- 17 Dec, 2019 11 commits
-
-
Daniele Ceraolo Spurio authored
For better isolation of the request tracking from the rest of the CT-related data. v2: split to separate patch, move next_fence to substructure (Michal) Signed-off-by:
Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: John Harrison <John.C.Harrison@Intel.com> Cc: Matthew Brost <matthew.brost@intel.com> Reviewed-by:
Michal Wajdeczko <michal.wajdeczko@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20191217012316.13271-4-daniele.ceraolospurio@intel.com
-
Daniele Ceraolo Spurio authored
The GuC supports having multiple CT buffer pairs and we designed our implementation with that in mind. However, the different channels are not processed in parallel within the GuC, so there is very little advantage in having multiple channels (independent locks?), compared to the drawbacks (one channel can starve the other if messages keep being submitted to it). Given this, it is unlikely we'll ever add a second channel and therefore we can simplify our code by removing the flexibility. v2: split substructure grouping to separate patch, improve docs (Michal) Signed-off-by:
-
Daniele Ceraolo Spurio authored
We track the status of the GuC much more closely now and we expect the enable/disable functions to be correctly called only once. If this isn't true we do want to flag it as a flow failure (via the BUG_ON in the ctch functions) and not silently ignore the call. Suggested-by:
-
Daniele Ceraolo Spurio authored
The only difference from the GuC POV between guc_communication_stop and guc_communication_disable is that the former can be called after GuC has been reset. Instead of having two separate paths, we can just skip the call into GuC in the disabling path and re-use that. Note that by using the disable() path instead of the stop() one there are two additional changes in SW side for the stop path: - interrupts are now disabled before disabling the CT, which is ok because we do not want interrupts with CT disabled; - guc_get_mmio_msg() is called in the stop case as well, which is ok because if there are errors before the reset we do want to record them. Signed-off-by:
-
Tvrtko Ursulin authored
Get_pid_task() needs to be paired with a put_pid or we leak a pid reference every time a banned client tries to create a context. v2: * task_pid_nr helper exists! (Chris) Signed-off-by:
Tvrtko Ursulin <tvrtko.ursulin@intel.com> Fixes: b083a087 ("drm/i915: Add per client max context ban limit") Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Reviewed-by:
Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by:
-
Chris Wilson authored
As we stash a pointer to the HWSP cacheline on the request, when reading it we only need confirm that the cacheline is still valid by checking that the request and timeline are still intact. v2: Protect hwsp_cachline with RCU Signed-off-by:
-
Chris Wilson authored
Since commit e5dadff4 ("drm/i915: Protect request retirement with timeline->mutex"), the request retirement can happen outside of the struct_mutex serialised only by the timeline->mutex. We drop the timeline->mutex on submitting the request (i915_request_add) so after that point, it is liable to be freed. Make sure our local reference is kept alive until we have finished attaching it to the signalers. (Note that this erodes the argument that i915_request_add should consume the reference, but that is a slightly larger patch!) Fixes: e5dadff4 ("drm/i915: Protect request retirement with timeline->mutex") Signed-off-by:
Matthew Auld <matthew.auld@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20191217134729.3297818-1-chris@chris-wilson.co.uk
-
Hans de Goede authored
At least Bay Trail (BYT) and Cherry Trail (CHT) devices can use 1 of 2 different PWM controllers for controlling the LCD's backlight brightness. Either the one integrated into the PMIC or the one integrated into the SoC (the 1st LPSS PWM controller). So far in the LPSS code on BYT we have skipped registering the LPSS PWM controller "pwm_backlight" lookup entry when a Crystal Cove PMIC is present, assuming that in this case the PMIC PWM controller will be used. On CHT we have been relying on only 1 of the 2 PWM controllers being enabled in the DSDT at the same time; and always registered the lookup. So far this has been working, but the correct way to determine which PWM controller needs to be used is by checking a bit in the VBT table and recently I've learned about 2 different BYT devices: Point of View MOBII TAB-P800W Acer Switch 10 SW5-012 Which use a Crystal Cove PMIC, yet the LCD is connected to the SoC/LPSS PWM controller (and the VBT correctly indicates this), so here our old heuristics fail. This commit fixes using the wrong PWM controller on these devices by calling pwm_get() for the right PWM controller based on the VBT dsi.config.pwm_blc bit. Note this is part of a series which contains 2 other patches which renames the PWM lookup for the 1st SoC/LPSS PWM from "pwm_backlight" to "pwm_pmic_backlight" and the PWM lookup for the Crystal Cove PMIC PWM from "pwm_backlight" to "pwm_pmic_backlight". Acked-by:
Jani Nikula <jani.nikula@intel.com> Reviewed-by:
Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by:
Hans de Goede <hdegoede@redhat.com> Link: https://patchwork.freedesktop.org/patch/msgid/20191216202906.1662893-4-hdegoede@redhat.com
-
Hans de Goede authored
At least Bay Trail (BYT) and Cherry Trail (CHT) devices can use 1 of 2 different PWM controllers for controlling the LCD's backlight brightness. Either the one integrated into the PMIC or the one integrated into the SoC (the 1st LPSS PWM controller). So far in the LPSS code on BYT we have skipped registering the LPSS PWM controller "pwm_backlight" lookup entry when a Crystal Cove PMIC is present, assuming that in this case the PMIC PWM controller will be used. On CHT we have been relying on only 1 of the 2 PWM controllers being enabled in the DSDT at the same time; and always registered the lookup. So far this has been working, but the correct way to determine which PWM controller needs to be used is by checking a bit in the VBT table and recently I've learned about 2 different BYT devices: Point of View MOBII TAB-P800W Acer Switch 10 SW5-012 Which use a Crystal Cove PMIC, yet the LCD is connected to the SoC/LPSS PWM controller (and the VBT correctly indicates this), so here our old heuristics fail. Since only the i915 driver has access to the VBT, this commit renames the "pwm_backlight" lookup entries for the Crystal Cove PMIC's PWM controller to "pwm_pmic_backlight" so that the i915 driver can do a pwm_get() for the right controller depending on the VBT bit, instead of the i915 driver relying on a "pwm_backlight" lookup getting registered which magically points to the right controller. Acked-by:
Lee Jones <lee.jones@linaro.org> Link: https://patchwork.freedesktop.org/patch/msgid/20191216202906.1662893-3-hdegoede@redhat.com
-
Hans de Goede authored
At least Bay Trail (BYT) and Cherry Trail (CHT) devices can use 1 of 2 different PWM controllers for controlling the LCD's backlight brightness. Either the one integrated into the PMIC or the one integrated into the SoC (the 1st LPSS PWM controller). So far in the LPSS code on BYT we have skipped registering the LPSS PWM controller "pwm_backlight" lookup entry when a Crystal Cove PMIC is present, assuming that in this case the PMIC PWM controller will be used. On CHT we have been relying on only 1 of the 2 PWM controllers being enabled in the DSDT at the same time; and always registered the lookup. So far this has been working, but the correct way to determine which PWM controller needs to be used is by checking a bit in the VBT table and recently I've learned about 2 different BYT devices: Point of View MOBII TAB-P800W Acer Switch 10 SW5-012 Which use a Crystal Cove PMIC, yet the LCD is connected to the SoC/LPSS PWM controller (and the VBT correctly indicates this), so here our old heuristics fail. Since only the i915 driver has access to the VBT, this commit renames the "pwm_backlight" lookup entries for the 1st BYT/CHT LPSS PWM controller to "pwm_soc_backlight" so that the i915 driver can do a pwm_get() for the right controller depending on the VBT bit, instead of the i915 driver relying on a "pwm_backlight" lookup getting registered which magically points to the right controller. Acked-by:
Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by:
-
Chris Wilson authored
Sandybridge is the gen that didn't handle multiple registers in a single LRI packet. Don't forget it! Fixes: 902eb748 ("drm/i915/gt: Tidy up full-ppgtt on Ivybridge") Signed-off-by:
Tomi Sarvela <tomi.p.sarvela@intel.com> Acked-by:
Mika Kuoppala <mika.kuoppala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20191217091328.3093551-1-chris@chris-wilson.co.uk
-
- 16 Dec, 2019 8 commits
-
-
Chris Wilson authored
We currently use an error-prone mutex_trylock to grab another timeline to find an earlier request along it. However, with a bit of a sleight-of-hand, we can reduce the mutex_trylock to a spin_lock on the immediate request and careful pointer chasing to acquire a reference on the previous request. Signed-off-by:
-
Maya Rashish authored
Signed-off-by:
Maya Rashish <coypu@sdf.org> Reviewed-by:
-
Chris Wilson authored
With a couple more memory barriers dotted around the place we can significantly reduce the MTBF on Ivybridge. Still doesn't really help Haswell though. Signed-off-by:
-
Chris Wilson authored
When creating a handle, it is just that, an abstract handle. The fact that we cannot currently support a handle larger than the size of the backing storage is an artifact of our whole-object-at-a-time handling in get_pages() and being an implementation limitation is best handled at that point -- similar to shmem, where we only barf when asked to populate the whole object if larger than RAM. (Pinning the whole object at a time is major hindrance that we are likely to have to overcome in the near future.) In the case of the buddy allocator, the late check is preferable as the request size may often be smaller than the required size. Signed-off-by:
-
Venkata Sandeep Dhanalakota authored
typecheck() macro creates an huge stack size causing issues with static analysis with coverity, addressing this with creating a local pointer. Fixes: 639f2f24 ("drm/i915: Introduce new macros for tracing") Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Jani Nikula <jani.nikula@intel.com> Signed-off-by:
Venkata Sandeep Dhanalakota <venkata.s.dhanalakota@intel.com> Reviewed-by:
-
zhengbin authored
Fixes coccicheck warning: drivers/gpu/drm/i915/gem/i915_gem_region.c:88:2-3: Unneeded semicolon drivers/gpu/drm/i915/gvt/gtt.c:1285:2-3: Unneeded semicolon Reported-by:
Hulk Robot <hulkci@huawei.com> Signed-off-by:
zhengbin <zhengbin13@huawei.com> Acked-by:
Zhenyu Wang <zhenyuw@linux.intel.com> Reviewed-by:
-
Vandita Kulkarni authored
In some cases like latency[level]==0, wm[level].res_lines>31, min_ddb_alloc can be U16_MAX, exclude it from the WARN_ON. v2: Specify the cases in which we hit U16_MAX, indentation (Ville) Fixes: 10a7e07b ("drm/i915: Make sure cursor has enough ddb for the selected wm level") Suggested-by:
Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by:
Vandita Kulkarni <vandita.kulkarni@intel.com> Reviewed-by:
-
Hans de Goede authored
According to both the old acpi_igd_opregion_spec_0.pdf and the newer skl_opregion_rev0p5.pdf opregion specification documents, if a driver handles hotplug events itself, it should set the opregion CHPD field to 1 to indicate this and the firmware should respond to this by no longer sending ACPI 0x00 notification events on e.g. lid-state changes. Specifically skl_opregion_rev0p5.pdf states thid in the documentation of the CHPD word: "Re-enumeration trigger logic in System BIOS MUST be disabled for all the Operating Systems supporting Hot-Plug (e.g., Windows* Longhorn and above)." Note the MUST in there. We ignore these notifications, so this should not be a problem but many recent DSTDs seem to all have the same copy-pasted bug in the GNOT() AML function which is used to send these notifications. Windows likely does not hit this bug as it presumably correcty sets CHPD to 1. Here is an example of the broken GNOT() method: Method (GNOT, 2, NotSerialized) { ... CEVT = Arg0 CSTS = 0x03 If (((CHPD == Zero) && (Arg1 == Zero))) { If (((OSYS > 0x07D0) || (OSYS < 0x07D6))) { Notify (PCI0, Arg1) } Else { Notify (GFX0, Arg1) } } ... Notice that the condition for the If is always true I believe that the || like needs to be an &&, but there is nothing we can do about this and in my own DSDT archive 55 of the 93 DSDTs have this issue. When the if is true the notification gets send to the PCI root instead of only to the GFX0 device. This causes Linux to re-enumerate PCI devices whenever the LID opens / closes, leading to unexpected messages in dmesg: Suspend through lid close: [ 313.598199] intel_atomisp2_pm 0000:00:03.0: Refused to change power state, currently in D3 [ 313.664453] intel_atomisp2_pm 0000:00:03.0: Refused to change power state, currently in D3 [ 313.737982] pci_bus 0000:01: Allocating resources [ 313.738036] pcieport 0000:00:1c.0: bridge window [io 0x1000-0x0fff] to [bus 01] add_size 1000 [ 313.738051] pcieport 0000:00:1c.0: bridge window [mem 0x00100000-0x000fffff 64bit pref] to [bus 01] add_size 200000 add_align 100000 [ 313.738111] pcieport 0000:00:1c.0: BAR 15: assigned [mem 0x91000000-0x911fffff 64bit pref] [ 313.738128] pcieport 0000:00:1c.0: BAR 13: assigned [io 0x1000-0x1fff] Resume: [ 813.623894] pci 0000:00:03.0: [8086:22b8] type 00 class 0x048000 [ 813.623955] pci 0000:00:03.0: reg 0x10: [mem 0x00000000-0x003fffff] [ 813.630477] pci 0000:00:03.0: BAR 0: assigned [mem 0x91c00000-0x91ffffff] [ 854.579101] intel_atomisp2_pm 0000:00:03.0: Refused to change power state, currently in D3 And more importantly this re-enumeration races with suspend/resume causing enumeration to not be complete when assert_isp_power_gated() from drivers/gpu/drm/i915/display/intel_display_power.c runs. This causes the !pci_dev_present(isp_ids) check in assert_isp_power_gated() to fail making the condition for the WARN true, leading to: [ 813.327886] ------------[ cut here ]------------ [ 813.327898] ISP not power gated [ 813.328028] WARNING: CPU: 2 PID: 2317 at drivers/gpu/drm/i915/display/intel_display_power.c:4870 intel_display_print_error_state+0x2b98/0x3a80 [i915] ... [ 813.328599] ---[ end trace f01e81b599596774 ]--- This commit fixes the unwanted ACPI notification on the PCI root device by setting CHPD to 1, so that the broken if condition in the AML never gets checked as notifications of type 0x00 are disabled altogether. Reviewed-by:
-
- 14 Dec, 2019 1 commit
-
-
Chris Wilson authored
Wait for the object to be idle before changing its cache-level and unbinding. This was dropped as supposedly superfluous from commit 8b1c78e0 ("drm/i915: Avoid calling i915_gem_object_unbind holding object lock"), but it turns out to prevent some cache dirt escaping. Smells like papering over a race... Closes: https://gitlab.freedesktop.org/drm/intel/issues/820 Fixes: 8b1c78e0 ("drm/i915: Avoid calling i915_gem_object_unbind holding object lock") Signed-off-by:
Andi Shyti <andi.shyti@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20191213223140.1830738-1-chris@chris-wilson.co.uk
-
- 13 Dec, 2019 11 commits
-
-
Jani Nikula authored
Commit 4d89adc7 ("drm/i915/display/dsi: Add support to pipe D") added pipe D support for DSI, but failed to update the state readout. Fixes: 4d89adc7 ("drm/i915/display/dsi: Add support to pipe D") Cc: Lucas De Marchi <lucas.demarchi@intel.com> Cc: José Roberto de Souza <jose.souza@intel.com> Cc: Vandita Kulkarni <vandita.kulkarni@intel.com> Signed-off-by:
José Roberto de Souza <jose.souza@intel.com> Signed-off-by:
Lucas De Marchi <lucas.demarchi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20191211110844.2996-1-jani.nikula@intel.com
-
Lucas De Marchi authored
Just like in commit 523e0cc8 ("drm/i915/tgl: allow DVI/HDMI on port A"), the port checks when reading the VBT can easily not match what the platform really exposes. However here we only have some additional debug messages that are not adding much value: in the previous debug message we already print everything we know about the VBT. Instead of keep fixing the possible port assignments according to the platform, just nuke the additional messages. Signed-off-by:
Matt Roper <matthew.d.roper@intel.com> Acked-by:
-
Andi Shyti authored
Add two helpers that for reading the actual GT's frequency. The two helpers are: - intel_rps_read_cagf: reads the frequency and returns it not normalized - intel_rps_read_actual_frequency: provides the frequency in Hz. Use the above helpers in sysfs and debugfs. Signed-off-by:
-
Chris Wilson authored
While not good behaviour, it is, however, established behaviour that we can punt EAGAIN to userspace if we need to retry the ioctl. When trying to acquire a mutex, prefer to use EAGAIN to propagate losing the race so that if it does end up back in userspace, we try again. Fixes: c81471f5 ("drm/i915: Copy across scheduler behaviour flags across submit fences") Closes: https://gitlab.freedesktop.org/drm/intel/issues/800Signed-off-by:
Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20191213160347.1789004-1-chris@chris-wilson.co.uk
-
Venkata Sandeep Dhanalakota authored
New macros ENGINE_TRACE(), CE_TRACE(), RQ_TRACE() and GT_TRACE() are introduce to tag device name and engine name with contexts and requests tracing in i915. Cc: Sudeep Dutt <sudeep.dutt@intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Jani Nikula <jani.nikula@intel.com> Signed-off-by:
-
Venkata Sandeep Dhanalakota authored
We do not require to register the sysctl paths per instance, so making registration global. v2: make sysctl path register and unregister function driver specific (Tvrtko and Lucas). Cc: Sudeep Dutt <sudeep.dutt@intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Jani Nikula <jani.nikula@intel.com> Signed-off-by: -
Matt Roper authored
Now that the combo PHY aux power well handlers are used exclusively on Icelake, we can drop a bunch of the extra tests. v2: Don't try to use intel_uncore_rmw for register updates yet; there's pending display uncore patches that need to land first. (Lucas) v3: Drop the combo phy assertion. It was backward before, but doesn't seem terribly necessary. I'm keeping the IS_ICELAKE assertion though since we often copy/paste/modify the power well tables when defining new platforms and it's too easy to cargo cult the ICL-specific handling to new platforms that shouldn't use it. (Lucas) v4: Fix build; forgot to commit all the changes. (CI) Cc: Lucas De Marchi <lucas.demarchi@intel.com> Signed-off-by: -
Matt Roper authored
The TGL workaround database no longer shows Wa #1178 (or anything similar under different workaround names/numbers) so we should be able to drop it. In fact Swati just discovered that applying this workaround is the root cause of some power well enable failures we've been seeing in CI (gitlab issue 498). Once we stop applying this WA, TGL no longer utilizes any of the special handling provided by icl_combo_phy_aux_power_well_ops so we can just drop back to using the standard hsw-style power well ops instead. v3: Drop now-unused _TGL_AUX_ANAOVRD1_C definition too. (Lucas) Closes: https://gitlab.freedesktop.org/drm/intel/issues/498 Fixes: deea06b4 ("drm/i915/tgl: apply Display WA #1178 to fix type C dongles") Cc: Lucas De Marchi <lucas.demarchi@intel.com> Cc: Swati Sharma <swati2.sharma@intel.com> Cc: Imre Deak <imre.deak@intel.com> Signed-off-by:
-
Matt Roper authored
Outputs C and D on EHL are combo PHY outputs and thus should not be using the same TC AUX power well handlers as ICL. And even though icl_combo_phy_aux_power_well_ops works okay for EHL/JSL combo PHYs none of its special handling is actually necessary for this platform: * EHL/JSL don't actually need to program PORT_CL_DW12 * Display WA #1178 does not apply to EHL/JSL Thus we can simply drop back to using our standard "hsw-style" power well ops for EHL AUX power wells. Bspec: 4301 Fixes: f722b8c1 ("drm/i915/ehl: All EHL ports are combo phys") Cc: Jose Souza <jose.souza@intel.com> Cc: Bob Paauwe <bob.j.paauwe@intel.com> Cc: Vivek Kasireddy <vivek.kasireddy@intel.com> Cc: Lucas De Marchi <lucas.demarchi@intel.com> Signed-off-by:
-
Dan Carpenter authored
We know that "err" is non-zero so there is no need to check. Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by:
-
Dan Carpenter authored
The "num_dtd" variable is the number of elements in the generic_dtd->dtd[] array so the > needs to be >= to prevent reading one element beyond the end of the array. Fixes: 33ef6d4f ("drm/i915/vbt: Handle generic DTD block") Reviewed-by:
-
- 12 Dec, 2019 9 commits
-
-
Ville Syrjälä authored
skl_commit_modeset_enables() is a bit of mess. Let's streamline it by simply tracking which pipes still need to be updated. As a bonus we get rid of the state->wm_results.dirty_pipes usage. v2: Rebase due to port sync Cc: José Roberto de Souza <jose.souza@intel.com> Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com> #v1 Signed-off-by:
-
Lee Shawn C authored
U series device need different DDI buffer setup for eDP and DP. If driver did not recognize ULT id proerply. The setting for H and S series would be used. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Jani Nikula <jani.nikula@intel.com> Cc: Anusha Srivatsa <anusha.srivatsa@intel.com> Cc: Cooper Chiou <cooper.chiou@intel.com> Signed-off-by:
Lee Shawn C <shawn.c.lee@intel.com> Signed-off-by:
-
Lee Shawn C authored
commit 'a7b4deeb ("drm/i915/cml: Add CML PCI IDS)' introduced new PCI ID that CML support. But some PCI IDs were removed in BSpec for CML. This patch is used to eliminate the unsed ID. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Jani Nikula <jani.nikula@intel.com> Cc: Anusha Srivatsa <anusha.srivatsa@intel.com> Cc: Cooper Chiou <cooper.chiou@intel.com> Signed-off-by:
-
Chris Wilson authored
drivers/gpu/drm/i915/gt/intel_rps.c:1726:24: error: incompatible types in comparison expression (different address spaces): drivers/gpu/drm/i915/gt/intel_rps.c:1726:24: struct drm_i915_private [noderef] <asn:4> * drivers/gpu/drm/i915/gt/intel_rps.c:1726:24: struct drm_i915_private * Signed-off-by:
-
Chris Wilson authored
Since dma_fence_init may call ops (because of a meaningless trace_dma_fence), we need to set the worker ops prior to that call. Reported-by:
Vince Weaver <vincent.weaver@maine.edu> Fixes: 8e458fe2 ("drm/i915: Generalise the clflush dma-worker") Signed-off-by:
-
Michal Wajdeczko authored
On non-debug builds we were not using i915 param and thus we may cause "unused variable" warning/error if caller was not using i915 elsewhere. Let compiler see this param. Signed-off-by:
-
Chris Wilson authored
Execute the cmdparser asynchronously as part of the submission pipeline. Using our dma-fences, we can schedule execution after an asynchronous piece of work, so we move the cmdparser out from under the struct_mutex inside execbuf as run it as part of the submission pipeline. The same security rules apply, we copy the user batch before validation and userspace cannot touch the validation shadow. The only caveat is that we will do request construction before we complete cmdparsing and so we cannot know the outcome of the validation step until later -- so the execbuf ioctl does not report -EINVAL directly, but we must cancel execution of the request and flag the error on the out-fence. Closes: https://gitlab.freedesktop.org/drm/intel/issues/611 Closes: https://gitlab.freedesktop.org/drm/intel/issues/412Signed-off-by:
-
Chris Wilson authored
The gen7 cmdparser is primarily a promotion-based system to allow access to additional registers beyond the HW validation, and allows fallback to normal execution of the user batch buffer if valid and requires chaining. In the next patch, we will do the cmdparser validation in the pipeline asynchronously and so at the point of request construction we will not know if we want to execute the privileged and validated batch, or the original user batch. The solution employed here is to execute both batches, one with raised privileges and one as normal. This is because the gen7 MI_BATCH_BUFFER_START command cannot change privilege level within a batch and must strictly use the current privilege level (or undefined behaviour kills the GPU). So in order to execute the original batch, we need a second non-priviledged batch buffer chain from the ring, i.e. we need to emit two batches for each user batch. Inside the two batches we determine which one should actually execute, we provide a conditional trampoline to call the original batch. Implementation-wise, we create a single buffer and write the shadow and the trampoline inside it at different offsets; and bind the buffer into both the kernel GGTT for the privileged execution of the shadow and into the user ppGTT for the non-privileged execution of the trampoline and original batch. One buffer, two batches and two vma. Signed-off-by:
-
Chris Wilson authored
An oversight in that we use rc6->ctl_enable to disable rc6 on gen9 and so it does not simply indicate indirect control via a PCU. Switch the rc6->ctl_enable check for a platform-based check. Fixes: 972745fd ("drm/i915/gt: Disable manual rc6 for Braswell/Baytrail") Signed-off-by:
-
