 140 141 DRI3 only works with acceleration 142 143 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94214 144 145 (Ported from radeon commit d21ac4669a8b2cdd4eec5e5a94d1950b7423b8b5) 146 147 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 148 149commit 3177fe817a5f2de4ed10860866a0dd6d6c6ba816 150Author: Michel Dänzer <michel.daenzer@amd.com> 151Date: Thu Mar 24 18:51:59 2016 +0900 152 153 Check for xf86CursorResetCursor 154 155 If it's available, Xorg calls it on each mode configuration change. It 156 does what xf86_reload_cursors does (and more), so we don't need to call 157 the latter anymore. 158 159 (Ported from radeon commit d670c5c9851b4eff21c845d26c7d7e4eb5ee0fa9) 160 161 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 162 163commit a3dfce7b24e1ea01c1aa62926025a545312cbe13 164Author: Michel Dänzer <michel.daenzer@amd.com> 165Date: Thu Mar 24 18:45:46 2016 +0900 166 167 Don't try DRI2/Present flipping while the HW cursor can't be used 168 169 Flipping doesn't interact correctly with SW cursor: A flip makes the SW 170 cursor disappear. It will only appear again when the cursor is moved, 171 but it will be surrounded by corruption, because the SW cursor code 172 will restore stale screen contents at the old cursor location before 173 drawing the cursor at the new location. 174 175 (Ported from radeon commit 7f3d0780ca65a90117c2a61362dbc0899bd9c0b0) 176 177 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 178 179commit ba9be8f32f0321689133e17c1681809dec8c6cf1 180Author: Michel Dänzer <michel.daenzer@amd.com> 181Date: Thu Mar 24 18:44:30 2016 +0900 182 183 Factor out HW cursor checking code into drmmode_can_use_hw_cursor 184 185 And add a check for RandR 1.4 multihead. 186 187 (Ported from radeon commit 3de480e83c0a1824838d662d6d67c9fe85277298) 188 189 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 190 191commit 4a60b4b1851a3cbc2d8ad9048d68eeb6947cf132 192Author: Michel Dänzer <michel.daenzer@amd.com> 193Date: Thu Mar 24 12:03:38 2016 +0900 194 195 Call AMDGPUBlockHandler_KMS before setting initial modes 196 197 Doing it the other way around meant that there was still a possibility 198 for the front buffer contents to be uninitialized when they start being 199 scanned out. 200 201 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 202 203commit 37bd79652a8ec612b94a1863e8c580b1cfaf3960 204Author: Michel Dänzer <michel.daenzer@amd.com> 205Date: Fri Mar 18 18:51:00 2016 +0900 206 207 present: Return rotated CRTCs from amdgpu_present_get_crtc 208 209 Sync-to-vblank works fine with rotation. We're still checking for 210 rotation in amdgpu_present_check_flip. 211 212 Returning NULL from here resulted in the xserver present code falling 213 back to the fake CRTC running at 1 fps. 214 215 (Ported from radeon commit a03271de5ecdaa7790d1316e993c4450b91fe936) 216 217 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 218 219commit 6b930fb3285dea4a6440e31099c96f08da508d49 220Author: Michel Dänzer <michel.daenzer@amd.com> 221Date: Fri Mar 18 18:47:10 2016 +0900 222 223 present: Clear drmmode->fb_id before calling set_mode_major for unflip 224 225 Without this, drmmode_set_mode_major may just re-set the FB for the 226 last flipped BO, in which case the display will probably freeze. 227 228 Reproduction recipe: Enable rotation while a fullscreen client is 229 flipping. 230 231 (Ported from radeon commit 40191d82370eb7e58bd34c44966cbf44c3703229) 232 233 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 234 235commit 6889e091442b6ba1b9351e72bd067425e87d96e9 236Author: Michel Dänzer <michel.daenzer@amd.com> 237Date: Fri Mar 18 18:18:04 2016 +0900 238 239 Make Option "TearFree" effective for rotated/reflected outputs as well 240 241 Support varies by xserver version: 242 243 < 1.12: No support for the driver handling rotation/reflection 244 1.12-1.15: Support for driver handling rotation/reflection, but there's 245 a bug preventing the HW cursor from being visible everywhere 246 it should be on rotated outputs, so we can only support 247 TearFree for reflection. 248 >= 1.16: While the bug above is still there (fixes pending review), 249 the driver can force SW cursor for rotated outputs, so we 250 can support TearFree for rotation as well. 251 252 (Ported from radeon commit 798c4fd16d339b1ad5fd729cc884be084c60e38b) 253 254 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 255 256commit da4e0c66fcbcf63143372720e3d606a462332e3a 257Author: Michel Dänzer <michel.daenzer@amd.com> 258Date: Fri Mar 18 18:15:34 2016 +0900 259 260 Consolidate pScreen usage in drmmode_set_mode_major 261 262 We were already relying on pScrn->pScreen being non-NULL in some cases, 263 which is supposedly always true ever since this function is no longer 264 getting called from ScreenInit. 265 266 (Ported from radeon commit eb611a2e4ecce7a1ab85fd72b9b78e3269311dd5) 267 268 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 269 270commit 0bbf09dd7ef54133b3e534becb3ba15c0cf3eed2 271Author: Michel Dänzer <michel.daenzer@amd.com> 272Date: Fri Mar 18 18:14:28 2016 +0900 273 274 Remove check for XF86_CRTC_VERSION 3 275 276 We require xserver >= 1.8, which was already at version 3. 277 278 (Ported from radeon commit 06602171386e538081c298645fb7ca1a70fe80cc) 279 280 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 281 282commit 3485ca0051a224d00135d4024a97a6c4e85a9644 283Author: Michel Dänzer <michel.daenzer@amd.com> 284Date: Fri Mar 18 18:07:07 2016 +0900 285 286 Deal with modesets and page flips crossing on a CRTC 287 288 If we set a mode while a flip is pending, the kernel driver may program 289 the flip to the hardware after the modeset. If that happens, the hardware 290 will display the BO from the flip, whereas we will assume it displays the 291 BO from the modeset. In other words, the display will most likely freeze, 292 at least until another modeset. 293 294 Prevent this condition by waiting for a pending flip to finish before 295 setting a mode. 296 297 Fixes display freezing when setting rotation or a transform with 298 TearFree enabled. 299 300 (Ported from radeon commit a88985f5d1e39caca49ceb65678aaa9cb622a0d2) 301 302 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 303 304commit b9d00fa7aaf946d985897380bfa42faafbf1b3fb 305Author: Michel Dänzer <michel.daenzer@amd.com> 306Date: Fri Mar 18 17:18:00 2016 +0900 307 308 Make DRM event queue xf86CrtcPtr based instead of ScrnInfoPtr based 309 310 This allows for a minor simplification of the code. 311 312 (Ported from radeon commit f5d968cbba3c9b7ec202161f2157d8d64778c817) 313 314 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 315 316commit e0ed26151bfeadf309da53d001751c0a014dbd24 317Author: Michel Dänzer <michel.daenzer@amd.com> 318Date: Fri Mar 18 17:11:47 2016 +0900 319 320 Remove amdgpu_scanout_flip_handler 321 322 No longer necessary now that amdgpu_drm_queue_handler can handle 323 e->handler == NULL. 324 325 (Ported from radeon commit d5dbb07db22d5420c81dfebc060f0dd86e7b8a20) 326 327 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 328 329commit acd5da56f502d6ad115501e77bce06fe72b1895c 330Author: Michel Dänzer <michel.daenzer@amd.com> 331Date: Fri Mar 18 17:14:49 2016 +0900 332 333 DRI2: Also clear dri2_flipping when client disconnects before event 334 335 Fixes the following problem: 336 337 With DRI3 enabled, run glxgears with LIBGL_DRI3_DISABLE=1, make it 338 fullscreen and press Escape while it's still fullscreen. This could 339 result in dri2_flipping not getting cleared, spuriously preventing apps 340 using DRI3 from flipping. 341 342 (Ported from radeon commit e87365117acbd80b7d80fbb5eb30890ef7153291) 343 344 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 345 346commit a58bfa98208cc092014d3f36a08714eb1e0d8814 347Author: Michel Dänzer <michel.daenzer@amd.com> 348Date: Fri Mar 18 17:07:47 2016 +0900 349 350 drm_queue: Don't abort events immediately from amdgpu_drm_abort_client 351 352 Keep them around until the DRM event arrives, but then call the abort 353 functions instead of the handler functions. 354 355 This is a prerequisite for the following fix. 356 357 (Ported from radeon commit 3989766edde85d1abe7024577b98fc9b007bc02a) 358 359 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 360 361commit e4888df6e32bb817bf0d6166a22b19c14e189a84 362Author: Michel Dänzer <michel.daenzer@amd.com> 363Date: Fri Mar 18 17:04:10 2016 +0900 364 365 Fix RandR CRTC transforms 366 367 Currently, Xorg will only transform the cursor as of the first time the 368 cursor image changes after a transform is set. 369 370 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=80678 371 372 (Ported from radeon commit 9483a3d777919b224f70c3b4d01e4b320a57db31) 373 374 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 375 376commit 43af92ede0968f2108f9562aa4c2c861ac703617 377Author: Michel Dänzer <michel.daenzer@amd.com> 378Date: Fri Mar 18 16:58:07 2016 +0900 379 380 Build RandR 1.4 provider name from chipset name and bus ID 381 382 Instead of just "amdgpu", it's now e.g. "TONGA @ pci:0000:01:00.0". 383 384 (Ported from radeon commit c7cf00487cd6d4a5d0f39d5b92ff04f6420d6a32) 385 386 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 387 388commit 5ec1797a2858d693d18d21326e2307d71555e1db 389Author: Michel Dänzer <michel.daenzer@amd.com> 390Date: Wed Feb 24 17:33:49 2016 +0900 391 392 DRI2: Use amdgpu_pixmap_get_handle 393 394 Now we can share pixmaps with no struct amdgpu_buffer via DRI2. 395 396 Fixes VDPAU video playback freezing when using an OpenGL compositor with 397 DRI3 enabled and mpv VAAPI hardware decoding with OpenGL output. 398 399 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89755 400 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93804 401 402 (ported from radeon commit f8b0f23e9f4af9f9097ee5e72d53b45173163c41) 403 404 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 405 406commit df60c635e1e632233de9dd4b01d63c2b963003f8 407Author: Michel Dänzer <michel.daenzer@amd.com> 408Date: Wed Feb 24 17:06:43 2016 +0900 409 410 glamor: Avoid generating GEM flink names for BOs shared via DRI3 (v2) 411 412 We can't create our own struct amdgpu_buffer representation in this case 413 because destroying that would make the GEM handle inaccessible to glamor 414 as well. So just get the handle directly via dma-buf. 415 416 (ported from radeon commit 391900a670addec39515f924265bfa9f8bfa9ec0, 417 extended to cache BO handles in the private for non-DRI3 pixmaps as 418 well) 419 420 v2: Swap whole pixmap privates instead of just BOs in 421 amdgpu_dri2_exchange_buffers to avoid invalidating cached BO handles 422 423 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 424 425commit e463b849f3e9d7b69e64a65619a22e00e78d297b 426Author: Michel Dänzer <michel.daenzer@amd.com> 427Date: Tue Feb 23 18:10:29 2016 +0900 428 429 Make amdgpu_do_pageflip take a pixmap instead of a BO 430 431 (inspired by radeon commit 7b4fc4a677d252d01c2bf80d162bc35814059eaa) 432 433 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 434 435commit 1ee341f9d909f3b7ba2984fc912dabdb98c34b19 436Author: Michel Dänzer <michel.daenzer@amd.com> 437Date: Tue Feb 23 18:42:19 2016 +0900 438 439 Add amdgpu_pixmap_get_handle helper 440 441 (inspired by radeon commits dfad91fffb5bd013785223b42d78886df839eacf 442 and ccbda955ebae1d457d35293833f12791e0f9fb0b) 443 444 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 445 446commit a36bbfd98b96426bbe0be3923c64da7ec0e565d0 447Author: Michel Dänzer <michel.daenzer@amd.com> 448Date: Mon Feb 15 18:41:51 2016 +0900 449 450 HAS_DIRTYTRACKING_ROTATION also supports multiple CRTCs 451 452 (ported from radeon commit ff9a6b6f079a8419f4e6fadfee778060618bf735) 453 454 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 455 456commit a37746ffceaed83e48e48fb05439be7e020dd2ea 457Author: Michel Dänzer <michel.daenzer@amd.com> 458Date: Mon Feb 15 18:35:54 2016 +0900 459 460 Load fb module before glamoregl/shadow modules 461 462 Fixes unresolved symbols on some systems. 463 464 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93105 465 (ported from radeon commit 78fbca095ae9887a2d3de48bb07975e2d1126e68) 466 467 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 468 469commit 59c0a6807110eca829c6708e16585a38f39a5c17 470Author: Michel Dänzer <michel.daenzer@amd.com> 471Date: Mon Feb 15 18:28:13 2016 +0900 472 473 Don't advertise any PRIME offloading capabilities without acceleration 474 475 Acceleration is required even for display offloading. Trying to enable 476 display offloading without acceleration resulted in a crash. 477 478 (ported from radeon commit b19417e2fddf4df725951aea5ad5e9558338f59e) 479 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 480 481commit a3eac85d812ecc605436e6bd5b9ee7ebf307e3d3 482Author: Michel Dänzer <michel.daenzer@amd.com> 483Date: Tue Jan 26 16:12:28 2016 +0900 484 485 Only map front buffer if glamor acceleration is disabled (v2) 486 487 Otherwise the front buffer may not be accessible by the CPU, because Mesa 488 sets the AMDGPU_GEM_CREATE_NO_CPU_ACCESS flag for tiled buffers, because 489 accessing tiled buffers with the CPU makes little sense. 490 491 v2: Also handle Option "AccelMethod" "none" 492 493 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 494 495commit 2fcb7dadd3c71cd405cbbaafc777697538ca9c29 496Author: jimqu <Jim.Qu@amd.com> 497Date: Mon Jan 25 09:47:00 2016 +0800 498 499 glamor: Return NullPixmap on failure to create shareable pixmap 500 501 If we were asked to create a shareable pixmap, it doesn't make sense 502 to return a pixmap which isn't shareable. Doing so caused trouble down 503 the line such as a crash with older versions of glamor when trying to 504 use GLX pixmaps of bpp < 32 via DRI2. 505 506 Signed-off-by: JimQu <jim.qu@amd.com> 507 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 508 509commit 5269a2228bff6023c1a7f3e8534027e1d7addc25 510Author: jimqu <Jim.Qu@amd.com> 511Date: Mon Jan 25 10:12:02 2016 +0800 512 513 Move amdgpu_glamor_destroy_pixmap before amdgpu_glamor_create_pixmap 514 515 The next commit will call the former from the latter. No functional 516 change. 517 518 Signed-off-by: JimQu <jim.qu@amd.com> 519 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 520 521commit 54c959c163288caa87f612911b70df73f87d29d6 522Author: Tom St Denis <tom.stdenis@amd.com> 523Date: Wed Jan 20 09:37:36 2016 -0500 524 525 Move memset() after variable declarations 526 527 To make the code more "C" like move the function calls 528 after the variable declarations. 529 530 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 531 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 532 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 533 534commit 8853b07ae8169c409740c40d45cd335bd608f2a7 535Author: Michel Dänzer <michel.daenzer@amd.com> 536Date: Tue Jan 19 17:35:11 2016 +0900 537 538 Set the RandR primary output on startup if Xorg hasn't 539 540 Fixes xrandr (XRRGetOutputPrimary) not reporting any output as primary 541 after startup. 542 543 (Ported from radeon commit b16856b25086ffb27365ac2249b8da921066ce62) 544 545 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 546 547commit bd5c65daceaf633c36fcec86ff061df10c364bc0 548Author: Michel Dänzer <michel.daenzer@amd.com> 549Date: Thu Jan 7 15:53:41 2016 +0900 550 551 Only call amdgpu_bus_id once in each probe path (v2) 552 553 Instead of up to twice as before. 554 555 v2: Remove free(busIdString) call from amdgpu_kernel_mode_enabled, the 556 bus ID string is now managed by its callers. 557 558 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v1) 559 560commit 6e42c58375a4c3229da93c27bbd104af145c6163 561Author: Michel Dänzer <michel.daenzer@amd.com> 562Date: Thu Jan 7 15:57:38 2016 +0900 563 564 Remove pci_dev test from amdgpu_get_scrninfo 565 566 The pci_dev parameter can never be NULL since we only support KMS. 567 568 Reported-by: Tom St Denis <tom.stdenis@amd.com> 569 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 570 571commit 8e09180798a06af5afa030d754938e4ca06e272f 572Author: Michel Dänzer <michel.daenzer@amd.com> 573Date: Thu Jan 7 15:35:35 2016 +0900 574 575 Re-use PCI bus ID code from kernel_open_fd in kernel_mode_enabled 576 577 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 578 579commit 4eb9cedca080b30c57ded349a397620ee7d0cd46 580Author: Mykola Lysenko <Mykola.Lysenko@amd.com> 581Date: Wed Jan 13 14:03:55 2016 +0800 582 583 Initialize drmmode_crtc dpms_mode to DPMSModeOff 584 585 This disables query of disabled pipes for drmWaitVBlank on X start 586 587 Signed-off-by: Mykola Lysenko <Mykola.Lysenko@amd.com> 588 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 589 590commit 1d0b0c1794e65e581a48aa9fb19679d928d82a5d 591Author: Michel Dänzer <michel.daenzer@amd.com> 592Date: Thu Dec 10 18:08:12 2015 +0900 593 594 sync: Check if miSyncShmScreenInit symbol is resolved at runtime 595 596 It may be disabled in the Xorg build, either explicitly or because the 597 xshmfence library isn't available. 598 599 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 600 601commit f4107f67f147e2500582fc36cf0f0f76bc1ef098 602Author: Mykola Lysenko <Mykola.Lysenko@amd.com> 603Date: Wed Dec 23 11:58:47 2015 -0500 604 605 Check for NULL koutput in drmmode_output_dpms 606 607 This situation happens whit start of usage of DRM DP MST framework, 608 when connectors created and destroyed dynamically. 609 610 Signed-off-by: Mykola Lysenko <Mykola.Lysenko@amd.com> 611 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 612 613commit ea558e645786b08d75307716036045170e97b43e 614Author: Jammy Zhou <Jammy.Zhou@amd.com> 615Date: Fri Nov 20 17:03:05 2015 +0800 616 617 Use render node for DRI3 if available 618 619 Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com> 620 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 621 622commit 43c2dc1aab682d5b6ad49d24983d6382c4f305bb 623Author: Michel Dänzer <michel.daenzer@amd.com> 624Date: Thu Nov 19 17:05:05 2015 +0900 625 626 glamor: Deal with glamor_egl_destroy_textured_pixmap being removed 627 628 When it's not available, it's safe to call down to the glamor 629 DestroyPixmap hook instead. 630 631 (ported from radeon commit 10b7c3def58bb34acc38f076bc230e25b454ab79) 632 633 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 634 635commit 84cab5738a315e9825bd0864c4f0fc5b03eb81a1 636Author: Michel Dänzer <michel.daenzer@amd.com> 637Date: Thu Nov 19 16:44:22 2015 +0900 638 639 glamor: Restore all ScreenRec hooks during CloseScreen 640 641 (ported from radeon commit 535e5438b2c32f774b9c8c27ee0289b4749548ef) 642 643 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 644 645commit a00c050c2e5667ed815c51979a3cadb5146136ff 646Author: Michel Dänzer <michel.daenzer@amd.com> 647Date: Thu Nov 19 17:55:53 2015 +0900 648 649 Post 1.0.0 release version bump 650 651commit 755e6ff2337cf615e3ba0854ccd533baec7144db 652Author: Michel Dänzer <michel.daenzer@amd.com> 653Date: Thu Nov 19 17:28:19 2015 +0900 654 655 Bump version for 1.0.0 release 656 657commit 49c7d2be99aaf6d040e553065bdc461ce8d4769a 658Author: Michel Dänzer <michel.daenzer@amd.com> 659Date: Thu Nov 19 17:14:54 2015 +0900 660 661 Add amdgpu_pixmap.h to src/Makefile.am's EXTRA_DIST 662 663 Fixes make distcheck. 664 665commit d069ec5d27f5c8d2ab17b759b85293ef4113acf3 666Author: Stephen Chandler Paul <cpaul@redhat.com> 667Date: Wed Nov 11 18:10:55 2015 +0900 668 669 Handle failures in setting a CRTC to a DRM mode properly 670 671 This fixes a bug where running the card out of PPLL's when hotplugging 672 another monitor would result in all of the displays going blank and 673 failing to work properly until X was restarted or the user switched to 674 another VT. 675 676 [Michel Dänzer: Pass errno instead of -ret to strerror()] 677 678 Signed-off-by: Stephen Chandler Paul <cpaul@redhat.com> 679 (ported from radeon commit 7186a8713ba004de4991f21c1a9fc4abc62aeff4) 680 681 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 682 683commit c8bddcf6c97b1338be3715f1fc5e0b17ce71c195 684Author: Michel Dänzer <michel.daenzer@amd.com> 685Date: Wed Nov 11 18:09:59 2015 +0900 686 687 Call xf86CrtcRotate from initial drmmode_set_desired_modes call 688 689 Fixes various problems when rotation is specified in xorg.conf. 690 691 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=92475 692 693 (ported from radeon commit 548e97b3b7d1e94075a54ca2bb4eb683025098a7) 694 695 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 696 697commit 12815156f38ce3357f03901a78402db834577d11 698Author: Emil Velikov <emil.l.velikov@gmail.com> 699Date: Wed Nov 11 18:04:01 2015 +0900 700 701 Do not link amdgpu_drv.so against libpciaccess 702 703 Not used directly. 704 705 Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com> 706 (ported from radeon commit fcb32231a38f9461d12720cbf72f63502197a711) 707 708 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 709 710commit a02982b0ae0b79d2f183a1628edc05cafed8703a 711Author: Michel Dänzer <michel.daenzer@amd.com> 712Date: Wed Nov 11 17:59:14 2015 +0900 713 714 Skip disabled CRTCs in amdgpu_scanout_(do_)update 715 716 The vblank / page flip ioctls don't work as expected for a disabled CRTC. 717 718 (ported from radeon commit acc11877423ecd81a6e0a7f38466f80e43efee20) 719 720 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 721 722commit 0ddd20600d0046afd17aa47ffebe86dfd91a2215 723Author: Michel Dänzer <michel.daenzer@amd.com> 724Date: Wed Nov 11 17:44:16 2015 +0900 725 726 Prefer drmModeSetCursor2 over drmModeSetCursor 727 728 The former includes information about the position of the hotspot within 729 the cursor image. 730 731 Copied from xf86-video-modesetting. 732 733 (ported from radeon commit c9f8f642fd495937400618a4fc25ecae3f8888fc) 734 735 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 736 737commit 83a47c0ebe17caa79d12a8b2f94b59cc945452f5 738Author: Michel Dänzer <michel.daenzer@amd.com> 739Date: Wed Nov 11 17:37:54 2015 +0900 740 741 PRIME: Don't advertise offload capabilities when acceleration is disabled 742 743 Xorg tends to crash if the user tries to actually use the offload 744 capabilities with acceleration disabled. 745 746 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57200 747 (ported from radeon commit c74de9fec13fac2c836bb2a07ae6f90e1d61e667) 748 749 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 750 751commit 560b7fe6dc66405762020f00e9a05918a36f3a17 752Author: Michel Dänzer <michel.daenzer@amd.com> 753Date: Wed Nov 11 17:31:34 2015 +0900 754 755 Rename Option "NoAccel" to "Accel" 756 757 Removes the need for a double negation when forcing acceleration on. 758 759 Note that this change is backwards compatible, as the option parser 760 automagically handles the 'No' prefix. 761 762 (ported from radeon commit cc615d06db0332fc6e673b55632bcc7bf957b44b) 763 764 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 765 766commit ad77ad32c4a723447d3191d527cfa6de9f54d7ce 767Author: Adam Jackson <ajax@redhat.com> 768Date: Wed Nov 11 17:20:21 2015 +0900 769 770 Use own thunk function instead of shadowUpdatePackedWeak 771 772 I plan to delete the Weak functions from a future server. 773 774 Signed-off-by: Adam Jackson <ajax@redhat.com> 775 (ported from radeon commit 851b2cf8714618843725f6d067915375485ade9d) 776 777 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 778 779commit f5ccea99c03b62acf3a25984aba617c665d80b7c 780Author: Michel Dänzer <michel.daenzer@amd.com> 781Date: Wed Nov 11 17:16:58 2015 +0900 782 783 dri2: Handle PRIME for source buffer as well in amdgpu_dri2_copy_region2 784 785 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=77810 786 787 (ported from radeon commit c84230d686c078aac1dc98d82153f8b02521b2e1) 788 789 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 790 791commit 92e7c93d2f9c3036da1a17d7fccccb6f9e9eaa3d 792Author: Michel Dänzer <michel.daenzer@amd.com> 793Date: Mon Nov 2 18:29:24 2015 +0900 794 795 Move scrn/info declaration inside USE_GLAMOR in amdgpu_dri3_fd_from_pixmap 796 797 Fixes warning when building with --disable-glamor: 798 799 ../../src/amdgpu_dri3.c: In function 'amdgpu_dri3_fd_from_pixmap': 800 ../../src/amdgpu_dri3.c:135:16: warning: unused variable 'info' [-Wunused-variable] 801 AMDGPUInfoPtr info = AMDGPUPTR(scrn); 802 ^ 803 804 Reported-by: Jammy Zhou <Jammy.Zhou@amd.com> 805 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 806 807commit c9bd1399a13cea2e1331af2c826ca054b88db071 808Author: Michel Dänzer <michel.daenzer@amd.com> 809Date: Mon Nov 2 18:21:50 2015 +0900 810 811 Call AMDGPUFreeRec from AMDGPUFreeScreen_KMS even if info == NULL 812 813 It's safe now. 814 815 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 816 817commit fb8444e731765588c0ff1e9053c1c7b73f5f0907 818Author: Michel Dänzer <michel.daenzer@amd.com> 819Date: Mon Nov 2 18:20:41 2015 +0900 820 821 Don't use AMDGPUEntPriv in AMDGPUFreeRec 822 823 It crashes if info == NULL. 824 825 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 826 827commit 8e7ee03f55c2f3874f6e84daeb5700f8b8037a51 828Author: Michel Dänzer <michel.daenzer@amd.com> 829Date: Wed Oct 28 17:53:27 2015 +0900 830 831 Remove amdgpu_reference_drm_fd 832 833 Increase pAMDGPUEnt->fd_ref in the probe code instead when we're reusing 834 the existing fd. 835 836 The previous reference counting was imbalanced, so pAMDGPUEnt->fd_ref 837 could never go to 0. 838 839 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 840 841commit 6bab8fabb37eb131e131ce59446c214ded28f779 842Author: Michel Dänzer <michel.daenzer@amd.com> 843Date: Wed Oct 28 17:44:09 2015 +0900 844 845 Remove info->dri2.drm_fd and info->drmmode->fd 846 847 Use pAMDGPUEnt->fd everywhere instead. 848 849 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 850 851commit 0530e39cc6b7340163e7f6bb6d82719d102ee6e9 852Author: Jammy Zhou <jammy.zhou@amd.com> 853Date: Thu Oct 29 17:08:01 2015 +0900 854 855 Pass struct pci_device *pci_dev directly to amdgpu_get_scrninfo 856 857 Instead of throwing away the type information by passing it as a void*. 858 859 Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> 860 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 861 862commit edf72afee3a25eae9827b4de3a013b541b78e213 863Author: Jammy Zhou <Jammy.Zhou@amd.com> 864Date: Wed Oct 28 21:24:29 2015 +0800 865 866 Fix crash in PCI probe path (v4) 867 868 The crash is caused by the NULL value returned by AMDGPUPTR(pScrn), 869 because the driverPrivate is not allocated yet in PciProbe phase, 870 and it is usually done in the PreInit phase. 871 872 Use pAMDGPUEnt->fd instead of info->dri2.drm_fd to avoid AMDGPUInfoPtr 873 related code in amdgpu_open_drm_master, so that the crash can be fixed. 874 875 v4: (md) Remove unused parameter entity_num, split out logically 876 separate changes 877 v3: some more cleanup 878 v2: switch to pAMDGPUEnt->fd, and update the commit message 879 880 Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com> 881 Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> 882 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v3) 883 884commit cef725121eb0e56aa54d9c4665e36047373f4db7 885Author: Michel Dänzer <michel.daenzer@amd.com> 886Date: Wed Oct 28 17:56:13 2015 +0900 887 888 Remove dead code from probe paths 889 890 amdgpu_get_scrninfo allocates the memory pointed to by pAMDGPUEnt just 891 before it calls amdgpu_open_drm_master, so pAMDGPUEnt->fd is always 0 892 in the latter. 893 894 Also, no need to clear pAMDGPUEnt->fd just before freeing the memory 895 it's stored in. 896 897 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 898 899commit 3b0a3c89b53b3ebe21a9d703a4dbff6e57c65a57 900Author: Samuel Li <samuel.li@amd.com> 901Date: Thu Oct 22 12:50:21 2015 -0400 902 903 Add Stoney support 904 905 (agd): rebase 906 907 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 908 Signed-off-by: Samuel Li <samuel.li@amd.com> 909 910commit 9c8b7ebe15eec7abd5dc10ad6ccecbc57225494a 911Author: Michel Dänzer <michel.daenzer@amd.com> 912Date: Wed Oct 21 17:18:44 2015 +0900 913 914 Revert "Handle RandR CRTC transforms properly" 915 916 This reverts commit 175251645fec1a3d19f498e1cd1e655374c67801. 917 918 I accidentally pushed this patch. 919 920commit 0a6ba4bf50128464a30951721b0c72e748fb89bc 921Author: Darren Powell <darren.powell@amd.com> 922Date: Tue Oct 20 16:56:54 2015 -0400 923 924 Add Option "TearFree" to manpage 925 926 This was missed in commit c57da33308a81fa575179238a0415abcb8b34908. 927 928 Signed-off-by: Darren Powell <darren.powell@amd.com> 929 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 930 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 931 932commit 175251645fec1a3d19f498e1cd1e655374c67801 933Author: Michel Dänzer <michel.daenzer@amd.com> 934Date: Thu Oct 15 16:35:51 2015 +0900 935 936 Handle RandR CRTC transforms properly 937 938commit 6000aef4e2f0a121b94023484406fb6f04688f74 939Author: Tom St Denis <tom.stdenis@amd.com> 940Date: Wed Oct 14 13:25:59 2015 -0400 941 942 Clean up amdgpu_dri2_create_buffer2() 943 944 Remove the depth_pixmap variable from the function and clear 945 out any dead/odd behaviour that results. 946 947 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 948 949commit 21e72fb2418b5cc7fc849a9cf951186e209036b0 950Author: Michel Dänzer <michel.daenzer@amd.com> 951Date: Fri Oct 9 18:38:47 2015 +0900 952 953 Properly handle drmModeAddFB failure in drmmode_crtc_scanout_allocate 954 955 We were printing an error message, but not propagating the failure. That 956 would probably lead to trouble down the road. 957 958 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 959 960commit 8da1d0c870e1081d77925807d6e3bbc61a23f54f 961Author: Michel Dänzer <michel.daenzer@amd.com> 962Date: Fri Oct 9 18:59:16 2015 +0900 963 964 Eliminate redundant data parameter from drmmode_crtc_scanout_create 965 966 drmmode_crtc_scanout_create just needs to call 967 drmmode_crtc_scanout_allocate when scanout->bo is NULL. 968 969 This makes it clearer to the reader / compiler that 970 drmmode_crtc_scanout_create doesn't dereference scanout->bo when it's 971 NULL. 972 973 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 974 975commit dc40582d5ff94d812cbc08f95cf14b80cd0f410d 976Author: Michel Dänzer <michel.daenzer@amd.com> 977Date: Wed Oct 7 16:19:22 2015 +0900 978 979 Don't advertise rotation support without hardware acceleration v2 980 981 Rotation currently doesn't work without acceleration (doesn't actually 982 rotate with Option "NoAccel", crashes with Option "AccelMethod" "none" 983 or when glamor fails to initialize) and would probably be too slow 984 anyway. 985 986 v2: Also remove now dead code checking for ShadowFB from 987 drmmode_crtc_scanout_allocate(). 988 989 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 990 991commit 460560502a1bdf26d06f3c30df46fa9f28ffb9e5 992Author: Tom St Denis <tom.stdenis@amd.com> 993Date: Tue Oct 6 08:49:54 2015 -0400 994 995 Simplify drmmode_set_mode_major() and avoid leaking memory. 996 997 The function would leak the memory allocated for output_ids. This 998 patch addresses that as well as simplifies the logic somewhat. 999 1000 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1001 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1002 1003commit 56398d6651dfc4935cbd117ad861e1800077c73c 1004Author: Tom St Denis <tom.stdenis@amd.com> 1005Date: Tue Oct 6 08:43:12 2015 -0400 1006 1007 Avoid NULL dereference if drmmode_crtc_scanout_allocate fails 1008 1009 This avoids a NULL dereference if the memory allocation fails. 1010 1011 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1012 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1013 1014commit 4b92b960c7705be8b3a5dee17b2341864d7ca9bb 1015Author: Tom St Denis <tom.stdenis@amd.com> 1016Date: Mon Oct 5 10:00:09 2015 -0400 1017 1018 cleanup the entity rec 1019 1020 Based on radeon commit: b32a0a3de84a44b9af4f1ca8be19f10d7fa31b12 1021 1022 Some of these were set, some of them were 1023 always opposites, so clean things up. 1024 1025 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1026 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1027 1028commit fe100fd6bf483228eaf64b959c56a68e8dac4447 1029Author: Tom St Denis <tom.stdenis@amd.com> 1030Date: Mon Oct 5 10:45:33 2015 -0400 1031 1032 present: Handle DPMS off in radeon_present_get_ust_msc 1033 1034 Based on radeon commit: 95f5d09e3667ded027ae648c97eb4737d8bf67c5 1035 1036 The DRM_IOCTL_WAIT_VBLANK ioctl may return an error during DPMS off, 1037 which would trigger an error message in drmmode_crtc_get_ust_msc. 1038 1039 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1040 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1041 1042commit bfa925a04815cee5fd57b99447cb2ee0e158036c 1043Author: Tom St Denis <tom.stdenis@amd.com> 1044Date: Mon Oct 5 10:10:51 2015 -0400 1045 1046 present: Look at all CRTCs to determine if we can flip 1047 1048 Based on radeon commit 211862b777d0be251a4662f5dd24f2d400544c09 1049 1050 Inspired by modesetting driver change by Kenneth Graunke. 1051 1052 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1053 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1054 1055commit a1e47e76322619ed037ebce27974a4e3792940c2 1056Author: Tom St Denis <tom.stdenis@amd.com> 1057Date: Mon Oct 5 10:41:22 2015 -0400 1058 1059 present: Fall back to modeset for unflip operation 1060 1061 Based on radeon commit: 802d33e474a82262d9cdf11b03568b0c4929cd0d 1062 1063 It's not always possible to use the page flip ioctl for this, e.g. 1064 during DPMS off. We were previously just skipping the unflip in that 1065 case, which could result in hangs when setting DPMS off while a 1066 fullscreen Present app is running, e.g. at the GNOME3 lock screen. 1067 1068 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1069 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1070 1071commit bac21dfc8e60a07f08158b13fab1f3a9b9d27d1b 1072Author: Tom St Denis <tom.stdenis@amd.com> 1073Date: Mon Oct 5 10:37:50 2015 -0400 1074 1075 Don't attempt a DRI2/Present page flip while the other one is flipping 1076 1077 Based on radeon commit 49f5b0bc301414df049e00d226034e3d6e56421b 1078 1079 Fixes corrupted display and hangs when switching between DRI2 and DRI3 1080 fullscreen apps, e.g. a compositor using DRI3 and a fullscreen app using 1081 DRI2 or vice versa. 1082 1083 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1084 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1085 1086commit a5f7f2e68bad1935f5ad52286033237467f77302 1087Author: Tom St Denis <tom.stdenis@amd.com> 1088Date: Mon Oct 5 13:12:23 2015 -0400 1089 1090 Move amdgpu_drm_handler/abort_proc fields to drmmode_flipdata_re 1091 1092 Based on radeon commit de5ddd09db82141b263338dcf0c28e01f58268ee 1093 1094 Their values are the same for all DRM flip ioctl calls within a single 1095 radeon_do_pageflip() call. 1096 1097 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1098 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1099 1100commit e14e3560bff2537d3ad4c93d2b31442a122cde66 1101Author: Tom St Denis <tom.stdenis@amd.com> 1102Date: Mon Oct 5 13:08:43 2015 -0400 1103 1104 Simplify amdgpu_do_pageflip() error handling slightly more 1105 1106 Based on radeon commit e8c0f6319fbf4c3ea11e22ab1a68837031bdec8c 1107 1108 We don't need the local variable old_fb_id. 1109 1110 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1111 1112 [ Michel Dänzer: fix up slightly to better match radeon formatting ] 1113 1114 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1115 1116commit e9621ec0e2400f62db320c560a739b29258edb87 1117Author: Tom St Denis <tom.stdenis@amd.com> 1118Date: Mon Oct 5 09:34:47 2015 -0400 1119 1120 Increase robustness against DRM page flip ioctl failures 1121 1122 Based on radeon commit 8fc22360d5520469c82092ccb0fcf2af330c573f 1123 1124 Centralize cleanup, only clean up things that have been allocated for 1125 the failed ioctl call. 1126 1127 Fixes double-free after a flip ioctl failure. 1128 1129 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89681 1130 1131 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1132 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1133 1134commit db3bb2061b9ac16b0922d9afae99874820356a04 1135Author: Tom St Denis <tom.stdenis@amd.com> 1136Date: Tue Sep 29 13:07:04 2015 -0400 1137 1138 Clean up allocation in AMDGPUInitVideo() 1139 1140 The allocation of the adapters should use the correct sizeof (even if 1141 allocating an array of pointers). 1142 1143 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1144 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1145 1146commit 94caf7ac777134b8396aa762a506053179bbb4c6 1147Author: Tom St Denis <tom.stdenis@amd.com> 1148Date: Thu Oct 1 13:08:41 2015 -0400 1149 1150 Avoid leaking memory on output. 1151 1152 Based on radeon commit 63dc36dc49f93cb00111b497ab6805194bc9d240 1153 1154 and 2nd patch: 1155 1156 Proper leak fix, previous leak fix was bogus. 1157 1158 Based on radeon commit b8ec9ed4fe86952763b963c86f0af0dcae69aa6c 1159 1160 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1161 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1162 1163commit f035faec041cb5df65c78effa58eb50197cedf88 1164Author: Tom St Denis <tom.stdenis@amd.com> 1165Date: Thu Oct 1 12:56:05 2015 -0400 1166 1167 add support for DP 1.2 display hotplug 1168 1169 Based on radeon commit 2f11dcd43966cf2ee26e61960fd72e6644f5e037 1170 1171 > This allows for dynamic creation of conneectors when the 1172 > kernel tells us. 1173 > 1174 > v2: fix dpms off crash 1175 1176 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1177 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1178 1179commit aee72b29210d79dbf41bde6eef16d7fe817e6cf4 1180Author: Tom St Denis <tom.stdenis@amd.com> 1181Date: Thu Oct 1 12:29:36 2015 -0400 1182 1183 move output name creation to its own function 1184 1185 Based on radeon commit c88424d1f4aaa78b569e5d44f0b4a47de2f422f4 1186 1187 > The secondary indent is deliberate to make the next patch more 1188 > parseable for mst support. 1189 1190 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1191 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1192 1193commit 0846abeace649d27a5f2c17373e717f92d246797 1194Author: Tom St Denis <tom.stdenis@amd.com> 1195Date: Thu Oct 1 12:13:21 2015 -0400 1196 1197 stop caching mode resources 1198 1199 Based on radeon commit 32b003cb7657e07d5af6338ad44d768eda87fd33 1200 1201 > This is step one towards MST connector hotplug support, 1202 > it stop caching the mode resources structure, and 1203 > just passes a pointer to it around. 1204 1205 With a few tweaks to match the state of the AMDGPU tree. 1206 1207 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1208 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1209 1210commit 4ca8f957e0b417b099f625470db98a54531a731d 1211Author: Tom St Denis <tom.stdenis@amd.com> 1212Date: Thu Oct 1 13:16:15 2015 -0400 1213 1214 Silence type mismatch warning. 1215 1216 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1217 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1218 1219commit a79735ab1499c1f7814036d1b19ff465705c5f45 1220Author: Tom St Denis <tom.stdenis@amd.com> 1221Date: Thu Oct 1 10:51:07 2015 -0400 1222 1223 Add support for server managed fds 1224 1225 Based on radeon commit ed0cfbb4fe77146b0b38f777bc28f3a4ea6da07f 1226 1227 and 2nd patch: 1228 1229 Fix building on older servers without xf86platformBus.h 1230 1231 Based on radeon commit b50da3b96c212086cb58501dbe988d64f1f35b6d 1232 1233 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1234 1235 [ Michel Dänzer: Fixed up amdgpu_kernel_open_fd() not to need 1236 AMDGPUEntPriv(), which doesn't work yet at that point ] 1237 1238 Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> 1239 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1240 1241commit b93934a9ed5e92f3a6eac6554c5c4fa2967a6dd0 1242Author: Tom St Denis <tom.stdenis@amd.com> 1243Date: Thu Oct 1 10:05:36 2015 -0400 1244 1245 Add amdgpu_open_drm_master helper function 1246 1247 Based on radeon commit 3d7861fe112f25874319d4cdc12b745fbcd359cf 1248 1249 > This is a preparation patch for adding server-managed-fd support without it 1250 > turning into a goto fest. 1251 1252 With appropriate modifications because the open call stack is different 1253 in the amdgpu tree. 1254 1255 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1256 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1257 1258commit f5c3fd0b57cf9e392bf591110568637937a1d338 1259Author: Tom St Denis <tom.stdenis@amd.com> 1260Date: Thu Oct 1 09:13:57 2015 -0400 1261 1262 Cleaning up for server-fd support 1263 1264 Based on radeon commit a63342ad15408071437c80b411d14196f3288aed 1265 1266 > radeon_open_drm_master get rid of unnecessary goto 1267 1268 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1269 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1270 1271commit 3055724aef76a624718f26d5f0f9e9d567ffbcfb 1272Author: Tom St Denis <tom.stdenis@amd.com> 1273Date: Thu Sep 24 13:08:31 2015 -0400 1274 1275 Simplify pick best crtc to fold two loops into one 1276 1277 This patch folds the two for loops from amdgpu_pick_best_crtc() into 1278 one to reduce the LOC and make the routine easier to read. 1279 1280 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1281 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1282 1283commit 9945b4ae1664ab815b39ff07e7b66cfa7f942dfa 1284Author: Tom St Denis <tom.stdenis@amd.com> 1285Date: Wed Sep 9 09:38:02 2015 -0400 1286 1287 Avoid use-after-free in drmmode_output_destroy() 1288 1289 The encoders array is freed before potentially all of the elements of 1290 the array are individually freed. 1291 1292 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1293 Reviewed-by: Christian König <christian.koenig@amd.com> 1294 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1295 Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> (cosmetic fixups) 1296 1297commit 36b3faebdd1d2090a286616eeeb131d15e9a1386 1298Author: Tom St Denis <tom.stdenis@amd.com> 1299Date: Wed Sep 9 09:36:59 2015 -0400 1300 1301 Avoid use-after-free in amdgpu_kernel_open_fd() 1302 1303 If the device cannot be opened avoid re-using busid after it has been 1304 freed. 1305 1306 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1307 Reviewed-by: Christian König <christian.koenig@amd.com> 1308 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1309 Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> (cosmetic fixups) 1310 1311commit 8823c3d4c6db70cff7699b31088f2d92db8faaf4 1312Author: Tom St Denis <tom.stdenis@amd.com> 1313Date: Wed Sep 9 09:34:38 2015 -0400 1314 1315 dri2: Avoid calculation with undefined msc value 1316 1317 If the get_msc() call fails for any reason we should avoid updating the 1318 vblank counter delta with undefined data. 1319 1320 Signed-off-by: Tom St Denis <tom.stdenis@amd.com> 1321 Reviewed-by: Christian König <christian.koenig@amd.com> 1322 Acked-by: Alex Deucher <alexander.deucher@amd.com> 1323 Signed-off-by: Michel Dänzer <michel.daenzer@amd.com> (minor fixups) 1324 1325commit 63948ea091a9b324327ade7ec4fc5d67ca7e6f6f 1326Author: Michel Dänzer <michel.daenzer@amd.com> 1327Date: Fri Aug 14 18:41:57 2015 +0900 1328 1329 DRI2: Keep MSC monotonic when moving window between CRTCs 1330 1331 This mirrors the DRI3 implementation in xserver. Fixes VDPAU video 1332 playback hanging when moving the window between CRTCs. 1333 1334 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66384 1335 1336 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1337 1338commit 55a4461bd95698cb8d52f9f6c28583f8f81afb4e 1339Author: Michel Dänzer <michel.daenzer@amd.com> 1340Date: Fri Aug 7 11:46:31 2015 +0900 1341 1342 Wait for scanout BO initialization to finish before setting mode 1343 1344 This should avoid intermittent artifacts which could sometimes be visible 1345 when setting a new scanout pixmap, e.g. on server startup or when 1346 changing resolutions. 1347 1348 (Ported from radeon commit 3791fceabf2cb037467dc41c15364e9f9ec1e47e) 1349 1350 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1351 1352commit 4c425e9c5c038504a0f0498dd800ab1fb40bf0c5 1353Author: Michel Dänzer <michel.daenzer@amd.com> 1354Date: Fri Aug 7 12:39:24 2015 +0900 1355 1356 glamor: Add amdgpu_glamor_finish to wait for glamor rendering to finish 1357 1358 This is a bit sneaky, because it calls glFinish directly from the driver, 1359 but it seems to work fine. 1360 1361 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1362 1363commit bb989e173dc364a7d68e50d7e819d0e0ee133d2f 1364Author: Michel Dänzer <michel.daenzer@amd.com> 1365Date: Fri Aug 7 11:43:48 2015 +0900 1366 1367 Only call drmmode_copy_fb (at most) once on server startup 1368 1369 It doesn't make sense to copy the screen contents from console when VT 1370 switching back to Xorg or when Xorg resets. 1371 1372 Fixes intermittent artifacts when VT switching back from console to the 1373 gdm login screen. 1374 1375 (Ported from radeon commit 4e3dfa69e4630df2e0ec0f5b81d61159757c4664) 1376 1377 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1378 1379commit ebe2c020fbf2ef8de01fc50b201ab23ddb9fb13b 1380Author: Mario Kleiner <mario.kleiner.de@gmail.com> 1381Date: Mon Aug 10 23:34:40 2015 +0200 1382 1383 Make selection between DRI2 and DRI3 consistent with other drivers. (v2) 1384 1385 Add Option "DRI" to allow selection of maximum DRI level. 1386 1387 This allows the user to select the maximum level of DRI 1388 implementation to use, DRI2 or DRI3. It replaces the old 1389 option "DRI3" which had exactly the same purpose, but 1390 differs from the method used in both intel ddx and nouveau ddx. 1391 Make this consistent before a new stable driver is released. 1392 1393 v2: Retain handling of old Option "DRI3" for backwards 1394 compatibility, but Option "DRI" will take precedence 1395 over "DRI3" if both are provided. 1396 1397 Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com> 1398 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1399 1400commit c9611a2aa0f8d3bb55c552353740d60f6e4f63a0 1401Author: Alex Deucher <alexander.deucher@amd.com> 1402Date: Tue Jul 7 22:46:34 2015 -0400 1403 1404 add fiji pci id 1405 1406 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1407 Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 1408 1409commit 2622ac1554761b8824bfbbb2e3051a632ee38ce7 1410Author: Alex Deucher <alexander.deucher@amd.com> 1411Date: Tue Jul 7 22:46:08 2015 -0400 1412 1413 Add fiji support 1414 1415 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1416 Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 1417 1418commit 7a49d8728d17875206a84fd1023f62b37c4a9f51 1419Author: Michel Dänzer <michel.daenzer@amd.com> 1420Date: Thu Aug 6 18:21:30 2015 +0900 1421 1422 On screen resize, clear the new buffer before displaying it 1423 1424 Fixes garbage being intermittently visible during a screen resize. 1425 1426 (Ported from radeon commit 80f3d727f93cb6efedd2b39338d2301035965fe2) 1427 1428 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1429 1430commit 9f988bf1dc9d4cb92926c051ed8f15e9ba58a016 1431Author: Michel Dänzer <michel.daenzer@amd.com> 1432Date: Thu Aug 6 17:50:11 2015 +0900 1433 1434 Make drmmode_copy_fb() work with glamor 1435 1436 Needed for Xorg -background none. 1437 1438 (Ported from radeon commit 3999bf88cdb192fe2f30b03bd2ed6f6a3f9f9057) 1439 1440 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1441 1442commit 13cf61bd8d46b0059f26120a8902da6f86e6bd11 1443Author: Michel Dänzer <michel.daenzer@amd.com> 1444Date: Thu Aug 6 17:46:38 2015 +0900 1445 1446 Update scanout pixmap contents before setting a mode with it 1447 1448 This ensures the scanout pixmaps used for Option "TearFree" and Option 1449 "ShadowPrimary" have been initialized when their initial mode is set. 1450 1451 (Ported from radeon commit a4a8cdbcc10c1c5f07485a2af9e9e81e490c3e1d) 1452 1453 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1454 1455commit 15050aabf256c17250d1fca0bfac97fc6707b195 1456Author: Michel Dänzer <michel.daenzer@amd.com> 1457Date: Thu Aug 6 17:37:11 2015 +0900 1458 1459 Defer initial modeset until the first BlockHandler invocation 1460 1461 This ensures that the screen pixmap contents have been initialized when 1462 the initial modes are set. 1463 1464 (Ported from radeon commits 673e1c7637687c74fc9bdeeeffb7ace0d04b734f and 1465 1584dc545c78e0bce8d4b4b9f26b568e2c211453) 1466 1467 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1468 1469commit 96b5364496222f1b3afb9caad458f16f156b6c47 1470Author: Michel Dänzer <michel.daenzer@amd.com> 1471Date: Thu Aug 6 17:32:45 2015 +0900 1472 1473 Defer initial drmmode_copy_fb call until root window creation 1474 1475 That's late enough for acceleration to be fully initialized, but still 1476 early enough to set pScreen->canDoBGNoneRoot. 1477 1478 (Ported from radeon commit 37874a4eeace5df04b02c8fc28f67b824e3f0f5f) 1479 1480 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1481 1482commit 0fb45f2bba89379ba25d4c863091937b6384bda9 1483Author: Michel Dänzer <michel.daenzer@amd.com> 1484Date: Thu Aug 6 17:25:53 2015 +0900 1485 1486 Only copy fbcon BO contents if bgNoneRoot is TRUE 1487 1488 Otherwise, the X server will initialize the screen pixmap contents 1489 anyway. 1490 1491 (Ported from radeon commit 39c497f3efca5ca08343b884f44c93215dcdef31) 1492 1493 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1494 1495commit cac553d3b691d26eaad24fbdcba06097b6728a6d 1496Author: Michel Dänzer <michel.daenzer@amd.com> 1497Date: Thu Aug 6 17:20:22 2015 +0900 1498 1499 Add .dir-locals.el file with Emacs indentation settings 1500 1501 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1502 1503commit ea32253541959cc36a40fb0118200a8f493dc98a 1504Author: Jammy Zhou <Jammy.Zhou@amd.com> 1505Date: Wed Jul 15 11:26:28 2015 +0800 1506 1507 Adapt to the interface change of amdgpu_bo_alloc v3 1508 1509 The amdgpu_bo_alloc_result structure is removed from libdrm_amdgpu, 1510 and the amdgpu_bo_handle is returned directly 1511 1512 v2: remove the va_map/unmap 1513 v3: simply the code a bit 1514 1515 Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com> 1516 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1517 1518commit 3010d3259d3bc74263d526e54e02bc169c8d4b4d 1519Author: Mario Kleiner <mario.kleiner.de@gmail.com> 1520Date: Wed Jul 15 09:54:59 2015 +0200 1521 1522 Allow/Fix use of multiple ZaphodHead outputs per x-screen. 1523 1524 Defining multiple ZaphodHead outputs per x-screen in a 1525 multiple x-screen's per gpu configuration caused all 1526 outputs except one per x-screen to go dark, because 1527 there was a fixed mapping x-screen number -> crtc number, 1528 limiting the number of crtc's per x-screen to one. 1529 1530 On a ZaphodHead's setup, be more clever and assign 1531 as many crtc's to a given x-screen as there are 1532 ZaphodHeads defined for that screen, assuming 1533 there are enough unused crtc's available. 1534 1535 (Ported from radeon commit afab7839fc15722dbaa7203d00fe7f6ce5336b9d) 1536 1537 Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com> 1538 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1539 1540commit 159c5d460a330cf0a24678f3c6c3e2fbaf23c571 1541Author: Dave Airlie <airlied@gmail.com> 1542Date: Tue Jul 14 17:04:14 2015 +0900 1543 1544 Adopt for new X server dirty tracking APIs. 1545 1546 Signed-off-by: Dave Airlie <airlied@redhat.com> 1547 1548 (Ported from radeon commit b6d871bf299c7d0f106c07ee4d8bd3b2337f53cc) 1549 1550 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1551 1552commit 7b3212e33cd36fb6f122774df27b56ec4e1a22b8 1553Author: Michel Dänzer <michel.daenzer@amd.com> 1554Date: Thu Jul 9 17:57:29 2015 +0900 1555 1556 DRI2: Don't ignore rotated CRTCs in amdgpu_dri2_drawable_crtc 1557 1558 Waiting for vblank interrupts works fine with rotated CRTCs. The only 1559 case we can't handle with rotation is page flipping, which is handled 1560 in can_exchange(). 1561 1562 This fixes gnome-shell hanging on rotation, probably because 1563 amdgpu_dri2_get_msc returned MSC/UST 0 for rotated CRTCs. 1564 1565 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 1566 1567commit 5587a7b43d02d6371ed4675a6260427492ebad94 1568Author: Piotr Redlewski <predlewski@gmail.com> 1569Date: Wed Jul 8 20:59:14 2015 +0200 1570 1571 Do not try to enable already enabled CRTCs in DPMS hook 1572 1573 (Ported from radeon commit a8ed62010d5012dfb27773595c446b217f3c00c5) 1574 1575 Signed-off-by: Piotr Redlewski <predlewski@gmail.com> 1576 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1577 1578commit b176e63df20b345cb378fe962afd14eed43421d3 1579Author: Piotr Redlewski <predlewski@gmail.com> 1580Date: Sun Jun 28 23:20:22 2015 +0200 1581 1582 Enable/disable CRTCs in DPMS hook 1583 1584 The CRTC DPMS hook hasn't enabled or disabled hardware CRTCs. 1585 1586 (Based on radeon commit 48e5be1d5a82c1e0ccf6b7d52924c92a630e52a8) 1587 1588 Signed-off-by: Piotr Redlewski <predlewski@gmail.com> 1589 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1590 1591commit d94d4a609c593b46ab718544ee24c25530732f22 1592Author: Michel Dänzer <michel.daenzer@amd.com> 1593Date: Thu Jun 11 17:49:33 2015 +0900 1594 1595 Handle CRTC DPMS from output DPMS hooks 1596 1597 This fixes at least two issues: 1598 1599 The CRTC DPMS hook isn't called after a modeset, so the vertical blank 1600 interrupt emulation code considered the CRTC disabled after a modeset. As 1601 a side effect, page flipping was no longer used after a modeset. 1602 1603 This change also makes sure the vertical blank interrupt emulation code 1604 runs before the hardware CRTC is disabled and after it's enabled from the 1605 output DPMS hook. The wrong order could cause gnome-shell to hang after 1606 a suspend/resume and/or DPMS off/on cycle. 1607 1608 (Ported from radeon commit c4ae0e2cbcc0e2ebf9f13ee92d59b5120254a1dc) 1609 1610 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1611 1612commit c57da33308a81fa575179238a0415abcb8b34908 1613Author: Michel Dänzer <michel.daenzer@amd.com> 1614Date: Tue Jun 9 12:39:21 2015 +0900 1615 1616 Add Option "TearFree" 1617 1618 Avoids tearing by flipping between two scanout BOs per (non-rotated) CRTC 1619 1620 (Cherry picked from radeon commit 43159ef400c3b18b9f4d3e6fa1c4aef2d60d38fe) 1621 1622 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1623 1624commit bd0aca09770543fa77b934e1728a832c9c2dc90c 1625Author: Michel Dänzer <michel.daenzer@amd.com> 1626Date: Tue Jun 9 11:57:59 2015 +0900 1627 1628 glamor: Remove the stride member of struct radeon_pixmap 1629 1630 Its value was always the same as that of the PixmapRec devKind member. 1631 1632 (Cherry picked from radeon commit ed401f5b4f07375db17ff05e294907ec95fc946d) 1633 1634 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1635 1636commit e5dfb6c2667994701ee451bf82c4142cbf343405 1637Author: Michel Dänzer <michel.daenzer@amd.com> 1638Date: Wed Mar 18 16:23:24 2015 +0900 1639 1640 glamor: Add Option "ShadowPrimary" 1641 1642 When this option is enabled, most pixmaps (including the screen pixmap) 1643 are allocated in system RAM and mostly accessed by the CPU. Changed areas 1644 of the screen pixmap are copied to dedicated per-CRTC scanout pixmaps 1645 regularly, triggered by the vblank interrupt. 1646 1647 (Cherry picked from radeon commits ae92d1765fa370a8d94c2856ad6c45d273ec3c69 1648 and 1af044d7eee211fd4b248c236280274a68334da5) 1649 1650 [ Michel Dänzer: Additional adjustements for the amdgpu driver ] 1651 1652 Signed-off-by: Darren Powell <darren.powell@amd.com> 1653 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1654 1655commit 08da7b691d556735dcc22b1351c886a5079dfd3f 1656Author: Michel Dänzer <michel.daenzer@amd.com> 1657Date: Wed Jun 10 16:21:21 2015 +0900 1658 1659 Add AMDGPU_CREATE_PIXMAP_GTT flag 1660 1661 When set, the pixmap memory is allocated in GTT instead of in VRAM. 1662 1663 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1664 1665commit 59bdb578266a2637fda8d11168b9332f6845157c 1666Author: Michel Dänzer <michel.daenzer@amd.com> 1667Date: Wed Jun 10 12:04:29 2015 +0900 1668 1669 Factor out amdgpu_bo_get_handle helper 1670 1671 The helper transparently handles BOs allocated from GBM and 1672 libdrm_amdgpu. 1673 1674 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1675 1676commit 9a6eff506b6804481a6e8139d362355fc5ffdbfb 1677Author: Michel Dänzer <michel.daenzer@amd.com> 1678Date: Wed Jun 10 12:10:24 2015 +0900 1679 1680 Set AMDGPU_BO_FLAGS_GBM for cursor buffers allocated from GBM 1681 1682 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1683 1684commit d3ea8a69b02b308f8f23662be6e0c7bd81c1a2c9 1685Author: Michel Dänzer <michel.daenzer@amd.com> 1686Date: Fri May 29 18:53:50 2015 +0900 1687 1688 glamor: Add wrappers for the X server rendering hooks 1689 1690 They can choose between using the GPU or CPU for the operation. 1691 1692 (cherry picked from radeon commits eea79472a84672ee4dc7adc4487cec6a4037048a 1693 and e58fc380ccf2a581d28f041fd74b963626ca5404) 1694 1695 Signed-off-by: Darren Powell <darren.powell@amd.com> 1696 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1697 1698commit 895e4d73d5f042afa13065b64a78f5625ecb5612 1699Author: Michel Dänzer <michel.daenzer@amd.com> 1700Date: Fri May 29 18:53:40 2015 +0900 1701 1702 glamor: Remove unused function radeon_glamor_pixmap_is_offscreen 1703 1704 (cherry picked from radeon commit 2fa021f77372ca93375a3d13a0c43a9089674899) 1705 1706 Signed-off-by: Darren Powell <darren.powell@amd.com> 1707 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1708 1709commit cc5671c587d575b2a7d2802d17e8af0384a2cea5 1710Author: Michel Dänzer <michel.daenzer@amd.com> 1711Date: Fri May 29 18:53:36 2015 +0900 1712 1713 Add RADEON_CREATE_PIXMAP_SCANOUT flag 1714 1715 It means that the pixmap is used for scanout exclusively. 1716 1717 (cherry picked from radeon commit e96349ba6281fd18b8bf9c76629128276b065e6c) 1718 1719 Signed-off-by: Darren Powell <darren.powell@amd.com> 1720 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1721 1722commit 21834953ee64920438dee1c94f3a1e53dc58b82d 1723Author: Michel Dänzer <michel.daenzer@amd.com> 1724Date: Fri May 29 18:53:32 2015 +0900 1725 1726 Split out struct drmmode_scanout for rotation shadow buffer information 1727 1728 Will be used for other kinds of dedicated scanout buffers as well. 1729 1730 (cherry picked from radeon commit 9be7dd382e86d2b804de81d4e2af7431b2e16843) 1731 1732 Signed-off-by: Darren Powell <darren.powell@amd.com> 1733 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1734 1735commit e4e4f7b83e7d7e43993fa0793d666d6dec2980f8 1736Author: Michel Dänzer <michel.daenzer@amd.com> 1737Date: Fri May 29 18:53:21 2015 +0900 1738 1739 Rename scanout_pixmap_x field to prime_pixmap_x 1740 1741 To avoid confusion with upcoming changes. 1742 1743 (cherry picked from radeon commit c32b0530302739f6512755bccf281c2300617376) 1744 1745 Signed-off-by: Darren Powell <darren.powell@amd.com> 1746 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1747 1748commit edfff6b1a3a19953644b8052b30076f76f7dc337 1749Author: Michel Dänzer <michel.daenzer@amd.com> 1750Date: Tue Jun 2 17:04:21 2015 +0900 1751 1752 Add DRI3 support 1753 1754 Must be enabled with 1755 1756 Option "DRI3" 1757 1758 in xorg.conf. 1759 1760 (Cherry picked from radeon commits 64e1e4dbdd3caee6f5d8f6b6c094b4533fa94953, 1761 694e04720b886060fe3eefdce59741f218c8269f, 1762 f940fd741b15f03393037c5bb904cd74f012de9d, 1763 fcd37f65f485291084c174666bd605e215bf1398, 1764 4b0997e56dec0053cb2cb793e0f4ae35055ff7e6, 1765 f68d9b5ba0c91a725b5eec9386c61bea8824c299 and 1766 98fb4199e63fedd4607cddee64bf602d6398df81) 1767 1768 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1769 1770commit d295b5b3310bc5c23d232c4be4170165a057c090 1771Author: Michel Dänzer <michel.daenzer@amd.com> 1772Date: Tue Jun 2 17:01:06 2015 +0900 1773 1774 amdgpu_set_shared_pixmap_backing: Add support for GBM / glamor v2 1775 1776 v2: Initialize reference count of imported GBM BOs to 1, fixes leaking 1777 them. 1778 1779 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> [v1] 1780 1781commit 03ad0fa0185d215f7d4234006e04406af1ab63ca 1782Author: Michel Dänzer <michel.daenzer@amd.com> 1783Date: Fri May 29 18:53:45 2015 +0900 1784 1785 glamor: Add radeon_pixmap parameter to radeon_glamor_create_textured_pixmap 1786 1787 (cherry picked from radeon commit 051d46382656ffc3e6cac1aab3aee7efdf5b623a) 1788 1789 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1790 Signed-off-by: Darren Powell <darren.powell@amd.com> 1791 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1792 1793commit fafb8c6ac925ad16073e5a60dbf60d5add11bb25 1794Author: Michel Dänzer <michel.daenzer@amd.com> 1795Date: Tue Jun 2 17:00:46 2015 +0900 1796 1797 Add support for the Present extension 1798 1799 (Cherry picked from radeon commits 3c65fb849e1ba9fb6454bcaa55b696548902f3fc, 1800 694e04720b886060fe3eefdce59741f218c8269f, 1801 e3be8b0a8cf484ff16597413a6172788178e80c8, 1802 80eede245d1eda27eaba108b0761a24bfd69aff6 and 1803 5f82a720374c9c1caebb42bfbeea1f0cf8847d28) 1804 1805 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1806 1807commit 5b51f0e7e396ea946ef85429a8e9be5c1d5c39c3 1808Author: Michel Dänzer <michel.daenzer@amd.com> 1809Date: Tue Jun 2 16:58:27 2015 +0900 1810 1811 Add support for SYNC extension fences 1812 1813 (Cherry picked from radeon commits 8fc9a241ab59ffbcdc178d6415332c88a54e85fe, 1814 af1862a37570fa512a525ab47d72b30400d2e2d6, 1815 aa7825eb29cdf6ac9d7b28ad18186807ff384687, 1816 af6076241c0d322b295a4e898407ae2472bd8eb4 and 1817 d64a13ebe0ecd241ee3260dbffd8f4a01e254183) 1818 1819 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1820 1821commit a30060d22a42688371166a861e5050fdd5ce8f7b 1822Author: Michel Dänzer <michel.daenzer@amd.com> 1823Date: Mon Jun 1 18:33:33 2015 +0900 1824 1825 DRI2: Split out helper for getting UST and MSC of a specific CRTC 1826 1827 (Cherry picked from radeon commits 76c2923ac5c7230a8b2f9f8329c308d28b44d9c0 1828 and d7c82731a8bf3d381bc571b94d80d9bb2dd6e40d) 1829 1830 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1831 1832commit 9a554a683b970660b467566cf05b921393705a20 1833Author: Michel Dänzer <michel.daenzer@amd.com> 1834Date: Mon Jun 1 17:32:56 2015 +0900 1835 1836 DRI2: Use helper functions for DRM event queue management 1837 1838 This is mostly in preparation for Present support, but it also simplifies 1839 the DRI2 specific code a little. 1840 1841 (Cherry picked from radeon commit 6c3a721cde9317233072b573f9502348dcd21b16) 1842 1843 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1844 1845commit e6164ad340f65ff8ee6f6a6934302591af875a43 1846Author: Michel Dänzer <michel.daenzer@amd.com> 1847Date: Mon Jun 1 17:29:30 2015 +0900 1848 1849 DRI2: Move amdgpu_dri2_flip_event_handler 1850 1851 In preparation for the next change, which will modify it to a static 1852 function which needs to be in the new place. No functional change. 1853 1854 (Cherry picked from radeon commit c3fa22a479e61d1899fa9d327d9c4e2a7f64b0c1) 1855 1856 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1857 1858commit 5419e13da7ec3cffd43510ac88106076ea81124c 1859Author: Michel Dänzer <michel.daenzer@amd.com> 1860Date: Mon Jun 1 17:25:23 2015 +0900 1861 1862 DRI2: Remove superfluous assignments to *_info->frame 1863 1864 That field is only used for page flipping. 1865 1866 (Cherry picked from radeon commit 65045112fdc8a9fa36e0e00f46739a6152b775ff) 1867 1868 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1869 1870commit f4c2b640be17ab1f8694b35d4cb74ccfce3d1385 1871Author: Michel Dänzer <michel.daenzer@amd.com> 1872Date: Mon Jun 1 17:11:30 2015 +0900 1873 1874 DRI2: Simplify blit fallback handling for scheduled swaps 1875 1876 Also use amdgpu_dri2_schedule_event when possible. 1877 1878 (Cherry picked from radeon commit ad27f16f308079d06a2b1c788b3cb0947531253a) 1879 1880 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1881 1882commit 13a7284e061081a12180b375d66f9b8394cf8753 1883Author: Michel Dänzer <michel.daenzer@amd.com> 1884Date: Mon Jun 1 16:58:00 2015 +0900 1885 1886 Add DRM event queue helpers 1887 1888 (Cherry picked from radeon commit b4af8a327ed8420f0ff4ea0f113f4a59406ed4d3) 1889 1890 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1891 1892commit eb7c6958dff5cb8b0aad02d1d5673483dae4e3d4 1893Author: Michel Dänzer <michel.daenzer@amd.com> 1894Date: Mon Jun 1 16:52:40 2015 +0900 1895 1896 Move xorg_list backwards compatibility to new amdgpu_list.h header 1897 1898 (Cherry picked from radeon commits 7c3470f4b659206ed23f761948936ede3a2dba3d 1899 and 4a98f60117c387a228d5cbaadb6e298fb4e865df) 1900 1901 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1902 1903commit 69d161a54b4ea0d8033a0873210f2857c91ceae8 1904Author: Michel Dänzer <michel.daenzer@amd.com> 1905Date: Mon Jun 1 16:46:30 2015 +0900 1906 1907 Require at least xserver 1.8 1908 1909 So we can rely on the list.h header. 1910 1911 xserver 1.8 was released in April 2010. 1912 1913 (Cherry picked from radeon commit 7388d0b6c54b9d536fdb161e3aa61b326627b939) 1914 1915 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 1916 1917commit 7363156b7c077def2aaf9a4573410817f5e92610 1918Author: Jammy Zhou <Jammy.Zhou@amd.com> 1919Date: Sat May 30 00:31:44 2015 +0800 1920 1921 Check GBM_BO_USE_LINEAR correctly v2 1922 1923 v2: remove the check for gbm.h 1924 1925 Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com> 1926 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1927 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> [v1] 1928 1929commit e75e9f39c95b8b563885882bf29d776861cd6ca3 1930Author: Brian Paterni <bpaterni@gmail.com> 1931Date: Sat May 16 15:00:14 2015 -0500 1932 1933 extend conditional group GBM_BO_USE_LINEAR 1934 over both usages 1935 1936 Fixes 'GBM_BO_USE_LINEAR' undeclared error when compiling against older 1937 libgbm 1938 1939 Signed-off-by: Brian Paterni <bpaterni@gmail.com> 1940 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1941 1942commit 37b389ee9e13f065fb080d1269f9a6aed616c210 1943Author: Michel Dänzer <michel.daenzer@amd.com> 1944Date: Fri May 15 10:24:24 2015 +0900 1945 1946 glamor: Deal with glamor_glyphs_init being removed from xserver 1947 1948 Port of radeon commit 818c180c8932233b214a35ba0647af82f7bcec3d. 1949 1950commit 22917044e419023d487f816e0d4f094695b55fa6 1951Author: Alex Deucher <alexander.deucher@amd.com> 1952Date: Tue May 12 13:29:00 2015 -0400 1953 1954 add some new tonga pci ids 1955 1956 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 1957 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1958 Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 1959 1960commit e71be4a22799ec4c02051b75c5fed16a3a953c7b 1961Author: Alex Deucher <alexander.deucher@amd.com> 1962Date: Tue May 12 13:25:02 2015 -0400 1963 1964 add new bonaire pci id 1965 1966 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 1967 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1968 Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 1969 1970commit b795d1e137b34a314b4b41d025d96ca9251d6bbe 1971Author: Michel Dänzer <michel.daenzer@amd.com> 1972Date: Thu May 7 18:05:32 2015 +0900 1973 1974 Link against libgbm 1975 1976 Fixes unresolved symbol "gbm_create_device". 1977 1978 Reported-and-Tested-by: Brian Paterni <bpaterni@gmail.com> 1979 1980commit 7e3b27390a03e423772717fca3c757cf5cc4d7b4 1981Author: Jammy Zhou <Jammy.Zhou@amd.com> 1982Date: Tue May 12 05:34:49 2015 +0800 1983 1984 Disable tiling for PRIME shared pixmap 1985 1986 Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com> 1987 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 1988 1989commit 4840f918ab7d61b4f55bcdff3afdac7b34e45d88 1990Author: Jammy Zhou <Jammy.Zhou@amd.com> 1991Date: Tue May 12 00:09:42 2015 +0800 1992 1993 Use gbm_bo_get_fd to get DMA_BUF fd 1994 1995 When GBM is used for buffer allocation, gbm_bo_get_fd should be 1996 used to get the DMA_BUF fd. 1997 1998 Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com> 1999 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 2000 2001commit b69c5b3cc2d7da3bb85acd687db9b5a021258914 2002Author: Marek Olšák <marek.olsak@amd.com> 2003Date: Fri Mar 27 22:56:37 2015 +0100 2004 2005 ddx: use amdgpu_query_crtc_from_id 2006 2007 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 2008 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 2009 2010commit 91aa694a7da7b690a3e5d59a1a8fa42cbb3ebda4 2011Author: Marek Olšák <marek.olsak@amd.com> 2012Date: Fri Mar 27 22:22:35 2015 +0100 2013 2014 ddx: remove AMDGPUIsAccelWorking 2015 2016 libdrm fails to initialize without acceleration, so this always returns true. 2017 2018 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 2019 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 2020 2021commit afc33040f862e2e13ba7f132bb363cf16fb6a1d7 2022Author: Marek Olšák <marek.olsak@amd.com> 2023Date: Fri Mar 27 22:14:37 2015 +0100 2024 2025 ddx: enable acceleration by default on Hawaii 2026 2027 Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com> 2028 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 2029 2030commit 8a34a8149860ac15e83ccdbd8d9a527d8d3e5997 2031Author: Jammy Zhou <Jammy.Zhou@amd.com> 2032Date: Mon Apr 27 14:27:34 2015 +0800 2033 2034 Remove throttling from amdgpu_dri2_copy_region2 2035 2036 Throttling should be handled by the client-side drivers. 2037 2038 Signed-off-by: Jammy Zhou <Jammy.Zhou@amd.com> 2039 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 2040 2041commit 9f61a5506b1028d30c99cb5866abcec35d5c9cb8 2042Author: Alex Deucher <alexander.deucher@amd.com> 2043Date: Fri Apr 24 11:47:32 2015 -0400 2044 2045 fixup README 2046 2047 Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> 2048 Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 2049 2050commit a49ad11af18dad74506c2f69d7bbda07b67529d2 2051Author: Michel Dänzer <michel.daenzer@amd.com> 2052Date: Fri Apr 24 09:57:27 2015 +0900 2053 2054 Add 10-amdgpu.conf xorg.conf.d snippet 2055 2056 This instructs Xorg >= 1.16 to try loading the amdgpu driver for devices 2057 managed by the amdgpu kernel driver. 2058 2059 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 2060 2061commit fa4aed6cf56048a6520eac57514e38db3685cd15 2062Author: Michel Dänzer <michel.daenzer@amd.com> 2063Date: Fri Apr 24 09:53:33 2015 +0900 2064 2065 Document Option "AccelMethod" in the manpage 2066 2067 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 2068 2069commit fe4a4b6836252cc8caa642a32fb3910c8590076b 2070Author: Michel Dänzer <michel.daenzer@amd.com> 2071Date: Fri Apr 24 09:52:04 2015 +0900 2072 2073 Fix build when gbm.h doesn't define GBM_BO_USE_LINEAR 2074 2075 Option "AccelMethod" "none" is ignored in that case. 2076 2077 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> 2078 2079commit 84df3e7114fb71b5e10c1a6f7869ab1505fef5b0 2080Author: Michel Dänzer <michel.daenzer@amd.com> 2081Date: Fri Apr 24 09:51:22 2015 +0900 2082 2083 glamor: Handle GLAMOR_* flags removed from xserver 2084 2085 The behaviour is the same as when the removed flags were passed in. 2086 2087 (cherry picked from radeon commit b16609b453bb1a181198cf27778f205dc23fb642) 2088 2089 Reviewed-by: Christian König <christian.koenig@amd.com> 2090 2091commit b947f4bf4efa8841bea4d306d0b0d21c7511c724 2092Author: Michel Dänzer <michel.daenzer@amd.com> 2093Date: Fri Apr 24 09:50:51 2015 +0900 2094 2095 Move #include "radeon_glamor.h" from amdgpu_drv.h to where it's needed 2096 2097 (cherry picked from radeon commit 4b8adebb80158bcf81ada83bb88517febe931b12) 2098 2099 Reviewed-by: Christian König <christian.koenig@amd.com> 2100 2101commit ff62bf6e9dce55dbde92baf4fa30193c7344ee8a 2102Author: Alex Deucher <alexander.deucher@amd.com> 2103Date: Mon Apr 20 11:57:52 2015 -0400 2104 2105 amdgpu: add the xf86-video-amdgpu driver 2106 2107 This adds the new xf86-video-amdgpu driver for 2108 newer AMD GPUs. 2109 2110 Signed-off-by: Alex Deucher <alexander.deucher@amd.com> 2111