swap_sample_id_all() calls BUG_ON(size % sizeof(u64)) which kills
perf on any event where the sample_id_all tail is not 8-byte aligned.
A crafted perf.data can trigger this trivially.
Replace BUG_ON with a bounds check: skip the swap if the data pointer
is past the end of the event, and only swap when there are bytes
remaining.
Note: the strlen calls in string-field swap handlers (comm,
mmap, mmap2, cgroup) are replaced with bounded strnlen by the
next patch in this series ("perf session: Add validated swap
infrastructure with null-termination checks").
Reported-by: sashiko-bot@kernel.org # Running on a local machine
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Assisted-by: Claude:claude-opus-4.6-1m
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The kernel dynamically sizes PERF_RECORD_READ based on
attr.read_format: only the fields enabled by PERF_FORMAT_TOTAL_TIME_ENABLED,
PERF_FORMAT_TOTAL_TIME_RUNNING, PERF_FORMAT_ID, and PERF_FORMAT_LOST
are emitted, packed with no gaps.
perf_event__read_swap() unconditionally byte-swapped time_enabled,
time_running, and id at their fixed struct offsets, causing
out-of-bounds access on smaller events and swapping the wrong
bytes when not all format fields are present. It also swapped
sample_id_all at a fixed offset past the full struct, which is
wrong for shorter events.
Replace the individual field swaps with a single mem_bswap_64()
over the entire tail from value onward. Since every field after
pid/tid is u64 regardless of which combination is present, this
correctly handles any read_format combination and any trailing
sample_id_all fields.
Similarly, dump_read() accessed optional fields via fixed struct
offsets, displaying values from wrong positions when not all
format bits are set. Walk the packed u64 array sequentially
instead, with bounds checks against event->header.size.
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Assisted-by: Claude:claude-opus-4.6-1m
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
zstd_decompress_stream() has two bugs in its multi-iteration loop:
1. After each ZSTD_decompressStream() call, the code advances
output.dst by output.pos but doesn't reset output.pos to 0.
ZSTD interprets output.pos relative to output.dst, so the
next iteration writes at (dst + pos) + pos = dst + 2*pos,
skipping a gap and potentially writing out of bounds.
2. On ZSTD_decompressStream() error, the loop executes break
and returns output.pos (which is > 0 if some bytes were
decompressed before the error). The caller checks
!decomp_size and skips the error, silently accepting
truncated or corrupted data.
Fix both by removing the output buffer adjustment — ZSTD
correctly accumulates output.pos across calls without it.
Return 0 on decompression error so the caller detects it.
Add a no-progress guard to prevent infinite loops if the
output buffer fills before all input is consumed.
Note: the compressed event data_size is validated against
header.size by a subsequent patch in this series
("perf tools: Harden compressed event processing").
Reported-by: sashiko-bot@kernel.org # Running on a local machine
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Assisted-by: Claude:claude-opus-4.6-1m
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The error fallback does memcpy(dst, src, src_size) intending to store
uncompressed data when compression fails, but this has three bugs:
1. dst has been advanced past the record header (and potentially
past earlier compressed records), so the copy writes to the
wrong offset in the output buffer.
2. src still points to the start of the input, not to the
remaining uncompressed data at src + input.pos. On a second
or later iteration, previously compressed data would be
duplicated.
3. No check that dst_size >= src_size — if the remaining output
space is smaller, this is an out-of-bounds write.
Replace with return -1 after resetting the ZSTD compression
context via ZSTD_initCStream(). The -1 propagates through
zstd_compress() -> record__pushfn() -> perf_mmap__push() to the
recording loop, which breaks out and terminates recording.
Add an out_child_no_flush label in __cmd_record() so the
mmap-read failure path skips the final record__mmap_read_all()
flush — retrying the same read that just failed would just fail
again, and the flush is only useful when the mmap data is intact
but the control path (auxtrace, switch_output) had an error.
Consolidate all error paths through a single 'reset' label to
ensure the compression context is always reset on failure —
including the output-buffer-full path, where a bare return
without resetting would leave stale stream state that corrupts
output if the caller retries.
Also guard against process_header() writing the event header
before the buffer-full check: add a sizeof(perf_event_header)
pre-check so the callback never writes past the output buffer.
Guard against ZSTD making no progress: if output.pos is zero
after ZSTD_compressStream(), calling process_header(record, 0)
would re-trigger header initialization, double-subtracting the
header size from dst_size and underflowing the unsigned counter.
Also fix two pre-existing issues in the same function:
- Add a dst_size guard before subtracting the record header
size: if the output buffer is nearly full, the unsigned
dst_size -= size underflows to a huge value, causing
ZSTD_compressStream to write past the buffer boundary.
- Check the ZSTD_initCStream() return value and log an error
if the context reset itself fails.
Reported-by: sashiko-bot@kernel.org # Running on a local machine
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Assisted-by: Claude:claude-opus-4.6-1m
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
event_contains() checked whether a field's start offset was within
the event (header.size > offsetof), but not whether the full field
fit. A crafted event with header.size = offsetof(field) + 1 would
pass the check, but an 8-byte access (bswap_64, direct read) would
overrun the event boundary by up to 7 bytes.
Fix the macro to verify the complete field:
header.size >= offsetof(field) + sizeof(field)
Also update all callers that check event_contains(time_cycles) but
access later fields (time_mask, cap_user_time_zero,
cap_user_time_short) to check for cap_user_time_short — the last
field accessed — so the entire extended block is verified:
tsc.c, arm-spe.c, cs-etm.c, jitdump.c.
Note: session.c's perf_event__time_conv_swap() also guards on
time_cycles but accesses time_mask — a pre-existing issue not
introduced by this macro change. It is fixed by a later patch
in this series ("perf session: Add validated swap
infrastructure with null-termination checks"), which decouples
time_cycles and time_mask into independent per-field
event_contains() checks. The struct assignment overread
(session->time_conv = event->time_conv copies sizeof on a
potentially shorter event) is separately fixed by "perf
session: Use bounded copy for PERF_RECORD_TIME_CONV".
Reported-by: sashiko-bot@kernel.org # Running on a local machine
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Assisted-by: Claude:claude-opus-4.6-1m
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf_session__peek_event() computes an event pointer directly from
file_offset when one_mmap is active, without verifying that file_offset
and the subsequent event->header.size fall within the mapped region.
A corrupted perf.data file could cause out-of-bounds memory reads.
Add one_mmap_size to the session struct and validate both the header
and full event fit within the mmap before dereferencing.
Reported-by: sashiko-bot@kernel.org # Running on a local machine
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Assisted-by: Claude:claude-opus-4.6-1m
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add a per-type minimum size table (perf_event__min_size[]) and
enforce it before swap and processing, so that both cross-endian
and native-endian paths are protected from accessing fields past
the event boundary.
The table uses offsetof() for types with trailing variable-length
fields (filenames, strings, msg arrays) and sizeof() for
fixed-size types. Zero entries mean no minimum beyond the 8-byte
header already enforced by the reader.
Undersized events are skipped with a warning in process_event
and rejected in peek_event — both checked before the swap
handler runs, preventing OOB access on crafted event fields.
Also reject events whose header.size is not 8-byte aligned. The
kernel aligns all event sizes to sizeof(u64) — see
perf_event_comm_event() (ALIGN), perf_event_mmap_event(),
perf_event_cgroup(), perf_event_ksymbol() (IS_ALIGNED loops),
and perf_event_text_poke() (ALIGN) in kernel/events/core.c.
An unaligned size means the file is corrupted or crafted; reject
early so downstream code that divides by sizeof(u64) to compute
array element counts gets exact results.
Three legacy user events are exempted from the alignment check:
TRACING_DATA (66) had a 12-byte struct before commit b39c915a4f
("libperf event: Ensure tracing data is multiple of 8 sized")
added padding, COMPRESSED (81) carries raw ZSTD output (already
superseded by COMPRESSED2 with PERF_ALIGN), and HEADER_FEATURE
(80) uses do_write_string() with a 4-byte length prefix.
Also guard event_swap() against crafted event types >=
PERF_RECORD_HEADER_MAX to prevent OOB reads on the
perf_event__swap_ops[] array.
Changes in v2:
- Fix double-skip for unsupported event types: return 0 instead
of event->header.size in perf_session__process_event() for
HEADER_MAX, since reader__read_event() already advances by
event->header.size (Reported-by: sashiko-bot@kernel.org)
- Exempt TRACING_DATA, COMPRESSED, and HEADER_FEATURE from the
alignment check — these legacy user events predate the 8-byte
alignment rule (Reported-by: sashiko-bot@kernel.org)
- peek_event: return 0 (skip) for unknown event types instead of
-1 (error), consistent with process_event which already skips
unsupported types gracefully (Reported-by: sashiko-bot@kernel.org)
Reported-by: sashiko-bot@kernel.org # Running on a local machine
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Assisted-by: Claude:claude-opus-4.6-1m
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Merge the mmc fixes for v7.1-rc[n] into the next branch, to allow them to
get tested together with the mmc changes that are targeted for the next
release.
Signed-off-by: Ulf Hansson <ulfh@kernel.org>
commit 42d99857d6 ("ASoC: core: Move all users to deferrable card binding")
converted the -EPROBE_DEFER return value of snd_soc_bind_card() to 0
which results in the machine driver probe return 0 and will not be
called again when any component is not yet registered.
We get the right component name from the registered components
and use it in the dai links. It will lead to bind fail if the default
component name is used. Return -EPROBE_DEFER to allow the machine driver
probe again after the components are registered.
Suggested-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://patch.msgid.link/20260529014259.2528048-1-yung-chuan.liao@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
None of the calls to trace_array_printk_buf() will do anything
if we don't initialize the buffer on instance creation (unless
some other tracer called it), so do that.
Add an osnoise_print() function to facilitate adding debug prints
(without tainting).
Use trace_array_printk() instead of trace_array_printk_buf(), as we're
only writing to the main buffer (of a non-main instance) anyway -- and
trace_array_printk_buf() skips the check to make sure we're not printing
to the global instance.
Link: https://patch.msgid.link/20260511223035.1475676-1-crwood@redhat.com
Signed-off-by: Crystal Wood <crwood@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Add catalog entry and register configuration for the Adreno 810
found in Qualcomm SM7635 (Milos) based devices.
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Signed-off-by: Alexander Koskovich <akoskovich@pm.me>
Patchwork: https://patchwork.freedesktop.org/patch/728812/
Message-ID: <20260528-adreno-810-v7-6-7fe7fdd97fc2@pm.me>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
A8XX GPUs have two sets of protect registers: 64 global slots and 16
pipe specific slots. The last-span-unbound feature is only available
on pipe protect registers, and should always target pipe slot 15.
This matches the downstream driver which hardcodes pipe slot 15 for
all A8XX GPUs (GRAPHICS.LA.15.0.r1) and resolves protect errors on
A810.
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Signed-off-by: Alexander Koskovich <akoskovich@pm.me>
Patchwork: https://patchwork.freedesktop.org/patch/728810/
Message-ID: <20260528-adreno-810-v7-5-7fe7fdd97fc2@pm.me>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Platforms without a LLCC (e.g. milos) still need to be able to read and
write to the cx_mem region. Previously if LLCC slices were unavailable
the cx_misc_mmio mapping was overwritten with ERR_PTR, causing a crash
when the GMU later accessed cx_mem.
Move the cx_misc_mmio mapping out of a6xx_llc_slices_init() into
a6xx_gpu_init() so that cx_mem mapping is independent of LLCC.
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Signed-off-by: Alexander Koskovich <akoskovich@pm.me>
Patchwork: https://patchwork.freedesktop.org/patch/728808/
Message-ID: <20260528-adreno-810-v7-4-7fe7fdd97fc2@pm.me>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
This region is used for more than just LLCC, it also provides access to
software fuse values (raytracing, etc).
Rename relevant symbols from _llc to _cx_misc for use in a follow up
change that decouples this from LLCC.
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Signed-off-by: Alexander Koskovich <akoskovich@pm.me>
Patchwork: https://patchwork.freedesktop.org/patch/728806/
Message-ID: <20260528-adreno-810-v7-3-7fe7fdd97fc2@pm.me>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Document the GPU compatible string used for the Adreno 810.
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Signed-off-by: Alexander Koskovich <akoskovich@pm.me>
Patchwork: https://patchwork.freedesktop.org/patch/728804/
Message-ID: <20260528-adreno-810-v7-2-7fe7fdd97fc2@pm.me>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Document Adreno 810 GMU in the dt-binding specification.
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Signed-off-by: Alexander Koskovich <akoskovich@pm.me>
Patchwork: https://patchwork.freedesktop.org/patch/728802/
Message-ID: <20260528-adreno-810-v7-1-7fe7fdd97fc2@pm.me>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Now that the dynamic pwrup reglist has SEL reg values to restore
appended, so that SEL regs are restored on IFPC exit, we can stop
completely disabling IFPC while global counter sampling is active.
To accomplish this, we re-use sysprof_setup() with a force_on param
to inhibit IFPC specifically while the counter regs are being read,
while leaving IFPC enabled the rest of the time.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728219/
Message-ID: <20260526145137.160554-17-robin.clark@oss.qualcomm.com>
This is needed so that SEL reg values are restored on exit from IFPC.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728218/
Message-ID: <20260526145137.160554-16-robin.clark@oss.qualcomm.com>
To make room for appending SEL reg programming. Without increasing the
size, we would overflow the pwrup_reglist at ~190 counters on gen8.
Or possibly fewer, considering that some gen8 counter groups also have
separate slice vs unslice SELectors.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728228/
Message-ID: <20260526145137.160554-15-robin.clark@oss.qualcomm.com>
Add new UABI and implementation of PERFCNTR_CONFIG ioctl.
A bit more work is required to configure the pwrup_reglist for the GMU
to restore SELect regs on exit of IFPC, before we can stop disabling
IFPC while global counter collection. This will follow in a later
commit, but will be transparent to userspace.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728217/
Message-ID: <20260526145137.160554-14-robin.clark@oss.qualcomm.com>
With the slice architecture, we need to flush the slice and unslice
counters to perf RAM before reading counters.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728216/
Message-ID: <20260526145137.160554-13-robin.clark@oss.qualcomm.com>
Add support to configure counter SELect regs. In some cases the reg
writes need to happen while the GPU is idle. And for a7xx+, in some
cases SEL regs need to be configured from BV or BR aperture. The
easiest way to deal with this is to configure from the RB.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728215/
Message-ID: <20260526145137.160554-12-robin.clark@oss.qualcomm.com>
Add the basic infrastructure for tracking assigned perfcntrs.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728212/
Message-ID: <20260526145137.160554-11-robin.clark@oss.qualcomm.com>
The upcoming PERFCNTR_CONFIG ioctl will allow for both global counter
collection, and per-context counter reservation for local (ie. within
a single GEM_SUBMIT ioctl) counter collection.
Any number of contexts can reserve the same counters, but we will need
to ensure that counters reserved for local counter collection do not
conflict with counters used for global counter collection.
So add tracking for per-context local counter reservations.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728211/
Message-ID: <20260526145137.160554-10-robin.clark@oss.qualcomm.com>
It's a common pattern, needing to insert a yield packet before flushing
the rb. And we'll need this once again for configuring perfcntr SEL
regs. So add a helper.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728208/
Message-ID: <20260526145137.160554-9-robin.clark@oss.qualcomm.com>
Currently the sysprof param serves two functions, (a) disabling perfcntr
clearing on context switch/preemption, and (b) disabling IFPC. In the
future, with kernel side global perfcntr collection/stream, the decision
about disabling IFPC will change.
To prepare for this, split out two helpers/accessors for the two
different cases. For now, they are the same thing, but this will
change.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728214/
Message-ID: <20260526145137.160554-8-robin.clark@oss.qualcomm.com>
Wire up the generated perfcntr tables for a6xx+. The PERFCNTR_CONFIG
ioctl will use this information to assign counters.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728213/
Message-ID: <20260526145137.160554-7-robin.clark@oss.qualcomm.com>
Pull in perfcntr json and wire up generation of perfcntr tables.
Sync from mesa commit a573e25b6dcd ("freedreno/registers: Gen8 perfcntr
fixes")
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728204/
Message-ID: <20260526145137.160554-6-robin.clark@oss.qualcomm.com>
Update gen_header.py to bring in support for generating perfcntr tables.
Sync from mesa commit 96c5179c02d1 ("freedreno/registers: Skip deprecated
warns for kernel")
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728203/
Message-ID: <20260526145137.160554-5-robin.clark@oss.qualcomm.com>
Most of the churn is just reworking the usage attribute on the mesa
side.
Sync from mesa commit ff41a00fab89 ("freedreno/registers: Correct
register name")
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728202/
Message-ID: <20260526145137.160554-4-robin.clark@oss.qualcomm.com>
Use perfmon_capable() which checks both CAP_SYS_ADMIN and CAP_PERFMON.
This matches what i915 and xe do, and seems more appropriate.
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728198/
Message-ID: <20260526145137.160554-3-robin.clark@oss.qualcomm.com>
Outside of a3xx, this was never really used. And it low-key gets in the
way of the new perfcntr support (or at least it is confusing to have two
things called "perf"). So lets remove it.
This drops the "perf" debugfs file. But these days, nvtop is a better
option. (Plus perfetto for newer gens.)
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Anna Maniscalco <anna.maniscalco2000@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/728200/
Message-ID: <20260526145137.160554-2-robin.clark@oss.qualcomm.com>
Adreno X2-185 GPU found in Glymur chipsets belongs to the A8x family.
It features a new slice architecture with 4 slices, significantly higher
bandwidth throughput compared to mobile counterparts, raytracing support,
and the highest GPU Fmax seen so far on an Adreno GPU (1850 Mhz), among
other improvements. Update the dt bindings documentation to describe this
GPU.
Signed-off-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/727119/
Message-ID: <20260522-glymur-gpu-dt-v5-2-562c406b210c@oss.qualcomm.com>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
In A8xx, the RSCC block is part of GPU's register space. Update the
virtual base address of rscc to point to the correct address.
Fixes: 50e8a557d8 ("drm/msm/a8xx: Add support for A8x GMU")
Signed-off-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/727117/
Message-ID: <20260522-glymur-gpu-dt-v5-1-562c406b210c@oss.qualcomm.com>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
In A8x GPUs, the GX GDSC is moved to a separate block called GXCLKCTL
which is under the GX power domain. Due to the way the support for this
block is implemented in its driver, pm_runtime votes result in a vote on
GX/GMxC/MxC rails from the APPS RSC. This is against the Adreno
architecture which require GMU to be the sole voter of these collapsible
rails on behalf of GPU, except during the GPU/GMU recovery.
To align with this architectural requirement and to realize the power
benefits of the IFPC feature, remove the GXPD votes during gmu resume
and suspend. And during the recovery sequence, enable/disable the GXPD
along with the 'synced_poweroff' genpd hint to force collapse this GDSC.
Signed-off-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Signed-off-by: Taniya Das <taniya.das@oss.qualcomm.com>
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/720979/
Message-ID: <20260427-gfx-clk-fixes-v2-6-797e54b3d464@oss.qualcomm.com>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Similar to a6xx_recover(), check the GX power domain status before
accessing mmio in GX domain a8xx_recover().
Fixes: 288a932008 ("drm/msm/adreno: Introduce A8x GPU Support")
Signed-off-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Signed-off-by: Taniya Das <taniya.das@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/720977/
Message-ID: <20260427-gfx-clk-fixes-v2-5-797e54b3d464@oss.qualcomm.com>
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
Sashiko reported an inconsistent use of NULL vs ERR_PTR()
returns in the stub helpers in xynos-acpm-protocol.h.
Since this only happens on dead code for COMPILE_TEST=y, this is not
really a bug though. Having stub functions that return NULL is a common
way to define optional interfaces, where callers still work when the
feature is disabled, though this clearly does not work for acpm because
some callers have a NULL pointer dereference when compile testing.
Since CONFIG_EXYNOS_ACPM_PROTOCOL already supports compile-testing itself,
and all (both) drivers using it clearly require the support, so this
just simplifies the option space without losing any build coverage.
Remove the stub functions entirely and adjust the one Kconfig
dependency to require EXYNOS_ACPM_PROTOCOL unconditionally.
Fixes: 6837c006d4 ("firmware: exynos-acpm: add empty method to allow compile test")
Closes: https://sashiko.dev/#/patchset/20260420-acpm-tmu-v3-0-3dc8e93f0b26%40linaro.org
Link: https://lore.kernel.org/all/a7994860-24a3-4f87-84bf-109ed653dda4@linaro.org/
Reviewed-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20260529134454.2147446-1-arnd@kernel.org
[krzk: Rebase on difference in devm_acpm_get_by_node()]
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Introduce devm_acpm_get_by_phandle() to standardize how consumer
drivers acquire a handle to the ACPM IPC interface. Enforce the
use of the "samsung,acpm-ipc" property name across the SoC and
simplify the boilerplate code in client drivers.
The first consumer of this helper is the Exynos ACPM Thermal Management
Unit (TMU) driver. The TMU utilizes a hybrid management approach: direct
register access from the Application Processor (AP) is restricted to the
interrupt pending (INTPEND) registers for event identification.
High-level functional tasks, such as sensor initialization, threshold
programming, and temperature reads, are delegated to the ACPM firmware
via this IPC interface.
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Link: https://patch.msgid.link/20260515-acpm-tmu-helpers-v2-6-8ca011d5a965@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
The Thermal Management Unit (TMU) on the Google GS101 SoC is managed
through a hybrid model shared between the kernel and the Alive Clock
and Power Manager (ACPM) firmware.
Add the protocol helpers required to communicate with the ACPM for
thermal operations, including initialization, threshold configuration,
temperature reading, and system suspend/resume handshakes.
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Link: https://patch.msgid.link/20260515-acpm-tmu-helpers-v2-5-8ca011d5a965@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Replace the embedded `struct acpm_ops` inside `struct acpm_handle` with
a pointer to a `const struct acpm_ops`.
Previously, the operations structure was embedded directly within the
handle and populated dynamically at runtime via `acpm_setup_ops()`.
This resulted in mutable function pointers and unnecessary per-instance
memory overhead.
By defining `exynos_acpm_driver_ops` statically as a `const` structure,
the function pointers are now safely housed in the read-only `.rodata`
section. This improves security by preventing function pointer
overwrites, saves memory, and slightly reduces initialization overhead
in `acpm_probe()`.
Consequently, update all consumer drivers (clk, mfd) to access the
operations via the new pointer indirection (`->ops->`). Finally, fix
the previously empty kernel-doc description for the ops member to
reflect its new pointer nature.
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Link: https://patch.msgid.link/20260515-acpm-tmu-helpers-v2-4-8ca011d5a965@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Rename the `n_cmd` member of `struct acpm_rx_data` to `cmdcnt` to
maintain consistent nomenclature across the driver (aligning with
`txcnt`, `rxcnt`, and transfer helpers).
With the member renamed, annotate the dynamically allocated `cmd`
pointer with the `__counted_by_ptr(cmdcnt)` macro to improve runtime
bounds checking.
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Link: https://patch.msgid.link/20260515-acpm-tmu-helpers-v2-2-8ca011d5a965@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Both the DVFS and PMIC ACPM sub-drivers implement similar local helper
functions (acpm_dvfs_set_xfer and acpm_pmic_set_xfer) to initialize the
acpm_xfer structure before sending an IPC message.
Move this logic into a single centralized helper, acpm_set_xfer(),
in the core ACPM driver to reduce boilerplate, eliminate code
duplication, and prepare for the upcoming ACPM TMU helper sub-driver
which will also utilize this method.
Note that there is no change in underlying functionality. While the old
acpm_pmic_set_xfer() unconditionally assigned the RX buffer parameters
(xfer->rxd and xfer->rxcnt), the new unified helper introduces a
'response' boolean. All updated PMIC call sites now explicitly pass
'true' for this argument. This ensures the unified helper takes the
'if (response)' branch, performing the exact same assignments and
preserving the original PMIC behavior.
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Link: https://patch.msgid.link/20260515-acpm-tmu-helpers-v2-1-8ca011d5a965@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
The ring_size field in struct dw_mci is misleadingly named.
Despite its name, it does not represent the size of the descriptor
ring buffer in bytes, but rather the number of descriptors allocated
within the fixed-size ring buffer.
The actual ring buffer size is fixed at PAGE_SIZE (or DESC_RING_BUF_SZ,
which equals PAGE_SIZE). Within this buffer, we allocate either
struct idmac_desc or struct idmac_desc_64addr descriptors, and
ring_size stores the count of these descriptors.
This naming has caused confusion, as it's also used to set
mmc->max_segs (the maximum number of scatter-gather segments),
which logically corresponds to the number of descriptors, not a
size in bytes.
No functional change is introduced by this naming-only patch.
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Ulf Hansson <ulfh@kernel.org>
This is the only Qcom binding that doesn't have 'qcom' prefix in the
bindings name. This doesn't match with the regex in MAINTAINERS file and
the 'get_maintainer.pl' script fails to list the 'linux-arm-msm' list:
Ulf Hansson <ulfh@kernel.org> (maintainer:MULTIMEDIA CARD (MMC), SECURE DIGITAL (SD) AND...)
Rob Herring <robh@kernel.org> (maintainer:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS)
Krzysztof Kozlowski <krzk+dt@kernel.org> (maintainer:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS)
Conor Dooley <conor+dt@kernel.org> (maintainer:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS)
Bjorn Andersson <andersson@kernel.org> (in file)
Konrad Dybcio <konradybcio@kernel.org> (in file)
linux-mmc@vger.kernel.org (open list:MULTIMEDIA CARD (MMC), SECURE DIGITAL (SD) AND...)
devicetree@vger.kernel.org (open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS)
linux-kernel@vger.kernel.org (open list)
Hence, rename the binding to include 'qcom' prefix so that the regex
matches correctly.
Reported-by: Mukesh Ojha <mukesh.ojha@oss.qualcomm.com>
Closes: https://lore.kernel.org/all/20260526151003.p4kn2vek3hpv4gzv@hu-mojha-hyd.qualcomm.com
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@oss.qualcomm.com>
Signed-off-by: Ulf Hansson <ulfh@kernel.org>
I met one suspend/resume issue with sdr104 capable sdio wifi card (with
"keep-power-in-suspend" set in DT property):
After resuming from suspend to ram, the sdio wifi card stops working.
Further debug shows that although ios shows the sdio card is at sdr104
mode, the voltage is still at 3V3. This is due to missing the calling
of ->start_signal_voltage_switch() in sdhci_resume_host().
Fix this issue by adding ->start_signal_voltage_switch() in
sdhci_resume_host(). This also matches what we do for
sdhci_runtime_resume_host().
Then the question is: why this issue hasn't reported and fixed for so
long time. IMHO, several reasons: Some host controllers just kick off
the runtime resume for system resume, so they benefit from the well
supported runtime pm code; Some platforms just use the old sdio wifi
card which doesn't need signal voltage switch at all, the default
voltage is 3v3 after resuming.
Fixes: 6308d2905b ("mmc: sdhci: add quirk for keeping card power during suspend")
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ulf Hansson <ulfh@kernel.org>
The really old controllers (rk2928, rk3066, rk3188) do not support UHS
speeds at all, and thus never handled phase data.
For that reason it never had a parse_dt callback and no driver private
data at all.
Commit ff6f0286c8 ("mmc: dw_mmc-rockchip: Add memory clock auto-gating
support") makes the private data sort of mandatory, because the init
function checks whether phases are configured internally or through the
clock controller.
This results in the old SoCs then experiencing NULL-pointer dereferences
when they try to access that private-data struct.
While we could have if (priv) conditionals in all places, it's way less
cluttery to just give the old types their private-data struct.
Fixes: ff6f0286c8 ("mmc: dw_mmc-rockchip: Add memory clock auto-gating support")
Cc: stable@vger.kernel.org
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Acked-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Ulf Hansson <ulfh@kernel.org>
Replace common pattern of dev_err() + return with dev_err_probe() in
probe functions and their callees. This macro provides standardized
error message format with symbolic error names and adds deferred probe
debugging information.
The conversion makes the code more compact and ensures consistent error
logging across all initialization paths.
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Artem Shimko <a.shimko.dev@gmail.com>
Signed-off-by: Ulf Hansson <ulfh@kernel.org>
