Preempt fences can sleep waiting for an exec queue suspend operation to
complete. If the system_unbound_wq is used for waiting and the number of
waiters exceeds max_active this will result in other users of the
system_unbound_wq getting starved. Use a device private work queue for
preempt fences to avoid starvation of the system_unbound_wq.
Even though suspend operations can complete out-of-order, all suspend
operations within a VM need to complete before the preempt rebind worker
can start. With that, use a device private ordered wq for preempt fence
waiting.
v2:
- Add comment about cleanup on failure (Matt R)
- Update commit message (Lucas)
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240401221913.139672-2-matthew.brost@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
(cherry picked from commit 37c15c4aae)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Rebinding might allocate page-table bos, causing evictions.
To support blocking locking during these evictions,
perform the rebinding in the drm_exec locking loop.
Also Reserve fence slots where actually needed rather than trying to
predict how many fence slots will be needed over a complete
wound-wait transaction.
v2:
- Remove a leftover call to xe_vm_rebind() (Matt Brost)
- Add a helper function xe_vm_validate_rebind() (Matt Brost)
v3:
- Add comments and squash with previous patch (Matt Brost)
Fixes: 24f947d58f ("drm/xe: Use DRM GPUVM helpers for external- and evicted objects")
Fixes: 29f424eb87 ("drm/xe/exec: move fence reservation")
Cc: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240327091136.3271-5-thomas.hellstrom@linux.intel.com
(cherry picked from commit 7ee7dd6f30)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
For each rebind we insert a GuC TLB invalidation and add a
corresponding unordered TLB invalidation fence. This might
add a huge number of TLB invalidation fences to wait for so
rather than doing that, defer the TLB invalidation to the
next ring ops for each affected exec queue. Since the TLB
is invalidated on exec_queue switch, we need to invalidate
once for each affected exec_queue.
v2:
- Simplify if-statements around the tlb_flush_seqno.
(Matthew Brost)
- Add some comments and asserts.
Fixes: 5387e865d9 ("drm/xe: Add TLB invalidation fence after rebinds issued from execs")
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: <stable@vger.kernel.org> # v6.8+
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240327091136.3271-2-thomas.hellstrom@linux.intel.com
(cherry picked from commit 4fc4899e86)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Pull Kbuild fixes from Masahiro Yamada:
- Deduplicate Kconfig entries for CONFIG_CXL_PMU
- Fix unselectable choice entry in MIPS Kconfig, and forbid this
structure
- Remove unused include/asm-generic/export.h
- Fix a NULL pointer dereference bug in modpost
- Enable -Woverride-init warning consistently with W=1
- Drop KCSAN flags from *.mod.c files
* tag 'kbuild-fixes-v6.9' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
kconfig: Fix typo HEIGTH to HEIGHT
Documentation/llvm: Note s390 LLVM=1 support with LLVM 18.1.0 and newer
kbuild: Disable KCSAN for autogenerated *.mod.c intermediaries
kbuild: make -Woverride-init warnings more consistent
modpost: do not make find_tosym() return NULL
export.h: remove include/asm-generic/export.h
kconfig: do not reparent the menu inside a choice block
MIPS: move unselectable FIT_IMAGE_FDT_EPM5 out of the "System type" choice
cxl: remove CONFIG_CXL_PMU entry in drivers/cxl/Kconfig
The -Woverride-init warn about code that may be intentional or not,
but the inintentional ones tend to be real bugs, so there is a bit of
disagreement on whether this warning option should be enabled by default
and we have multiple settings in scripts/Makefile.extrawarn as well as
individual subsystems.
Older versions of clang only supported -Wno-initializer-overrides with
the same meaning as gcc's -Woverride-init, though all supported versions
now work with both. Because of this difference, an earlier cleanup of
mine accidentally turned the clang warning off for W=1 builds and only
left it on for W=2, while it's still enabled for gcc with W=1.
There is also one driver that only turns the warning off for newer
versions of gcc but not other compilers, and some but not all the
Makefiles still use a cc-disable-warning conditional that is no
longer needed with supported compilers here.
Address all of the above by removing the special cases for clang
and always turning the warning off unconditionally where it got
in the way, using the syntax that is supported by both compilers.
Fixes: 2cd3271b7a ("kbuild: avoid duplicate warning options")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Andrew Jeffery <andrew@codeconstruct.com.au>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
dEQP-VK.sparse_resources.image_rebind.2d_array.r64i.128_128_8
was causing a remap operation like the below.
op_remap: prev: 0000003fffed0000 00000000000f0000 00000000a5abd18a 0000000000000000
op_remap: next:
op_remap: unmap: 0000003fffed0000 0000000000100000 0
op_map: map: 0000003ffffc0000 0000000000010000 000000005b1ba33c 00000000000e0000
This was resulting in an unmap operation from 0x3fffed0000+0xf0000, 0x100000
which was corrupting the pagetables and oopsing the kernel.
Fixes the prev + unmap range calcs to use start/end and map back to addr/range.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Fixes: b88baab828 ("drm/nouveau: implement new VM_BIND uAPI")
Cc: Danilo Krummrich <dakr@redhat.com>
Signed-off-by: Danilo Krummrich <dakr@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240328024317.2041851-1-airlied@gmail.com
Increase the timeout value to prevent system logs on Amlogic boards flooding
with power transition warnings:
[ 13.047638] panfrost ffe40000.gpu: shader power transition timeout
[ 13.048674] panfrost ffe40000.gpu: l2 power transition timeout
[ 13.937324] panfrost ffe40000.gpu: shader power transition timeout
[ 13.938351] panfrost ffe40000.gpu: l2 power transition timeout
...
[39829.506904] panfrost ffe40000.gpu: shader power transition timeout
[39829.507938] panfrost ffe40000.gpu: l2 power transition timeout
[39949.508369] panfrost ffe40000.gpu: shader power transition timeout
[39949.509405] panfrost ffe40000.gpu: l2 power transition timeout
The 2000 value has been found through trial and error testing with devices
using G52 and G31 GPUs.
Fixes: 22aa1a2090 ("drm/panfrost: Really power off GPU cores in panfrost_gpu_power_off()")
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240322164525.2617508-1-christianshewitt@gmail.com
Object debugging tools were sporadically reporting illegal attempts to
free a still active i915 VMA object when parking a GT believed to be idle.
[161.359441] ODEBUG: free active (active state 0) object: ffff88811643b958 object type: i915_active hint: __i915_vma_active+0x0/0x50 [i915]
[161.360082] WARNING: CPU: 5 PID: 276 at lib/debugobjects.c:514 debug_print_object+0x80/0xb0
...
[161.360304] CPU: 5 PID: 276 Comm: kworker/5:2 Not tainted 6.5.0-rc1-CI_DRM_13375-g003f860e5577+ #1
[161.360314] Hardware name: Intel Corporation Rocket Lake Client Platform/RocketLake S UDIMM 6L RVP, BIOS RKLSFWI1.R00.3173.A03.2204210138 04/21/2022
[161.360322] Workqueue: i915-unordered __intel_wakeref_put_work [i915]
[161.360592] RIP: 0010:debug_print_object+0x80/0xb0
...
[161.361347] debug_object_free+0xeb/0x110
[161.361362] i915_active_fini+0x14/0x130 [i915]
[161.361866] release_references+0xfe/0x1f0 [i915]
[161.362543] i915_vma_parked+0x1db/0x380 [i915]
[161.363129] __gt_park+0x121/0x230 [i915]
[161.363515] ____intel_wakeref_put_last+0x1f/0x70 [i915]
That has been tracked down to be happening when another thread is
deactivating the VMA inside __active_retire() helper, after the VMA's
active counter has been already decremented to 0, but before deactivation
of the VMA's object is reported to the object debugging tool.
We could prevent from that race by serializing i915_active_fini() with
__active_retire() via ref->tree_lock, but that wouldn't stop the VMA from
being used, e.g. from __i915_vma_retire() called at the end of
__active_retire(), after that VMA has been already freed by a concurrent
i915_vma_destroy() on return from the i915_active_fini(). Then, we should
rather fix the issue at the VMA level, not in i915_active.
Since __i915_vma_parked() is called from __gt_park() on last put of the
GT's wakeref, the issue could be addressed by holding the GT wakeref long
enough for __active_retire() to complete before that wakeref is released
and the GT parked.
I believe the issue was introduced by commit d939397303 ("drm/i915:
Remove the vma refcount") which moved a call to i915_active_fini() from
a dropped i915_vma_release(), called on last put of the removed VMA kref,
to i915_vma_parked() processing path called on last put of a GT wakeref.
However, its visibility to the object debugging tool was suppressed by a
bug in i915_active that was fixed two weeks later with commit e92eb246fe
("drm/i915/active: Fix missing debug object activation").
A VMA associated with a request doesn't acquire a GT wakeref by itself.
Instead, it depends on a wakeref held directly by the request's active
intel_context for a GT associated with its VM, and indirectly on that
intel_context's engine wakeref if the engine belongs to the same GT as the
VMA's VM. Those wakerefs are released asynchronously to VMA deactivation.
Fix the issue by getting a wakeref for the VMA's GT when activating it,
and putting that wakeref only after the VMA is deactivated. However,
exclude global GTT from that processing path, otherwise the GPU never goes
idle. Since __i915_vma_retire() may be called from atomic contexts, use
async variant of wakeref put. Also, to avoid circular locking dependency,
take care of acquiring the wakeref before VM mutex when both are needed.
v7: Add inline comments with justifications for:
- using untracked variants of intel_gt_pm_get/put() (Nirmoy),
- using async variant of _put(),
- not getting the wakeref in case of a global GTT,
- always getting the first wakeref outside vm->mutex.
v6: Since __i915_vma_active/retire() callbacks are not serialized, storing
a wakeref tracking handle inside struct i915_vma is not safe, and
there is no other good place for that. Use untracked variants of
intel_gt_pm_get/put_async().
v5: Replace "tile" with "GT" across commit description (Rodrigo),
- avoid mentioning multi-GT case in commit description (Rodrigo),
- explain why we need to take a temporary wakeref unconditionally inside
i915_vma_pin_ww() (Rodrigo).
v4: Refresh on top of commit 5e4e06e408 ("drm/i915: Track gt pm
wakerefs") (Andi),
- for more easy backporting, split out removal of former insufficient
workarounds and move them to separate patches (Nirmoy).
- clean up commit message and description a bit.
v3: Identify root cause more precisely, and a commit to blame,
- identify and drop former workarounds,
- update commit message and description.
v2: Get the wakeref before VM mutex to avoid circular locking dependency,
- drop questionable Fixes: tag.
Fixes: d939397303 ("drm/i915: Remove the vma refcount")
Closes: https://gitlab.freedesktop.org/drm/intel/issues/8875
Signed-off-by: Janusz Krzysztofik <janusz.krzysztofik@linux.intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Nirmoy Das <nirmoy.das@intel.com>
Cc: Andi Shyti <andi.shyti@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: stable@vger.kernel.org # v5.19+
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240305143747.335367-6-janusz.krzysztofik@linux.intel.com
(cherry picked from commit f3c71b2ded)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Since commit 0c65dc0626 ("drm/i915/jsl: s/JSL/JASPERLAKE for
platform/subplatform defines"), boot freezes on a Jasper Lake tablet
(Librem 11), usually with graphical corruption on the eDP display,
but sometimes just a black screen. This commit was included in 6.6 and
later.
That commit was intended to refactor EHL and JSL macros, but the change
to ehl_combo_pll_div_frac_wa_needed() started matching JSL incorrectly
when it was only intended to match EHL.
It replaced:
return ((IS_PLATFORM(i915, INTEL_ELKHARTLAKE) &&
IS_JSL_EHL_DISPLAY_STEP(i915, STEP_B0, STEP_FOREVER)) ||
with:
return (((IS_ELKHARTLAKE(i915) || IS_JASPERLAKE(i915)) &&
IS_DISPLAY_STEP(i915, STEP_B0, STEP_FOREVER)) ||
Remove IS_JASPERLAKE() to fix the regression.
Signed-off-by: Jonathon Hall <jonathon.hall@puri.sm>
Cc: stable@vger.kernel.org
Fixes: 0c65dc0626 ("drm/i915/jsl: s/JSL/JASPERLAKE for platform/subplatform defines")
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240313135424.3731410-1-jonathon.hall@puri.sm
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
(cherry picked from commit 1ef4885931)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In i915 hwmon sysfs getter path we now take a hwmon_lock, then acquire an
rpm wakeref. That results in lock inversion:
<4> [197.079335] ======================================================
<4> [197.085473] WARNING: possible circular locking dependency detected
<4> [197.091611] 6.8.0-rc7-Patchwork_129026v7-gc4dc92fb1152+ #1 Not tainted
<4> [197.098096] ------------------------------------------------------
<4> [197.104231] prometheus-node/839 is trying to acquire lock:
<4> [197.109680] ffffffff82764d80 (fs_reclaim){+.+.}-{0:0}, at: __kmalloc+0x9a/0x350
<4> [197.116939]
but task is already holding lock:
<4> [197.122730] ffff88811b772a40 (&hwmon->hwmon_lock){+.+.}-{3:3}, at: hwm_energy+0x4b/0x100 [i915]
<4> [197.131543]
which lock already depends on the new lock.
...
<4> [197.507922] Chain exists of:
fs_reclaim --> >->reset.mutex --> &hwmon->hwmon_lock
<4> [197.518528] Possible unsafe locking scenario:
<4> [197.524411] CPU0 CPU1
<4> [197.528916] ---- ----
<4> [197.533418] lock(&hwmon->hwmon_lock);
<4> [197.537237] lock(>->reset.mutex);
<4> [197.543376] lock(&hwmon->hwmon_lock);
<4> [197.549682] lock(fs_reclaim);
...
<4> [197.632548] Call Trace:
<4> [197.634990] <TASK>
<4> [197.637088] dump_stack_lvl+0x64/0xb0
<4> [197.640738] check_noncircular+0x15e/0x180
<4> [197.652968] check_prev_add+0xe9/0xce0
<4> [197.656705] __lock_acquire+0x179f/0x2300
<4> [197.660694] lock_acquire+0xd8/0x2d0
<4> [197.673009] fs_reclaim_acquire+0xa1/0xd0
<4> [197.680478] __kmalloc+0x9a/0x350
<4> [197.689063] acpi_ns_internalize_name.part.0+0x4a/0xb0
<4> [197.694170] acpi_ns_get_node_unlocked+0x60/0xf0
<4> [197.720608] acpi_ns_get_node+0x3b/0x60
<4> [197.724428] acpi_get_handle+0x57/0xb0
<4> [197.728164] acpi_has_method+0x20/0x50
<4> [197.731896] acpi_pci_set_power_state+0x43/0x120
<4> [197.736485] pci_power_up+0x24/0x1c0
<4> [197.740047] pci_pm_default_resume_early+0x9/0x30
<4> [197.744725] pci_pm_runtime_resume+0x2d/0x90
<4> [197.753911] __rpm_callback+0x3c/0x110
<4> [197.762586] rpm_callback+0x58/0x70
<4> [197.766064] rpm_resume+0x51e/0x730
<4> [197.769542] rpm_resume+0x267/0x730
<4> [197.773020] rpm_resume+0x267/0x730
<4> [197.776498] rpm_resume+0x267/0x730
<4> [197.779974] __pm_runtime_resume+0x49/0x90
<4> [197.784055] __intel_runtime_pm_get+0x19/0xa0 [i915]
<4> [197.789070] hwm_energy+0x55/0x100 [i915]
<4> [197.793183] hwm_read+0x9a/0x310 [i915]
<4> [197.797124] hwmon_attr_show+0x36/0x120
<4> [197.800946] dev_attr_show+0x15/0x60
<4> [197.804509] sysfs_kf_seq_show+0xb5/0x100
Acquire the wakeref before the lock and hold it as long as the lock is
also held. Follow that pattern across the whole source file where similar
lock inversion can happen.
v2: Keep hardware read under the lock so the whole operation of updating
energy from hardware is still atomic (Guenter),
- instead, acquire the rpm wakeref before the lock and hold it as long
as the lock is held,
- use the same aproach for other similar places across the i915_hwmon.c
source file (Rodrigo).
Fixes: 1b44019a93 ("drm/i915/guc: Disable PL1 power limit when loading GuC firmware")
Signed-off-by: Janusz Krzysztofik <janusz.krzysztofik@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: <stable@vger.kernel.org> # v6.5+
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240311203500.518675-2-janusz.krzysztofik@linux.intel.com
(cherry picked from commit 71b2187714)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Looks like the undelayed vblank gets signalled exactly when
the active period ends. That is a problem for DSB+VRR when
we are already in vblank and expect DSB to start executing
as soon as we send the push. Instead of starting, the DSB
just keeps on waiting for the undelayed vblank which won't
signal until the end of the next frame's active period,
which is far too late.
The end result is that DSB won't have even started
executing by the time the flips/etc. have completed.
We then wait for an extra 1ms, after which we terminate
the DSB and report a timeout:
[drm] *ERROR* [CRTC:80:pipe A] DSB 0 timed out waiting for idle (current head=0xfedf4000, head=0x0, tail=0x1080)
To fix this let's configure DSB to use the so called VRR
"safe window" instead of the undelayed vblank to trigger
the DSB vblank logic, when VRR is enabled.
Cc: stable@vger.kernel.org
Fixes: 34d8311f4a ("drm/i915/dsb: Re-instate DSB for LUT updates")
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9927
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240306040806.21697-3-ville.syrjala@linux.intel.com
Reviewed-by: Animesh Manna <animesh.manna@intel.com>
(cherry picked from commit 41429d9b68)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reinstate commit 88b065943c ("drm/i915/dsi: Do display on
sequence later on icl+"), for the most part. Turns out some
machines (eg. Chuwi Minibook X) really do need that updated order.
It is also the order the Windows driver uses.
However we can't just undo the revert since that would again
break Lenovo 82TQ. After staring at the VBT sequences for both
machines I've concluded that the Lenovo 82TQ sequences look
somewhat broken:
- INIT_OTP is not present at all
- what should be in INIT_OTP is found in DISPLAY_ON
- what should be in DISPLAY_ON is found in BACKLIGHT_ON
(along with the actual backlight stuff)
The Chuwi Minibook X on the other hand has a full complement
of sequences in its VBT.
So let's try to deal with the broken sequences in the
Lenovo 82TQ VBT by simply swapping the (non-existent)
INIT_OTP sequence with the DISPLAY_ON sequence. Thus we
execute DISPLAY_ON when intending to execute INIT_OTP,
and execute nothing at all when intending to execute
DISPLAY_ON. That should be 100% equivalent to the
revert, for such broken VBTs.
Cc: stable@vger.kernel.org
Fixes: 6992eb815d ("Revert "drm/i915/dsi: Do display on sequence later on icl+"")
References: https://gitlab.freedesktop.org/drm/intel/-/issues/10071
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/10334
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240305083659.8396-1-ville.syrjala@linux.intel.com
Acked-by: Jani Nikula <jani.nikula@intel.com>
(cherry picked from commit 94ae4612ea)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This is stopping me building here for some reason,
/home/airlied/devel/kernel/dim/drm-fixes/drivers/gpu/drm/i915/i915_memcpy.c: In function ‘i915_unaligned_memcpy_from_wc’:
/home/airlied/devel/kernel/dim/drm-fixes/drivers/gpu/drm/i915/i915_memcpy.c:33:25: error: implicit declaration of function ‘BUG_ON’; did you mean ‘CI_BUG_ON’? [-Werror=implicit-function-declaration]
33 | #define CI_BUG_ON(expr) BUG_ON(expr)
| ^~~~~~
/home/airlied/devel/kernel/dim/drm-fixes/drivers/gpu/drm/i915/i915_memcpy.c:144:9: note: in expansion of macro ‘CI_BUG_ON’
144 | CI_BUG_ON(!i915_has_memcpy_from_wc());
| ^~~~~~~~~
engage maintainer overrides :-)
Signed-off-by: Dave Airlie <airlied@redhat.com>
