[RFC 5/5] drm/xe: Support devcoredump capture from dead CT handler
John.C.Harrison at Intel.com
John.C.Harrison at Intel.com
Sat Nov 9 01:59:34 UTC 2024
From: John Harrison <John.C.Harrison at Intel.com>
The dead CT handler was doing a locally rolled fake devcoredump
capture and print. Replace that with support for calling the genuine
devcoredump code to do the capture and print instead.
Signed-off-by: John Harrison <John.C.Harrison at Intel.com>
---
drivers/gpu/drm/xe/xe_devcoredump.c | 8 +++-
drivers/gpu/drm/xe/xe_devcoredump.h | 2 +
drivers/gpu/drm/xe/xe_guc_ct.c | 57 +++++++++++++---------------
drivers/gpu/drm/xe/xe_guc_ct_types.h | 6 +--
4 files changed, 38 insertions(+), 35 deletions(-)
diff --git a/drivers/gpu/drm/xe/xe_devcoredump.c b/drivers/gpu/drm/xe/xe_devcoredump.c
index 146b5cd4fbe6..d3a5cb82bf72 100644
--- a/drivers/gpu/drm/xe/xe_devcoredump.c
+++ b/drivers/gpu/drm/xe/xe_devcoredump.c
@@ -223,7 +223,7 @@ static ssize_t xe_devcoredump_read(char *buffer, loff_t offset,
return byte_copied;
}
-static void xe_devcoredump_free(void *data)
+void xe_devcoredump_free(void *data)
{
struct xe_devcoredump *coredump = data;
@@ -376,6 +376,12 @@ static void devcoredump_snapshot_for_thing(struct xe_devcoredump *coredump,
dma_fence_end_signalling(cookie);
}
+void xe_devcoredump_snapshot_gt(struct xe_devcoredump *coredump, struct xe_gt *gt)
+{
+ coredump->xe = gt_to_xe(gt);
+ devcoredump_snapshot_for_thing(coredump, gt, NULL, false);
+}
+
static void devcoredump_for_thing(struct xe_device *_xe, struct xe_gt *gt,
struct xe_sched_job *job, bool want_work)
{
diff --git a/drivers/gpu/drm/xe/xe_devcoredump.h b/drivers/gpu/drm/xe/xe_devcoredump.h
index 1d2f9bbd4ebe..91337432bf78 100644
--- a/drivers/gpu/drm/xe/xe_devcoredump.h
+++ b/drivers/gpu/drm/xe/xe_devcoredump.h
@@ -31,6 +31,8 @@ static inline int xe_devcoredump_init(struct xe_device *xe)
#endif
void xe_devcoredump_print(struct drm_printer *p, struct xe_devcoredump *coredump);
+void xe_devcoredump_snapshot_gt(struct xe_devcoredump *coredump, struct xe_gt *gt);
+void xe_devcoredump_free(void *data);
void xe_print_blob_ascii85(struct drm_printer *p, const char *prefix,
const void *blob, size_t offset, size_t size);
diff --git a/drivers/gpu/drm/xe/xe_guc_ct.c b/drivers/gpu/drm/xe/xe_guc_ct.c
index 2eeadcdf7bc3..a738dcf90aa9 100644
--- a/drivers/gpu/drm/xe/xe_guc_ct.c
+++ b/drivers/gpu/drm/xe/xe_guc_ct.c
@@ -1800,11 +1800,10 @@ void xe_guc_ct_print(struct xe_guc_ct *ct, struct drm_printer *p, bool want_ctb)
#if IS_ENABLED(CONFIG_DRM_XE_DEBUG)
static void ct_dead_capture(struct xe_guc_ct *ct, struct guc_ctb *ctb, u32 reason_code)
{
- struct xe_guc_log_snapshot *snapshot_log;
- struct xe_guc_ct_snapshot *snapshot_ct;
- struct xe_guc *guc = ct_to_guc(ct);
+ struct xe_devcoredump *coredump;
unsigned long flags;
bool have_capture;
+ struct xe_gt *gt = ct_to_gt(ct);
if (ctb)
ctb->info.broken = true;
@@ -1825,29 +1824,36 @@ static void ct_dead_capture(struct xe_guc_ct *ct, struct guc_ctb *ctb, u32 reaso
if (have_capture)
return;
- snapshot_log = xe_guc_log_snapshot_capture(&guc->log, true);
- snapshot_ct = xe_guc_ct_snapshot_capture((ct));
+ coredump = kzalloc(sizeof(*coredump), GFP_ATOMIC);
+ if (!coredump) {
+ xe_gt_err(gt, "Failed to allocate coredump for dead CT capture!\n");
+ return;
+ }
+
+ xe_devcoredump_snapshot_gt(coredump, gt);
spin_lock_irqsave(&ct->dead.lock, flags);
- if (ct->dead.snapshot_log || ct->dead.snapshot_ct) {
- xe_gt_err(ct_to_gt(ct), "Got unexpected dead CT capture!\n");
- xe_guc_log_snapshot_free(snapshot_log);
- xe_guc_ct_snapshot_free(snapshot_ct);
+ if (ct->dead.coredump) {
+ xe_gt_err(gt, "Got unexpected dead CT capture!\n");
} else {
- ct->dead.snapshot_log = snapshot_log;
- ct->dead.snapshot_ct = snapshot_ct;
+ ct->dead.coredump = coredump;
+ coredump = NULL;
}
spin_unlock_irqrestore(&ct->dead.lock, flags);
- queue_work(system_unbound_wq, &(ct)->dead.worker);
+ queue_work(system_unbound_wq, &ct->dead.worker);
+
+ if (coredump) {
+ xe_devcoredump_free(coredump);
+ kfree(coredump);
+ }
}
static void ct_dead_print(struct xe_dead_ct *dead)
{
struct xe_guc_ct *ct = container_of(dead, struct xe_guc_ct, dead);
- struct xe_device *xe = ct_to_xe(ct);
struct xe_gt *gt = ct_to_gt(ct);
static int g_count;
struct drm_printer ip = xe_gt_info_printer(gt);
@@ -1860,18 +1866,8 @@ static void ct_dead_print(struct xe_dead_ct *dead)
drm_printf(&lp, "CTB is dead - reason=0x%X\n", dead->reason);
- /* Can't generate a genuine core dump at this point, so just do the good bits */
- drm_puts(&lp, "**** Xe Device Coredump ****\n");
- xe_device_snapshot_print(xe, &lp);
-
- drm_printf(&lp, "**** GT #%d ****\n", gt->info.id);
- drm_printf(&lp, "\tTile: %d\n", gt->tile->id);
-
- drm_puts(&lp, "**** GuC Log ****\n");
- xe_guc_log_snapshot_print(dead->snapshot_log, &lp);
-
- drm_puts(&lp, "**** GuC CT ****\n");
- xe_guc_ct_snapshot_print(dead->snapshot_ct, &lp);
+ if (dead->coredump)
+ xe_devcoredump_print(&lp, dead->coredump);
drm_puts(&lp, "Done.\n");
}
@@ -1879,6 +1875,7 @@ static void ct_dead_print(struct xe_dead_ct *dead)
static void ct_dead_worker_func(struct work_struct *w)
{
struct xe_guc_ct *ct = container_of(w, struct xe_guc_ct, dead.worker);
+ struct xe_devcoredump *coredump;
if (!ct->dead.reported) {
ct->dead.reported = true;
@@ -1886,11 +1883,8 @@ static void ct_dead_worker_func(struct work_struct *w)
}
spin_lock_irq(&ct->dead.lock);
-
- xe_guc_log_snapshot_free(ct->dead.snapshot_log);
- ct->dead.snapshot_log = NULL;
- xe_guc_ct_snapshot_free(ct->dead.snapshot_ct);
- ct->dead.snapshot_ct = NULL;
+ coredump = ct->dead.coredump;
+ ct->dead.coredump = NULL;
if (ct->dead.reason & (1 << CT_DEAD_STATE_REARM)) {
/* A reset has occurred so re-arm the error reporting */
@@ -1899,5 +1893,8 @@ static void ct_dead_worker_func(struct work_struct *w)
}
spin_unlock_irq(&ct->dead.lock);
+
+ xe_devcoredump_free(coredump);
+ kfree(coredump);
}
#endif
diff --git a/drivers/gpu/drm/xe/xe_guc_ct_types.h b/drivers/gpu/drm/xe/xe_guc_ct_types.h
index 8e1b9d981d61..7131dd6a8366 100644
--- a/drivers/gpu/drm/xe/xe_guc_ct_types.h
+++ b/drivers/gpu/drm/xe/xe_guc_ct_types.h
@@ -99,10 +99,8 @@ struct xe_dead_ct {
bool reported;
/** @worker: worker thread to get out of interrupt context before dumping */
struct work_struct worker;
- /** snapshot_ct: copy of CT state and CTB content at point of error */
- struct xe_guc_ct_snapshot *snapshot_ct;
- /** snapshot_log: copy of GuC log at point of error */
- struct xe_guc_log_snapshot *snapshot_log;
+ /** coredump: copy of system state at point of error */
+ struct xe_devcoredump *coredump;
};
#endif
--
2.47.0
More information about the Intel-xe
mailing list