- 23 Dec, 2021 4 commits
-
-
John Harrison authored
A fault injection probe test hit a BUG_ON in a GuC error path. It showed that the GuC code could potentially attempt to do many things when the device is actually wedged. So, add a check in to prevent that. v2: Use intel_gt_is_wedged instead of testing bits directly in the GuC submission code (review feedback from Tvrtko). Signed-off-by:
John Harrison <John.C.Harrison@Intel.com> Reviewed-by:
Matthew Brost <matthew.brost@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20211221210212.1438670-1-John.C.Harrison@Intel.com
-
Matthew Brost authored
A weak implementation of parallel submission (multi-bb execbuf IOCTL) for execlists. Doing as little as possible to support this interface for execlists - basically just passing submit fences between each request generated and virtual engines are not allowed. This is on par with what is there for the existing (hopefully soon deprecated) bonding interface. We perma-pin these execlists contexts to align with GuC implementation. v2: (John Harrison) - Drop siblings array as num_siblings must be 1 v3: (John Harrison) - Drop single submission v4: (John Harrison) - Actually drop single submission - Use IS_ERR check on return value from intel_context_create - Set last request to NULL on unpin Signed-off-by:
-
John Harrison authored
Don't silently drop reset notifications from the GuC. It might not be safe to do an error capture but we still want some kind of report that the reset happened. Signed-off-by:
-
Maarten Lankhorst authored
Convert free_work into delayed_work, similar to ttm to allow converting the blocking lock in __i915_gem_free_objects to a trylock. Unlike ttm, the object should already be idle, as it's kept alive by a reference through struct i915_vma->active, which is dropped after all vma's are idle. Because of this, we can use a no wait by default, or when the lock is contested, we use ttm's 10 ms. The trylock should only fail when the object is sharing it's resv with other objects, and typically objects are not kept locked for a long time, so we can safely retry on failure. Fixes: be7612fd ("drm/i915: Require object lock when freeing pages during destruction") Testcase: igt/gem_exec_alignment/pi* Signed-off-by:
Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Reviewed-by:
Thomas Hellström <thomas.hellstrom@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20211222155622.2960379-1-maarten.lankhorst@linux.intel.com
-
- 22 Dec, 2021 4 commits
-
-
Thomas Hellström authored
Protect updates of struct i915_vma flags and async binding / unbinding with the vm::mutex. This means that i915_vma_bind() needs to assert vm::mutex held. In order to make that possible drop the caching of kmap_atomic() maps around i915_vma_bind(). An alternative would be to use kmap_local() but since we block cpu unplugging during sleeps inside kmap_local() sections this may have unwanted side-effects. Particularly since we might wait for gpu while holding the vm mutex. This change may theoretically increase execbuf cpu-usage on snb, but at least on non-highmem systems that increase should be very small. Signed-off-by:
Matthew Auld <matthew.auld@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20211221200050.436316-5-thomas.hellstrom@linux.intel.com
-
Thomas Hellström authored
Since it's starting to be used outside the i915 TTM move code, move it to a separate set of files. v2: - Update the documentation. v4: - Rebase. Signed-off-by:
-
Christian König authored
First of all as discussed multiple times now kernel copies *must* always wait for all fences in a BO before actually doing the copy. This is mandatory. Additional to that drop the handling when there can't be a shared slot allocated on the source BO and just properly return an error code. Otherwise this code path would only be tested under out of memory conditions. Signed-off-by:
Christian König <christian.koenig@amd.com> Reviewed-by:
-
Thomas Hellström authored
Since the gt migration code was using only a single fence for dependencies, these were collected in a dma_fence_array. However, it turns out that it's illegal to use some dma_fences in a dma_fence_array, in particular other dma_fence_arrays and dma_fence_chains, and this causes trouble for us moving forward. Have the gt migration code instead take a const struct i915_deps for dependencies. This means we can skip the dma_fence_array creation and instead pass the struct i915_deps instead to circumvent the problem. v2: - Make the prev_deps() function static. (kernel test robot <lkp@intel.com>) - Update the struct i915_deps kerneldoc. v4: - Rebase. Fixes: 5652df82 ("drm/i915/ttm: Update i915_gem_obj_copy_ttm() to be asynchronous") Signed-off-by:
-
- 21 Dec, 2021 4 commits
-
-
Vinay Belgaumkar authored
By default, GT (and GuC) run at RPn. Requesting for RP0 before firmware load can speed up DMA and HuC auth as well. In addition to writing to 0xA008, we also need to enable swreq in 0xA024 so that Punit will pay heed to our request. SLPC will restore the frequency back to RPn after initialization, but we need to manually do that for the non-SLPC path. We don't need a manual override in the SLPC disabled case, just use the intel_rps_set function to ensure consistent RPS state. Signed-off-by:
Vinay Belgaumkar <vinay.belgaumkar@intel.com> Reviewed-by:
Sujaritha Sundaresan <sujaritha.sundaresan@intel.com> Signed-off-by:
-
Maarten Lankhorst authored
This is required for i915_gem_evict_vm, to be able to evict the entire VM, including objects that are already locked to the current ww ctx. Signed-off-by:
-
Maarten Lankhorst authored
TTM already requires this, and we require it for delayed destroy. Signed-off-by:
-
Maarten Lankhorst authored
We're working on requiring the obj->resv lock during unbind, fix the shrinker to take the object lock. Signed-off-by:
-
- 20 Dec, 2021 10 commits
-
-
John Harrison authored
If GuC encounters an error during engine reset, the i915 driver promotes to full GT reset. This includes an info message about why the reset is happening. However, that is not treated as a failure by any of the CI systems because resets are an expected occurrance during testing. This kind of failure is a major problem and should never happen. So, complain more loudly and make sure CI notices. Signed-off-by:
Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20211211065859.2248188-4-John.C.Harrison@Intel.com
-
John Harrison authored
Lots of testing is done with the DEBUG_GEM config option enabled but not the DEBUG_GUC option. That means we only get teeny-tiny GuC logs which are not hugely useful. Enabling full DEBUG_GUC also spews lots of other detailed output that is not generally desired. However, bigger GuC logs are extremely useful for almost any regression debug. So enable bigger logs for DEBUG_GEM builds as well. Signed-off-by:
-
John Harrison authored
Add support for telling the debugfs interface the size of the GuC log dump in advance. Without that, the underlying framework keeps calling the 'show' function with larger and larger buffer allocations until it fits. That means reading the log from graphics memory many times - 16 times with the full 18MB log size. v2: Don't return error codes from size query. Report overflow in the error dump as well (review feedback from Daniele). Signed-off-by:
-
Maarten Lankhorst authored
Now that we require locking to evict, multiple vmas from the same object might not be evicted. This is expected and required, because execbuf will move to short-term pinning by using the lock only. This will cause these tests to fail, because they create a ton of vma's for the same object. Unbind manually to prevent spurious -ENOSPC in those mock tests. Signed-off-by:
-
Maarten Lankhorst authored
In the next commits, we may not evict when refcount = 0. igt_vm_isolation() continuously tries to pin/unpin at same address, but also calls put() on the object, which means the object may not be unpinned in time. Instead of this, re-use the same object over and over, so they can be unbound as required. Changes since v1: - Fix cleaning up obj_b on failure. (Matt) Signed-off-by:
-
Maarten Lankhorst authored
We will need the lock to unbind the vma, and wait for bind to complete. Remove the special casing for the !ww path, and force ww locking for all. Changes since v1: - Pass err to for_i915_gem_ww handling for -EDEADLK handling. Signed-off-by:
-
Maarten Lankhorst authored
i915_vma_wait_for_bind needs the vma lock held, fix the caller. Signed-off-by:
-
Maarten Lankhorst authored
Big delta, but boils down to moving set_pages to i915_vma.c, and removing the special handling, all callers use the defaults anyway. We only remap in ggtt, so default case will fall through. Because we still don't require locking in i915_vma_unpin(), handle this by using xchg in get_pages(), as it's locked with obj->mutex, and cmpxchg in unpin, which only fails if we race a against a new pin. Changes since v1: - aliasing gtt sets ZERO_SIZE_PTR, not -ENODEV, remove special case from __i915_vma_get_pages(). (Matt) Changes since v2: - Free correct old pages in __i915_vma_get_pages(). (Matt) Remove race of clearing vma->pages accidentally from put, free it but leave it set, as only get has the lock. Signed-off-by:
-
Maarten Lankhorst authored
Call drop_pages with the gem object lock held, instead of the other way around. This will allow us to drop the vma bindings with the gem object lock held. We plan to require the object lock for unpinning in the future, and this is an easy target. Signed-off-by:
Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20211216142749.1966107-3-maarten.lankhorst@linux.intel.com
-
Maarten Lankhorst authored
When reworking the code to move the eviction fence to the object, the best code is removed code. Remove some functions that are unused, and change the function definition if it's only used in 1 place. Signed-off-by:
-
- 18 Dec, 2021 10 commits
-
-
Andi Shyti authored
In preparation of the multitile support, highlight the root GT by calling it gt0 inside the drm i915 private data. Signed-off-by:
Andi Shyti <andi.shyti@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Lucas De Marchi <lucas.demarchi@intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Reviewed-by:
Matt Roper <matthew.d.roper@intel.com> Signed-off-by:
-
Michał Winiarski authored
Use to_gt() helper consistently throughout the codebase. Pure mechanical s/i915->gt/to_gt(i915). No functional changes. Signed-off-by:
Michał Winiarski <michal.winiarski@intel.com> Signed-off-by:
-
Andi Shyti authored
Use to_gt() helper consistently throughout the codebase. Pure mechanical s/i915->gt/to_gt(i915). No functional changes. Signed-off-by:
-
Andi Shyti authored
Use to_gt() helper consistently throughout the codebase. Pure mechanical s/i915->gt/to_gt(i915). No functional changes. Signed-off-by:
-
Michał Winiarski authored
Use to_gt() helper consistently throughout the codebase. Pure mechanical s/i915->gt/to_gt(i915). No functional changes. Signed-off-by:
-
Michał Winiarski authored
Use to_gt() helper consistently throughout the codebase. Pure mechanical s/i915->gt/to_gt(i915). No functional changes. Signed-off-by:
-
Michał Winiarski authored
Use to_gt() helper consistently throughout the codebase. Pure mechanical s/i915->gt/to_gt(i915). No functional changes. Signed-off-by:
-
Michał Winiarski authored
Use to_gt() helper consistently throughout the codebase. Pure mechanical s/i915->gt/to_gt(i915). No functional changes. Signed-off-by:
-
Michał Winiarski authored
To allow further refactoring and abstract away the fact that GT is stored inside i915 private. No functional changes. Signed-off-by:
-
Michał Winiarski authored
We now support a per-gt uncore, yet we're not able to infer which GT we're operating upon. Let's store a backpointer for now. At this point the early initialization of the gt needs to be broken in two parts where the first is needed to assign to the gt the i915 private data pointer and the uncore. A temporary function has been made and the two parts are __intel_gt_init_early() and intel_gt_init_early(). This split will be fixed in the future with the multitile patch. Signed-off-by:
-
- 16 Dec, 2021 7 commits
-
-
Matthew Brost authored
Testing the stealing of guc ids is hard from user space as we have 64k guc_ids. Add a selftest, which artificially reduces the number of guc ids, and forces a steal. The test creates a spinner which is used to block all subsequent submissions until it completes. Next, a loop creates a context and a NOP request each iteration until the guc_ids are exhausted (request creation returns -EAGAIN). The spinner is ended, unblocking all requests created in the loop. At this point all guc_ids are exhausted but are available to steal. Try to create another request which should successfully steal a guc_id. Wait on last request to complete, idle GPU, verify a guc_id was stolen via a counter, and exit the test. Test also artificially reduces the number of guc_ids so the test runs in a timely manner. v2: (John Harrison) - s/stole/stolen - Fix some wording in test description - Rework indexing into context array - Add test description to commit message - Fix typo in commit message (Checkpatch) - s/guc/(guc) in NUMBER_MULTI_LRC_GUC_ID v3: (John Harrison) - Set array value to NULL after extracting error - Fix a few typos in comments / error messages - Delete redundant comment in commit message Signed-off-by:
-
Matthew Brost authored
Let's be paranoid and kick the G2H tasklet, which dequeues messages, if G2H credits are exhausted. Signed-off-by:
-
Matthew Brost authored
Print CT state (H2G + G2H head / tail pointers, credits) on CT deadlock. v2: (John Harrison) - Add units to debug messages Reviewed-by:
-
John Harrison authored
While attempting to debug a CT deadlock issue in various CI failures (most easily reproduced with gem_ctx_create/basic-files), I was seeing CPU deadlock errors being reported. This were because the context destroy loop was blocking waiting on H2G space from inside an IRQ spinlock. There no was deadlock as such, it's just that the H2G queue was full of context destroy commands and GuC was taking a long time to process them. However, the kernel was seeing the large amount of time spent inside the IRQ lock as a dead CPU. Various Bad Things(tm) would then happen (heartbeat failures, CT deadlock errors, outstanding H2G WARNs, etc.). Re-working the loop to only acquire the spinlock around the list management (which is all it is meant to protect) rather than the entire destroy operation seems to fix all the above issues. v2: (John Harrison) - Fix typo in comment message Signed-off-by:
-
Matthew Brost authored
A full GT reset can race with the last context put resulting in the context ref count being zero but the destroyed bit not yet being set. Remove GEM_BUG_ON in scrub_guc_desc_for_outstanding_g2h that asserts the destroyed bit must be set in ref count is zero. Reviewed-by:
-
Matthew Brost authored
Previously assigned whole guc_id structure (list, spin lock) which is incorrect, only assign the guc_id.id. Fixes: 0f797650 ("drm/i915/guc: Rework and simplify locking") Signed-off-by:
-
Matthew Brost authored
s/ce/cn/ when grabbing guc_state.lock before calling clr_context_registered. Fixes: 0f797650 ("drm/i915/guc: Rework and simplify locking") Signed-off-by:
-
- 14 Dec, 2021 1 commit
-
-
Lucas De Marchi authored
PAT can be disabled on boot with "nopat" in the command line. Replace one x86-ism with another, which is slightly more correct to prepare for supporting other architectures. Cc: Matt Roper <matthew.d.roper@intel.com> Signed-off-by:
Lucas De Marchi <lucas.demarchi@intel.com> Reviewed-by:
José Roberto de Souza <jose.souza@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20211202003048.1015511-1-lucas.demarchi@intel.com
-
