On Fri, Jul 29, 2022 at 09:03:55AM +0200, Mauro Carvalho Chehab wrote:
Add a description for the TLB cache invalidation algorithm and for
the related kAPI functions.
Signed-off-by: Mauro Carvalho Chehab <mchehab@xxxxxxxxxx>
---
To avoid mailbombing on a large number of people, only mailing lists were C/C on the cover.
See [PATCH v2 0/2] at: https://lore.kernel.org/all/cover.1659077372.git.mchehab@xxxxxxxxxx/
Documentation/gpu/i915.rst | 7 ++
drivers/gpu/drm/i915/gt/intel_tlb.c | 25 +++++++
drivers/gpu/drm/i915/gt/intel_tlb.h | 101 ++++++++++++++++++++++++++++
3 files changed, 133 insertions(+)
diff --git a/Documentation/gpu/i915.rst b/Documentation/gpu/i915.rst
index 4e59db1cfb00..46911fdd79e8 100644
--- a/Documentation/gpu/i915.rst
+++ b/Documentation/gpu/i915.rst
@@ -58,6 +58,13 @@ Intel GVT-g Host Support(vGPU device model)
.. kernel-doc:: drivers/gpu/drm/i915/intel_gvt.c
:internal:
+TLB cache invalidation
+----------------------
+
+.. kernel-doc:: drivers/gpu/drm/i915/gt/intel_tlb.h
+
+.. kernel-doc:: drivers/gpu/drm/i915/gt/intel_tlb.c
+
Workarounds
-----------
diff --git a/drivers/gpu/drm/i915/gt/intel_tlb.c b/drivers/gpu/drm/i915/gt/intel_tlb.c
index af8cae979489..4873b7ecc015 100644
--- a/drivers/gpu/drm/i915/gt/intel_tlb.c
+++ b/drivers/gpu/drm/i915/gt/intel_tlb.c
@@ -145,6 +145,18 @@ static void mmio_invalidate_full(struct intel_gt *gt)
intel_uncore_forcewake_put_delayed(uncore, FORCEWAKE_ALL);
}
+/**
+ * intel_gt_invalidate_tlb_full - do full TLB cache invalidation
+ * @gt: GT structure
+ * @seqno: sequence number
+ *
+ * Do a full TLB cache invalidation if the @seqno is bigger than the last
+ * full TLB cache invalidation.
+ *
+ * Note:
+ * The TLB cache invalidation logic depends on GEN-specific registers.
+ * It currently supports MMIO-based TLB flush for GEN8 to GEN12.
+ */
void intel_gt_invalidate_tlb_full(struct intel_gt *gt, u32 seqno)
{
intel_wakeref_t wakeref;
@@ -171,12 +183,25 @@ void intel_gt_invalidate_tlb_full(struct intel_gt *gt, u32 seqno)
}
}
+/**
+ * intel_gt_init_tlb - initialize TLB-specific vars
+ * @gt: GT structure
+ *
+ * TLB cache invalidation logic internally uses some resources that require
+ * initialization. Should be called before doing any TLB cache invalidation.
+ */
void intel_gt_init_tlb(struct intel_gt *gt)
{
mutex_init(>->tlb.invalidate_lock);
seqcount_mutex_init(>->tlb.seqno, >->tlb.invalidate_lock);
}
+/**
+ * intel_gt_fini_tlb - initialize TLB-specific vars
Free TLB-specific vars
+ * @gt: GT structure
+ *
+ * Frees any resources needed by TLB cache invalidation logic.
+ */
void intel_gt_fini_tlb(struct intel_gt *gt)
{
mutex_destroy(>->tlb.invalidate_lock);
diff --git a/drivers/gpu/drm/i915/gt/intel_tlb.h b/drivers/gpu/drm/i915/gt/intel_tlb.h
index 46ce25bf5afe..dca70c33bd61 100644
--- a/drivers/gpu/drm/i915/gt/intel_tlb.h
+++ b/drivers/gpu/drm/i915/gt/intel_tlb.h
@@ -11,16 +11,117 @@
#include "intel_gt_types.h"
+/**
+ * DOC: TLB cache invalidation logic
+ *
+ * The way the current algorithm works is that a struct drm_i915_gem_object can
+ * be created on any order. At unbind/evict time, the object is warranted that
+ * it won't be used anymore. So, a sequence number provided by
+ * intel_gt_next_invalidate_tlb_full() is stored on it. This can happen either
+ * at __vma_put_pages() - for VMA sync unbind, or at ppgtt_unbind_vma() - for
+ * VMA async VMA bind.
+ *
+ * At __i915_gem_object_unset_pages(), intel_gt_invalidate_tlb_full() is called,
+ * where it checks if the sequence number of the object was already invalidated
+ * or not. If not, it flushes the TLB and increments the sequence number::
+ *
+ * void intel_gt_invalidate_tlb_full(struct intel_gt *gt, u32 seqno)
+ * {
+ * ...
+ * with_intel_gt_pm_if_awake(gt, wakeref) {
+ * mutex_lock(>->tlb.invalidate_lock);
+ * if (tlb_seqno_passed(gt, seqno))
+ * goto unlock;
+ *
+ * // Some code to do TLB invalidation
+ * ...
+ *
+ * write_seqcount_invalidate(>->tlb.seqno); // increment seqno
+ * mutex_lock(>->tlb.invalidate_lock);
+ * }
+ *
+ * So, let's say the current seqno is 2 and 3 new objects were created,
+ * on this order::
+ *
+ * obj1
+ * obj2
+ * obj3
+ *
+ * They can be unbind/evict on a different order. At unbind/evict time,
+ * the mm.tlb will be stamped with the sequence number, using the number
+ * from the last TLB flush, plus 1.
I am trying to get my head around the below function.
void vma_invalidate_tlb(struct i915_address_space *vm, u32 tlb)
{
WRITE_ONCE(tlb, intel_gt_next_invalidate_tlb_full(vm->gt));
}
Though we pass obj->mm.tlb for 'tlb' while calling this function,
aren't we writing to local 'tlb' variable here instead of obj->mm.tlb?
+ *
+ * Different threads may be used on unbind/evict and/or unset pages.
+ * As the logic at void intel_gt_invalidate_tlb_full() is protected by a mutex,
May be we can skip 'void' and just keep function name here.
+ * for simplicity, let's consider just two threads:
+ *
+ * +-------------------+-------------------------+---------------------------------+
+ * | sequence number | Thread 0 | Thread 1 +
+ * +===================+=========================+=================================+
+ * | seqno=2 | | |
+ * | +-------------------------+---------------------------------+
+ * | | unbind/evict obj3. | |
+ * | | | |
+ * | | obj3.mm.tlb = seqno | 1 | |
+ * | | // obj3.mm.tlb = 3 | |
+ * | +-------------------------+---------------------------------+
+ * | | unbind/evict obj1. | |
+ * | | | |
+ * | | obj1.mm.tlb = seqno | 1 | |
+ * | | // obj1.mm.tlb = 3 | |
+ * | +-------------------------+---------------------------------+
+ * | | | __i915_gem_object_unset_pages() |
+ * | | | called for obj3 => TLB flush |
+ * | | | invalidating both obj1 and obj2.|
+ * | | | |
+ * | | | seqno += 2 |
+ * +-------------------+-------------------------+---------------------------------+
+ * | seqno=4 | | |
+ * | +-------------------------+---------------------------------+
+ * | | unbind/evict obj2. | |
+ * | | | |
+ * | | obj2.mm.tlb = seqno | 1 | |
+ * | | // obj2.mm.tlb = 5 | |
+ * | +-------------------------+---------------------------------+
+ * | | | __i915_gem_object_unset_pages() |
+ * | | | called for obj1, don't flush |
+ * | | | as past flush invalidated obj1. |
+ * | +-------------------------+---------------------------------+
+ * | | | __i915_gem_object_unset_pages() |
+ * | | | called for obj2 => TLB flush. |
+ * | | | invalidating obj2. |
+ * | | | |
+ * | | | seqno += 2 |
+ * +-------------------+-------------------------+---------------------------------+
+ * | seqno=6 | | |
+ * +-------------------+-------------------------+---------------------------------+
+ */
+
void intel_gt_invalidate_tlb_full(struct intel_gt *gt, u32 seqno);
void intel_gt_init_tlb(struct intel_gt *gt);
void intel_gt_fini_tlb(struct intel_gt *gt);
+/**
+ * intel_gt_tlb_seqno - Returns the current TLB invlidation sequence number
+ *
Probably this empty comment line needs to be removed before the parameter
description below?
+ * @gt: GT structure
+ *
+ * There's no need to lock while calling it, as seqprop_sequence is thread-safe
+ */
static inline u32 intel_gt_tlb_seqno(const struct intel_gt *gt)
{
return seqprop_sequence(>->tlb.seqno);
}
+/**
+ * intel_gt_next_invalidate_tlb_full - Returns the next TLB full invalidation
+ * sequence number
+ *
Same here.
-Niranjana
+ * @gt: GT structure
+ *
+ * There's no need to lock while calling it, as seqprop_sequence is thread-safe
+ */
static inline u32 intel_gt_next_invalidate_tlb_full(const struct intel_gt *gt)
{
return intel_gt_tlb_seqno(gt) | 1;
--
2.36.1
