It is really easy to introduce subtle deadlocks in
preempt_fence_work_func() since we operate on single global ordered-wq
for signalling our preempt fences behind the scenes, so even though we
signal a particular fence, everything in the callback should be in the
fence critical section, since blocking in the callback will prevent
other published fences from signalling. If we enlarge the fence critical
section to cover the entire callback, then lockdep should be able to
understand this better, and complain if we grab a sensitive lock like
vm->lock, which is also held when waiting on preempt fences.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240418144630.299531-2-matthew.auld@intel.com
In many validation situations when debugging GPU Hangs,
it is useful to preserve the GT situation from the moment
that the timeout occurred.
This patch introduces a module parameter that could be used
on situations like this.
If xe.wedged module parameter is set to 2, Xe will be declared
wedged on every single execution timeout (a.k.a. GPU hang) right
after devcoredump snapshot capture and without attempting any
kind of GT reset and blocking entirely any kind of execution.
v2: Really block gt_reset from guc side. (Lucas)
s/wedged/busted (Lucas)
v3: - s/busted/wedged
- Really use global_flags (Dafna)
- More robust timeout handling when wedging it.
v4: A really robust clean exit done by Matt Brost.
No more kernel warns on unbind.
v5: Simplify error message (Lucas)
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Dafna Hirschfeld <dhirschfeld@habana.ai>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Alan Previn <alan.previn.teres.alexis@intel.com>
Cc: Himanshu Somaiya <himanshu.somaiya@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240423221817.1285081-3-rodrigo.vivi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Introduce a very simple 'wedged' state where any attempt
to access the GPU is entirely blocked.
On some critical cases, like on gt_reset failure, we need to
block any other attempt to use the GPU. Otherwise we are at
a risk of reaching cases that would force us to reboot the machine.
So, when this cases are identified we corner and block any GPU
access. No IOCTL and not even another GT reset should be attempted.
The 'wedged' state in Xe is an end state with no way back.
Only a device "re-probe" (unbind + bind) can restore the GPU access.
v2: - s/wedged/busted (Lucas)
- use unbind+bind instead of module reload (Lucas)
- added more info on unbind operations and instruction on bug report
- only print the message once.
v3: - s/busted/wedged (Ashutosh, Tvrtko, Thomas)
- don't assume user has sudo and tee available (Lucas)
v4: - remove unnecessary cases around ct communication or migration.
Cc: Ashutosh Dixit <ashutosh.dixit@intel.com>
Cc: Tvrtko Ursulin <tursulin@ursulin.net>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Anshuman Gupta <anshuman.gupta@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com> #v2
Link: https://patchwork.freedesktop.org/patch/msgid/20240423221817.1285081-1-rodrigo.vivi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We already have functions to configure VF resources and to print
actual provisioning details. Expose this functionality in debugfs
to allow experiment with different settings or inspect details in
case of unexpected issues with the provisioning.
As debugfs attributes are per-VF, we use parent d_inode->i_private
to store VFID, similarly how we did for per-GT attributes.
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240423131244.2045-2-michal.wajdeczko@intel.com
We already have mechanism that allows a VF driver to communicate
with the PF driver, now add PF side handlers for VF2PF requests
defined in version 1.0 of VF/PF GuC Relay ABI specification.
The VF2PF_HANDSHAKE request must be used by the VF driver to
negotiate the ABI version prior to sending any other request.
We will reset any negotiated version later during FLR.
The outcome of the VF2PF_QUERY_RUNTIME requests depends on actual
platform, for legacy platforms used as SDV is provided as-is, for
latest platforms it is preliminary, and might be changed.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240423180436.2089-5-michal.wajdeczko@intel.com
Let's simply convert all the current callers towards direct
xe_pm_runtime access and remove this extra layer of indirection.
No functional change is expected with this patch since
xe_mem_access_get was already using the xe_pm_runtime_get_noresume
at this point.
v2: Convert all the current callers instead of a big refactor
at once.
v3: - Rebased
- Squashed the GSC/HDCP
- Added a new case: sriov_pf_policy
- Improved commit message to highlight that
there's no functional change in this patch.
Reviewed-by: Matthew Auld <matthew.auld@intel.com> #v2
Cc: Suraj Kandpal <suraj.kandpal@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Suraj Kandpal <suraj.kandpal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240418143049.43231-1-rodrigo.vivi@intel.com
Rather than trying to identify exactly which engines are available on
each platform in the IP descriptor, just include the list of all media
engines that the IP could theoretically support (i.e., 8 VCS + 4 VECS).
We still rely on the media fuse registers to tell us which specific
engine instances are actually present on a given platform, so there
shouldn't be any functional change. This will help prevent mistakes
with engine numbering (for example ambiguity about whether the 2nd VCS
engine on a platform with exactly two engines is numbered "VCS1" or
"VCS2") and will also future-proof the code a bit more in case new SKUs
or platform refreshes extend the engine list in the future.
Note that the media fuse register technically has an 8-bit field for
VECS engine presence starting on Xe2. However there's still no MMIO
register range reserved for VE engines above VECS3, so VE0-VE3 is still
consider the "maximum" VE engine mask that the driver can support for
now.
Bspec: 52614, 52615, 62567
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240417152621.3357990-2-matthew.d.roper@intel.com
In the past, the noresume function was used by the GEM code to ensure
wakelocks were held and bump its usage. This is no longer the case
and this function was totally unused until it started to be used again
by display with commit 77e619a82f ("drm/i915/display: convert inner
wakeref get towards get_if_in_use")
However, on the display code, most of the callers are using the
raw wakeref, rather then the wakelock version. What caused a
major regression caught by CI.
Another option to this patch is to go with the original plan and
use the get_if_in_use variant in the display code, what is enough
to fulfil our needs. Then, an extra patch to delete the unused
_noresume variant.
v2: Keep grabbing wakelock but only assert for wakeref. (Imre)
Cc: Imre Deak <imre.deak@intel.com>
Cc: Francois Dugast <francois.dugast@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Fixes: 77e619a82f ("drm/i915/display: convert inner wakeref get towards get_if_in_use")
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/10875
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240418223756.68427-1-rodrigo.vivi@intel.com
When both hwmon and hwmon drvdata (on which hwmon depends) are device
managed resources, the expectation, on device unbind, is that hwmon will be
released before drvdata. However, in i915 there are two separate code
paths, which both release either drvdata or hwmon and either can be
released before the other. These code paths (for device unbind) are as
follows (see also the bug referenced below):
Call Trace:
release_nodes+0x11/0x70
devres_release_group+0xb2/0x110
component_unbind_all+0x8d/0xa0
component_del+0xa5/0x140
intel_pxp_tee_component_fini+0x29/0x40 [i915]
intel_pxp_fini+0x33/0x80 [i915]
i915_driver_remove+0x4c/0x120 [i915]
i915_pci_remove+0x19/0x30 [i915]
pci_device_remove+0x32/0xa0
device_release_driver_internal+0x19c/0x200
unbind_store+0x9c/0xb0
and
Call Trace:
release_nodes+0x11/0x70
devres_release_all+0x8a/0xc0
device_unbind_cleanup+0x9/0x70
device_release_driver_internal+0x1c1/0x200
unbind_store+0x9c/0xb0
This means that in i915, if use devm, we cannot gurantee that hwmon will
always be released before drvdata. Which means that we have a uaf if hwmon
sysfs is accessed when drvdata has been released but hwmon hasn't.
The only way out of this seems to be do get rid of devm_ and release/free
everything explicitly during device unbind.
v2: Change commit message and other minor code changes
v3: Cleanup from i915_hwmon_register on error (Armin Wolf)
v4: Eliminate potential static analyzer warning (Rodrigo)
Eliminate fetch_and_zero (Jani)
v5: Restore previous logic for ddat_gt->hwmon_dev error return (Andi)
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/10366
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240417145646.793223-1-ashutosh.dixit@intel.com
