- 19 Dec, 2017 8 commits
-
-
Chris Wilson authored
Useful for verifying our bookkeeper when we encounter is knowing whether we think the engine is idle at the time of the GPU hang. References: https://bugs.freedesktop.org/show_bug.cgi?id=104305Signed-off-by:
Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Reviewed-by:
Rodrigo Vivi <rodrigo.vivi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171219131419.13117-1-chris@chris-wilson.co.uk
-
Rafael Antognolli authored
There seems to be another clock gating issue which the workaround is described as: "WA: Set 0xE4F0[1] = 1 to disable Early EOT of thread." Signed-off-by:
Rafael Antognolli <rafael.antognolli@intel.com> Reviewed-by:
Lucas De Marchi <lucas.demarchi@intel.com> Signed-off-by:
-
Rafael Antognolli authored
This workaround supposedly fixes some hangs in the VF unit. Signed-off-by:
-
Michal Wajdeczko authored
We dump modparams in few places (debugfs, gpu_error) using different functions. Lets add reusable function to avoid code duplication. add/remove: 1/0 grow/shrink: 0/2 up/down: 1096/-2339 (-1243) Function old new delta i915_params_dump - 1096 +1096 i915_capabilities 1353 185 -1168 i915_error_state_to_str 5507 4336 -1171 Total: Before=1285716, After=1284473, chg -0.10% v2: use forward decl rather than include (Chris) Signed-off-by:
Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by:
-
Michal Wajdeczko authored
Convert intel_device_info_dump into pretty printer to be consistent with the rest of the driver code. Suggested-by:
-
Michal Wajdeczko authored
We dump device flags in few places (init_early, debugfs, gpu_error) using different functions. Lets add reusable function to avoid code duplication. add/remove: 1/0 grow/shrink: 0/3 up/down: 1296/-3572 (-2276) Function old new delta intel_device_info_dump_flags - 1296 +1296 i915_capabilities 2435 1353 -1082 i915_error_state_to_str 6642 5507 -1135 intel_device_info_dump 1507 152 -1355 Total: Before=1287992, After=1285716, chg -0.18% Signed-off-by:
-
Chris Wilson authored
drivers/gpu/drm/i915/intel_ddi.c:2098 intel_ddi_clk_select() warn: inconsistent indenting References: 8edcda12 ("drm/i915: Protect DDI port to DPLL map from theoretical race.") Signed-off-by:
Paulo Zanoni <paulo.r.zanoni@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171219112649.9388-1-chris@chris-wilson.co.uk
-
Chris Wilson authored
After a reset, the state of the CSB registers are scrubbed and not valid until a powercontext is reloaded. We only know when a powercontext has been reloaded once we see a CS-interrupt, before then we must ignore the CSB registers within the execlists_submission_tasklet. However, glk is sporadically dying with an illegal CSB pointer value (both in the HSWP and mmio) suggesting that it is running with the CS-interrupt bit set before the powercontext has been reloaded. Make sure the clearing of that bit is serialised on reset with the re-enabling of the tasklet. References: https://bugs.freedesktop.org/show_bug.cgi?id=104262Signed-off-by:
Tvrtko Ursulin <tvrtko.ursulin@intel.com>
-
- 18 Dec, 2017 8 commits
-
-
Joonas Lahtinen authored
CNL supports horizontal plane flipping on non-linear plane formats. v2: - Avoid BUG unlike elsewhere in the code (Ville) - Hoist the rotation-tiling restriction check (Ville) v3 (Rodrigo): - Rebased after a while. - Fix small indentation issues. Bspec: 7656 Suggested-by:
Anusha Srivatsa <anusha.srivatsa@intel.com> Cc: Ville Syrjala <ville.syrjala@linux.intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Anusha Srivatsa <anusha.srivatsa@intel.com> Signed-off-by:
Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Reviewed-by:
Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by:
-
Rodrigo Vivi authored
In case we have multiple modesets for different connectors happening in parallel we could have a race on the RMW on these shared registers. This possibility was initially raised by Paulo when reviewing commit '555e38d2 ("drm/i915/cnl: DDI - PLL mapping")' but the original possibility comes from commit '5416d871 ("drm/i915/skl: Set the eDP link rate on DPLL0")'. Or maybe later when atomic commits entered into picture. Apparently the discussion around this topic showed that the right solution would be on serializing the atomic commits in a way that we don't have the possibility of races here since if that parallel modeset happenings apparently many other things will be on fire. Code is there since SKL and there was no report of issue, but since we never looked back to that serialization possibility, and also we don't have an igt case for that it is better to at least protect this corner. Suggested-by:
-
Chris Wilson authored
Now that we skip a per-engine reset on an idle engine, we need to update the selftest to take that into account. In the process, we find that we were not stressing the per-engine reset very hard, so add those missing active resets. v2: Actually test i915_reset_engine() by loading it with requests. Fixes: f6ba181a ("drm/i915: Skip an engine reset if it recovered before our preparations") Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104313Signed-off-by:
Michel Thierry <michel.thierry@intel.com>
-
Lionel Landwerlin authored
When monitoring the GPU with i915 perf, reports are tagged with a hw id. Gem context creation tracepoints already have a hw_id field, unfortunately you only get this correlation between a process id and a hw context id once when the context is created. It doesn't help if you started monitoring after the process was initialized or if the drm fd was transfered from one process to another. This change adds the hw_id field to gem requests, so that correlation can also be done on submission. v2: Place hw_id at the end of the tracepoint to not disrupt too much existing tools (Chris) v3: Reorder hw_id field again (Chris) v4: Add missing hw_id to i915_gem_request_wait_begin tracepoint (Chris) Signed-off-by:Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by:
-
Lionel Landwerlin authored
Let's make the order of the fields of the tracepoints involving gem request match across i915. This makes userspace processing of tracepoint a bit easier. Suggested-by:
-
Chris Wilson authored
A useful bit of information for inspecting GPU stalls from intel_engine_dump() are the error registers, IPEIR and IPEHR. v2: Fixup gen changes in register offsets (Tvrtko) v3: Old FADDR location as well v4: Use I915_READ64_2x32 Signed-off-by:
-
Matthew Auld authored
We have an existing helper for testing obj->mm.pages, so use it. Signed-off-by:
Matthew Auld <matthew.auld@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/20171218103855.25274-1-matthew.auld@intel.comReviewed-by:
-
Chris Wilson authored
Inside i915_gem_reset(), we start touching the HW and so require the low-level HW to be re-enabled, in particular the PCI BARs. Fixes: 7b6da818 ("drm/i915: Restore the kernel context after a GPU reset on an idle engine") References: 0db8c961 ("drm/i915: Re-enable GTT following a device reset") Testcase: igt/drv_hangman #i915g/i915gm Signed-off-by:
-
- 17 Dec, 2017 1 commit
-
-
Michal Wajdeczko authored
Instead of trying different seq_puts messages, lets use common -ENODEV error code to indicate missing/unsupported feature. v2: don't forget about guc_log_control fops (Sagar) Suggested-by:
-
- 16 Dec, 2017 2 commits
-
-
Chris Wilson authored
As part of the system requirement for powersaving is that we always have a context loaded. Upon boot and resume, we load the kernel_context to ensure that some valid state is set before powersaving kicks in, we should do so after a full GPU reset as well. We only need to do so for an idle engine, as any active engines will restart by executing the stuck request, loading its context. For the idle engine, we create a new request to load the kernel_context instead. For whatever reason, perfoming a dummy execute on the idle engine after reset papers over a subsequent GPU hang in rare circumstances, even on machines not using contexts (e.g. Pineview). Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104259 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104261Signed-off-by:
-
Chris Wilson authored
At the beginning of a reset, we disable the submission method and find the stuck request. We expect to find a stuck request for we have declared the engine stalled. However, if we find no active request, the engine must have recovered from its stall before we could issue a reset, so let the engine continue on without a reset. If the engine is truly stuck, we will back soon enough with the next reset attempt. v2: Remove the stale debug message. Signed-off-by:
-
- 15 Dec, 2017 3 commits
-
-
Lucas De Marchi authored
CFL was missing from intel_early_ids[]. The PCI ID needs to be there to allow the memory region to be stolen, otherwise we could have RAM being arbitrarily overwritten if for example we keep using the UEFI framebuffer, depending on how BIOS has set up the e820 map. Fixes: b056f8f3 ("drm/i915/cfl: Add Coffee Lake PCI IDs for S Skus.") Signed-off-by:
Jani Nikula <jani.nikula@intel.com> Acked-by:
Ingo Molnar <mingo@kernel.org> Signed-off-by:
-
Chris Wilson authored
Just printk the string, or at least do not double up on the newlines! Fixes: eef57324 ("drm/i915: setup bridge for HDMI LPE audio driver") Signed-off-by:
Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171213182858.2159-1-chris@chris-wilson.co.uk
-
Chris Wilson authored
Internal objects consistent of scratch pages not subject to the persistence guarantees of user facing objects. They are used for example, in ring buffers where they are only required for temporary storage of commands that will be rewritten every time. As they are temporary constructs, quietly report -ENOMEM back along the callchain rather than subject the system to oomkiller if an allocation fails. Signed-off-by:
-
- 14 Dec, 2017 12 commits
-
-
Rodrigo Vivi authored
Signed-off-by: -
https://github.com/intel/gvt-linuxRodrigo Vivi authored
gvt-next-2017-12-14: - fixes for two coverity scan errors (Colin) - mmio switch code refine (Changbin) - more virtual display dmabuf fixes (Tina/Gustavo) - misc cleanups (Pei) Signed-off-by:
-
Sebastian Andrzej Siewior authored
The code has an ifdef and uses two functions to either init the bare spinlock or init it and set a lock-class. It is possible to do the same thing without an ifdef. With this patch (in debug case) we first use the "default" lock class which is later overwritten to the supplied one. Without lockdep the set name/class function vanishes. Reported-by:
kbuild test robot <fengguang.wu@intel.com> Signed-off-by:
Sebastian Andrzej Siewior <bigeasy@linutronix.de> Reviewed-by:
-
Chris Wilson authored
Knowing the state of the engine when hangcheck thinks it is stalling is useful for both debugging hangcheck itself and the potential cause of an unwanted stall. Signed-off-by:
-
Lionel Landwerlin authored
As suggested by Chris, we should make this more obvious for people working with newer generations. Suggested-by:
-
Michał Winiarski authored
We have the selftest that's checking doorbell create/destroy, so there's no need to check all doorbells delaying the reset every time. We do want to have that extra sanity check at module load/unload though. Signed-off-by:
Michał Winiarski <michal.winiarski@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Reviewed-by:
-
Michał Winiarski authored
We can now move the clients allocation to submission_init path, rather than keeping the condition inside submission_enable called on every reset. Signed-off-by:
-
Michał Winiarski authored
Full GPU reset causes GuC to be reset. This means that every time we're doing a reset, we need to talk to GuC and tell it about doorbells. Let's separate the communication part (create_doorbell) from our internal bookkeeping (reserve_doorbell) so that we can cleanly separate the initialization done at module load from reinitialization done at reset in the following patch. While I'm here, let's also add a proper (although slightly asymetric) cleanup that doesn't try to communicate with GuC after it's already gone, getting rid of "expected" warnings caused by GuC action failures on module unload. Note that I've also removed one of the tests (bitmap out of sync), since it doesn't make much sense anymore - bitmaps are now not expected to change during the lifetime of a client. Signed-off-by:
-
Michał Winiarski authored
To make this operation a bit cleaner, we should make sure that the HW can catch up by calling the new implementation right away. Note that currently we're only touching the vfunc at module load time (before GuC is even loaded), so this shouldn't cause any functional changes. Suggested-by:
-
Michał Winiarski authored
After GPU reset, GuC HW needs to be reinitialized (with FW reload). Unfortunately, we're doing some extra work there (mostly allocating stuff), work that can be moved to guc_init and called once at driver load time. As a side effect we're no longer hitting an assert in i915_ggtt_enable_guc on suspend/resume. v2: Do not duplicate disable_communication / reset_guc_interrupts v3: Add proper teardown after rebase References: 04f7b24e ("drm/i915/guc: Assert that we switch between known ggtt->invalidate functions") Signed-off-by:
-
Michał Winiarski authored
This gets rid of the following lockdep splat: ====================================================== WARNING: possible circular locking dependency detected 4.15.0-rc2-CI-Patchwork_7428+ #1 Not tainted ------------------------------------------------------ debugfs_test/1351 is trying to acquire lock: (&dev->struct_mutex){+.+.}, at: [<000000009d90d1a3>] i915_mutex_lock_interruptible+0x47/0x130 [i915] but task is already holding lock: (&mm->mmap_sem){++++}, at: [<000000005df01c1e>] __do_page_fault+0x106/0x560 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #6 (&mm->mmap_sem){++++}: __might_fault+0x63/0x90 _copy_to_user+0x1e/0x70 filldir+0x8c/0xf0 dcache_readdir+0xeb/0x160 iterate_dir+0xe6/0x150 SyS_getdents+0xa0/0x130 entry_SYSCALL_64_fastpath+0x1c/0x89 -> #5 (&sb->s_type->i_mutex_key#5){++++}: lockref_get+0x9/0x20 -> #4 ((completion)&req.done){+.+.}: wait_for_common+0x54/0x210 devtmpfs_create_node+0x130/0x150 device_add+0x5ad/0x5e0 device_create_groups_vargs+0xd4/0xe0 device_create+0x35/0x40 msr_device_create+0x22/0x40 cpuhp_invoke_callback+0xc5/0xbf0 cpuhp_thread_fun+0x167/0x210 smpboot_thread_fn+0x17f/0x270 kthread+0x173/0x1b0 ret_from_fork+0x24/0x30 -> #3 (cpuhp_state-up){+.+.}: cpuhp_issue_call+0x132/0x1c0 __cpuhp_setup_state_cpuslocked+0x12f/0x2a0 __cpuhp_setup_state+0x3a/0x50 page_writeback_init+0x3a/0x5c start_kernel+0x393/0x3e2 secondary_startup_64+0xa5/0xb0 -> #2 (cpuhp_state_mutex){+.+.}: __mutex_lock+0x81/0x9b0 __cpuhp_setup_state_cpuslocked+0x4b/0x2a0 __cpuhp_setup_state+0x3a/0x50 page_alloc_init+0x1f/0x26 start_kernel+0x139/0x3e2 secondary_startup_64+0xa5/0xb0 -> #1 (cpu_hotplug_lock.rw_sem){++++}: cpus_read_lock+0x34/0xa0 apply_workqueue_attrs+0xd/0x40 __alloc_workqueue_key+0x2c7/0x4e1 intel_guc_submission_init+0x10c/0x650 [i915] intel_uc_init_hw+0x29e/0x460 [i915] i915_gem_init_hw+0xca/0x290 [i915] i915_gem_init+0x115/0x3a0 [i915] i915_driver_load+0x9a8/0x16c0 [i915] i915_pci_probe+0x2e/0x90 [i915] pci_device_probe+0x9c/0x120 driver_probe_device+0x2a3/0x480 __driver_attach+0xd9/0xe0 bus_for_each_dev+0x57/0x90 bus_add_driver+0x168/0x260 driver_register+0x52/0xc0 do_one_initcall+0x39/0x150 do_init_module+0x56/0x1ef load_module+0x231c/0x2d70 SyS_finit_module+0xa5/0xe0 entry_SYSCALL_64_fastpath+0x1c/0x89 -> #0 (&dev->struct_mutex){+.+.}: lock_acquire+0xaf/0x200 __mutex_lock+0x81/0x9b0 i915_mutex_lock_interruptible+0x47/0x130 [i915] i915_gem_fault+0x201/0x760 [i915] __do_fault+0x15/0x70 __handle_mm_fault+0x85b/0xe40 handle_mm_fault+0x14f/0x2f0 __do_page_fault+0x2d1/0x560 page_fault+0x22/0x30 other info that might help us debug this: Chain exists of: &dev->struct_mutex --> &sb->s_type->i_mutex_key#5 --> &mm->mmap_sem Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&mm->mmap_sem); lock(&sb->s_type->i_mutex_key#5); lock(&mm->mmap_sem); lock(&dev->struct_mutex); *** DEADLOCK *** 1 lock held by debugfs_test/1351: #0: (&mm->mmap_sem){++++}, at: [<000000005df01c1e>] __do_page_fault+0x106/0x560 stack backtrace: CPU: 2 PID: 1351 Comm: debugfs_test Not tainted 4.15.0-rc2-CI-Patchwork_7428+ #1 Hardware name: /NUC6i5SYB, BIOS SYSKLi35.86A.0057.2017.0119.1758 01/19/2017 Call Trace: dump_stack+0x5f/0x86 print_circular_bug+0x230/0x3b0 check_prev_add+0x439/0x7b0 ? lockdep_init_map_crosslock+0x20/0x20 ? unwind_get_return_address+0x16/0x30 ? __lock_acquire+0x1385/0x15a0 __lock_acquire+0x1385/0x15a0 lock_acquire+0xaf/0x200 ? i915_mutex_lock_interruptible+0x47/0x130 [i915] __mutex_lock+0x81/0x9b0 ? i915_mutex_lock_interruptible+0x47/0x130 [i915] ? i915_mutex_lock_interruptible+0x47/0x130 [i915] ? i915_mutex_lock_interruptible+0x47/0x130 [i915] i915_mutex_lock_interruptible+0x47/0x130 [i915] ? __pm_runtime_resume+0x4f/0x80 i915_gem_fault+0x201/0x760 [i915] __do_fault+0x15/0x70 __handle_mm_fault+0x85b/0xe40 handle_mm_fault+0x14f/0x2f0 __do_page_fault+0x2d1/0x560 page_fault+0x22/0x30 RIP: 0033:0x7f98d6f49116 RSP: 002b:00007ffd6ffc3278 EFLAGS: 00010283 RAX: 00007f98d39a2bc0 RBX: 0000000000000000 RCX: 0000000000001680 RDX: 0000000000001680 RSI: 00007ffd6ffc3400 RDI: 00007f98d39a2bc0 RBP: 00007ffd6ffc33a0 R08: 0000000000000000 R09: 00000000000005a0 R10: 000055e847c2a830 R11: 0000000000000002 R12: 0000000000000001 R13: 000055e847c1d040 R14: 00007ffd6ffc3400 R15: 00007f98d6752ba0 v2: Init preempt_work unconditionally (Chris) v3: Mention that we need the enable_guc=1 for lockdep splat (Chris) Testcase: igt/debugfs_test/read_all_entries # with i915.enable_guc=1 Signed-off-by: -
Michał Winiarski authored
We need shared data for actions (e.g. guc suspend/resume), and we're using those with GuC submission disabled. Let's introduce intel_guc_init and move shared data alloc there. This fixes GPF during module unload with HuC, but without GuC submission: BUG: unable to handle kernel NULL pointer dereference at 000000005aee7809 IP: intel_guc_suspend+0x34/0x140 [i915] PGD 0 P4D 0 Oops: 0000 [#1] PREEMPT SMP Modules linked in: i915(O-) netconsole x86_pkg_temp_thermal intel_powerclamp coretemp crct10dif_pclmul crc32_pclmul ghash_clmulni_intel mei_me i2c_i801 mei prime_numbers [last unloaded: i915] CPU: 2 PID: 2794 Comm: rmmod Tainted: G U W O 4.15.0-rc2+ #297 Hardware name: /NUC6i5SYB, BIOS SYSKLi35.86A.0054.2016.0930.1102 09/30/2016 task: 0000000055945c61 task.stack: 00000000264ccb43 RIP: 0010:intel_guc_suspend+0x34/0x140 [i915] RSP: 0018:ffffc90000483df8 EFLAGS: 00010286 RAX: 0000000000000000 RBX: ffff880829180000 RCX: 0000000000000000 RDX: 0000000000000006 RSI: ffff880844c2c938 RDI: ffff880844c2c000 RBP: ffff880829180000 R08: 00000000a29c58c1 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: ffffffffa040ba40 R13: ffffffffa040bab0 R14: ffff88084a195060 R15: 000055df3ef357a0 FS: 00007ff43c043740(0000) GS:ffff88084e200000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000000f9 CR3: 000000083f179005 CR4: 00000000003606e0 Call Trace: i915_gem_suspend+0x9d/0x130 [i915] ? i915_driver_unload+0x68/0x180 [i915] i915_driver_unload+0x70/0x180 [i915] i915_pci_remove+0x15/0x20 [i915] pci_device_remove+0x36/0xb0 device_release_driver_internal+0x15f/0x220 driver_detach+0x3a/0x80 bus_remove_driver+0x58/0xd0 pci_unregister_driver+0x29/0x90 SyS_delete_module+0x150/0x1e0 entry_SYSCALL_64_fastpath+0x23/0x9a RIP: 0033:0x7ff43b51b5c7 RSP: 002b:00007ffe6825a758 EFLAGS: 00000206 ORIG_RAX: 00000000000000b0 RAX: ffffffffffffffda RBX: 0000000000000003 RCX: 00007ff43b51b5c7 RDX: 000000000000000a RSI: 0000000000000800 RDI: 000055df3ef35808 RBP: 0000000000000000 R08: 00007ffe682596d1 R09: 0000000000000000 R10: 00007ff43b594880 R11: 0000000000000206 R12: 000055df3ef357a0 R13: 00007ffe68259740 R14: 000055df3ef35260 R15: 000055df3ef357a0 Code: 00 00 02 74 03 31 c0 c3 53 48 89 fb 48 83 ec 10 e8 52 0f f8 ff 48 b8 01 05 00 00 02 00 00 00 48 89 44 24 04 48 8b 83 00 12 00 00 <f6> 80 f9 00 00 00 01 0f 84 a7 00 00 00 f6 80 98 00 00 00 01 0f RIP: intel_guc_suspend+0x34/0x140 [i915] RSP: ffffc90000483df8 CR2: 00000000000000f9 ---[ end trace 23a192a61d937a3e ]--- Fixes: b8e5eb96 ("drm/i915/guc: Allocate separate shared data object for GuC communication") Signed-off-by:
-
- 13 Dec, 2017 6 commits
-
-
Chris Wilson authored
Since Michal introduced new user controllable errors other than -EIO during i915_gem_init(), we need to actually unwind on the error path as we have to abort the module load (and we expect to do so cleanly!). As we now teardown key state and then mark the driver as wedged (on EIO), we have to be careful to not allow ourselves to resume and unwedge, thus attempting to use the uninitialised driver. v2: Try not to free driver state for the suppressed EIO v3: Use load-fault-injection to test both error/recovery paths. References: 8620eb1d ("drm/i915/uc: Don't use -EIO to report missing firmware") Signed-off-by:
-
Chris Wilson authored
If we fail to allocate a request, we can reap the outstanding requests and push them to the request's slab's freelist before trying again. This forces us to ratelimit malicious clients that tie up all of the system resources in requests, instead of causing a system-wide oom. Testcase: igt/gem_shrink/execbuf1 Signed-off-by:
-
Chris Wilson authored
If a fence allocation fails in a blocking context, we will sleep on the fence as a last resort. We can therefore allow ourselves to fail and sleep on the fence instead of triggering a system-wide oom. This allows us to throttle malicious clients that are consuming lots of system resources by capping the amount of memory used by fences. Testcase: igt/gem_shrink/execbufX Signed-off-by:
-
Chris Wilson authored
As kmalloc is allowed to block (if given the right flags), mark up the two i915_sw_fence routines that may call kmalloc as potential sleeping routines. Signed-off-by:
-
Chris Wilson authored
i915_gem_wait_for_idle() is called from inside the shrinker, to ensure that we drain the last resources from the GPU in dire circumstances (OOM). As we may allocate whilst building a request, it is then possible to hit the shrinker with a request under construction, and so we must account for the incomplete request whilst waiting. In particular, we preincrement (in reserve_engine) the i915->gt.active_requests counter and mark the GPU as busy, therefore we can not use that counter for shortcircuiting the wait-for-idle. [ 950.859024] GEM_BUG_ON(i915->gt.active_requests) [ 950.859041] WARNING: CPU: 2 PID: 2178 at drivers/gpu/drm/i915/i915_gem.c:3615 i915_gem_wait_for_idle.part.56+0x166/0x4e0 [ 950.859041] Modules linked in: ccm tun fuse nf_conntrack_netbios_ns nf_conntrack_broadcast ip6t_rpfilter ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack libcrc32c ebtable_nat ebtable_broute bridge stp llc ebtable_filter ebtables ip6table_mangle ip6table_security ip6table_raw ip6table_filter ip6_tables iptable_mangle iptable_security iptable_raw arc4 iwldvm mac80211 snd_hda_codec_hdmi snd_hda_codec_idt snd_hda_codec_generic snd_hda_intel snd_hda_codec btusb snd_hda_core btrtl btbcm iwlwifi snd_hwdep btintel bluetooth snd_seq snd_seq_device snd_pcm ecdh_generic x86_pkg_temp_thermal tpm_infineon coretemp tpm_tis crc32_pclmul wmi_bmof crc32c_intel iTCO_wdt hp_wmi snd_timer iTCO_vendor_support sparse_keymap tpm_tis_core mei_me cfg80211 [ 950.859082] snd joydev tpm mei rfkill pcspkr wmi soundcore lpc_ich hp_accel lis3lv02d input_polldev binfmt_misc e1000e ptp serio_raw pps_core [ 950.859094] CPU: 2 PID: 2178 Comm: gem_exec_nop Tainted: G U 4.15.0-rc2+ #900 [ 950.859102] Hardware name: Hewlett-Packard HP ProBook 6360b/1620, BIOS 68SCF Ver. B.42 12/29/2010 [ 950.859107] task: c5119cb4 task.stack: f3ccb8d8 [ 950.859112] EIP: i915_gem_wait_for_idle.part.56+0x166/0x4e0 [ 950.859113] EFLAGS: 00010296 CPU: 2 [ 950.859114] EAX: 00000024 EBX: f36c1888 ECX: f777a044 EDX: 00000007 [ 950.859115] ESI: f36c1888 EDI: edd53958 EBP: edd53970 ESP: edd53938 [ 950.859116] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 [ 950.859117] CR0: 80050033 CR2: b7f39000 CR3: 2f2b3000 CR4: 000406d0 [ 950.859118] Call Trace: [ 950.859125] ? drm_printk+0x70/0x70 [ 950.859129] i915_gem_wait_for_idle+0x18/0x30 [ 950.859133] i915_gem_shrink+0x360/0x410 [ 950.859138] ? vmpressure+0xa8/0xf0 [ 950.859142] ? ktime_get+0x4a/0x100 [ 950.859147] i915_gem_shrink_all+0x21/0x40 [ 950.859151] i915_gem_shrinker_oom+0x23/0x130 [ 950.859156] notifier_call_chain+0x4e/0x70 [ 950.859160] __blocking_notifier_call_chain+0x2f/0x60 [ 950.859164] blocking_notifier_call_chain+0x11/0x20 [ 950.859169] out_of_memory+0x207/0x280 [ 950.859174] __alloc_pages_nodemask+0xd47/0xe60 [ 950.859179] new_slab+0x32d/0x450 [ 950.859183] ___slab_alloc.constprop.81+0x358/0x4e0 [ 950.859189] ? i915_sw_fence_await_dma_fence+0x53/0x160 [ 950.859193] ? __slab_free+0x1fe/0x310 [ 950.859197] ? native_sched_clock+0x1e/0xc0 [ 950.859201] ? i915_gem_request_alloc+0xcf/0x510 [ 950.859205] ? sched_clock+0x9/0x10 [ 950.859209] __slab_alloc.constprop.80+0x29/0x40 [ 950.859212] ? __slab_alloc.constprop.80+0x29/0x40 [ 950.859216] kmem_cache_alloc_trace+0x160/0x1a0 [ 950.859220] ? i915_sw_fence_await_dma_fence+0x53/0x160 [ 950.859224] i915_sw_fence_await_dma_fence+0x53/0x160 [ 950.859229] i915_gem_request_await_dma_fence+0x1eb/0x390 [ 950.859233] i915_gem_request_await_object+0xee/0x230 [ 950.859239] i915_gem_do_execbuffer+0xc16/0x1200 [ 950.859246] ? irqtime_account_irq+0x3e/0xc0 [ 950.859251] ? irq_exit+0x4f/0xb0 [ 950.859257] ? smp_apic_timer_interrupt+0x5f/0x110 [ 950.859261] ? apic_timer_interrupt+0x35/0x3c [ 950.859266] i915_gem_execbuffer2_ioctl+0x212/0x440 [ 950.859270] ? apic_timer_interrupt+0x35/0x3c [ 950.859274] ? i915_gem_do_execbuffer+0x1200/0x1200 [ 950.859279] ? insn_get_seg_base+0x1b/0x50 [ 950.859283] ? i915_gem_do_execbuffer+0x1200/0x1200 [ 950.859287] drm_ioctl_kernel+0x51/0xa0 [ 950.859291] drm_ioctl+0x2a3/0x350 [ 950.859294] ? i915_gem_do_execbuffer+0x1200/0x1200 [ 950.859300] ? sched_clock+0x9/0x10 [ 950.859303] ? drm_getunique+0x70/0x70 [ 950.859308] do_vfs_ioctl+0x7d/0x640 [ 950.859311] ? native_sched_clock+0x1e/0xc0 [ 950.859315] ? sched_clock+0x9/0x10 [ 950.859319] ? sched_clock_cpu+0x13/0x120 [ 950.859323] SyS_ioctl+0x4e/0x80 [ 950.859326] do_fast_syscall_32+0x75/0x250 [ 950.859331] ? irq_exit+0x4f/0xb0 [ 950.859334] entry_SYSENTER_32+0x47/0x71 [ 950.859338] EIP: 0xb7f81d11 [ 950.859339] EFLAGS: 00000296 CPU: 2 [ 950.859340] EAX: ffffffda EBX: 00000003 ECX: 40406469 EDX: bfde4c20 [ 950.859340] ESI: 00000003 EDI: 40406469 EBP: 00000003 ESP: bfde4b38 [ 950.859341] DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 007b [ 950.859343] Code: e8 30 60 01 00 83 c4 10 83 c3 04 39 f3 75 e0 8b 45 d8 8b 80 14 37 00 00 85 c0 74 13 68 dd 33 e4 c0 68 49 6f e3 c0 e8 4a 55 be ff <0f> ff 5e 5f b8 fe ff ff 3f bb 0a 00 00 00 e8 b7 14 c4 ff 8b 15 Signed-off-by:
-
Chris Wilson authored
It is illegal to perform an immediate free of the struct irq_work from inside the irq_work callback (as irq_work_run_list modifies work->flags after execution of the work->func()). As we use the irq_work to coordinate the freeing of the callback from two different softirq paths, we need to defer the kfree from inside our irq_work callback, for which we can use kfree_rcu. Fixes: 81c0ed21 ("drm/i915/fence: Avoid del_timer_sync() from inside a timer") Signed-off-by:
-
