[PATCH 1/1] drm/xe/guc/capture: Maintenence of devcoredump <-> GuC-Err-Capture plumbing
Teres Alexis, Alan Previn
alan.previn.teres.alexis at intel.com
Tue Dec 3 03:12:19 UTC 2024
On Tue, 2024-11-26 at 12:09 -0500, Dong, Zhanjun wrote:
> See my comments inline below:
>
> Regards,
> Zhanjun
>
> On 2024-11-17 1:44 p.m., Alan Previn wrote:
> > The order of the devcoredump event flow is:
> > drm-scheduler -> guc-submission-execq-timed-out-job ->
> > guc-submission-kill-job -> xe-devcoredump (once the work
> > is confirmed to have been killed).
> >
> > As we are aware, the GuC-FW IRQ for error-capture delivery
> > and extraction could have happenned before the start of
> happened?
alan: will fix.
> > guc-execq-timed-out-job or the middle of it (before or
> > during the explicit kill) or not at all. Thus, today, the
> > above flow takes a manual capture first before triggering
> > the kill-job just in case we need it.
> >
> > The structure layering of devcoredump internals are:
> > xe_devcoredump_snapshot -> xe_foo_snapshot (where foo
> > can be any data dump associated to the job was killed).
> > Foo includes the xe_hw_engine_snapshot. Since GuC-Error-Capture
> > provides just the register dump of an engine, GuC-Err-Capture
> > snapshots should be managed by the xe_hw_engine_snapshot.
> > That isn't the case today.
> >
> > Furthermode, neither xe_devcoredump_snapshot nor
> Furthermore?
alan: will fix
> > xe_hw_engine_snapshot even exists at the start of
> > guc-submission-execq-timed-out-job. Thus, the first
> > manual capture node has no home. However, today,
> > GuC-Error-Capture stores capture snapshots off the
> > top-level xe_devcoredump_snapshot's matched_node.
> > GuC-Error-Capture also had absorbed the function for
> > xe_hw_engine_snapshot generation.
> > NOTE: Existing code isn't broken because xe_devcoredump
> > is not dynamically allocated and designed to hold a
> > single event at a time (i.e. single engine dump).
> > But it's not scalable for future improvement.
> >
> > Thus this patch:
> > 1. Moves "matched_node" from xe_devcoredump_snapshot to
> > xe_hw_engine_snapshot.
> > 2. Relocates the functions for xe_hw_engine_snapshot generation
> > and printing back to xe_hw_engine.c. However, split out the
> > register dump printing so it stays within GuC-Error-Capture
> > (so we don't need to maintain two sets of register lists).
> > 3. Keep both the manual and firmware capture nodes within
> > GuC-Error-Capture subsystem's linked list until
> > xe_hw_engine_snapshot gets and puts them later.
> > 4. Give xe_hw_engine_snapshot the control and ability to
> > query GuC-Error-Capture for matching snapshots while choosing
> > between manual vs firmware capture getting/putting node.
> > 5. While at it, relocate (and rename) key structures, enums
> > and function protos to xe_guc_capture_snapshot_types.h
> > (as an inter-module header) for xe_hw_engine_snapshot to use.
> > 6. Since xe_hw_engine_snapshot can also be called by via debugfs
> > without a job, create a new function that does a manual capture
> > of engine registers without any associated job.
> >
> > Signed-off-by: Alan Previn <alan.previn.teres.alexis at intel.com>
> > ---
> > drivers/gpu/drm/xe/xe_devcoredump.c | 3 -
> > drivers/gpu/drm/xe/xe_devcoredump_types.h | 6 -
> > drivers/gpu/drm/xe/xe_guc_capture.c | 397 ++++++++----------
> > drivers/gpu/drm/xe/xe_guc_capture.h | 10 +-
> > .../drm/xe/xe_guc_capture_snapshot_types.h | 68 +++
> > drivers/gpu/drm/xe/xe_guc_submit.c | 21 +-
> > drivers/gpu/drm/xe/xe_hw_engine.c | 117 ++++--
> > drivers/gpu/drm/xe/xe_hw_engine.h | 4 +-
> > drivers/gpu/drm/xe/xe_hw_engine_types.h | 13 +-
> > 9 files changed, 353 insertions(+), 286 deletions(-)
> > create mode 100644 drivers/gpu/drm/xe/xe_guc_capture_snapshot_types.h
> >
> > diff --git a/drivers/gpu/drm/xe/xe_devcoredump.c b/drivers/gpu/drm/xe/xe_devcoredump.c
> > index 0e5edf14a241..b98d71e670ca 100644
> > --- a/drivers/gpu/drm/xe/xe_devcoredump.c
> > +++ b/drivers/gpu/drm/xe/xe_devcoredump.c
> > @@ -144,9 +144,6 @@ static void xe_devcoredump_snapshot_free(struct xe_devcoredump_snapshot *ss)
> > xe_guc_ct_snapshot_free(ss->guc.ct);
> > ss->guc.ct = NULL;
> >
> > - xe_guc_capture_put_matched_nodes(&ss->gt->uc.guc);
> > - ss->matched_node = NULL;
> > -
> > xe_guc_exec_queue_snapshot_free(ss->ge);
> > ss->ge = NULL;
> >
> > diff --git a/drivers/gpu/drm/xe/xe_devcoredump_types.h b/drivers/gpu/drm/xe/xe_devcoredump_types.h
> > index be4d59ea9ac8..06ac75ce63dd 100644
> > --- a/drivers/gpu/drm/xe/xe_devcoredump_types.h
> > +++ b/drivers/gpu/drm/xe/xe_devcoredump_types.h
> > @@ -49,12 +49,6 @@ struct xe_devcoredump_snapshot {
> > struct xe_hw_engine_snapshot *hwe[XE_NUM_HW_ENGINES];
> > /** @job: Snapshot of job state */
> > struct xe_sched_job_snapshot *job;
> > - /**
> > - * @matched_node: The matched capture node for timedout job
> > - * this single-node tracker works because devcoredump will always only
> > - * produce one hw-engine capture per devcoredump event
> > - */
> > - struct __guc_capture_parsed_output *matched_node;
> > /** @vm: Snapshot of VM state */
> > struct xe_vm_snapshot *vm;
> >
> > diff --git a/drivers/gpu/drm/xe/xe_guc_capture.c b/drivers/gpu/drm/xe/xe_guc_capture.c
> > index f87755af545f..691fc72a5c9e 100644
> > --- a/drivers/gpu/drm/xe/xe_guc_capture.c
> > +++ b/drivers/gpu/drm/xe/xe_guc_capture.c
> > @@ -26,6 +26,7 @@
> > #include "xe_guc_ads.h"
> > #include "xe_guc_capture.h"
> > #include "xe_guc_capture_types.h"
> > +#include "xe_guc_capture_snapshot_types.h"
> > #include "xe_guc_ct.h"
> > #include "xe_guc_exec_queue_types.h"
> > #include "xe_guc_log.h"
> > @@ -53,40 +54,6 @@ struct __guc_capture_bufstate {
> > u32 wr;
> > };
> >
> > -/*
> > - * struct __guc_capture_parsed_output - extracted error capture node
> > - *
> > - * A single unit of extracted error-capture output data grouped together
> > - * at an engine-instance level. We keep these nodes in a linked list.
> > - * See cachelist and outlist below.
> > - */
> > -struct __guc_capture_parsed_output {
> > - /*
> > - * A single set of 3 capture lists: a global-list
> > - * an engine-class-list and an engine-instance list.
> > - * outlist in __guc_capture_parsed_output will keep
> > - * a linked list of these nodes that will eventually
> > - * be detached from outlist and attached into to
> > - * xe_codedump in response to a context reset
> > - */
> > - struct list_head link;
> > - bool is_partial;
> > - u32 eng_class;
> > - u32 eng_inst;
> > - u32 guc_id;
> > - u32 lrca;
> > - u32 type;
> > - bool locked;
> > - enum xe_hw_engine_snapshot_source_id source;
> > - struct gcap_reg_list_info {
> > - u32 vfid;
> > - u32 num_regs;
> > - struct guc_mmio_reg *regs;
> > - } reginfo[GUC_STATE_CAPTURE_TYPE_MAX];
> > -#define GCAP_PARSED_REGLIST_INDEX_GLOBAL BIT(GUC_STATE_CAPTURE_TYPE_GLOBAL)
> > -#define GCAP_PARSED_REGLIST_INDEX_ENGCLASS BIT(GUC_STATE_CAPTURE_TYPE_ENGINE_CLASS)
> > -};
> > -
> > /*
> > * Define all device tables of GuC error capture register lists
> > * NOTE:
> > @@ -286,7 +253,7 @@ struct xe_guc_state_capture {
> >
> > static void
> > guc_capture_remove_stale_matches_from_list(struct xe_guc_state_capture *gc,
> > - struct __guc_capture_parsed_output *node);
> > + struct xe_guc_capture_snapshot *node);
> >
> > static const struct __guc_mmio_reg_descr_group *
> > guc_capture_get_device_reglist(struct xe_device *xe)
> > @@ -840,7 +807,7 @@ static void check_guc_capture_size(struct xe_guc *guc)
> > }
> >
> > static void
> > -guc_capture_add_node_to_list(struct __guc_capture_parsed_output *node,
> > +guc_capture_add_node_to_list(struct xe_guc_capture_snapshot *node,
> > struct list_head *list)
> > {
> > list_add(&node->link, list);
> > @@ -848,7 +815,7 @@ guc_capture_add_node_to_list(struct __guc_capture_parsed_output *node,
> >
> > static void
> > guc_capture_add_node_to_outlist(struct xe_guc_state_capture *gc,
> > - struct __guc_capture_parsed_output *node)
> > + struct xe_guc_capture_snapshot *node)
> > {
> > guc_capture_remove_stale_matches_from_list(gc, node);
> > guc_capture_add_node_to_list(node, &gc->outlist);
> > @@ -856,17 +823,17 @@ guc_capture_add_node_to_outlist(struct xe_guc_state_capture *gc,
> >
> > static void
> > guc_capture_add_node_to_cachelist(struct xe_guc_state_capture *gc,
> > - struct __guc_capture_parsed_output *node)
> > + struct xe_guc_capture_snapshot *node)
> > {
> > guc_capture_add_node_to_list(node, &gc->cachelist);
> > }
> >
> > static void
> > guc_capture_free_outlist_node(struct xe_guc_state_capture *gc,
> > - struct __guc_capture_parsed_output *n)
> > + struct xe_guc_capture_snapshot *n)
> > {
> > if (n) {
> > - n->locked = 0;
> > + n->locked = false;
> > list_del(&n->link);
> > /* put node back to cache list */
> > guc_capture_add_node_to_cachelist(gc, n);
> > @@ -875,9 +842,9 @@ guc_capture_free_outlist_node(struct xe_guc_state_capture *gc,
> >
> > static void
> > guc_capture_remove_stale_matches_from_list(struct xe_guc_state_capture *gc,
> > - struct __guc_capture_parsed_output *node)
> > + struct xe_guc_capture_snapshot *node)
> > {
> > - struct __guc_capture_parsed_output *n, *ntmp;
> > + struct xe_guc_capture_snapshot *n, *ntmp;
> > int guc_id = node->guc_id;
> >
> > list_for_each_entry_safe(n, ntmp, &gc->outlist, link) {
> > @@ -887,7 +854,7 @@ guc_capture_remove_stale_matches_from_list(struct xe_guc_state_capture *gc,
> > }
> >
> > static void
> > -guc_capture_init_node(struct xe_guc *guc, struct __guc_capture_parsed_output *node)
> > +guc_capture_init_node(struct xe_guc *guc, struct xe_guc_capture_snapshot *node)
> > {
> > struct guc_mmio_reg *tmp[GUC_STATE_CAPTURE_TYPE_MAX];
> > int i;
> > @@ -934,24 +901,31 @@ guc_capture_init_node(struct xe_guc *guc, struct __guc_capture_parsed_output *no
> > * This node is created from a pre-allocated list of blank nodes in
> > * guc->capture->cachelist and populated with the error-capture
> > * data from GuC and then it's added into guc->capture->outlist linked
> > - * list. This list is used for matchup and printout by xe_devcoredump_read
> > - * and xe_engine_snapshot_print, (when user invokes the devcoredump sysfs).
> > + * list. This node (dumped-register-lists) will be requested later by
> > + * xe_hwe_engine_snapshot creation.
> > *
> > - * GUC --> notify context reset:
> > - * -----------------------------
> > + * DRM Scheduler job-timeout OR GuC-notify guc-id reset:
> > + * -----------------------------------------------------
> > * --> guc_exec_queue_timedout_job
> > - * L--> xe_devcoredump
> > - * L--> devcoredump_snapshot
> > - * --> xe_hw_engine_snapshot_capture
> > - * --> xe_engine_manual_capture(For manual capture)
> > + * L--> alloc D: xe_guc_capture_snapshot_store_manual_job
> > + * (only if "alloc C" above didn't happen)
> > + * L--> devcoredump_snapshot
> > + * --> xe_engine_snapshot_capture_for_queue(q) ..
> > + * --> xe_guc_capture_snapshot_get(q)
> > + * use "alloc C" or else "alloc D"
> > + * --> store in xe_hw_engine_snapshot->matched_node
> > *
> > - * User Sysfs / Debugfs
> > - * --------------------
> > - * --> xe_devcoredump_read->
> > + * User Devcoredump Sysfs
> > + * ----------------------
> > + * --> xe_devcoredump_read-> (user cats devcoredump)
> > * L--> xxx_snapshot_print
> > * L--> xe_engine_snapshot_print
> > - * Print register lists values saved at
> > - * guc->capture->outlist
> > + * L --> xe_guc_capture_snapshot_print (for printing register
> > + * lists values of xe_hw_engine_snapshot->matched_node)
> > + * --> xe_devcoredump_free (user clears devcoredump)
> > + * L--> xe_devcoredump_free -> xe_devcoredump_snapshot_free
> > + * L--> xe_hw_engine_snapshot_free
> > + * L--> xe_guc_capture_snapshot_put(matched_node)
> > *
> > */
> >
> > @@ -1066,13 +1040,13 @@ guc_capture_log_get_register(struct xe_guc *guc, struct __guc_capture_bufstate *
> > return 0;
> > }
> >
> > -static struct __guc_capture_parsed_output *
> > +static struct xe_guc_capture_snapshot *
> > guc_capture_get_prealloc_node(struct xe_guc *guc)
> > {
> > - struct __guc_capture_parsed_output *found = NULL;
> > + struct xe_guc_capture_snapshot *found = NULL;
> >
> > if (!list_empty(&guc->capture->cachelist)) {
> > - struct __guc_capture_parsed_output *n, *ntmp;
> > + struct xe_guc_capture_snapshot *n, *ntmp;
> >
> > /* get first avail node from the cache list */
> > list_for_each_entry_safe(n, ntmp, &guc->capture->cachelist, link) {
> > @@ -1080,7 +1054,7 @@ guc_capture_get_prealloc_node(struct xe_guc *guc)
> > break;
> > }
> > } else {
> > - struct __guc_capture_parsed_output *n, *ntmp;
> > + struct xe_guc_capture_snapshot *n, *ntmp;
> >
> > /*
> > * traverse reversed and steal back the oldest node already
> > @@ -1099,11 +1073,11 @@ guc_capture_get_prealloc_node(struct xe_guc *guc)
> > return found;
> > }
> >
> > -static struct __guc_capture_parsed_output *
> > -guc_capture_clone_node(struct xe_guc *guc, struct __guc_capture_parsed_output *original,
> > +static struct xe_guc_capture_snapshot *
> > +guc_capture_clone_node(struct xe_guc *guc, struct xe_guc_capture_snapshot *original,
> > u32 keep_reglist_mask)
> > {
> > - struct __guc_capture_parsed_output *new;
> > + struct xe_guc_capture_snapshot *new;
> > int i;
> >
> > new = guc_capture_get_prealloc_node(guc);
> > @@ -1145,7 +1119,7 @@ guc_capture_extract_reglists(struct xe_guc *guc, struct __guc_capture_bufstate *
> > struct xe_gt *gt = guc_to_gt(guc);
> > struct guc_state_capture_group_header_t ghdr = {0};
> > struct guc_state_capture_header_t hdr = {0};
> > - struct __guc_capture_parsed_output *node = NULL;
> > + struct xe_guc_capture_snapshot *node = NULL;
> > struct guc_mmio_reg *regs = NULL;
> > int i, numlists, numregs, ret = 0;
> > enum guc_state_capture_type datatype;
> > @@ -1438,11 +1412,11 @@ void xe_guc_capture_process(struct xe_guc *guc)
> > __guc_capture_process_output(guc);
> > }
> >
> > -static struct __guc_capture_parsed_output *
> > +static struct xe_guc_capture_snapshot *
> > guc_capture_alloc_one_node(struct xe_guc *guc)
> > {
> > struct drm_device *drm = guc_to_drm(guc);
> > - struct __guc_capture_parsed_output *new;
> > + struct xe_guc_capture_snapshot *new;
> > int i;
> >
> > new = drmm_kzalloc(drm, sizeof(*new), GFP_KERNEL);
> > @@ -1467,7 +1441,7 @@ guc_capture_alloc_one_node(struct xe_guc *guc)
> > static void
> > __guc_capture_create_prealloc_nodes(struct xe_guc *guc)
> > {
> > - struct __guc_capture_parsed_output *node = NULL;
> > + struct xe_guc_capture_snapshot *node = NULL;
> > int i;
> >
> > for (i = 0; i < PREALLOC_NODES_MAX_COUNT; ++i) {
> All above is about renaming structure or comments, that's fine.
alan: yes, thats right, this plumbing refactor allows the xe_hw_engine_capture
to get and put a guc-matching-node or manual-capture-node and exposing a handle
type to an external subsystem means we should prefix it with the standard
naming "xe_guc" in front.
>
> > @@ -1564,35 +1538,18 @@ read_reg_to_node(struct xe_hw_engine *hwe, const struct __guc_mmio_reg_descr_gro
> > }
> > }
> >
> > -/**
> > - * xe_engine_manual_capture - Take a manual engine snapshot from engine.
> > - * @hwe: Xe HW Engine.
> > - * @snapshot: The engine snapshot
> > - *
> > - * Take engine snapshot from engine read.
> > - *
> > - * Returns: None
> > - */
> > -void
> > -xe_engine_manual_capture(struct xe_hw_engine *hwe, struct xe_hw_engine_snapshot *snapshot)
> > +static struct xe_guc_capture_snapshot *
> > +guc_capture_get_manual_snapshot(struct xe_guc *guc, struct xe_hw_engine *hwe)
> > {
> > struct xe_gt *gt = hwe->gt;
> > - struct xe_device *xe = gt_to_xe(gt);
> > - struct xe_guc *guc = >->uc.guc;
> > - struct xe_devcoredump *devcoredump = &xe->devcoredump;
> > enum guc_capture_list_class_type capture_class;
> > const struct __guc_mmio_reg_descr_group *list;
> > - struct __guc_capture_parsed_output *new;
> > + struct xe_guc_capture_snapshot *new;
> > enum guc_state_capture_type type;
> > - u16 guc_id = 0;
> > - u32 lrca = 0;
> > -
> > - if (IS_SRIOV_VF(xe))
> > - return;
> >
> > new = guc_capture_get_prealloc_node(guc);
> > if (!new)
> > - return;
> > + return NULL;
> >
> > capture_class = xe_engine_class_to_guc_capture_class(hwe->class);
> > for (type = GUC_STATE_CAPTURE_TYPE_GLOBAL; type < GUC_STATE_CAPTURE_TYPE_MAX; type++) {
> > @@ -1626,26 +1583,83 @@ xe_engine_manual_capture(struct xe_hw_engine *hwe, struct xe_hw_engine_snapshot
> > }
> > }
> >
> > - if (devcoredump && devcoredump->captured) {
> > - struct xe_guc_submit_exec_queue_snapshot *ge = devcoredump->snapshot.ge;
> > + new->eng_class = xe_engine_class_to_guc_class(hwe->class);
> > + new->eng_inst = hwe->instance;
> >
> > - if (ge) {
> > - guc_id = ge->guc.id;
> > - if (ge->lrc[0])
> > - lrca = ge->lrc[0]->context_desc;
> > - }
> > + return new;
> > +}
> > +
> > +/**
> > + * xe_guc_capture_snapshot_store_and_get_manual_hwe - Generate and get manual engine register dump
> > + * @guc: Target GuC for manual capture
> > + * @hwe: The engine instance to capture from
> > + *
> > + * Generate a manual GuC-Error-Capture snapshot of engine instance + engine class registers
> > + * without any queue association. This capture node is not stored in outlist or cachelist,
> > + * Returns: New capture node and caller must "put"
> > + */
> > +struct xe_guc_capture_snapshot *
> > +xe_guc_capture_snapshot_store_and_get_manual_hwe(struct xe_guc *guc, struct xe_hw_engine *hwe)
> > +{
> > + struct xe_guc_capture_snapshot *new;
> > +
> > + new = guc_capture_get_manual_snapshot(guc, hwe);
> > + if (!new)
> > + return NULL;
> > +
> > + new->guc_id = 0;
> > + new->lrca = 0;
> > + new->is_partial = 0;
> > + new->source = XE_ENGINE_CAPTURE_SOURCE_MANUAL;
> > +
> > + return new;
> > +}
> > +
> > +/**
> > + * xe_guc_capture_snapshot_store_manual_job - Generate and store a manual engine register dump
> > + * @guc: Target GuC for manual capture
> > + * @q: Associated xe_exec_queue to simulate a manual capture on its behalf.
> > + *
> > + * Generate a manual GuC-Error-Capture snapshot of engine instance + engine class registers
> > + * for the engine of the given exec queue. Stores this node in internal outlist for future
> > + * retrieval with the ability to match up against the same queue.
> > + *
> > + * Returns: None
> > + */
> > +void
> > +xe_guc_capture_snapshot_store_manual_job(struct xe_guc *guc, struct xe_exec_queue *q)
> > +{
> > + struct xe_guc_capture_snapshot *new;
> > + struct xe_gt *gt = guc_to_gt(guc);
> > + struct xe_hw_engine *hwe;
> > + enum xe_hw_engine_id id;
> > +
> > + if (q) {
> > + xe_gt_warn(gt, "Manual GuC Error capture requested with invalid job\n");
> > + return;
> > }
> Need to do SRIOV check here or at timedout job
> if (IS_SRIOV_VF(xe))
> return;
> for compareration, the existing implementation is
> guc_exec_queue_timedout_job
> xe_engine_snapshot_capture_for_queue
> which did the SRIOV check.
>
alan: good catch - thanks - I'll add that bailout.
> >
> > - new->eng_class = xe_engine_class_to_guc_class(hwe->class);
> > - new->eng_inst = hwe->instance;
> > - new->guc_id = guc_id;
> > - new->lrca = lrca;
> > + /* Find hwe for the queue */
> > + for_each_hw_engine(hwe, gt, id) {
> > + if (hwe != q->hwe)
> > + continue;
> > + break;
> > + }
> > +
> > + new = guc_capture_get_manual_snapshot(guc, hwe);
> > + if (!new)
> > + return;
> > +
> > + new->guc_id = q->guc->id;
> > + new->lrca = xe_lrc_ggtt_addr(q->lrc[0]);
> > new->is_partial = 0;
> > - new->locked = 1;
> > + /* lock manual captures until devcoredump-engine puts it */
> > + new->locked = true;
> > new->source = XE_ENGINE_CAPTURE_SOURCE_MANUAL;
> >
> > guc_capture_add_node_to_outlist(guc->capture, new);
> > - devcoredump->snapshot.matched_node = new;
> > +
> > + return;
> > }
> >
> > static struct guc_mmio_reg *
> > @@ -1666,24 +1680,16 @@ guc_capture_find_reg(struct gcap_reg_list_info *reginfo, u32 addr, u32 flags)
> > }
> >
> > static void
> > -snapshot_print_by_list_order(struct xe_hw_engine_snapshot *snapshot, struct drm_printer *p,
> > - u32 type, const struct __guc_mmio_reg_descr_group *list)
> > +print_noderegs_by_list_order(struct xe_guc *guc, struct gcap_reg_list_info *reginfo,
> > + const struct __guc_mmio_reg_descr_group *list, struct drm_printer *p)
> > {
> > - struct xe_gt *gt = snapshot->hwe->gt;
> > - struct xe_device *xe = gt_to_xe(gt);
> > - struct xe_guc *guc = >->uc.guc;
> > - struct xe_devcoredump *devcoredump = &xe->devcoredump;
> > - struct xe_devcoredump_snapshot *devcore_snapshot = &devcoredump->snapshot;
> > - struct gcap_reg_list_info *reginfo = NULL;
> > u32 last_value, i;
> > bool is_ext;
> >
> > if (!list || list->num_regs == 0)
> > return;
> > - XE_WARN_ON(!devcore_snapshot->matched_node);
> >
> > is_ext = list == guc->capture->extlists;
> > - reginfo = &devcore_snapshot->matched_node->reginfo[type];
> >
> > /*
> > * loop through descriptor first and find the register in the node
> > @@ -1717,7 +1723,7 @@ snapshot_print_by_list_order(struct xe_hw_engine_snapshot *snapshot, struct drm_
> >
> > group = FIELD_GET(GUC_REGSET_STEERING_GROUP, reg_desc->flags);
> > instance = FIELD_GET(GUC_REGSET_STEERING_INSTANCE, reg_desc->flags);
> > - dss = xe_gt_mcr_steering_info_to_dss_id(gt, group, instance);
> > + dss = xe_gt_mcr_steering_info_to_dss_id(guc_to_gt(guc), group, instance);
> >
> > drm_printf(p, "\t%s[%u]: 0x%08x\n", reg_desc->regname, dss, value);
> > } else {
> > @@ -1727,74 +1733,65 @@ snapshot_print_by_list_order(struct xe_hw_engine_snapshot *snapshot, struct drm_
> > }
> >
> > /**
> > - * xe_engine_snapshot_print - Print out a given Xe HW Engine snapshot.
> > - * @snapshot: Xe HW Engine snapshot object.
> > + * xe_guc_capture_snapshot_print - Print out a the contents of a provided Guc-Err-Capture node
> > + * @guc : Target GuC for operation.
> > + * @node: GuC Error Capture register dump node.
> > * @p: drm_printer where it will be printed out.
> > *
> > - * This function prints out a given Xe HW Engine snapshot object.
> > + * This function prints out a register dump of a GuC-Err-Capture node that was retrieved
> > + * earlier either by GuC-FW reporting or by manual capture depending on how the
> > + * caller (typically xe_hw_engine_snapshot) was invoked and used.
> > */
> > -void xe_engine_snapshot_print(struct xe_hw_engine_snapshot *snapshot, struct drm_printer *p)
> > +
> > +void xe_guc_capture_snapshot_print(struct xe_guc *guc, struct xe_guc_capture_snapshot *node,
> > + struct drm_printer *p)
> > {
> > const char *grptype[GUC_STATE_CAPTURE_GROUP_TYPE_MAX] = {
> > "full-capture",
> > "partial-capture"
> > };
> > - int type;
> > const struct __guc_mmio_reg_descr_group *list;
> > - enum guc_capture_list_class_type capture_class;
> > -
> > struct xe_gt *gt;
> > - struct xe_device *xe;
> > - struct xe_devcoredump *devcoredump;
> > - struct xe_devcoredump_snapshot *devcore_snapshot;
> > + int type;
> >
> > - if (!snapshot)
> > + if (!guc)
> > return;
> > -
> > - gt = snapshot->hwe->gt;
> > - xe = gt_to_xe(gt);
> > - devcoredump = &xe->devcoredump;
> > - devcore_snapshot = &devcoredump->snapshot;
> > -
> > - if (!devcore_snapshot->matched_node)
> > + gt = guc_to_gt(guc);
> > + if (!node) {
> > + xe_gt_warn(gt, "GuC Capture printing without node!\n");
> > return;
> > + }
> > + if (!p) {
> > + xe_gt_warn(gt, "GuC Capture printing without printer!\n");
> > + return;
> > + }
> >
> > - xe_gt_assert(gt, snapshot->hwe);
> > -
> > - capture_class = xe_engine_class_to_guc_capture_class(snapshot->hwe->class);
> > -
> > - drm_printf(p, "%s (physical), logical instance=%d\n",
> > - snapshot->name ? snapshot->name : "",
> > - snapshot->logical_instance);
> > drm_printf(p, "\tCapture_source: %s\n",
> > - devcore_snapshot->matched_node->source == XE_ENGINE_CAPTURE_SOURCE_GUC ?
> > + node->source == XE_ENGINE_CAPTURE_SOURCE_GUC ?
> > "GuC" : "Manual");
> > - drm_printf(p, "\tCoverage: %s\n", grptype[devcore_snapshot->matched_node->is_partial]);
> > - drm_printf(p, "\tForcewake: domain 0x%x, ref %d\n",
> > - snapshot->forcewake.domain, snapshot->forcewake.ref);
> > - drm_printf(p, "\tReserved: %s\n",
> > - str_yes_no(snapshot->kernel_reserved));
> I see this change seperate capture related print into this function.
> I like this idea.
alan: yeah, it makes it better for the layering design while allowing
xe_hw_engine_snapshot to be the owner of guc-error-capture node get/print/put.
>
> > + drm_printf(p, "\tCoverage: %s\n", grptype[node->is_partial]);
> >
> > for (type = GUC_STATE_CAPTURE_TYPE_GLOBAL; type < GUC_STATE_CAPTURE_TYPE_MAX; type++) {
> > list = xe_guc_capture_get_reg_desc_list(gt, GUC_CAPTURE_LIST_INDEX_PF, type,
> > - capture_class, false);
> > - snapshot_print_by_list_order(snapshot, p, type, list);
> > + node->eng_class, false);
> > + print_noderegs_by_list_order(guc, &node->reginfo[type], list, p);
> > }
> >
> > - if (capture_class == GUC_CAPTURE_LIST_CLASS_RENDER_COMPUTE) {
> > + if (node->eng_class == GUC_CAPTURE_LIST_CLASS_RENDER_COMPUTE) {
> > + type = GUC_STATE_CAPTURE_TYPE_ENGINE_CLASS;
> > list = xe_guc_capture_get_reg_desc_list(gt, GUC_CAPTURE_LIST_INDEX_PF,
> > - GUC_STATE_CAPTURE_TYPE_ENGINE_CLASS,
> > - capture_class, true);
> > - snapshot_print_by_list_order(snapshot, p, GUC_STATE_CAPTURE_TYPE_ENGINE_CLASS,
> > - list);
> > + type, node->eng_class, true);
> > + print_noderegs_by_list_order(guc, &node->reginfo[type], list, p);
> > }
> > -
> > drm_puts(p, "\n");
> > }
> >
> > /**
> > - * xe_guc_capture_get_matching_and_lock - Matching GuC capture for the queue.
> > - * @q: The exec queue object
> > + * xe_guc_capture_snapshot_get - Matching GuC capture for the queue.
> > + *
> > + * @guc: The GuC being searched for a matching guc-error-capture snapshot.
> > + * @q: The exec queue object that provides engine, guc-id and lrca to match (can be NULL!)
> > + * @src: The source of the GuC-error-Capture snapshot to retrieve;
> > *
> > * Search within the capture outlist for the queue, could be used for check if
> > * GuC capture is ready for the queue.
> > @@ -1802,28 +1799,31 @@ void xe_engine_snapshot_print(struct xe_hw_engine_snapshot *snapshot, struct drm
> > *
> > * Returns: found guc-capture node ptr else NULL
> > */
> > -struct __guc_capture_parsed_output *
> > -xe_guc_capture_get_matching_and_lock(struct xe_exec_queue *q)
> > +struct xe_guc_capture_snapshot *
> > +xe_guc_capture_snapshot_get(struct xe_guc *guc, struct xe_exec_queue *q,
> > + enum xe_guc_capture_snapshot_source src)
> > {
> > struct xe_hw_engine *hwe;
> > enum xe_hw_engine_id id;
> > + struct xe_gt *gt;
> > struct xe_device *xe;
> > u16 guc_class = GUC_LAST_ENGINE_CLASS + 1;
> > - struct xe_devcoredump_snapshot *ss;
> >
> > - if (!q || !q->gt)
> > + if (!guc)
> > return NULL;
> >
> > - xe = gt_to_xe(q->gt);
> > - if (xe->wedged.mode >= 2 || !xe_device_uc_enabled(xe) || IS_SRIOV_VF(xe))
> > + gt = guc_to_gt(guc);
> > + if (q && q->gt != gt) {
> > + xe_gt_warn(gt, "Guc-Err-Capture being querried with incorrect queue's GT!");
> > return NULL;
> > + }
> >
> > - ss = &xe->devcoredump.snapshot;
> > - if (ss->matched_node && ss->matched_node->source == XE_ENGINE_CAPTURE_SOURCE_GUC)
> > - return ss->matched_node;
> > + xe = gt_to_xe(gt);
> > + if (xe->wedged.mode >= 2 || !xe_device_uc_enabled(xe) || IS_SRIOV_VF(xe))
> > + return NULL;
> >
> > /* Find hwe for the queue */
> > - for_each_hw_engine(hwe, q->gt, id) {
> > + for_each_hw_engine(hwe, gt, id) {
> > if (hwe != q->hwe)
> > continue;
> > guc_class = xe_engine_class_to_guc_class(hwe->class);
> > @@ -1831,8 +1831,7 @@ xe_guc_capture_get_matching_and_lock(struct xe_exec_queue *q)
> > }
> >
> > if (guc_class <= GUC_LAST_ENGINE_CLASS) {
> > - struct __guc_capture_parsed_output *n, *ntmp;
> > - struct xe_guc *guc = &q->gt->uc.guc;
> > + struct xe_guc_capture_snapshot *n, *ntmp;
> > u16 guc_id = q->guc->id;
> > u32 lrca = xe_lrc_ggtt_addr(q->lrc[0]);
> >
> > @@ -1844,8 +1843,8 @@ xe_guc_capture_get_matching_and_lock(struct xe_exec_queue *q)
> > list_for_each_entry_safe(n, ntmp, &guc->capture->outlist, link) {
> > if (n->eng_class == guc_class && n->eng_inst == hwe->instance &&
> > n->guc_id == guc_id && n->lrca == lrca &&
> > - n->source == XE_ENGINE_CAPTURE_SOURCE_GUC) {
> > - n->locked = 1;
> > + n->source == src) {
> > + n->locked = true;
> > return n;
> > }
> > }
> > @@ -1853,77 +1852,19 @@ xe_guc_capture_get_matching_and_lock(struct xe_exec_queue *q)
> > return NULL;
> > }
> >
> > -/**
> > - * xe_engine_snapshot_capture_for_queue - Take snapshot of associated engine
> > - * @q: The exec queue object
> > - *
> > - * Take snapshot of associated HW Engine
> > - *
> > - * Returns: None.
> > - */
> > -void
> > -xe_engine_snapshot_capture_for_queue(struct xe_exec_queue *q)
> > -{
> > - struct xe_device *xe = gt_to_xe(q->gt);
> > - struct xe_devcoredump *coredump = &xe->devcoredump;
> > - struct xe_hw_engine *hwe;
> > - enum xe_hw_engine_id id;
> > - u32 adj_logical_mask = q->logical_mask;
> > -
> > - if (IS_SRIOV_VF(xe))
> > - return;
> > -
> > - for_each_hw_engine(hwe, q->gt, id) {
> > - if (hwe->class != q->hwe->class ||
> > - !(BIT(hwe->logical_instance) & adj_logical_mask)) {
> > - coredump->snapshot.hwe[id] = NULL;
> > - continue;
> > - }
> > -
> > - if (!coredump->snapshot.hwe[id]) {
> > - coredump->snapshot.hwe[id] =
> > - xe_hw_engine_snapshot_capture(hwe, q);
> > - } else {
> > - struct __guc_capture_parsed_output *new;
> > -
> > - new = xe_guc_capture_get_matching_and_lock(q);
> > - if (new) {
> > - struct xe_guc *guc = &q->gt->uc.guc;
> > -
> > - /*
> > - * If we are in here, it means we found a fresh
> > - * GuC-err-capture node for this engine after
> > - * previously failing to find a match in the
> > - * early part of guc_exec_queue_timedout_job.
> > - * Thus we must free the manually captured node
> > - */
> > - guc_capture_free_outlist_node(guc->capture,
> > - coredump->snapshot.matched_node);
> > - coredump->snapshot.matched_node = new;
> > - }
> > - }
> > -
> > - break;
> > - }
> > -}
> > -
> > /*
> > - * xe_guc_capture_put_matched_nodes - Cleanup macthed nodes
> > + * xe_guc_capture_snapshot_put - Release GuC Capture matched node from a prior snapshot_get
> > * @guc: The GuC object
> > + * @node: The node to release back to GuC
> > *
> > * Free matched node and all nodes with the equal guc_id from
> > * GuC captured outlist
> > */
> > -void xe_guc_capture_put_matched_nodes(struct xe_guc *guc)
> > +void xe_guc_capture_snapshot_put(struct xe_guc *guc, struct xe_guc_capture_snapshot *n)
> > {
> > - struct xe_device *xe = guc_to_xe(guc);
> > - struct xe_devcoredump *devcoredump = &xe->devcoredump;
> > - struct __guc_capture_parsed_output *n = devcoredump->snapshot.matched_node;
> > -
> > if (n) {
> > guc_capture_remove_stale_matches_from_list(guc->capture, n);
> > guc_capture_free_outlist_node(guc->capture, n);
> > - devcoredump->snapshot.matched_node = NULL;
> > }
> > }
> >
> > diff --git a/drivers/gpu/drm/xe/xe_guc_capture.h b/drivers/gpu/drm/xe/xe_guc_capture.h
> > index 20a078dc4b85..e85af277be9c 100644
> > --- a/drivers/gpu/drm/xe/xe_guc_capture.h
> > +++ b/drivers/gpu/drm/xe/xe_guc_capture.h
> > @@ -11,10 +11,7 @@
> > #include "xe_guc.h"
> > #include "xe_guc_fwif.h"
> >
> > -struct xe_exec_queue;
> > struct xe_guc;
> > -struct xe_hw_engine;
> > -struct xe_hw_engine_snapshot;
> >
> > static inline enum guc_capture_list_class_type xe_guc_class_to_capture_class(u16 class)
> > {
> > @@ -44,18 +41,13 @@ void xe_guc_capture_process(struct xe_guc *guc);
> > int xe_guc_capture_getlist(struct xe_guc *guc, u32 owner, u32 type,
> > enum guc_capture_list_class_type capture_class, void **outptr);
> > int xe_guc_capture_getlistsize(struct xe_guc *guc, u32 owner, u32 type,
> > - enum guc_capture_list_class_type capture_class, size_t *size);
> > + enum guc_capture_list_class_type capture_class, size_t *size);
> > int xe_guc_capture_getnullheader(struct xe_guc *guc, void **outptr, size_t *size);
> > size_t xe_guc_capture_ads_input_worst_size(struct xe_guc *guc);
> > const struct __guc_mmio_reg_descr_group *
> > xe_guc_capture_get_reg_desc_list(struct xe_gt *gt, u32 owner, u32 type,
> > enum guc_capture_list_class_type capture_class, bool is_ext);
> > -struct __guc_capture_parsed_output *xe_guc_capture_get_matching_and_lock(struct xe_exec_queue *q);
> > -void xe_engine_manual_capture(struct xe_hw_engine *hwe, struct xe_hw_engine_snapshot *snapshot);
> > -void xe_engine_snapshot_print(struct xe_hw_engine_snapshot *snapshot, struct drm_printer *p);
> > -void xe_engine_snapshot_capture_for_queue(struct xe_exec_queue *q);
> > void xe_guc_capture_steered_list_init(struct xe_guc *guc);
> > -void xe_guc_capture_put_matched_nodes(struct xe_guc *guc);
> > int xe_guc_capture_init(struct xe_guc *guc);
> >
> > #endif
> > diff --git a/drivers/gpu/drm/xe/xe_guc_capture_snapshot_types.h b/drivers/gpu/drm/xe/xe_guc_capture_snapshot_types.h
> > new file mode 100644
> > index 000000000000..76159cb8fcab
> > --- /dev/null
> > +++ b/drivers/gpu/drm/xe/xe_guc_capture_snapshot_types.h
> > @@ -0,0 +1,68 @@
> > +/* SPDX-License-Identifier: MIT */
> > +/*
> > + * Copyright © 2021-2024 Intel Corporation
> > + */
> > +
> > +#ifndef _XE_GUC_CAPTURE_SNAPSHOT_TYPES_H
> > +#define _XE_GUC_CAPTURE_SNAPSHOT_TYPES_H
> > +
> > +#include <linux/types.h>
> > +#include <abi/guc_capture_abi.h>
> > +#include "xe_guc_fwif.h"
> > +
> > +struct drm_printer;
> > +struct guc_mmio_reg;
> > +struct xe_guc;
> > +struct xe_exec_queue;
> > +struct xe_hw_engine;
> > +
> > +enum xe_guc_capture_snapshot_source {
> > + XE_ENGINE_CAPTURE_SOURCE_MANUAL,
> > + XE_ENGINE_CAPTURE_SOURCE_GUC
> > +};
> > +
> > +/*
> > + * struct xe_guc_capture_snapshot - extracted error capture node
> > + *
> > + * A single unit of extracted error-capture output data grouped together
> > + * at an engine-instance level. We keep these nodes in a linked list.
> > + * See cachelist and outlist below.
> > + */
> > +struct xe_guc_capture_snapshot {
> > + /*
> > + * A single set of 3 capture lists: a global-list
> > + * an engine-class-list and an engine-instance list.
> > + * outlist in __guc_capture_parsed_output will keep
> > + * a linked list of these nodes that will eventually
> > + * be detached from outlist and attached into to
> > + * xe_codedump in response to a context reset
> > + */
> > + struct list_head link;
> > + bool is_partial;
> > + u32 eng_class;
> > + u32 eng_inst;
> > + u32 guc_id;
> > + u32 lrca;
> > + u32 type;
> > + bool locked;
> > + enum xe_guc_capture_snapshot_source source;
> > + struct gcap_reg_list_info {
> > + u32 vfid;
> > + u32 num_regs;
> > + struct guc_mmio_reg *regs;
> > + } reginfo[GUC_STATE_CAPTURE_TYPE_MAX];
> > +#define GCAP_PARSED_REGLIST_INDEX_GLOBAL BIT(GUC_STATE_CAPTURE_TYPE_GLOBAL)
> > +#define GCAP_PARSED_REGLIST_INDEX_ENGCLASS BIT(GUC_STATE_CAPTURE_TYPE_ENGINE_CLASS)
> > +};
> > +
> > +struct xe_guc_capture_snapshot *
> > +xe_guc_capture_snapshot_get(struct xe_guc *guc, struct xe_exec_queue *q,
> > + enum xe_guc_capture_snapshot_source src);
> > +void xe_guc_capture_snapshot_print(struct xe_guc *guc, struct xe_guc_capture_snapshot *node,
> > + struct drm_printer *p);
> > +void xe_guc_capture_snapshot_put(struct xe_guc *guc, struct xe_guc_capture_snapshot *snapshot);
> > +void xe_guc_capture_snapshot_store_manual_job(struct xe_guc *guc, struct xe_exec_queue *q);
> > +struct xe_guc_capture_snapshot *
> > +xe_guc_capture_snapshot_store_and_get_manual_hwe(struct xe_guc *guc, struct xe_hw_engine *hwe);
> > +
> > +#endif
> > diff --git a/drivers/gpu/drm/xe/xe_guc_submit.c b/drivers/gpu/drm/xe/xe_guc_submit.c
> > index f9ecee5364d8..30c2bdf51958 100644
> > --- a/drivers/gpu/drm/xe/xe_guc_submit.c
> > +++ b/drivers/gpu/drm/xe/xe_guc_submit.c
> > @@ -28,6 +28,7 @@
> > #include "xe_gt_printk.h"
> > #include "xe_guc.h"
> > #include "xe_guc_capture.h"
> > +#include "xe_guc_capture_snapshot_types.h"
> > #include "xe_guc_ct.h"
> > #include "xe_guc_exec_queue_types.h"
> > #include "xe_guc_id_mgr.h"
> > @@ -1033,7 +1034,6 @@ guc_exec_queue_timedout_job(struct drm_sched_job *drm_job)
> > struct xe_gpu_scheduler *sched = &q->guc->sched;
> > struct xe_guc *guc = exec_queue_to_guc(q);
> > const char *process_name = "no process";
> > - struct xe_device *xe = guc_to_xe(guc);
> > unsigned int fw_ref;
> > int err = -ETIME;
> > pid_t pid = -1;
> > @@ -1062,18 +1062,23 @@ guc_exec_queue_timedout_job(struct drm_sched_job *drm_job)
> > exec_queue_destroyed(q);
> >
> > /*
> > - * If devcoredump not captured and GuC capture for the job is not ready
> > - * do manual capture first and decide later if we need to use it
> > + * If the queue has't been killed yet or we do not have a firmware-reported
> > + * GuC-Error-Capture node for the matching job, request GuC-Error-Capture to
> > + * store a manual capture within its internal list with a job-match.
> > + * xe_hw_engine_snapshot will decide later if it's needed.
> > */
> > - if (!exec_queue_killed(q) && !xe->devcoredump.captured &&
> > - !xe_guc_capture_get_matching_and_lock(q)) {
> > + if (!exec_queue_killed(q) ||
> > > or && ?
> If we already got GuC capture, no need for manual capture
>
> > + !xe_guc_capture_snapshot_get(guc, q, XE_ENGINE_CAPTURE_SOURCE_GUC)) {
> I see !xe->devcoredump.captured check is missing, any concerns?
> Right now, if devcoredump already captured, we don't need the manual
> capture.
>
alan: I was attempting to clean up the logic to make guc-error-capture
logic modular and agnostic, but on second thoughts, i realize its a very
quick and simple optimization that you have already done due diligence in
testing and verifying. That said, i will fix this and revert back to the
original logic.
> > /* take force wake before engine register manual capture */
> > fw_ref = xe_force_wake_get(gt_to_fw(q->gt), XE_FORCEWAKE_ALL);
> > if (!xe_force_wake_ref_has_domain(fw_ref, XE_FORCEWAKE_ALL))
> > xe_gt_info(q->gt, "failed to get forcewake for coredump capture\n");
> > -
> > - xe_engine_snapshot_capture_for_queue(q);
> > -
> > + /*
> > + * This will generate a manual capture node and store it in
> > + * This GuC Error Capture link-list as if it came from GuC
> > + * but with a source-id == manual-capture
> > + */
> > + xe_guc_capture_snapshot_store_manual_job(guc, q);
> > xe_force_wake_put(gt_to_fw(q->gt), fw_ref);
> > }
> >
> > diff --git a/drivers/gpu/drm/xe/xe_hw_engine.c b/drivers/gpu/drm/xe/xe_hw_engine.c
> > index c4b0dc3be39c..ddd91627e623 100644
> > --- a/drivers/gpu/drm/xe/xe_hw_engine.c
> > +++ b/drivers/gpu/drm/xe/xe_hw_engine.c
> > @@ -24,7 +24,8 @@
> > #include "xe_gt_printk.h"
> > #include "xe_gt_mcr.h"
> > #include "xe_gt_topology.h"
> > -#include "xe_guc_capture.h"
> > +//#include "xe_guc_capture.h"
> > +#include "xe_guc_capture_snapshot_types.h"
> > #include "xe_hw_engine_group.h"
> > #include "xe_hw_fence.h"
> > #include "xe_irq.h"
> > @@ -827,9 +828,9 @@ void xe_hw_engine_handle_irq(struct xe_hw_engine *hwe, u16 intr_vec)
> > }
> >
> > /**
> > - * xe_hw_engine_snapshot_capture - Take a quick snapshot of the HW Engine.
> > + * hw_engine_snapshot_capture - Take a quick snapshot of the HW Engine.
> > * @hwe: Xe HW Engine.
> > - * @q: The exec queue object.
> > + * @q: The exec queue object. (can be NULL for debugfs engine-register dump)
> > *
> > * This can be printed out in a later stage like during dev_coredump
> > * analysis.
> > @@ -837,11 +838,12 @@ void xe_hw_engine_handle_irq(struct xe_hw_engine *hwe, u16 intr_vec)
> > * Returns: a Xe HW Engine snapshot object that must be freed by the
> > * caller, using `xe_hw_engine_snapshot_free`.
> > */
> > -struct xe_hw_engine_snapshot *
> > -xe_hw_engine_snapshot_capture(struct xe_hw_engine *hwe, struct xe_exec_queue *q)
> > +static struct xe_hw_engine_snapshot *
> > +hw_engine_snapshot_capture(struct xe_hw_engine *hwe, struct xe_exec_queue *q)
> > {
> > struct xe_hw_engine_snapshot *snapshot;
> > - struct __guc_capture_parsed_output *node;
> > + struct xe_guc_capture_snapshot *manual_node;
> > + struct xe_guc *guc;
> >
> > if (!xe_hw_engine_is_valid(hwe))
> > return NULL;
> > @@ -865,25 +867,66 @@ xe_hw_engine_snapshot_capture(struct xe_hw_engine *hwe, struct xe_exec_queue *q)
> > return snapshot;
> >
> > if (q) {
> > - /* If got guc capture, set source to GuC */
> > - node = xe_guc_capture_get_matching_and_lock(q);
> > - if (node) {
> > - struct xe_device *xe = gt_to_xe(hwe->gt);
> > - struct xe_devcoredump *coredump = &xe->devcoredump;
> > -
> > - coredump->snapshot.matched_node = node;
> > - xe_gt_dbg(hwe->gt, "Found and locked GuC-err-capture node");
> > - return snapshot;
> > + guc = &q->gt->uc.guc;
> > + /* First find the pre-kill manual GuC-Err-Capture node for this job */
> > + manual_node = xe_guc_capture_snapshot_get(guc, q,
> > + XE_ENGINE_CAPTURE_SOURCE_MANUAL);
> > +
> > + /* Next, look for the GuC-Firmware reported node for this job */
> > + snapshot->matched_node = xe_guc_capture_snapshot_get(guc, q,
> > + XE_ENGINE_CAPTURE_SOURCE_GUC);
> > + if (!snapshot->matched_node) {
> > + xe_gt_dbg(hwe->gt, "Can't find GUC-Sourced err-capture node");
> > + snapshot->matched_node = manual_node;
> > + } else if (manual_node) {
> > + /* looks like we don't need the manually-captured node, return it */
> > + xe_guc_capture_snapshot_put(guc, manual_node);
> > }
> > }
> >
> > - /* otherwise, do manual capture */
> > - xe_engine_manual_capture(hwe, snapshot);
> > - xe_gt_dbg(hwe->gt, "Proceeding with manual engine snapshot");
> > + if (!snapshot->matched_node) {
> > + guc = &hwe->gt->uc.guc;
> > + /*
> > + * Fallback path - do an immediate jobless manual engine capture.
> > + * This will happen when debugfs is triggered to force an engine dump.
> > + */
> > + snapshot->matched_node = xe_guc_capture_snapshot_store_and_get_manual_hwe(guc, hwe);
> > + xe_gt_dbg(hwe->gt, "Fallback to jobless-manual-err-capture node");
> > + }
> >
> > return snapshot;
> > }
> >
> > +/**
> > + * xe_engine_snapshot_capture_for_queue - Take snapshot of associated engine
> > + * @q: The exec queue object
> > + *
> > + * Take snapshot of associated HW Engine
> > + *
> > + * Returns: None.
> > + */
> > +void
> > +xe_engine_snapshot_capture_for_queue(struct xe_exec_queue *q)
> > +{
> > + struct xe_device *xe = gt_to_xe(q->gt);
> > + struct xe_devcoredump *coredump = &xe->devcoredump;
> > + struct xe_hw_engine *hwe;
> > + enum xe_hw_engine_id id;
> > + u32 adj_logical_mask = q->logical_mask;
> > +
> > + if (IS_SRIOV_VF(xe))
> > + return;
> > +
> > + for_each_hw_engine(hwe, q->gt, id) {
> > + if (hwe->class != q->hwe->class ||
> > + !(BIT(hwe->logical_instance) & adj_logical_mask)) {
> > + coredump->snapshot.hwe[id] = NULL;
> > + continue;
> > + }
> > + coredump->snapshot.hwe[id] = hw_engine_snapshot_capture(hwe, q);
> > + }
> > +}
> > +
> > /**
> > * xe_hw_engine_snapshot_free - Free all allocated objects for a given snapshot.
> > * @snapshot: Xe HW Engine snapshot object.
> > @@ -898,17 +941,41 @@ void xe_hw_engine_snapshot_free(struct xe_hw_engine_snapshot *snapshot)
> > return;
> >
> > gt = snapshot->hwe->gt;
> > - /*
> > - * xe_guc_capture_put_matched_nodes is called here and from
> > - * xe_devcoredump_snapshot_free, to cover the 2 calling paths
> > - * of hw_engines - debugfs and devcoredump free.
> > - */
> > - xe_guc_capture_put_matched_nodes(>->uc.guc);
> > + xe_guc_capture_snapshot_put(>->uc.guc, snapshot->matched_node);
> > + snapshot->matched_node = NULL;
> >
> > kfree(snapshot->name);
> > kfree(snapshot);
> > }
> >
> > +/**
> > + * xe_engine_snapshot_print - Print out a given Xe HW Engine snapshot.
> > + * @snapshot: Xe HW Engine snapshot object.
> > + * @p: drm_printer where it will be printed out.
> > + *
> > + * This function prints out a given Xe HW Engine snapshot object.
> > + */
> > +void xe_engine_snapshot_print(struct xe_hw_engine_snapshot *snapshot, struct drm_printer *p)
> > +{
> > + struct xe_gt *gt;
> > +
> > + if (!snapshot)
> > + return;
> > +
> > + gt = snapshot->hwe->gt;
> > +
> > + drm_printf(p, "%s (physical), logical instance=%d\n",
> > + snapshot->name ? snapshot->name : "",
> > + snapshot->logical_instance);
> > + drm_printf(p, "\tForcewake: domain 0x%x, ref %d\n",
> > + snapshot->forcewake.domain, snapshot->forcewake.ref);
> > + drm_printf(p, "\tReserved: %s\n",
> > + str_yes_no(snapshot->kernel_reserved));
> > + drm_puts(p, "\n");
> > +
> > + xe_guc_capture_snapshot_print(>->uc.guc, snapshot->matched_node, p);
> > +}
> > +
> > /**
> > * xe_hw_engine_print - Xe HW Engine Print.
> > * @hwe: Hardware Engine.
> > @@ -920,7 +987,7 @@ void xe_hw_engine_print(struct xe_hw_engine *hwe, struct drm_printer *p)
> > {
> > struct xe_hw_engine_snapshot *snapshot;
> >
> > - snapshot = xe_hw_engine_snapshot_capture(hwe, NULL);
> > + snapshot = hw_engine_snapshot_capture(hwe, NULL);
> > xe_engine_snapshot_print(snapshot, p);
> > xe_hw_engine_snapshot_free(snapshot);
> > }
> > diff --git a/drivers/gpu/drm/xe/xe_hw_engine.h b/drivers/gpu/drm/xe/xe_hw_engine.h
> > index 6b5f9fa2a594..845153fbc149 100644
> > --- a/drivers/gpu/drm/xe/xe_hw_engine.h
> > +++ b/drivers/gpu/drm/xe/xe_hw_engine.h
> > @@ -55,9 +55,9 @@ void xe_hw_engine_handle_irq(struct xe_hw_engine *hwe, u16 intr_vec);
> > void xe_hw_engine_enable_ring(struct xe_hw_engine *hwe);
> > u32 xe_hw_engine_mask_per_class(struct xe_gt *gt,
> > enum xe_engine_class engine_class);
> > -struct xe_hw_engine_snapshot *
> > -xe_hw_engine_snapshot_capture(struct xe_hw_engine *hwe, struct xe_exec_queue *q);
> > +void xe_engine_snapshot_capture_for_queue(struct xe_exec_queue *q);
> > void xe_hw_engine_snapshot_free(struct xe_hw_engine_snapshot *snapshot);
> > +void xe_engine_snapshot_print(struct xe_hw_engine_snapshot *snapshot, struct drm_printer *p);
> > void xe_hw_engine_print(struct xe_hw_engine *hwe, struct drm_printer *p);
> > void xe_hw_engine_setup_default_lrc_state(struct xe_hw_engine *hwe);
> >
> > diff --git a/drivers/gpu/drm/xe/xe_hw_engine_types.h b/drivers/gpu/drm/xe/xe_hw_engine_types.h
> > index e14bee95e364..ea6b60c819d4 100644
> > --- a/drivers/gpu/drm/xe/xe_hw_engine_types.h
> > +++ b/drivers/gpu/drm/xe/xe_hw_engine_types.h
> > @@ -152,11 +152,7 @@ struct xe_hw_engine {
> > struct xe_hw_engine_group *hw_engine_group;
> > };
> >
> > -enum xe_hw_engine_snapshot_source_id {
> > - XE_ENGINE_CAPTURE_SOURCE_MANUAL,
> > - XE_ENGINE_CAPTURE_SOURCE_GUC
> > -};
> > -
> > +struct xe_guc_capture_snapshot;
> > /**
> > * struct xe_hw_engine_snapshot - Hardware engine snapshot
> > *
> > @@ -180,6 +176,13 @@ struct xe_hw_engine_snapshot {
> > u32 mmio_base;
> > /** @kernel_reserved: Engine reserved, can't be used by userspace */
> > bool kernel_reserved;
> > + /**
> > + * @matched_node: GuC Capture snapshot:
> > + * The matched capture node for the timedout job
> > + * this single-node tracker works because devcoredump will always only
> > + * produce one hw-engine capture per devcoredump event
> > + */
> > + struct xe_guc_capture_snapshot *matched_node;
> > };
> >
> > #endif
>
More information about the dri-devel
mailing list