Here is the failure stack:
[ 12.988209] ------------[ cut here ]------------
[ 12.988216] UBSAN: shift-out-of-bounds in ./include/linux/log2.h:57:13
[ 12.988232] shift exponent 64 is too large for 64-bit type 'long unsigned int'
[ 12.988235] CPU: 4 PID: 1310 Comm: gnome-shell Tainted: G U 6.9.0-rc6+prerelease1158+ #19
[ 12.988237] Hardware name: Intel Corporation Raptor Lake Client Platform/RPL-S ADP-S DDR5 UDIMM CRB, BIOS RPLSFWI1.R00.3301.A02.2208050712 08/05/2022
[ 12.988239] Call Trace:
[ 12.988240] <TASK>
[ 12.988242] dump_stack_lvl+0xd7/0xf0
[ 12.988248] dump_stack+0x10/0x20
[ 12.988250] ubsan_epilogue+0x9/0x40
[ 12.988253] __ubsan_handle_shift_out_of_bounds+0x10e/0x170
[ 12.988260] dma_resv_reserve_fences.cold+0x2b/0x48
[ 12.988262] ? ww_mutex_lock_interruptible+0x3c/0x110
[ 12.988267] drm_exec_prepare_obj+0x45/0x60 [drm_exec]
[ 12.988271] ? vm_bind_ioctl_ops_execute+0x5b/0x740 [xe]
[ 12.988345] vm_bind_ioctl_ops_execute+0x78/0x740 [xe]
It is caused by the value 0 of parameter num_fences in function
drm_exec_prepare_obj. And lead to in function __rounddown_pow_of_two,
"0 - 1" causes the shift-out-of-bounds.
By design drm_exec_prepare_obj() should be called only when there are
fences to be reserved. If num_fences is 0, calling drm_exec_lock_obj()
is sufficient as was done in commit 9377de4cb3 ("drm/xe/vm: Avoid
reserving zero fences")
Cc: Nirmoy Das <nirmoy.das@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/all/24d4a9a9-c622-4f56-8672-21f4c6785476@amd.com
Link: https://patchwork.freedesktop.org/patch/msgid/20240507130411.630361-1-shuicheng.lin@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
The default behavior of device atomics depends on the
VM type and buffer allocation types. Device atomics are
expected to function with all types of allocations for
traditional applications/APIs. Additionally, in compute/SVM
API scenarios with fault mode or LR mode VMs, device atomics
must work with single-region allocations. In all other cases
device atomics should be disabled by default also on platforms
where we know device atomics doesn't on work on particular
allocations types.
v3: fault mode requires LR mode so only check for LR mode
to determine compute API(Jose).
Handle SMEM+LMEM BO's migration to LMEM where device
atomics is expected to work. (Brian).
v2: Fix platform checks to correct atomics behaviour on PVC.
Acked-by: Michal Mrozek <michal.mrozek@intel.com>
Reviewed-by: Oak Zeng <oak.zeng@intel.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240430162529.21588-6-nirmoy.das@intel.com
Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Currently we dma_map on ttm_tt population and dma_unmap when
the pages are released in ttm_tt unpopulate.
Strictly, the dma_map is not needed until the bo is moved to the
XE_PL_TT placement, so perform the dma_mapping on such moves
instead, and remove the dma_mappig when moving to XE_PL_SYSTEM.
This is desired for the upcoming shrinker series where shrinking
of a ttm_tt might fail. That would lead to an odd construct where
we first dma_unmap, then shrink and if shrinking fails dma_map
again. If dma_mapping instead is performed on move like this,
shrinking does not need to care at all about dma mapping.
Finally, where a ttm_tt is destroyed while bound to a different
memory type than XE_PL_SYSTEM, we keep the dma_unmap in
unpopulate().
v2:
- Don't accidently unmap the dma-buf's sgtable.
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/20240502183251.10170-1-thomas.hellstrom@linux.intel.com
What needs to be asserted is that the pattern fits in the number
of bits provided by the user in patternbits, otherwise it would
be truncated when replicated according to the mask, which is
likely not the intended use of this function.
The pattern argument is a bitmap so use find_last_bit() instead
of fls(). The bit position starts at index 0 so remove "or equal"
from the comparison. XE_MAX_L3_BANK_MASK_BITS would be the
returned value if the pattern is 0, which can be the case on some
platforms.
v2: Check the result does not overflow the array (Lucas De Marchi)
v3: Use __fls() for long and handle mask == 0 (Lucas De Marchi)
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240502124311.159695-1-francois.dugast@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
UAPI Changes:
- drm/i915/guc: Use context hints for GT frequency
Allow user to provide a low latency context hint. When set, KMD
sends a hint to GuC which results in special handling for this
context. SLPC will ramp the GT frequency aggressively every time
it switches to this context. The down freq threshold will also be
lower so GuC will ramp down the GT freq for this context more slowly.
We also disable waitboost for this context as that will interfere with
the strategy.
We need to enable the use of SLPC Compute strategy during init, but
it will apply only to contexts that set this bit during context
creation.
Userland can check whether this feature is supported using a new param-
I915_PARAM_HAS_CONTEXT_FREQ_HINT. This flag is true for all guc submission
enabled platforms as they use SLPC for frequency management.
The Mesa usage model for this flag is here -
https://gitlab.freedesktop.org/sushmave/mesa/-/commits/compute_hint
- drm/i915/gt: Enable only one CCS for compute workload
Enable only one CCS engine by default with all the compute sices
allocated to it.
While generating the list of UABI engines to be exposed to the
user, exclude any additional CCS engines beyond the first
instance
***
NOTE: This W/A will make all DG2 SKUs appear like single CCS SKUs by
default to mitigate a hardware bug. All the EUs will still remain
usable, and all the userspace drivers have been confirmed to be able
to dynamically detect the change in number of CCS engines and adjust.
For the smaller percent of applications that get perf benefit from
letting the userspace driver dispatch across all 4 CCS engines we will
be introducing a sysfs control as a later patch to choose 4 CCS each
with 25% EUs (or 50% if 2 CCS).
NOTE: A regression has been reported at
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/10895
However Andi has been triaging the issue and we're closing in a fix
to the gap in the W/A implementation:
https://lists.freedesktop.org/archives/intel-gfx/2024-April/348747.html
Driver Changes:
- Add new and fix to existing workarounds: Wa_14018575942 (MTL),
Wa_16019325821 (Gen12.70), Wa_14019159160 (MTL), Wa_16015675438,
Wa_14020495402 (Gen12.70) (Tejas, John, Lucas)
- Fix UAF on destroy against retire race and remove two earlier
partial fixes (Janusz)
- Limit the reserved VM space to only the platforms that need it (Andi)
- Reset queue_priority_hint on parking for execlist platforms (Chris)
- Fix gt reset with GuC submission is disabled (Nirmoy)
- Correct capture of EIR register on hang (John)
- Remove usage of the deprecated ida_simple_xx() API
- Refactor confusing __intel_gt_reset() (Nirmoy)
- Fix the fix for GuC reset lock confusion (John)
- Simplify/extend platform check for Wa_14018913170 (John)
- Replace dev_priv with i915 (Andi)
- Add and use gt_to_guc() wrapper (Andi)
- Remove bogus null check (Rodrigo, Dan)
. Selftest improvements (Janusz, Nirmoy, Daniele)
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZitVBTvZmityDi7D@jlahtine-mobl.ger.corp.intel.com
Thomas needs the defio fixes, Maíra needs the vkms fixes and Joonas
has some fun with i915-gem conflicts.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
In order to detect duplicate implementations for the same workaround,
early in the implementation of RTP it was decided to error out even if
the values set are exactly the same. With the introduction of 18034896535
in commit 74671d23ca ("drm/xe/xe2: Add workaround 18034896535"), LNL
stepping with graphics stepping A1 now gives the following error on
module load:
xe 0000:00:02.0: [drm] *ERROR* GT0: [GT OTHER] \
discarding save-restore reg e48c (clear: 00000200, set: 00000200,\
masked: yes, mcr: yes): ret=-22
RTP may be improved in the future, but for now simply join the entries
like done with e.g. "1607297627, 1607030317, 1607186500".
Fixes: 74671d23ca ("drm/xe/xe2: Add workaround 18034896535")
Cc: Bommu Krishnaiah <krishnaiah.bommu@intel.com>
Cc: Tejas Upadhyay <tejas.upadhyay@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240427135339.3485559-1-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Pull scheduler fixes from Ingo Molnar:
- Fix EEVDF corner cases
- Fix two nohz_full= related bugs that can cause boot crashes
and warnings
* tag 'sched-urgent-2024-04-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched/isolation: Fix boot crash when maxcpus < first housekeeping CPU
sched/isolation: Prevent boot crash when the boot CPU is nohz_full
sched/eevdf: Prevent vlag from going out of bounds in reweight_eevdf()
sched/eevdf: Fix miscalculation in reweight_entity() when se is not curr
sched/eevdf: Always update V if se->on_rq when reweighting
Pull x86 fixes from Ingo Molnar:
- Make the CPU_MITIGATIONS=n interaction with conflicting
mitigation-enabling boot parameters a bit saner.
- Re-enable CPU mitigations by default on non-x86
- Fix TDX shared bit propagation on mprotect()
- Fix potential show_regs() system hang when PKE initialization
is not fully finished yet.
- Add the 0x10-0x1f model IDs to the Zen5 range
- Harden #VC instruction emulation some more
* tag 'x86-urgent-2024-04-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
cpu: Ignore "mitigations" kernel parameter if CPU_MITIGATIONS=n
cpu: Re-enable CPU mitigations by default for !X86 architectures
x86/tdx: Preserve shared bit on mprotect()
x86/cpu: Fix check for RDPKRU in __show_regs()
x86/CPU/AMD: Add models 0x10-0x1f to the Zen5 range
x86/sev: Check for MWAITX and MONITORX opcodes in the #VC handler
Pull irq fix from Ingo Molnar:
"Fix a double free bug in the init error path of the GICv3 irqchip
driver"
* tag 'irq-urgent-2024-04-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
irqchip/gic-v3-its: Prevent double free on error
