- 27 Apr, 2021 2 commits
-
-
Venkata Ramana Nayana authored
Use I915_MAP_WC when default state object is allocated in LMEM. Signed-off-by:
Venkata Ramana Nayana <venkata.ramana.nayana@intel.com> Reviewed-by:
Matthew Auld <matthew.auld@intel.com> Signed-off-by:
-
Lv Yunlong authored
Our code analyzer reported a double free bug. In gen8_preallocate_top_level_pdp, pde and pde->pt.base are allocated via alloc_pd(vm) with one reference. If pin_pt_dma() failed, pde->pt.base is freed by i915_gem_object_put() with a reference dropped. Then free_pd calls free_px() defined in intel_ppgtt.c, which calls i915_gem_object_put() to put pde->pt.base again. As pde->pt.base is protected by refcount, so the second put will not free pde->pt.base actually. But, maybe it is better to remove the first put? Fixes: 82adf901 ("drm/i915/gt: Shrink i915_page_directory's slab bucket") Signed-off-by:
Lv Yunlong <lyl2019@mail.ustc.edu.cn> Reviewed-by:
-
- 26 Apr, 2021 3 commits
-
-
Ville Syrjälä authored
These are the 965g/g45/g33 specific DRB registers. Give them a suitable suffix so we can add their counterparts for other platforms. Reviewed-by:
Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by:
Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20210421153401.13847-4-ville.syrjala@linux.intel.com
-
Ville Syrjälä authored
We've defined C0DRB3/C1DRB3 as 16 bit registers, so access them as such. Fixes: 1c8242c3 ("drm/i915: Use unchecked writes for setting up the fences") Reviewed-by:
-
Ville Syrjälä authored
Gen2 tiles are 2KiB in size so i915_gem_object_get_tile_row_size() can in fact return <4KiB, which leads to div-by-zero here. Avoid that. Not sure i915_gem_object_get_tile_row_size() is entirely sane anyway since it doesn't account for the different tile layouts on i8xx/i915... I'm not able to hit this before commit 6846895f ("drm/i915: Replace PIN_NONFAULT with calls to PIN_NOEVICT") and it looks like I also need to run recent version of Mesa. With those in place xonotic trips on this quite easily on my 85x. Cc: stable@vger.kernel.org Reviewed-by:
-
- 23 Apr, 2021 2 commits
-
-
Maarten Lankhorst authored
Fixes the following htmldocs warnings: drivers/gpu/drm/i915/i915_cmd_parser.c:1420: warning: Excess function parameter 'trampoline' description in 'intel_engine_cmd_parser' drivers/gpu/drm/i915/i915_cmd_parser.c:1420: warning: Function parameter or member 'jump_whitelist' not described in 'intel_engine_cmd_parser' drivers/gpu/drm/i915/i915_cmd_parser.c:1420: warning: Function parameter or member 'shadow_map' not described in 'intel_engine_cmd_parser' drivers/gpu/drm/i915/i915_cmd_parser.c:1420: warning: Function parameter or member 'batch_map' not described in 'intel_engine_cmd_parser' drivers/gpu/drm/i915/i915_cmd_parser.c:1420: warning: Excess function parameter 'trampoline' description in 'intel_engine_cmd_parser' Reported-by:
Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by:
Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20210421120353.544518-1-maarten.lankhorst@linux.intel.comReviewed-by:
Daniel Vetter <daniel.vetter@ffwll.ch>
-
Maarten Lankhorst authored
Fixes the following htmldocs warning: drivers/gpu/drm/i915/gem/i915_gem_shrinker.c:102: warning: Function parameter or member 'ww' not described in 'i915_gem_shrink' Fixes: cf41a8f1 ("drm/i915: Finally remove obj->mm.lock.") Reported-by:
-
- 22 Apr, 2021 4 commits
-
-
Matthew Auld authored
Stolen memory is always allocated as physically contiguous pages, so mark the object flags as such. It looks like the flags were previously just ignored so this had no effect. In the future we might to add the proper plumbing for passing the flags all over the way down from the caller, but for now we don't have a use for that. Signed-off-by:
Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20210421104658.304142-4-matthew.auld@intel.com
-
CQ Tang authored
Since stolen can now be device local-memory underneath, we should try to enforce any min_page_size restrictions when allocating pages. Signed-off-by:
CQ Tang <cq.tang@intel.com> Signed-off-by:
-
Matthew Auld authored
Underneath it's the same stuff, so things like the PTE_LM bits for the GTT should just keep working as-is. Signed-off-by:
-
CQ Tang authored
Add "REGION_STOLEN" device info to dg1, create stolen memory region from upper portion of local device memory, starting from DSMBASE. v2: - s/drm_info/drm_dbg; userspace likely doesn't care about stolen. - mem->type is only setup after the region probe, so setting the name as stolen-local or stolen-system based on this value won't work. Split system vs local stolen setup to fix this. - kill all the region->devmem/is_devmem stuff. We already differentiate the different types of stolen so such things shouldn't be needed anymore. v3: - split stolen lmem vs smem ops(Tvrtko) - add shortcut for stolen region in i915(Tvrtko) - sanity check dsm base vs bar size(Xinyun) v4(Tvrtko): - more cleanup - add some TODOs Signed-off-by:
-
- 21 Apr, 2021 3 commits
-
-
Matt Roper authored
Boot firmware performs memory training and health assessment during startup. If the memory training fails, the firmware will consider the GPU unusable and will instruct the punit to keep the GT powered down. If this happens, our driver will be unable to communicate with the GT (all GT registers will read back as 0, forcewake requests will timeout, etc.) so we should abort driver initialization if this happens. We can confirm that LMEM was initialized successfully via sgunit register GU_CNTL. Bspec: 53111 Signed-off-by:
Matt Roper <matthew.d.roper@intel.com> Cc: Caz Yokoyama <Caz.Yokoyama@intel.com> Reviewed-by:
-
Matthew Auld authored
Add section for drm/i915 uAPI and pull in i915_drm.h. Suggested-by:
Daniel Vetter <daniel@ffwll.ch> Signed-off-by:
-
Hsin-Yi Wang authored
pm_resume and pm_suspend might be conflict with the ones defined in include/linux/suspend.h. Rename all pm_* to igt_pm_* in selftests since they are only used here. v2 by Jani: - Use igt_ prefix instead of i915_ to avoid colliding with existing i915_pm_* functions - Rename all pm_ prefixed functions in the file Signed-off-by:
Hsin-Yi Wang <hsinyi@chromium.org> Reported-by:
kernel test robot <lkp@intel.com> Reviewed-by:
Rodrigo Vivi <rodrigo.vivi@intel.com> Signed-off-by:
Jani Nikula <jani.nikula@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20210420130853.10573-1-jani.nikula@intel.com
-
- 20 Apr, 2021 6 commits
-
-
Chris Wilson authored
RAPL provides an on-package power measurements which does not encompass discrete graphics, so let's avoid using the igfx masurements when testing dgfx. Later we will abstract the simple librapl interface over hwmon so that we can verify basic power consumption scenarios. Signed-off-by:
-
Chris Wilson authored
If there is no mappable aperture, we cannot remap it for access, and the selftest is void. Signed-off-by:
Imre Deak <imre.deak@intel.com> Signed-off-by:
-
Joonas Lahtinen authored
Merge tag 'topic/intel-gen-to-ver-2021-04-19' of git://anongit.freedesktop.org/drm/drm-intel into drm-intel-gt-next Gen to ver conversions across the driver The main change is Lucas' series [1], with Ville's GLK fixes [2] and a cherry-pick of Matt's commit [3] from drm-intel-next as a base to avoid conflicts. [1] https://patchwork.freedesktop.org/series/88825/ [2] https://patchwork.freedesktop.org/series/88938/ [3] 70bfb307 ("drm/i915/display: Eliminate IS_GEN9_{BC,LP}") Signed-off-by:
Joonas Lahtinen <joonas.lahtinen@linux.intel.com> From: Jani Nikula <jani.nikula@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/878s5ebny0.fsf@intel.com
-
Matthew Auld authored
Add a note about the two-step process. v2(Tvrtko): - Also document the other method of just passing in a buffer which is large enough, which avoids two ioctl calls. Can make sense for smaller query items. v3: prefer kernel-doc references for structs and members Suggested-by:Jason Ekstrand <jason@jlekstrand.net> Link: https://patchwork.freedesktop.org/patch/msgid/20210419105741.27844-4-matthew.auld@intel.com
-
Matthew Auld authored
Add some example usage for the extension chaining also, which is quite nifty. v2: (Daniel) - clarify that the name is just some integer, also document that the name space is not global v3: prefer kernel-doc references for structs Suggested-by: -
Matthew Auld authored
Fix the cases where it is almost already valid kernel doc, for the others just nerf the warnings for now. Signed-off-by:
-
- 14 Apr, 2021 18 commits
-
-
Lucas De Marchi authored
Make them independent so we can use DGFX_FEATURES more generically. For future platforms that do not use the GEN nomenclature we will define graphics, media and display separately, so we avoid setting graphics_ver with the GEN() macro. Reviewed-by:
Lucas De Marchi <lucas.demarchi@intel.com> Signed-off-by:
-
Lucas De Marchi authored
Since we are now converting from a single gen version to graphics_ver, media_ver and display_ver, add the last 2 when printing the device info. Reviewed-by:
-
Lucas De Marchi authored
Now that it's not being used anymore, finish its removal. Like for gen_mask, we replace INTEL_GEN() and IS_GEN() macros to use the new field. Reviewed-by:
-
Lucas De Marchi authored
Replace gen with the new graphics_ver value and use GRAPHICS_VER() in those places. Reviewed-by:
-
Lucas De Marchi authored
Now that it's not used anywhere, remove it from struct intel_device_info. To allow a period in which code will be converted to the new macro, keep IS_GEN_RANGE() around, just redefining it to use the new fields. The size advantage from IS_GEN_RANGE() using a mask is not that big as it has pretty limited use througout the driver: text data bss dec hex filename 2758497 95965 6496 2860958 2ba79e drivers/gpu/drm/i915/i915.ko.old 2758586 95953 6496 2861035 2ba7eb drivers/gpu/drm/i915/i915.ko.new Reviewed-by:
-
Lucas De Marchi authored
Remove the remaining uses of INTEL_GEN_MASK() and the correspondent gen_mask in struct intel_device_info. This will allow the removal of gen_mask later since it's incompatible with the new per-IP versioning scheme. Reviewed-by:
-
Lucas De Marchi authored
Since its introduction 2 years ago, we never used the mask to span more than one gen. Replace gen_mask a single number and start using the new GRAPHICS_VER(). Reviewed-by:
-
Lucas De Marchi authored
Start using the new fields graphics_version for the previous gen checks. Here we rename the "gen" field and replace the comparisons using it to start using the new GRAPHICS_VER(). Other uses of INTEL_GEN() were left as is for automatic conversion later. Reviewed-by:
-
Lucas De Marchi authored
Like it was done in commit 01eb15c9 ("drm/i915: Add DISPLAY_VER() and related macros") add the correspondent macros for graphics and media. Going forward we will prefer checking the versions for the specific IPs (graphics, media and display) rather than grouping everything under a "gen" version. For consistency and to make the maintenance easier, it'd be preferred not to mix the *GEN* macros with the new ones. For older platforms we can simply consider that the previous "gen" number will extend to all 3 IPs. Then we can start replacing its use in the driver. Right now this replacement is not done and only the infrastructure is put in place. We also leave gen and gen_mask inside struct intel_device_info while it's still being used throughout the code. v2: Repurpose IS_{GRAPHICS,MEDIA}_VER() macros to work with a range Signed-off-by:
José Roberto de Souza <jose.souza@intel.com> Reviewed-by:
-
Lucas De Marchi authored
While converting the rest of the driver to use GRAPHICS_VER() and MEDIA_VER(), following what was done for display, some discussions went back on what we did for display: 1) Why is the == comparison special that deserves a separate macro instead of just getting the version and comparing directly like is done for >, >=, <=? 2) IS_DISPLAY_RANGE() is weird in that it omits the "_VER" for brevity. If we remove the current users of IS_DISPLAY_VER(), we could actually repurpose it for a range check With (1) there could be an advantage if we used gen_mask since multiple conditionals be combined by the compiler in a single and instruction and check the result. However a) INTEL_GEN() doesn't use the mask since it would make the code bigger everywhere else and b) in the cases it made sense, it also made sense to convert to the _RANGE() variant. So here we repurpose IS_DISPLAY_VER() to work with a [ from, to ] range like was the IS_DISPLAY_RANGE() and convert the current IS_DISPLAY_VER() users to use == and != operators. Aside from the definition changes, this was done by the following semantic patch: @@ expression dev_priv, E1; @@ - !IS_DISPLAY_VER(dev_priv, E1) + DISPLAY_VER(dev_priv) != E1 @@ expression dev_priv, E1; @@ - IS_DISPLAY_VER(dev_priv, E1) + DISPLAY_VER(dev_priv) == E1 @@ expression dev_priv, from, until; @@ - IS_DISPLAY_RANGE(dev_priv, from, until) + IS_DISPLAY_VER(dev_priv, from, until) Cc: Jani Nikula <jani.nikula@intel.com> Cc: Matt Roper <matthew.d.roper@intel.com> Reviewed-by:
-
Lucas De Marchi authored
The macro we use to check is called DISPLAY_VER(). While using this macro and the new ones being added in following changes I made the mistake multiple times when mixing both "ver" and "version". Although it's usually better to prefer the complete name, the shorhand DISPLAY_VER() / GRAPHICS_VER / MEDIA_VER are clear and cause less visual polution. Another issue is when copying the variable to other places. "display.version" would be copied to a "display_version" variable which is long and would make people abbreviate as "version", or "display_ver". In the first case it's not always clear what version refers to, and in the second case it just hints it should be the name in the first place. So, in the same way use used "gen" rather than "generation", use "ver" instead of "version". Signed-off-by:
-
Lucas De Marchi authored
Commit 989634fb ("drm/i915/audio: set HDA link parameters in driver") added INTEL_GEN() in the display code, where it should actually be using DISPLAY_VER(). Switch to the new macro. Cc: Kai Vehmanen <kai.vehmanen@linux.intel.com> Reviewed-by:
-
Ville Syrjälä authored
Now that glk display version is 10 we can drop a few more glk checks. Reviewed-by:
-
Ville Syrjälä authored
Just let bxt/glk fall back to intel_hpd_pin_default() instead of using skl_hpd_pin() or cnl_hpd_pin(). Doesn't really matter since both functions will end up returning the correct hpd pin anyway, but I find it a bit less confusing when bxt/glk are fully separated from the logic for the other platforms. Reviewed-by:
-
Ville Syrjälä authored
The glk display version change is causing us to again attempt LTTPR detection on glk. We must not do tha since glk doesn't have a long enough AUX timeout. Restore the correct logic to skip the detection. Cc: Matt Roper <matthew.d.roper@intel.com> Fixes: 2b5a4562 ("drm/i915/display: Simplify GLK display version tests") Reviewed-by:
-
Ville Syrjälä authored
We lost a CCS related w/a on glk when the display version became 10 instead of 9. Restore the correct check. Cc: Matt Roper <matthew.d.roper@intel.com> Fixes: 2b5a4562 ("drm/i915/display: Simplify GLK display version tests") Reviewed-by:
-
Ville Syrjälä authored
We lost the FBC 16bpp 512byte stride requirement on glk when we switched from display version 9 to 10. Restore the w/a to avoid enabling FBC with a bad stride and thus display garbage. Cc: Matt Roper <matthew.d.roper@intel.com> Fixes: 2b5a4562 ("drm/i915/display: Simplify GLK display version tests") Reviewed-by:
-
Matt Roper authored
Now that we've eliminated INTEL_GEN(), IS_GEN_RANGE(), etc. from the display code, we should also kill off our use of the IS_GEN9_* macros too. We'll do the conversion manually this time instead of using Coccinelle since the most logical substitution can depend heavily on the code context, and sometimes we can keep the code simpler if we make additional adjustments such as swapping the order of if/else arms. v2: - Restore a lost negation in intel_pll_is_valid(). Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Lucas De Marchi <lucas.demarchi@intel.com> Signed-off-by:
Radhakrishna Sripada <radhakrishna.sripada@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20210407203945.1432531-1-matthew.d.roper@intel.com (cherry picked from commit 70bfb307) [Jani: cherry picked to topic branch to reduce conflicts] Signed-off-by:
-
- 09 Apr, 2021 2 commits
-
-
Tvrtko Ursulin authored
RC6 support cannot be simply established by looking at the static device HAS_RC6() flag. There are cases which disable RC6 at driver load time so use the status of those check when deciding whether to enumerate the rc6 counter. Signed-off-by:
Eero T Tamminen <eero.t.tamminen@intel.com> Reviewed-by:
-
Tvrtko Ursulin authored
Reference needs to be taken before arming the timer. Luckily, given the default timer period of 20s, the potential to hit the race is extremely unlikely. Signed-off-by:
Mika Kuoppala <mika.kuoppala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20210326105759.2387104-1-tvrtko.ursulin@linux.intel.com
-
