i915/pmu: Wire GuC backend to per-client busyness
Subsystems (2)
PATCH 1/2 i915/pmu: Wire GuC backend to per-client busyness
From: John Harrison <[email protected]>
GuC provides engine_id and last_switch_in ticks for an active context in the pphwsp. The context image provides a 32 bit total ticks which is the accumulated by the context (a.k.a. context[CTX_TIMESTAMP]). This information is used to calculate the context busyness as follows:
If the engine_id is valid, then busyness is the sum of accumulated total ticks and active ticks. Active ticks is calculated with current gt time as reference.
If engine_id is invalid, busyness is equal to accumulated total ticks.
Since KMD (CPU) retrieves busyness data from 2 sources - GPU and GuC, a potential race was highlighted in an earlier review that can lead to double accounting of busyness. While the solution to this is a wip, busyness is still usable for platforms running GuC submission.
Remaining work: Enable and test context busyness for virtual_parent_context_ops and virtual_child_context_ops.
v2: (Tvrtko)
- Use COPS_RUNTIME_ACTIVE_TOTAL
- Add code comment for the race
- Undo local variables initializations
v3:
- Add support for virtual engines based on
https://patchwork.freedesktop.org/series/105227/
v4:
- Update commit message with remaining work.
- Rebase
--- drivers/gpu/drm/i915/gt/intel_context.c | 12 +++- drivers/gpu/drm/i915/gt/intel_context.h | 6 +- drivers/gpu/drm/i915/gt/intel_context_types.h | 6 ++ drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h | 5 ++ .../gpu/drm/i915/gt/uc/intel_guc_submission.c | 65 ++++++++++++++++++- drivers/gpu/drm/i915/i915_drm_client.c | 6 +- 6 files changed, 89 insertions(+), 11 deletions(-)
drivers/gpu/drm/i915/gt/intel_context.c +10 -2
@@ -576,16 +576,24 @@ void intel_context_bind_parent_child(struct intel_context *parent, child->parallel.parent = parent; } -u64 intel_context_get_total_runtime_ns (const struct intel_context *ce) +u64 intel_context_get_total_runtime_ns (struct intel_context *ce) {u64 total, active; +if (ce->ops->update_stats) + ce->ops->update_stats (ce); + total = ce->stats.runtime.total;if (ce->ops->flags &COPS_RUNTIME_CYCLES ) total *= ce->engine->gt->clock_period_ns; active =READ_ONCE (ce->stats.active); -if (active) +/* +* When COPS_RUNTIME_ACTIVE_TOTAL is set for ce->cops, the backend +* already provides the total active time of the context, so skip this +* calculation when this flag is set. +*/ +if (active && !(ce->ops->flags &COPS_RUNTIME_ACTIVE_TOTAL )) active =intel_context_clock () - active;return total + active;
drivers/gpu/drm/i915/gt/intel_context.h +3 -3
@@ -58,7 +58,7 @@ static inline bool intel_context_is_parent(struct intel_context *ce)return !!ce->parallel.number_children; } -static inline bool intel_context_is_pinned (struct intel_context *ce); +static inline bool intel_context_is_pinned (const struct intel_context *ce);static inline struct intel_context *intel_context_to_parent (struct intel_context *ce)
@@ -118,7 +118,7 @@ static inline int intel_context_lock_pinned(struct intel_context *ce)* Returns: true if the context is currently pinned for use by the GPU. */ static inline bool -intel_context_is_pinned (struct intel_context *ce) +intel_context_is_pinned (const struct intel_context *ce) {return atomic_read (&ce->pin_count); }
@@ -362,7 +362,7 @@ intel_context_clear_nopreempt(struct intel_context *ce)clear_bit (CONTEXT_NOPREEMPT , &ce->flags); } -u64 intel_context_get_total_runtime_ns (const struct intel_context *ce); +u64 intel_context_get_total_runtime_ns (struct intel_context *ce);u64 intel_context_get_avg_runtime_ns (struct intel_context *ce);static inline u64 intel_context_clock (void )
drivers/gpu/drm/i915/gt/intel_context_types.h +6 -0
@@ -38,6 +38,9 @@ struct intel_context_ops {#define COPS_RUNTIME_CYCLES_BIT 1#define COPS_RUNTIME_CYCLES BIT(COPS_RUNTIME_CYCLES_BIT) +#define COPS_RUNTIME_ACTIVE_TOTAL_BIT 2 +#define COPS_RUNTIME_ACTIVE_TOTAL BIT(COPS_RUNTIME_ACTIVE_TOTAL_BIT) +int (*alloc)(struct intel_context *ce); void (*revoke)(struct intel_context *ce, struct i915_request *rq,
@@ -56,6 +59,8 @@ struct intel_context_ops {void (*sched_disable)(struct intel_context *ce); +void (*update_stats)(struct intel_context *ce); +void (*reset)(struct intel_context *ce);void (*destroy)(struct kref *kref);
@@ -148,6 +153,7 @@ struct intel_context {struct ewma_runtime avg;u64 total;u32 last; +u64 start_gt_clk; I915_SELFTEST_DECLARE(u32 num_underflow); I915_SELFTEST_DECLARE(u32 max_underflow); } runtime;
drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h +5 -0
@@ -196,6 +196,11 @@ static inline u8 guc_class_to_engine_class(u8 guc_class)return guc_class_engine_class_map[guc_class]; } +/* Per context engine usage stats: */ +#define PPHWSP_GUC_CONTEXT_USAGE_STAMP_LO (0x500 / sizeof(u32)) +#define PPHWSP_GUC_CONTEXT_USAGE_STAMP_HI (PPHWSP_GUC_CONTEXT_USAGE_STAMP_LO + 1) +#define PPHWSP_GUC_CONTEXT_USAGE_ENGINE_ID (PPHWSP_GUC_CONTEXT_USAGE_STAMP_HI + 1) +/* Work item for submitting workloads into work queue of GuC. */ struct guc_wq_item {u32 header;
drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c +64 -1
126 lines not on this page · raw message
drivers/gpu/drm/i915/i915_drm_client.c +1 -5
@@ -147,11 +147,7 @@ void i915_drm_client_fdinfo(struct seq_file *m, struct file *f) PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn));seq_printf (m,"drm-client-id:\t%u\n" , client->id); -/* -* Temporarily skip showing client engine information with GuC submission till -* fetching engine busyness is implemented in the GuC submission backend -*/ -if (GRAPHICS_VER (i915) <8 ||intel_uc_uses_guc_submission (&i915->gt0.uc)) +if (GRAPHICS_VER (i915) <8 )return ;for (i =0 ; i <ARRAY_SIZE (uabi_class_names); i++)
Signature · 1 line
2.34.1
RFC PATCH v2 2/2 Fix per client busyness locking
1. Do all ce->stats updates and reads under guc->timestamp.lock
2. Pin context image before reading
3. Merge __guc_context_update_clks and guc_context_update_stats into a single function
4. Call lrc_update_runtime() unconditionally in guc_context_update_stats
5. Seems no need to update ce->stats.active with this approach
Some of the above steps may not be correct or complete but the idea is to discuss/improve the original patch.
v2 (Address review comments from Umesh):
* Don't pin context when called from intel_context_get_total_runtime_ns
* Don't pin context in lrc_update_runtime_locked
* ce->stats.active is now used to cache previous active value
* Move lrc_update_runtime to else in __guc_context_update_stats
--- drivers/gpu/drm/i915/gt/intel_context.c | 9 +-- drivers/gpu/drm/i915/gt/intel_context_types.h | 2 +- .../gpu/drm/i915/gt/uc/intel_guc_submission.c | 57 ++++++++++++------- 3 files changed, 40 insertions(+), 28 deletions(-)
drivers/gpu/drm/i915/gt/intel_context.c +2 -7
@@ -581,19 +581,14 @@ u64 intel_context_get_total_runtime_ns(struct intel_context *ce)u64 total, active;if (ce->ops->update_stats) - ce->ops->update_stats (ce); +return ce->ops->update_stats (ce, false); total = ce->stats.runtime.total;if (ce->ops->flags &COPS_RUNTIME_CYCLES ) total *= ce->engine->gt->clock_period_ns; active =READ_ONCE (ce->stats.active); -/* -* When COPS_RUNTIME_ACTIVE_TOTAL is set for ce->cops, the backend -* already provides the total active time of the context, so skip this -* calculation when this flag is set. -*/ -if (active && !(ce->ops->flags &COPS_RUNTIME_ACTIVE_TOTAL )) +if (active) active =intel_context_clock () - active;return total + active;
drivers/gpu/drm/i915/gt/intel_context_types.h +1 -1
@@ -59,7 +59,7 @@ struct intel_context_ops {void (*sched_disable)(struct intel_context *ce); -void (*update_stats)(struct intel_context *ce); +u64 (*update_stats)(struct intel_context *ce,bool pin);void (*reset)(struct intel_context *ce);void (*destroy)(struct kref *kref);
drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c +37 -20
103 lines not on this page · raw message
Signature · 1 line
2.34.1
kernel test robot
build botRe: [Intel-gfx] [RFC PATCH v2 2/2] Fix per client busyness locking
reproduce (this is a W=1 build): wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross #…
Hi Ashutosh,
[FYI, it's a private test report for your RFC patch.] [auto build test WARNING on drm-tip/drm-tip]
url: https://github.com/intel-lab-lkp/linux/commits/Ashutosh-Dixit/i915-pmu-Wire-GuC-backend-to-per-client-busyness/20220902-075534 base: git://anongit.freedesktop.org/drm/drm-tip drm-tip config: i386-randconfig-a004 (https://download.01.org/0day-ci/archive/20220902/[email protected]/config) compiler: clang version 14.0.6 (https://github.com/llvm/llvm-project f28c006a5895fc0e329fe15fead81e37457cb1d1)
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# https://github.com/intel-lab-lkp/linux/commit/b1db15edb2416229fdde749425de6a6ae7a4b94a
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Ashutosh-Dixit/i915-pmu-Wire-GuC-backend-to-per-client-busyness/20220902-075534 git checkout b1db15edb2416229fdde749425de6a6ae7a4b94a
# save the config file
mkdir build_dir && cp config build_dir/.config
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 O=build_dir ARCH=i386 SHELL=/bin/bash drivers/gpu/drm/i915/
If you fix the issue, kindly add following tag where applicable
All warnings (new ones prefixed by >>):
>> drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c:1512:5: warning: no previous prototype for function 'guc_context_update_stats' [-Wmissing-prototypes]
u64 guc_context_update_stats(struct intel_context *ce, bool pin)
^ drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c:1512:1: note: declare 'static' if the function is not intended to be used outside of this translation unit u64 guc_context_update_stats(struct intel_context *ce, bool pin)
^
static
1 warning generated.
vim +/guc_context_update_stats +1512 drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
1511
> 1512 u64 guc_context_update_stats(struct intel_context *ce, bool pin)
1513 {
1514 struct intel_guc *guc = ce_to_guc(ce);
1515 unsigned long flags;
1516 u64 ret;
1517
1518 if (pin) {
1519 intel_context_pin(ce);
1520 } else {
1521 /* Return previously computed value */
1522 spin_lock_irqsave(&guc->timestamp.lock, flags);
1523 ret = ce->stats.runtime.total + ce->stats.active;
1524 spin_unlock_irqrestore(&guc->timestamp.lock, flags);
1525 return ret;
1526 }
1527
1528 spin_lock_irqsave(&guc->timestamp.lock, flags);
1529 ret = __guc_context_update_stats(ce);
1530 spin_unlock_irqrestore(&guc->timestamp.lock, flags);
1531
1532 intel_context_unpin(ce);
1533 return ret;
1534 }
1535
Signature · 2 lines
0-DAY CI Kernel Test Service https://01.org/lkp
Patchwork
CI[Intel-gfx] ✗ Fi.CI.BUILD: failure for series starting with [1/2] i915/pmu: Wire GuC backend to per-client busyness
In reply to Ashutosh Dixit== Series Details ==
== Series Details ==
Series: series starting with [1/2] i915/pmu: Wire GuC backend to per-client busyness
URL : https://patchwork.freedesktop.org/series/108053/
State : failure
== Summary ==
Error: make failed
CALL scripts/checksyscalls.sh
CALL scripts/atomic/check-atomics.sh
DESCEND objtool
CHK include/generated/compile.h
CC [M] drivers/gpu/drm/i915/gt/uc/intel_guc_submission.odrivers/gpu/drm/i915/gt/uc/intel_guc_submission.c:1512:5: error: no previous prototype for ‘guc_context_update_stats’ [-Werror=missing-prototypes]
u64 guc_context_update_stats(struct intel_context *ce, bool pin)
^~~~~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errorsscripts/Makefile.build:249: recipe for target 'drivers/gpu/drm/i915/gt/uc/intel_guc_submission.o' failed make[4]: *** [drivers/gpu/drm/i915/gt/uc/intel_guc_submission.o] Error 1 scripts/Makefile.build:465: recipe for target 'drivers/gpu/drm/i915' failed
make[3]: *** [drivers/gpu/drm/i915] Error 2scripts/Makefile.build:465: recipe for target 'drivers/gpu/drm' failed
make[2]: *** [drivers/gpu/drm] Error 2scripts/Makefile.build:465: recipe for target 'drivers/gpu' failed
make[1]: *** [drivers/gpu] Error 2
Makefile:1853: recipe for target 'drivers' failed
make: *** [drivers] Error 2