Mali: bifrost: disable tiler heap reclaiming temporally

Comes from mail "230423: 15:04: ARM support: Yao: ".

This patch and fw_230426 can resolve the occurrence
of "Unhandled Page fault" and CS_FAULTs exceptions
in the stress test "Dvfs+Gl_retrace+Rockx+emmc".

In another mail, Yao said "before any further suggestion or conclusion from RD,
you could temporally WA the issue by disable tiler heap reclaim".

Signed-off-by: Zhen Chen <chenzhen@rock-chips.com>
Change-Id: If06aa8e207b05f1a277aae9a86d1ff76b9a9f93b
This commit is contained in:
Zhen Chen
2023-04-24 10:48:30 +08:00
committed by Tao Huang
parent c5cdbfc2eb
commit eaf544c6d9

View File

@@ -344,14 +344,6 @@ void kbase_csf_tiler_heap_reclaim_mgr_init(struct kbase_device *kbdev)
reclaim->scan_objects = kbase_csf_tiler_heap_reclaim_scan_objects;
reclaim->seeks = HEAP_SHRINKER_SEEKS;
reclaim->batch = HEAP_SHRINKER_BATCH;
#if !defined(CONFIG_MALI_VECTOR_DUMP)
#if KERNEL_VERSION(6, 0, 0) > LINUX_VERSION_CODE
register_shrinker(reclaim);
#else
register_shrinker(reclaim, "mali-csf-tiler-heap");
#endif
#endif
}
void kbase_csf_tiler_heap_reclaim_mgr_term(struct kbase_device *kbdev)
@@ -359,10 +351,6 @@ void kbase_csf_tiler_heap_reclaim_mgr_term(struct kbase_device *kbdev)
struct kbase_csf_scheduler *scheduler = &kbdev->csf.scheduler;
u8 prio;
#if !defined(CONFIG_MALI_VECTOR_DUMP)
unregister_shrinker(&scheduler->reclaim_mgr.heap_reclaim);
#endif
for (prio = KBASE_QUEUE_GROUP_PRIORITY_REALTIME; prio < KBASE_QUEUE_GROUP_PRIORITY_COUNT;
prio++)
WARN_ON(!list_empty(&scheduler->reclaim_mgr.ctx_lists[prio]));