- 07 Mar, 2020 5 commits
-
-
Chris Wilson authored
Be sure to wait for the vma to be in place before we tell the GPU to execute from the wa batch. Since initialisation is mostly synchronous (or rather at some point during start up we will need to sync anyway), we can affort to do an explicit i915_vma_sync() during wa batch construction rather than check for a required await on every context switch. (We don't expect to change the wa bb at run time so paying the cost once up front seems preferrable.) Fixes: ee2413ee ("drm/i915: Add mechanism to submit a context WA on ring submission") Signed-off-by:
Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Reviewed-by:
Mika Kuoppala <mika.kuoppala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20200307122425.29114-1-chris@chris-wilson.co.uk
-
Chris Wilson authored
If the cacheline may still be busy, atomically mark it for future release, and only if we can determine that it will never be used again, immediately free it. Closes: https://gitlab.freedesktop.org/drm/intel/issues/1392 Fixes: ebece753 ("drm/i915: Keep timeline HWSP allocated until idle across the system") Signed-off-by:
-
Chris Wilson authored
If we stop filling the ELSP due to an incompatible virtual engine request, check if we should enable the timeslice on behalf of the queue. This fixes the case where we are inspecting the last->next element when we know that the last element is the last request in the execution queue, and so decided we did not need to enable timeslicing despite the intent to do so! Fixes: 8ee36e04 ("drm/i915/execlists: Minimalistic timeslicing") Signed-off-by:
-
Chris Wilson authored
Due to the ordering of cmpxchg()/dma_fence_signal() inside node_retire(), we must also use the xchg() as our primary memory barrier to flush the outstanding callbacks after expected completion of the i915_active. Signed-off-by:
-
Chris Wilson authored
Do not poison the timeline link on the i915_request to allow both forward/backward list traversal under RCU. [ 9759.139229] RIP: 0010:active_request+0x2a/0x90 [i915] [ 9759.139240] Code: 41 56 41 55 41 54 55 48 89 fd 53 48 89 f3 48 83 c5 60 e8 49 de dc e0 48 8b 83 e8 01 00 00 48 39 c5 74 12 48 8d 90 20 fe ff ff <48> 8b 80 50 fe ff ff a8 01 74 11 e8 66 20 dd e0 48 89 d8 5b 5d 41 [ 9759.139251] RSP: 0018:ffffc9000014ce80 EFLAGS: 00010012 [ 9759.139260] RAX: dead000000000122 RBX: ffff888817cac040 RCX: 0000000000022000 [ 9759.139267] RDX: deacffffffffff42 RSI: ffff888817cac040 RDI: ffff888851fee900 [ 9759.139275] RBP: ffff888851fee960 R08: 000000000000001a R09: ffffffffa04702e0 [ 9759.139282] R10: ffffffff82187ea0 R11: 0000000000000002 R12: 0000000000000004 [ 9759.139289] R13: ffffffffa04d5179 R14: ffff8887f994ae40 R15: ffff888857b9a068 [ 9759.139296] FS: 0000000000000000(0000) GS:ffff88885ed80000(0000) knlGS:0000000000000000 [ 9759.139304] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 9759.139311] CR2: 00007fff5bdec000 CR3: 00000008534fe001 CR4: 00000000001606e0 [ 9759.139318] Call Trace: [ 9759.139325] <IRQ> [ 9759.139389] execlists_reset+0x14d/0x310 [i915] [ 9759.139400] ? _raw_spin_unlock_irqrestore+0xf/0x30 [ 9759.139445] ? fwtable_read32+0x90/0x230 [i915] [ 9759.139499] execlists_submission_tasklet+0xf6/0x150 [i915] [ 9759.139508] tasklet_action_common.isra.17+0x32/0xa0 [ 9759.139516] __do_softirq+0x114/0x3dc [ 9759.139525] ? handle_irq_event_percpu+0x59/0x70 [ 9759.139533] irq_exit+0xa1/0xc0 [ 9759.139540] do_IRQ+0x76/0x150 [ 9759.139547] common_interrupt+0xf/0xf [ 9759.139554] </IRQ> Signed-off-by:
-
- 06 Mar, 2020 12 commits
-
-
Swathi Dhanavanthri authored
This workaround is to disable FF DOP Clock gating. The fix in B0 was backed out due to timing reasons and decided to be made permanent. Bspec: 52890 Signed-off-by:
Swathi Dhanavanthri <swathi.dhanavanthri@intel.com> Reviewed-by:
José Roberto de Souza <jose.souza@intel.com> Signed-off-by:
-
Vivek Kasireddy authored
On some platforms such as Elkhart Lake, although we may use DDI D to drive a connector, we have to use PHY A (Combo Phy PORT A) to detect the hotplug interrupts as per the spec because there is no one-to-one mapping between DDIs and PHYs. Therefore, use the function intel_port_to_phy() which contains the logic for such mapping(s) to find the correct hpd_pin. This change should not affect other platforms as there is always a one-to-one mapping between DDIs and PHYs. v2: - Convert the case statements to use PHYs instead of PORTs (Jani) v3: - Refactor the function to reduce the number of return statements by lumping all the case statements together except PHY_F which needs special handling (Jose) v4: - Add a comment describing how the HPD pin value associated with any port can be retrieved using port or phy enum value. (Jani) v5: - Use case ranges instead of individual labels and also normalize the return statement by adding -PHY_A to the expression (Ville) Cc: Jani Nikula <jani.nikula@intel.com> Cc: Matt Roper <matthew.d.roper@intel.com> Cc: José Roberto de Souza <jose.souza@intel.com> Cc: Ville Syrjala <ville.syrjala@linux.intel.com> Signed-off-by:
Vivek Kasireddy <vivek.kasireddy@intel.com> Reviewed-by:
-
Matthew Auld authored
Depending on RNG we might try to fill an 8G region for every possible order, using the smallest possible chunk size of 4K, which seems to be very slow. Try to remedy the situation by adding an overall timeout for the test, while also selecting each order level in a random fashion. Closes: https://gitlab.freedesktop.org/drm/intel/issues/1310Signed-off-by:
Matthew Auld <matthew.auld@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by:
-
Matthew Auld authored
Be careful not to mark an already free node as free again. Signed-off-by:
-
Matthew Auld authored
Check the edge case where batch_start_offset sits exactly on the batch size. v2: add new range_overflows variant to capture the special case where the size is permitted to be zero, like with batch_len. v3: other way around. the common case is the exclusive one which should just be >=, with that we then just need to convert the three odd ball cases that don't apply to use the new inclusive _end version. Testcase: igt/gem_exec_params/invalid-batch-start-offset Fixes: 0b537272 ("drm/i915/cmdparser: Use cached vmappings") Signed-off-by:
-
Chris Wilson authored
We only need to serialise the multiple pinning during the eb_reserve phase. Ideally this would be using the vm->mutex as an outer lock, or using a composite global mutex (ww_mutex), but at the moment we are using struct_mutex for the group. Closes: https://gitlab.freedesktop.org/drm/intel/issues/1381 Fixes: 003d8b91 ("drm/i915/gem: Only call eb_lookup_vma once during execbuf ioctl") Signed-off-by:
-
Chris Wilson authored
We only call i915_schedule() when we know we have changed the priority on a request and so require to propagate any change in priority to its signalers (for PI). By unconditionally checking all of our signalers, we avoid skipping changes made prior to construction of the request (as the request may be waited upon before submission when used in parallel). References: https://gitlab.freedesktop.org/drm/intel/issues/1318Signed-off-by:
-
Chris Wilson authored
Check the flow of requests into the hardware to verify that are submitted in order along their timeline. Signed-off-by:
-
Matthew Auld authored
The alignment is u64, and yet is_power_of_2() assumes unsigned long, which might give different results between 32b and 64b kernel. Signed-off-by:
-
Abdiel Janulgue authored
The release method will undo what we did at creation, and so we shouldn't care if we have pages or not. Fixes a small leak in the mock_phys selftest. Signed-off-by:
Abdiel Janulgue <abdiel.janulgue@linux.intel.com> Signed-off-by:
-
Prathap Kumar Valsan authored
On gen7 and gen7.5 devices, there could be leftover data residuals in EU/L3 from the retiring context. This patch introduces workaround to clear that residual contexts, by submitting a batch buffer with dedicated HW context to the GPU with ring allocation for each context switching. This security mitigation changes does not triggers any performance regression. Performance is on par with current drm-tips. v2: Add igt generated header file for CB kernel assembled with Mesa tool and addressed use of Kernel macro for ptr_align comment. v3: Resolve Sparse warnings with newly generated, and imported CB kernel. v4: Include new igt generated CB kernel for gen7 and gen7.5. Also add code formatting and compiler warnings changes (Chris Wilson) Signed-off-by:
Prathap Kumar Valsan <prathap.kumar.valsan@intel.com> Signed-off-by:
Akeem G Abodunrin <akeem.g.abodunrin@intel.com> Cc: Chris Wilson <chris.p.wilson@intel.com> Cc: Balestrieri Francesco <francesco.balestrieri@intel.com> Cc: Bloomfield Jon <jon.bloomfield@intel.com> Cc: Dutt Sudeep <sudeep.dutt@intel.com> Acked-by:
Chris Wilson <chris@chris-wilso.co.uk> Signed-off-by:
-
Mika Kuoppala authored
This patch adds framework to submit an arbitrary batchbuffer on each context switch to clear residual state for render engine on Gen7/7.5 devices. The idea of always emitting the context and vm setup around each request is primary to make reset recovery easy, and not require rewriting the ringbuffer. As each request would set up its own context, leaving it to the HW to notice and elide no-op context switches, we could restart the ring at any point, and reorder the requests freely. However, to avoid emitting clear_residuals() between consecutive requests in the ringbuffer of the same context, we do want to track the current context in the ring. In doing so, we need to be careful to only record a context switch when we are sure the next request will be emitted. This security mitigation change does not trigger any performance regression. Performance is on par with current mainline/drm-tip. v2: Update vm_alias params to point to correct address space "vm" due to changes made in the patch "f2161379" v3-v4: none Signed-off-by:
-
- 05 Mar, 2020 9 commits
-
-
Chris Wilson authored
Show the timeslicing priority hint in engine dumps to aide debugging. Signed-off-by:
-
Chris Wilson authored
Requests within a timeline are ordered by that timeline, so awaiting for the start of a request within the timeline is a no-op. This used to work by falling out of the mutex_trylock() as the signaler and waiter had the same timeline and not returning an error. Fixes: 6a79d848 ("drm/i915: Lock signaler timeline while navigating") Signed-off-by:
Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20200305134822.2750496-1-chris@chris-wilson.co.uk
-
Chris Wilson authored
Fix the inverted test to emit the wait on the end of the previous request if we /haven't/ already. Fixes: 6a79d848 ("drm/i915: Lock signaler timeline while navigating") Signed-off-by:
-
Swati Sharma authored
Converting error to debug print if sink fails to configure scrambling or TMDS bit clock ratio. In this case, we are timing out while disabling the scrambling and setting the SCDC ratio, as there is no response to the I2C SCDC write from the sink device. Error isn't due to something wrong done from driver side. Signed-off-by:
Swati Sharma <swati2.sharma@intel.com> Suggested-by:
Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by:
Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
-
Ville Syrjälä authored
Display w/a #1140 tells us we have to program the transition watermark to the minimum value on glk/cnl. Let's do that. Signed-off-by:
Rodrigo Vivi <rodrigo.vivi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20200228203552.30273-4-ville.syrjala@linux.intel.com
-
Ville Syrjälä authored
We are mistakenly skipping transition watermarks on glk. Fix up the condition for glk, and toss in the w/a name from the database. v2: Reorder the ipc enabled vs. platform check to be more sensible Signed-off-by:
-
Ville Syrjälä authored
Currently we're comparing the watermarks between the old and new states before we've fully computed the new watermarks. In particular skl_build_pipe_wm() will not account for the amount of ddb space we'll have. That information is only available during skl_compute_ddb() which will proceed to zero out any watermark level exceeding the ddb allocation. If we're short on ddb space this will end up adding the plane to the state due erronously determining that the watermarks have changed. Fix the problem by deferring skl_wm_add_affected_planes() until we have the final watermarks computed. Noticed this when trying enable transition watermarks on glk. We now computed the trans_wm as 28, but we only had 14 blocks of ddb, and thus skl_compute_ddb() ended up disabling the cursor trans_wm every time. Thus we ended up adding the cursor to every commit that didn't actually affect the cursor at all. Signed-off-by:
-
Ville Syrjälä authored
The hardware never sees the uv_wm values (apart from uv_wm.min_ddb_alloc affecting the ddb allocation). Thus there is no point in comparing uv_wm to determine if we need to reprogram the watermark registers. So let's check only the rgb/y watermark in skl_plane_wm_equals(). But let's leave a comment behind so that the next person reading this doesn't get as confused as I did when I added this check. If the ddb allocation ends up changing due to uv_wm skl_ddb_add_affected_planes() takes care of adding the plane to the state. TODO: we should perhaps just eliminate uv_wm from the state and simply track the min_ddb_alloc for uv instead. Cc: Matt Roper <matthew.d.roper@intel.com> Signed-off-by:
-
Tvrtko Ursulin authored
Enable FtrPerCtxtPreemptionGranularityControl bit and select thread- group as the default preemption level. v2: * Remove register whitelisting (Rafael, Tony). Signed-off-by:
Rafael Antognolli <rafael.antognolli@intel.com> Acked-by:
Jason Ekstrand <jason@jlekstrand.net> Acked-by:
Tony Ye <tony.ye@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20200304153144.10675-1-tvrtko.ursulin@linux.intel.com
-
- 04 Mar, 2020 12 commits
-
-
Chris Wilson authored
As we started marking the ce->gem_context as NULL on closure, we can no longer use that to carry closure information. Instead, we can look at whether the context was killed on closure instead. Fixes: 130a95e9 ("drm/i915/gem: Consolidate ctx->engines[] release") Closes: https://gitlab.freedesktop.org/drm/intel/issues/1379Signed-off-by:
-
Hans de Goede authored
The Thundersoft TST178 tablet uses a DSI panel with an external PWM controller (as all DSI panels do). But unlike other DSI panels a duty-cycle of 100% turns the backlight off and 0% sets it to maximum brightness. I've checked the VBT and there is a BDB_LVDS_BACKLIGHT section, but it does not set the active_low_pwm flag. This tablet re-uses the main PCI vendor and product ids for the subsystem ids, so I see no other option then to add a DMI based quirk to fix this. Note that the PWM backlight code in intel_panel.c currently does not honor the vbt.active_low_pwm flag, but that does not matter in this case. Signed-off-by:
Hans de Goede <hdegoede@redhat.com> Reviewed-by:
Jani Nikula <jani.nikula@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20200221172927.510027-2-hdegoede@redhat.com
-
Hans de Goede authored
Use intel_panel_compute_brightness() from pwm_setup_backlight() so that we correctly take i915_modparams.invert_brightness and/or QUIRK_INVERT_BRIGHTNESS into account when setting + getting the initial brightness value. Signed-off-by:
-
Chris Wilson authored
As we release the head requests back into the queue, propagate any change in error status that may have occurred while the requests were temporarily suspended. Closes: https://gitlab.freedesktop.org/drm/intel/issues/1277Signed-off-by:
-
Chris Wilson authored
Trying to use i915_request_skip() prior to i915_request_add() causes us to try and fill the ring upto request->postfix, which has not yet been set, and so may cause us to memset() past the end of the ring. Instead of skipping the request immediately, just flag the error on the request (only accepting the first fatal error we see) and then clear the request upon submission. Signed-off-by:
-
Matt Roper authored
intel_ddi_clock_get() tests the DPLL ID against DPLL_ID_ICL_TBTPLL (2) to determine whether to try to descend into a TBT-specific handler. However this test will also be true when DPLL4 on EHL is used since that shares the same DPLL ID (2). Add an extra check to ensure the PHY is actually a Type-C PHY before descending into the TBT handling. This should ensure EHL still takes the correct code path and somewhat future-proof the code as well. v2: Drop the gen+ check since only gen11+ platforms can have Type-C outputs. (Imre) Cc: José Roberto de Souza <jose.souza@intel.com> Cc: Imre Deak <imre.deak@intel.com> Closes: https://gitlab.freedesktop.org/drm/intel/issues/1369 Fixes: 45e4728b ("drm/i915: Move DPLL frequency calculation to intel_dpll_mgr.c") Signed-off-by:Matt Roper <matthew.d.roper@intel.com> Reviewed-by:
Imre Deak <imre.deak@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20200303195043.959913-1-matthew.d.roper@intel.com
-
Chris Wilson authored
drivers/gpu/drm/i915/gvt/gvt.c:264:6: error: no previous prototype for ‘intel_gvt_clean_device’ [-Werror=missing-prototypes] drivers/gpu/drm/i915/gvt/gvt.c:301:5: error: no previous prototype for ‘intel_gvt_init_device’ [-Werror=missing-prototypes] Signed-off-by:
-
Lyude Paul authored
According to Dell, trying to match their panels via OUI is not reliable enough and we've been told that we should check against the EDID instead. As well, Dell seems to have some panels that are actually intended to switch between using PWM for backlight controls and DPCD for backlight controls depending on whether or not the panel is in HDR or SDR mode. Yikes. Regardless, we need to add quirks for these so that DPCD backlight controls get enabled by default, since without additional driver support that's the only form of brightness control that will work. Hopefully in the future we can remove these quirks once we have a better way of probing for this. Changes since v1: * Add one more EDID per Dell's request * Remove model number (which is possibly wrong) and replace with Dell CML 2020 systems Signed-off-by:
Lyude Paul <lyude@redhat.com> Cc: Jani Nikula <jani.nikula@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20200211183358.157448-4-lyude@redhat.comReviewed-by:
Adam Jackson <ajax@redhat.com>
-
Lyude Paul authored
The X1 Extreme is one of the systems that lies about which backlight interface that it uses in its VBIOS as PWM backlight controls don't work at all on this machine. It's possible that this panel could be one of the infamous ones that can switch between PWM mode and DPCD backlight control mode, but we haven't gotten any more details on this from Lenovo just yet. For the time being though, making sure the backlight 'just works' is a bit more important. So, add a quirk to force DPCD backlight controls on for these systems based on EDID (since this panel doesn't appear to fill in the device ID). Hopefully in the future we'll figure out a better way of probing this. Changes since v2: * The bugzilla URL is deprecated, bug reporting happens on gitlab now. Update the messages we print to reflect this * Also, take the opportunity to move FDO_BUG_URL out of i915_utils.c and into i915_utils.h so that other places which print things that aren't traditional errors but are worth filing bugs about, can actually use it. Signed-off-by:
-
Lyude Paul authored
The whole point of using OUIs is so that we can recognize certain devices and potentially apply quirks for them. Normally this should work quite well, but there appears to be quite a number of laptop panels out there that will fill the OUI but not the device ID. As such, for devices like this I can't imagine it's a very good idea to try relying on OUIs for applying quirks. As well, some laptop vendors have confirmed to us that their panels have this exact issue. So, let's introduce the ability to apply DP quirks based on EDID identification. We reuse the same quirk bits for OUI-based quirks, so that callers can simply check all possible quirks using drm_dp_has_quirk(). Signed-off-by:
-
Anshuman Gupta authored
DP shim's config_stream_type considered to be succeeded when return value of intel_dp_hdcp2_write_msg() equals to size of message to be written, it makes config_stream_type to return a zero success value in order to succeed the HDCP auth. v2: - config_stream_type() returns 0 on success. [Ram] CC: Ramalingam C <ramalingam.c@intel.com> Signed-off-by:
Anshuman Gupta <anshuman.gupta@intel.com> Reviewed-by:
Ramalingam C <ramalingam.c@intel.com> Signed-off-by:
-
Anshuman Gupta authored
HDCP Repeater initializes seq_num_V to 0 at the beginning of hdcp Session i.e. after AKE_init received, refer HDCP 2.2 Spec HDMI PAGE 19, DP PAGE 20. HDCP 2.2 Comp specs 1B-06 test verifies that whether DUT considers failure of authentication if the repeater provides a non-zero value in seq_num_V in the first, RepeaterAuth_Send_ReceiverID_List message. Make sure that HDCP repeater initializes seq_num_V to zero at beginning of session i.e. after AKE_Init, fail the Auth if there is non zero seq_num_V. v2: - Used existing hdcp2_encrypted flag instead of declaring new flag. [Ram] Cc: Ramalingam C <ramalingam.c@intel.com> Reviewed-by:
-
- 03 Mar, 2020 2 commits
-
-
Chris Wilson authored
As we no longer stash anything inside i915_vma under the exclusive protection of struct_mutex, we do not need to revoke the i915_vma stashes before dropping struct_mutex to handle pagefaults. Knowing that we must drop the struct_mutex while keeping the eb->vma around, means that we are required to hold onto to the object reference until we have marked the vma as active. Fixes: 155ab883 ("drm/i915: Move object close under its own lock") Signed-off-by:
Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20200303204345.1859734-3-chris@chris-wilson.co.uk
-
Chris Wilson authored
For our convenience, and to avoid frequent allocations, we placed some lists we use for execbuf inside the common i915_vma struct. As we look to parallelise execbuf, such fields guarded by the struct_mutex BKL must be pulled under local control. Instead of using the i915_vma as our primary means of tracking the user's list of objects and their virtual mappings, we use a local eb_vma with the same lists as before (just now local not global). This should allow us to only perform the lookup of vma used for execution once during the execbuf ioctl, as currently we need to remove our secrets from inside i915_vma everytime we drop the struct_mutex as another execbuf may use the shared locations. Once potential user visible consequence is that we can remove the requirement that the execobj[] be unique, and only require that they do not conflict (i.e. you cannot softpin the same object into two locations. Signed-off-by:
-
