diff mbox series

[v5] drm/i915/gt: Add selftests for TLB invalidation

Message ID 20230130101436.924402-1-andrzej.hajda@intel.com (mailing list archive)
State New, archived
Headers show
Series [v5] drm/i915/gt: Add selftests for TLB invalidation | expand

Commit Message

Andrzej Hajda Jan. 30, 2023, 10:14 a.m. UTC
From: Chris Wilson <chris@chris-wilson.co.uk>

Check that we invalidate the TLB cache, the updated physical addresses
are immediately visible to the HW, and there is no retention of the old
physical address for concurrent HW access.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[ahajda: adjust to upstream driver, v2+]
Signed-off-by: Andrzej Hajda <andrzej.hajda@intel.com>
---
v2:
    - addressed comments (Tvrtko),
    - changed pin/sample address calculation,
    - removed checks for platforms older than 8,
    - use low ints in MI_DO_COMPARE to be more clear,
    - continue test if physical addresses have the same uppper 32 bits,
    - consolidate two calls to pte_tlbinv into one
v3:
    - skip pages not supported by vm (CI reported EINVAL),
    - fix dw size in MI_CONDITIONAL_BATCH_BUFFER_END for gen8 (CI reported EIO),
    - remove aggressive allocation to get different upper halves of physical
      address (CI reported OOM).
v4:
    - align address in MI_CONDITIONAL_BATCH_BUFFER_END to 8b,
    - set QWORD pointed by addr in above cmd, as required by Gen8/VCS.
v5:
    - set dw size again to 2 (CI reports EIO due to semaphore sanitycheck).

Sorry for spamming, but apparently CI is the only way to test Gen8.
---
 drivers/gpu/drm/i915/gt/intel_gpu_commands.h  |   1 +
 drivers/gpu/drm/i915/gt/intel_gt.c            |   4 +
 drivers/gpu/drm/i915/gt/selftest_tlb.c        | 379 ++++++++++++++++++
 .../drm/i915/selftests/i915_live_selftests.h  |   1 +
 4 files changed, 385 insertions(+)
 create mode 100644 drivers/gpu/drm/i915/gt/selftest_tlb.c

Comments

Tvrtko Ursulin Jan. 30, 2023, 12:35 p.m. UTC | #1
On 30/01/2023 10:14, Andrzej Hajda wrote:
> From: Chris Wilson <chris@chris-wilson.co.uk>
> 
> Check that we invalidate the TLB cache, the updated physical addresses
> are immediately visible to the HW, and there is no retention of the old
> physical address for concurrent HW access.
> 
> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> [ahajda: adjust to upstream driver, v2+]
> Signed-off-by: Andrzej Hajda <andrzej.hajda@intel.com>
> ---
> v2:
>      - addressed comments (Tvrtko),
>      - changed pin/sample address calculation,
>      - removed checks for platforms older than 8,
>      - use low ints in MI_DO_COMPARE to be more clear,
>      - continue test if physical addresses have the same uppper 32 bits,
>      - consolidate two calls to pte_tlbinv into one
> v3:
>      - skip pages not supported by vm (CI reported EINVAL),
>      - fix dw size in MI_CONDITIONAL_BATCH_BUFFER_END for gen8 (CI reported EIO),
>      - remove aggressive allocation to get different upper halves of physical
>        address (CI reported OOM).
> v4:
>      - align address in MI_CONDITIONAL_BATCH_BUFFER_END to 8b,
>      - set QWORD pointed by addr in above cmd, as required by Gen8/VCS.
> v5:
>      - set dw size again to 2 (CI reports EIO due to semaphore sanitycheck).
> 
> Sorry for spamming, but apparently CI is the only way to test Gen8.
> ---
>   drivers/gpu/drm/i915/gt/intel_gpu_commands.h  |   1 +
>   drivers/gpu/drm/i915/gt/intel_gt.c            |   4 +
>   drivers/gpu/drm/i915/gt/selftest_tlb.c        | 379 ++++++++++++++++++
>   .../drm/i915/selftests/i915_live_selftests.h  |   1 +
>   4 files changed, 385 insertions(+)
>   create mode 100644 drivers/gpu/drm/i915/gt/selftest_tlb.c
> 
> diff --git a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
> index 2af1ae3831df98..e10507fa71ce63 100644
> --- a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
> +++ b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
> @@ -394,6 +394,7 @@
>   #define MI_LOAD_URB_MEM         MI_INSTR(0x2C, 0)
>   #define MI_STORE_URB_MEM        MI_INSTR(0x2D, 0)
>   #define MI_CONDITIONAL_BATCH_BUFFER_END MI_INSTR(0x36, 0)
> +#define  MI_DO_COMPARE		REG_BIT(21)
>   
>   #define STATE_BASE_ADDRESS \
>   	((0x3 << 29) | (0x0 << 27) | (0x1 << 24) | (0x1 << 16))
> diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c
> index f0dbfc434e0773..001a7ec5b86182 100644
> --- a/drivers/gpu/drm/i915/gt/intel_gt.c
> +++ b/drivers/gpu/drm/i915/gt/intel_gt.c
> @@ -1205,3 +1205,7 @@ void intel_gt_invalidate_tlb(struct intel_gt *gt, u32 seqno)
>   		mutex_unlock(&gt->tlb.invalidate_lock);
>   	}
>   }
> +
> +#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
> +#include "selftest_tlb.c"
> +#endif
> diff --git a/drivers/gpu/drm/i915/gt/selftest_tlb.c b/drivers/gpu/drm/i915/gt/selftest_tlb.c
> new file mode 100644
> index 00000000000000..166d18a614d51d
> --- /dev/null
> +++ b/drivers/gpu/drm/i915/gt/selftest_tlb.c
> @@ -0,0 +1,379 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright © 2022 Intel Corporation
> + */
> +
> +#include "i915_selftest.h"
> +
> +#include "gem/i915_gem_internal.h"
> +#include "gem/i915_gem_region.h"
> +
> +#include "gen8_engine_cs.h"
> +#include "i915_gem_ww.h"
> +#include "intel_engine_regs.h"
> +#include "intel_gpu_commands.h"
> +#include "intel_context.h"
> +#include "intel_gt.h"
> +#include "intel_ring.h"
> +
> +#include "selftests/igt_flush_test.h"
> +#include "selftests/i915_random.h"
> +
> +static void vma_set_qw(struct i915_vma *vma, u64 addr, u64 val)
> +{
> +	GEM_BUG_ON(addr < i915_vma_offset(vma));
> +	GEM_BUG_ON(addr >= i915_vma_offset(vma) + i915_vma_size(vma) + sizeof(val));
> +	memset64(page_mask_bits(vma->obj->mm.mapping) +
> +		 (addr - i915_vma_offset(vma)), val, 1);
> +}
> +
> +static int
> +pte_tlbinv(struct intel_context *ce,
> +	   struct i915_vma *va,
> +	   struct i915_vma *vb,
> +	   u64 align,
> +	   void (*tlbinv)(struct i915_address_space *vm, u64 addr, u64 length),
> +	   u64 length,
> +	   struct rnd_state *prng)
> +{
> +	struct drm_i915_gem_object *batch;
> +	struct i915_request *rq;
> +	struct i915_vma *vma;
> +	u64 addr;
> +	int err;
> +	u32 *cs;
> +
> +	batch = i915_gem_object_create_internal(ce->vm->i915, 4096);
> +	if (IS_ERR(batch))
> +		return PTR_ERR(batch);
> +
> +	vma = i915_vma_instance(batch, ce->vm, NULL);
> +	if (IS_ERR(vma)) {
> +		err = PTR_ERR(vma);
> +		goto out;
> +	}
> +
> +	err = i915_vma_pin(vma, 0, 0, PIN_USER);
> +	if (err)
> +		goto out;
> +
> +	/* Pin va at random but aligned offset after vma */
> +	addr = round_up(vma->node.start + vma->node.size, align);
> +	/* MI_CONDITIONAL_BATCH_BUFFER_END limits address to 48b */
> +	addr = igt_random_offset(prng, addr, min(ce->vm->total, BIT_ULL(48)),
> +				 va->size, align);
> +	err = i915_vma_pin(va,  0, 0, addr | PIN_OFFSET_FIXED | PIN_USER);
> +	if (err) {
> +		pr_err("Cannot pin at %llx+%llx\n", addr, va->size);
> +		goto out;
> +	}
> +	GEM_BUG_ON(i915_vma_offset(va) != addr);
> +	vb->node = va->node; /* overwrites the _same_ PTE  */
> +
> +	/*
> +	 * Now choose random dword at the 1st pinned page.
> +	 *
> +	 * SZ_64K pages on dg1 require that the whole PT be marked
> +	 * containing 64KiB entries. So we make sure that vma
> +	 * covers the whole PT, despite being randomly aligned to 64KiB
> +	 * and restrict our sampling to the 2MiB PT within where
> +	 * we know that we will be using 64KiB pages.
> +	 */
> +	if (align == SZ_64K)
> +		addr = round_up(addr, SZ_2M);
> +	addr = igt_random_offset(prng, addr, addr + align, 8, 8);
> +
> +	pr_info("%s(%s): Sampling %llx, with alignment %llx, using PTE size %x (phys %x, sg %x), invalidate:%llx+%llx\n",
> +		ce->engine->name, va->obj->mm.region->name ?: "smem",
> +		addr, align, va->resource->page_sizes_gtt, va->page_sizes.phys,
> +		va->page_sizes.sg, addr & -length, length);

Worth skipping this log if va == vb? Or logging something different like 
"... Sanity checking ...".

> +
> +	cs = i915_gem_object_pin_map_unlocked(batch, I915_MAP_WC);
> +	*cs++ = MI_NOOP; /* for later termination */
> +	/*
> +	 * Sample the target to see if we spot an incorrect page.

s/incorrect page/updated backing store/? Or something like that.

> +	 * Gen8 VCS compares immediate value with bitwise-and of two
> +	 * consecutive DWORDS pointed by addr, other gen/engines compare value
> +	 * with DWORD pointed by addr. Moreover we want to exercise DWORD size
> +	 * invalidations. To fulfill all these requirements below values
> +	 * has been chosen.

s/has/have/

> +	 */
> +	*cs++ = MI_CONDITIONAL_BATCH_BUFFER_END | MI_DO_COMPARE | 2;
> +	*cs++ = 0; /* break if *addr == 0 */
> +	*cs++ = lower_32_bits(addr);
> +	*cs++ = upper_32_bits(addr);
> +	vma_set_qw(va, addr, -1);
> +	vma_set_qw(vb, addr, 0);
> +
> +	/* Keep sampling until we get bored */
> +	*cs++ = MI_BATCH_BUFFER_START | BIT(8) | 1;
> +	*cs++ = lower_32_bits(i915_vma_offset(vma));
> +	*cs++ = upper_32_bits(i915_vma_offset(vma));
> +
> +	i915_gem_object_flush_map(batch);
> +
> +	rq = i915_request_create(ce);
> +	if (IS_ERR(rq)) {
> +		err = PTR_ERR(rq);
> +		goto out_va;
> +	}
> +
> +	err = rq->engine->emit_bb_start(rq, i915_vma_offset(vma), 0, 0);
> +	if (err) {
> +		i915_request_add(rq);
> +		goto out_va;
> +	}
> +
> +	i915_request_get(rq);
> +	i915_request_add(rq);
> +
> +	/* Short sleep to sanitycheck the batch is spinning before we begin */
> +	msleep(10);
> +	if (va == vb) {
> +		if (!i915_request_completed(rq)) {
> +			pr_err("Semaphore sanitycheck failed\n");
> +			err = -EIO;
> +		}
> +	} else if (!i915_request_completed(rq)) {
> +		struct i915_vma_resource vb_res = {
> +			.bi.pages = vb->obj->mm.pages,
> +			.bi.page_sizes = vb->obj->mm.page_sizes,
> +			.start = i915_vma_offset(vb),
> +			.vma_size = i915_vma_size(vb)
> +		};
> +		unsigned int pte_flags = 0;
> +
> +		/* Flip the PTE between A and B */
> +		if (i915_gem_object_is_lmem(vb->obj))
> +			pte_flags |= PTE_LM;
> +		ce->vm->insert_entries(ce->vm, &vb_res, 0, pte_flags);
> +
> +		/* Flush the PTE update to concurrent HW */
> +		tlbinv(ce->vm, addr & -length, length);
> +
> +		if (wait_for(i915_request_completed(rq), HZ / 2)) {
> +			pr_err("%s: Request did not complete; the COND_BBE did not read the updated PTE\n",
> +			       ce->engine->name);
> +			err = -EINVAL;
> +		}
> +	} else {
> +		pr_err("Spinner ended unexpectedly\n");
> +		err = -EIO;
> +	}
> +	i915_request_put(rq);
> +
> +	cs = page_mask_bits(batch->mm.mapping);
> +	*cs = MI_BATCH_BUFFER_END;
> +	wmb();
> +
> +out_va:
> +	if (vb != va)
> +		memset(&vb->node, 0, sizeof(vb->node));

This can maybe be a bit fragile. Store a local copy of vb->node and 
restore from that?

> +	i915_vma_unpin(va);
> +	if (i915_vma_unbind_unlocked(va))
> +		err = -EIO;
> +out:
> +	i915_gem_object_put(batch);
> +	return err;
> +}
> +
> +static struct drm_i915_gem_object *create_lmem(struct intel_gt *gt)
> +{
> +	/*
> +	 * Allocation of largest possible page size allows to test all types
> +	 * of pages.
> +	 */
> +	return i915_gem_object_create_lmem(gt->i915, SZ_1G, I915_BO_ALLOC_CONTIGUOUS);
> +}
> +
> +static struct drm_i915_gem_object *create_smem(struct intel_gt *gt)
> +{
> +	/*
> +	 * SZ_64K pages require covering the whole 2M PT (gen8 to tgl/dg1).
> +	 * While that does not require the whole 2M block to be contiguous
> +	 * it is easier to make it so, since we need that for SZ_2M pagees.
> +	 * Since we randomly offset the start of the vma, we need a 4M object
> +	 * so that there is a 2M range within it is suitable for SZ_64K PTE.
> +	 */
> +	return i915_gem_object_create_internal(gt->i915, SZ_4M);
> +}
> +
> +static int
> +mem_tlbinv(struct intel_gt *gt,
> +	   struct drm_i915_gem_object *(*create_fn)(struct intel_gt *),
> +	   void (*tlbinv)(struct i915_address_space *vm, u64 addr, u64 length))
> +{
> +	unsigned int ppgtt_size = RUNTIME_INFO(gt->i915)->ppgtt_size;
> +	struct intel_engine_cs *engine;
> +	struct drm_i915_gem_object *A, *B;
> +	struct i915_ppgtt *ppgtt;
> +	struct i915_vma *va, *vb;
> +	enum intel_engine_id id;
> +	I915_RND_STATE(prng);
> +	void *vaddr;
> +	int err;
> +
> +	/*
> +	 * Check that the TLB invalidate is able to revoke an active
> +	 * page. We load a page into a spinning COND_BBE loop and then
> +	 * remap that page to a new physical address. The old address, and
> +	 * so the loop keeps spinning, is retained in the TLB cache until
> +	 * we issue an invalidate.
> +	 */
> +
> +	A = create_fn(gt);
> +	if (IS_ERR(A))
> +		return PTR_ERR(A);
> +
> +	vaddr = i915_gem_object_pin_map_unlocked(A, I915_MAP_WC);
> +	if (IS_ERR(vaddr)) {
> +		err = PTR_ERR(vaddr);
> +		goto out_a;
> +	}
> +
> +	B = create_fn(gt);
> +	if (IS_ERR(B)) {
> +		err = PTR_ERR(B);
> +		goto out_a;
> +	}
> +
> +	vaddr = i915_gem_object_pin_map_unlocked(B, I915_MAP_WC);
> +	if (IS_ERR(vaddr)) {
> +		err = PTR_ERR(vaddr);
> +		goto out_b;
> +	}
> +
> +	GEM_BUG_ON(A->base.size != B->base.size);
> +	if ((A->mm.page_sizes.phys | B->mm.page_sizes.phys) & (A->base.size - 1))
> +		pr_warn("Failed to allocate contiguous pages for size %zx\n",
> +			A->base.size);
> +
> +	ppgtt = i915_ppgtt_create(gt, 0);
> +	if (IS_ERR(ppgtt)) {
> +		err = PTR_ERR(ppgtt);
> +		goto out_b;
> +	}
> +
> +	va = i915_vma_instance(A, &ppgtt->vm, NULL);
> +	if (IS_ERR(va)) {
> +		err = PTR_ERR(va);
> +		goto out_vm;
> +	}
> +
> +	vb = i915_vma_instance(B, &ppgtt->vm, NULL);
> +	if (IS_ERR(vb)) {
> +		err = PTR_ERR(vb);
> +		goto out_vm;
> +	}
> +
> +	err = 0;
> +	for_each_engine(engine, gt, id) {
> +		struct i915_gem_ww_ctx ww;
> +		struct intel_context *ce;
> +		int bit;
> +
> +		ce = intel_context_create(engine);
> +		if (IS_ERR(ce)) {
> +			err = PTR_ERR(ce);
> +			break;
> +		}
> +
> +		i915_vm_put(ce->vm);
> +		ce->vm = i915_vm_get(&ppgtt->vm);
> +
> +		for_i915_gem_ww(&ww, err, true)
> +			err = intel_context_pin_ww(ce, &ww);
> +		if (err)
> +			goto err_put;
> +
> +		for_each_set_bit(bit,
> +				 (unsigned long *)&RUNTIME_INFO(gt->i915)->page_sizes,
> +				 BITS_PER_TYPE(RUNTIME_INFO(gt->i915)->page_sizes)) {
> +			unsigned int len;
> +
> +			if (BIT_ULL(bit) < i915_vm_obj_min_alignment(va->vm, va->obj))
> +				continue;
> +
> +			/* sanitycheck the semaphore wake up */
> +			err = pte_tlbinv(ce, va, va,
> +					 BIT_ULL(bit),
> +					 NULL, SZ_4K,
> +					 &prng);
> +			if (err)
> +				goto err_unpin;
> + 
> +			for (len = 2; len <= ppgtt_size; len = min(2 * len, ppgtt_size)) {
> +				err = pte_tlbinv(ce, va, vb,
> +						BIT_ULL(bit),
> +						tlbinv,
> +						BIT_ULL(len),
> +						&prng);
> +				if (err)
> +					goto err_unpin;
> +				if (len == ppgtt_size)
> +					break;
> +			}
> +		}
> +err_unpin:
> +		intel_context_unpin(ce);
> +err_put:
> +		intel_context_put(ce);
> +		if (err)
> +			break;
> +	}
> +
> +	if (igt_flush_test(gt->i915))
> +		err = -EIO;
> +
> +out_vm:
> +	i915_vm_put(&ppgtt->vm);
> +out_b:
> +	i915_gem_object_put(B);
> +out_a:
> +	i915_gem_object_put(A);
> +	return err;
> +}
> +
> +static void tlbinv_full(struct i915_address_space *vm, u64 addr, u64 length)
> +{
> +	intel_gt_invalidate_tlb(vm->gt, intel_gt_tlb_seqno(vm->gt) | 1);
> +}
> +
> +static int invalidate_full(void *arg)
> +{
> +	struct intel_gt *gt = arg;
> +	int err;
> +
> +	if (GRAPHICS_VER(gt->i915) < 8)
> +		return 0; /* TLB invalidate not implemented */
> +
> +	err = mem_tlbinv(gt, create_smem, tlbinv_full);
> +	if (err == 0)
> +		err = mem_tlbinv(gt, create_lmem, tlbinv_full);
> +	if (err == -ENODEV || err == -ENXIO)
> +		err = 0;
> +
> +	return err;
> +}
> +
> +int intel_tlb_live_selftests(struct drm_i915_private *i915)
> +{
> +	static const struct i915_subtest tests[] = {
> +		SUBTEST(invalidate_full),
> +	};
> +	struct intel_gt *gt;
> +	unsigned int i;
> +
> +	for_each_gt(gt, i915, i) {
> +		int err;
> +
> +		if (intel_gt_is_wedged(gt))
> +			continue;
> +
> +		err = intel_gt_live_subtests(tests, gt);
> +		if (err)
> +			return err;
> +	}
> +
> +	return 0;
> +}
> diff --git a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
> index aaf8a380e5c789..5aee6c9a8295ce 100644
> --- a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
> +++ b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
> @@ -25,6 +25,7 @@ selftest(gt_lrc, intel_lrc_live_selftests)
>   selftest(gt_mocs, intel_mocs_live_selftests)
>   selftest(gt_pm, intel_gt_pm_live_selftests)
>   selftest(gt_heartbeat, intel_heartbeat_live_selftests)
> +selftest(gt_tlb, intel_tlb_live_selftests)
>   selftest(requests, i915_request_live_selftests)
>   selftest(migrate, intel_migrate_live_selftests)
>   selftest(active, i915_active_live_selftests)

Okay I can follow it and it looks plausible to me. No obvious errors 
stick out and if CI says it's solid I'm happy.

Have you tried to do a run or wo with TLB invalidation nerfed to see how 
the detection rate is looking?

Regards,

Tvrtko
Andrzej Hajda Jan. 30, 2023, 4:53 p.m. UTC | #2
On 30.01.2023 13:35, Tvrtko Ursulin wrote:
> 
> On 30/01/2023 10:14, Andrzej Hajda wrote:
>> From: Chris Wilson <chris@chris-wilson.co.uk>
>>
>> Check that we invalidate the TLB cache, the updated physical addresses
>> are immediately visible to the HW, and there is no retention of the old
>> physical address for concurrent HW access.
>>
>> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
>> [ahajda: adjust to upstream driver, v2+]
>> Signed-off-by: Andrzej Hajda <andrzej.hajda@intel.com>
>> ---
>> v2:
>>      - addressed comments (Tvrtko),
>>      - changed pin/sample address calculation,
>>      - removed checks for platforms older than 8,
>>      - use low ints in MI_DO_COMPARE to be more clear,
>>      - continue test if physical addresses have the same uppper 32 bits,
>>      - consolidate two calls to pte_tlbinv into one
>> v3:
>>      - skip pages not supported by vm (CI reported EINVAL),
>>      - fix dw size in MI_CONDITIONAL_BATCH_BUFFER_END for gen8 (CI 
>> reported EIO),
>>      - remove aggressive allocation to get different upper halves of 
>> physical
>>        address (CI reported OOM).
>> v4:
>>      - align address in MI_CONDITIONAL_BATCH_BUFFER_END to 8b,
>>      - set QWORD pointed by addr in above cmd, as required by Gen8/VCS.
>> v5:
>>      - set dw size again to 2 (CI reports EIO due to semaphore 
>> sanitycheck).
>>
>> Sorry for spamming, but apparently CI is the only way to test Gen8.
>> ---
>>   drivers/gpu/drm/i915/gt/intel_gpu_commands.h  |   1 +
>>   drivers/gpu/drm/i915/gt/intel_gt.c            |   4 +
>>   drivers/gpu/drm/i915/gt/selftest_tlb.c        | 379 ++++++++++++++++++
>>   .../drm/i915/selftests/i915_live_selftests.h  |   1 +
>>   4 files changed, 385 insertions(+)
>>   create mode 100644 drivers/gpu/drm/i915/gt/selftest_tlb.c
>>
>> diff --git a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h 
>> b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
>> index 2af1ae3831df98..e10507fa71ce63 100644
>> --- a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
>> +++ b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
>> @@ -394,6 +394,7 @@
>>   #define MI_LOAD_URB_MEM         MI_INSTR(0x2C, 0)
>>   #define MI_STORE_URB_MEM        MI_INSTR(0x2D, 0)
>>   #define MI_CONDITIONAL_BATCH_BUFFER_END MI_INSTR(0x36, 0)
>> +#define  MI_DO_COMPARE        REG_BIT(21)
>>   #define STATE_BASE_ADDRESS \
>>       ((0x3 << 29) | (0x0 << 27) | (0x1 << 24) | (0x1 << 16))
>> diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c 
>> b/drivers/gpu/drm/i915/gt/intel_gt.c
>> index f0dbfc434e0773..001a7ec5b86182 100644
>> --- a/drivers/gpu/drm/i915/gt/intel_gt.c
>> +++ b/drivers/gpu/drm/i915/gt/intel_gt.c
>> @@ -1205,3 +1205,7 @@ void intel_gt_invalidate_tlb(struct intel_gt 
>> *gt, u32 seqno)
>>           mutex_unlock(&gt->tlb.invalidate_lock);
>>       }
>>   }
>> +
>> +#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
>> +#include "selftest_tlb.c"
>> +#endif
>> diff --git a/drivers/gpu/drm/i915/gt/selftest_tlb.c 
>> b/drivers/gpu/drm/i915/gt/selftest_tlb.c
>> new file mode 100644
>> index 00000000000000..166d18a614d51d
>> --- /dev/null
>> +++ b/drivers/gpu/drm/i915/gt/selftest_tlb.c
>> @@ -0,0 +1,379 @@
>> +// SPDX-License-Identifier: MIT
>> +/*
>> + * Copyright © 2022 Intel Corporation
>> + */
>> +
>> +#include "i915_selftest.h"
>> +
>> +#include "gem/i915_gem_internal.h"
>> +#include "gem/i915_gem_region.h"
>> +
>> +#include "gen8_engine_cs.h"
>> +#include "i915_gem_ww.h"
>> +#include "intel_engine_regs.h"
>> +#include "intel_gpu_commands.h"
>> +#include "intel_context.h"
>> +#include "intel_gt.h"
>> +#include "intel_ring.h"
>> +
>> +#include "selftests/igt_flush_test.h"
>> +#include "selftests/i915_random.h"
>> +
>> +static void vma_set_qw(struct i915_vma *vma, u64 addr, u64 val)
>> +{
>> +    GEM_BUG_ON(addr < i915_vma_offset(vma));
>> +    GEM_BUG_ON(addr >= i915_vma_offset(vma) + i915_vma_size(vma) + 
>> sizeof(val));
>> +    memset64(page_mask_bits(vma->obj->mm.mapping) +
>> +         (addr - i915_vma_offset(vma)), val, 1);
>> +}
>> +
>> +static int
>> +pte_tlbinv(struct intel_context *ce,
>> +       struct i915_vma *va,
>> +       struct i915_vma *vb,
>> +       u64 align,
>> +       void (*tlbinv)(struct i915_address_space *vm, u64 addr, u64 
>> length),
>> +       u64 length,
>> +       struct rnd_state *prng)
>> +{
>> +    struct drm_i915_gem_object *batch;
>> +    struct i915_request *rq;
>> +    struct i915_vma *vma;
>> +    u64 addr;
>> +    int err;
>> +    u32 *cs;
>> +
>> +    batch = i915_gem_object_create_internal(ce->vm->i915, 4096);
>> +    if (IS_ERR(batch))
>> +        return PTR_ERR(batch);
>> +
>> +    vma = i915_vma_instance(batch, ce->vm, NULL);
>> +    if (IS_ERR(vma)) {
>> +        err = PTR_ERR(vma);
>> +        goto out;
>> +    }
>> +
>> +    err = i915_vma_pin(vma, 0, 0, PIN_USER);
>> +    if (err)
>> +        goto out;
>> +
>> +    /* Pin va at random but aligned offset after vma */
>> +    addr = round_up(vma->node.start + vma->node.size, align);
>> +    /* MI_CONDITIONAL_BATCH_BUFFER_END limits address to 48b */
>> +    addr = igt_random_offset(prng, addr, min(ce->vm->total, 
>> BIT_ULL(48)),
>> +                 va->size, align);
>> +    err = i915_vma_pin(va,  0, 0, addr | PIN_OFFSET_FIXED | PIN_USER);
>> +    if (err) {
>> +        pr_err("Cannot pin at %llx+%llx\n", addr, va->size);
>> +        goto out;
>> +    }
>> +    GEM_BUG_ON(i915_vma_offset(va) != addr);
>> +    vb->node = va->node; /* overwrites the _same_ PTE  */
>> +
>> +    /*
>> +     * Now choose random dword at the 1st pinned page.
>> +     *
>> +     * SZ_64K pages on dg1 require that the whole PT be marked
>> +     * containing 64KiB entries. So we make sure that vma
>> +     * covers the whole PT, despite being randomly aligned to 64KiB
>> +     * and restrict our sampling to the 2MiB PT within where
>> +     * we know that we will be using 64KiB pages.
>> +     */
>> +    if (align == SZ_64K)
>> +        addr = round_up(addr, SZ_2M);
>> +    addr = igt_random_offset(prng, addr, addr + align, 8, 8);
>> +
>> +    pr_info("%s(%s): Sampling %llx, with alignment %llx, using PTE 
>> size %x (phys %x, sg %x), invalidate:%llx+%llx\n",
>> +        ce->engine->name, va->obj->mm.region->name ?: "smem",
>> +        addr, align, va->resource->page_sizes_gtt, va->page_sizes.phys,
>> +        va->page_sizes.sg, addr & -length, length);
> 
> Worth skipping this log if va == vb? Or logging something different like 
> "... Sanity checking ...".
> 
>> +
>> +    cs = i915_gem_object_pin_map_unlocked(batch, I915_MAP_WC);
>> +    *cs++ = MI_NOOP; /* for later termination */
>> +    /*
>> +     * Sample the target to see if we spot an incorrect page.
> 
> s/incorrect page/updated backing store/? Or something like that.
> 
>> +     * Gen8 VCS compares immediate value with bitwise-and of two
>> +     * consecutive DWORDS pointed by addr, other gen/engines compare 
>> value
>> +     * with DWORD pointed by addr. Moreover we want to exercise DWORD 
>> size
>> +     * invalidations. To fulfill all these requirements below values
>> +     * has been chosen.
> 
> s/has/have/
> 
>> +     */
>> +    *cs++ = MI_CONDITIONAL_BATCH_BUFFER_END | MI_DO_COMPARE | 2;
>> +    *cs++ = 0; /* break if *addr == 0 */
>> +    *cs++ = lower_32_bits(addr);
>> +    *cs++ = upper_32_bits(addr);
>> +    vma_set_qw(va, addr, -1);
>> +    vma_set_qw(vb, addr, 0);
>> +
>> +    /* Keep sampling until we get bored */
>> +    *cs++ = MI_BATCH_BUFFER_START | BIT(8) | 1;
>> +    *cs++ = lower_32_bits(i915_vma_offset(vma));
>> +    *cs++ = upper_32_bits(i915_vma_offset(vma));
>> +
>> +    i915_gem_object_flush_map(batch);
>> +
>> +    rq = i915_request_create(ce);
>> +    if (IS_ERR(rq)) {
>> +        err = PTR_ERR(rq);
>> +        goto out_va;
>> +    }
>> +
>> +    err = rq->engine->emit_bb_start(rq, i915_vma_offset(vma), 0, 0);
>> +    if (err) {
>> +        i915_request_add(rq);
>> +        goto out_va;
>> +    }
>> +
>> +    i915_request_get(rq);
>> +    i915_request_add(rq);
>> +
>> +    /* Short sleep to sanitycheck the batch is spinning before we 
>> begin */
>> +    msleep(10);
>> +    if (va == vb) {
>> +        if (!i915_request_completed(rq)) {
>> +            pr_err("Semaphore sanitycheck failed\n");
>> +            err = -EIO;
>> +        }
>> +    } else if (!i915_request_completed(rq)) {
>> +        struct i915_vma_resource vb_res = {
>> +            .bi.pages = vb->obj->mm.pages,
>> +            .bi.page_sizes = vb->obj->mm.page_sizes,
>> +            .start = i915_vma_offset(vb),
>> +            .vma_size = i915_vma_size(vb)
>> +        };
>> +        unsigned int pte_flags = 0;
>> +
>> +        /* Flip the PTE between A and B */
>> +        if (i915_gem_object_is_lmem(vb->obj))
>> +            pte_flags |= PTE_LM;
>> +        ce->vm->insert_entries(ce->vm, &vb_res, 0, pte_flags);
>> +
>> +        /* Flush the PTE update to concurrent HW */
>> +        tlbinv(ce->vm, addr & -length, length);
>> +
>> +        if (wait_for(i915_request_completed(rq), HZ / 2)) {
>> +            pr_err("%s: Request did not complete; the COND_BBE did 
>> not read the updated PTE\n",
>> +                   ce->engine->name);
>> +            err = -EINVAL;
>> +        }
>> +    } else {
>> +        pr_err("Spinner ended unexpectedly\n");
>> +        err = -EIO;
>> +    }
>> +    i915_request_put(rq);
>> +
>> +    cs = page_mask_bits(batch->mm.mapping);
>> +    *cs = MI_BATCH_BUFFER_END;
>> +    wmb();
>> +
>> +out_va:
>> +    if (vb != va)
>> +        memset(&vb->node, 0, sizeof(vb->node));
> 
> This can maybe be a bit fragile. Store a local copy of vb->node and 
> restore from that?
> 
>> +    i915_vma_unpin(va);
>> +    if (i915_vma_unbind_unlocked(va))
>> +        err = -EIO;
>> +out:
>> +    i915_gem_object_put(batch);
>> +    return err;
>> +}
>> +
>> +static struct drm_i915_gem_object *create_lmem(struct intel_gt *gt)
>> +{
>> +    /*
>> +     * Allocation of largest possible page size allows to test all types
>> +     * of pages.
>> +     */
>> +    return i915_gem_object_create_lmem(gt->i915, SZ_1G, 
>> I915_BO_ALLOC_CONTIGUOUS);
>> +}
>> +
>> +static struct drm_i915_gem_object *create_smem(struct intel_gt *gt)
>> +{
>> +    /*
>> +     * SZ_64K pages require covering the whole 2M PT (gen8 to tgl/dg1).
>> +     * While that does not require the whole 2M block to be contiguous
>> +     * it is easier to make it so, since we need that for SZ_2M pagees.
>> +     * Since we randomly offset the start of the vma, we need a 4M 
>> object
>> +     * so that there is a 2M range within it is suitable for SZ_64K PTE.
>> +     */
>> +    return i915_gem_object_create_internal(gt->i915, SZ_4M);
>> +}
>> +
>> +static int
>> +mem_tlbinv(struct intel_gt *gt,
>> +       struct drm_i915_gem_object *(*create_fn)(struct intel_gt *),
>> +       void (*tlbinv)(struct i915_address_space *vm, u64 addr, u64 
>> length))
>> +{
>> +    unsigned int ppgtt_size = RUNTIME_INFO(gt->i915)->ppgtt_size;
>> +    struct intel_engine_cs *engine;
>> +    struct drm_i915_gem_object *A, *B;
>> +    struct i915_ppgtt *ppgtt;
>> +    struct i915_vma *va, *vb;
>> +    enum intel_engine_id id;
>> +    I915_RND_STATE(prng);
>> +    void *vaddr;
>> +    int err;
>> +
>> +    /*
>> +     * Check that the TLB invalidate is able to revoke an active
>> +     * page. We load a page into a spinning COND_BBE loop and then
>> +     * remap that page to a new physical address. The old address, and
>> +     * so the loop keeps spinning, is retained in the TLB cache until
>> +     * we issue an invalidate.
>> +     */
>> +
>> +    A = create_fn(gt);
>> +    if (IS_ERR(A))
>> +        return PTR_ERR(A);
>> +
>> +    vaddr = i915_gem_object_pin_map_unlocked(A, I915_MAP_WC);
>> +    if (IS_ERR(vaddr)) {
>> +        err = PTR_ERR(vaddr);
>> +        goto out_a;
>> +    }
>> +
>> +    B = create_fn(gt);
>> +    if (IS_ERR(B)) {
>> +        err = PTR_ERR(B);
>> +        goto out_a;
>> +    }
>> +
>> +    vaddr = i915_gem_object_pin_map_unlocked(B, I915_MAP_WC);
>> +    if (IS_ERR(vaddr)) {
>> +        err = PTR_ERR(vaddr);
>> +        goto out_b;
>> +    }
>> +
>> +    GEM_BUG_ON(A->base.size != B->base.size);
>> +    if ((A->mm.page_sizes.phys | B->mm.page_sizes.phys) & 
>> (A->base.size - 1))
>> +        pr_warn("Failed to allocate contiguous pages for size %zx\n",
>> +            A->base.size);
>> +
>> +    ppgtt = i915_ppgtt_create(gt, 0);
>> +    if (IS_ERR(ppgtt)) {
>> +        err = PTR_ERR(ppgtt);
>> +        goto out_b;
>> +    }
>> +
>> +    va = i915_vma_instance(A, &ppgtt->vm, NULL);
>> +    if (IS_ERR(va)) {
>> +        err = PTR_ERR(va);
>> +        goto out_vm;
>> +    }
>> +
>> +    vb = i915_vma_instance(B, &ppgtt->vm, NULL);
>> +    if (IS_ERR(vb)) {
>> +        err = PTR_ERR(vb);
>> +        goto out_vm;
>> +    }
>> +
>> +    err = 0;
>> +    for_each_engine(engine, gt, id) {
>> +        struct i915_gem_ww_ctx ww;
>> +        struct intel_context *ce;
>> +        int bit;
>> +
>> +        ce = intel_context_create(engine);
>> +        if (IS_ERR(ce)) {
>> +            err = PTR_ERR(ce);
>> +            break;
>> +        }
>> +
>> +        i915_vm_put(ce->vm);
>> +        ce->vm = i915_vm_get(&ppgtt->vm);
>> +
>> +        for_i915_gem_ww(&ww, err, true)
>> +            err = intel_context_pin_ww(ce, &ww);
>> +        if (err)
>> +            goto err_put;
>> +
>> +        for_each_set_bit(bit,
>> +                 (unsigned long *)&RUNTIME_INFO(gt->i915)->page_sizes,
>> +                 BITS_PER_TYPE(RUNTIME_INFO(gt->i915)->page_sizes)) {
>> +            unsigned int len;
>> +
>> +            if (BIT_ULL(bit) < i915_vm_obj_min_alignment(va->vm, 
>> va->obj))
>> +                continue;
>> +
>> +            /* sanitycheck the semaphore wake up */
>> +            err = pte_tlbinv(ce, va, va,
>> +                     BIT_ULL(bit),
>> +                     NULL, SZ_4K,
>> +                     &prng);
>> +            if (err)
>> +                goto err_unpin;
>> + +            for (len = 2; len <= ppgtt_size; len = min(2 * len, 
>> ppgtt_size)) {
>> +                err = pte_tlbinv(ce, va, vb,
>> +                        BIT_ULL(bit),
>> +                        tlbinv,
>> +                        BIT_ULL(len),
>> +                        &prng);
>> +                if (err)
>> +                    goto err_unpin;
>> +                if (len == ppgtt_size)
>> +                    break;
>> +            }
>> +        }
>> +err_unpin:
>> +        intel_context_unpin(ce);
>> +err_put:
>> +        intel_context_put(ce);
>> +        if (err)
>> +            break;
>> +    }
>> +
>> +    if (igt_flush_test(gt->i915))
>> +        err = -EIO;
>> +
>> +out_vm:
>> +    i915_vm_put(&ppgtt->vm);
>> +out_b:
>> +    i915_gem_object_put(B);
>> +out_a:
>> +    i915_gem_object_put(A);
>> +    return err;
>> +}
>> +
>> +static void tlbinv_full(struct i915_address_space *vm, u64 addr, u64 
>> length)
>> +{
>> +    intel_gt_invalidate_tlb(vm->gt, intel_gt_tlb_seqno(vm->gt) | 1);
>> +}
>> +
>> +static int invalidate_full(void *arg)
>> +{
>> +    struct intel_gt *gt = arg;
>> +    int err;
>> +
>> +    if (GRAPHICS_VER(gt->i915) < 8)
>> +        return 0; /* TLB invalidate not implemented */
>> +
>> +    err = mem_tlbinv(gt, create_smem, tlbinv_full);
>> +    if (err == 0)
>> +        err = mem_tlbinv(gt, create_lmem, tlbinv_full);
>> +    if (err == -ENODEV || err == -ENXIO)
>> +        err = 0;
>> +
>> +    return err;
>> +}
>> +
>> +int intel_tlb_live_selftests(struct drm_i915_private *i915)
>> +{
>> +    static const struct i915_subtest tests[] = {
>> +        SUBTEST(invalidate_full),
>> +    };
>> +    struct intel_gt *gt;
>> +    unsigned int i;
>> +
>> +    for_each_gt(gt, i915, i) {
>> +        int err;
>> +
>> +        if (intel_gt_is_wedged(gt))
>> +            continue;
>> +
>> +        err = intel_gt_live_subtests(tests, gt);
>> +        if (err)
>> +            return err;
>> +    }
>> +
>> +    return 0;
>> +}
>> diff --git a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h 
>> b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
>> index aaf8a380e5c789..5aee6c9a8295ce 100644
>> --- a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
>> +++ b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
>> @@ -25,6 +25,7 @@ selftest(gt_lrc, intel_lrc_live_selftests)
>>   selftest(gt_mocs, intel_mocs_live_selftests)
>>   selftest(gt_pm, intel_gt_pm_live_selftests)
>>   selftest(gt_heartbeat, intel_heartbeat_live_selftests)
>> +selftest(gt_tlb, intel_tlb_live_selftests)
>>   selftest(requests, i915_request_live_selftests)
>>   selftest(migrate, intel_migrate_live_selftests)
>>   selftest(active, i915_active_live_selftests)
> 
> Okay I can follow it and it looks plausible to me. No obvious errors 
> stick out and if CI says it's solid I'm happy.

v6 sent with comments addresssed.

> 
> Have you tried to do a run or wo with TLB invalidation nerfed to see how 
> the detection rate is looking?

Yes, nerfing TLB_INV addresses works for every engine.

Regards
Andrzej


> 
> Regards,
> 
> Tvrtko
diff mbox series

Patch

diff --git a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
index 2af1ae3831df98..e10507fa71ce63 100644
--- a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
+++ b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
@@ -394,6 +394,7 @@ 
 #define MI_LOAD_URB_MEM         MI_INSTR(0x2C, 0)
 #define MI_STORE_URB_MEM        MI_INSTR(0x2D, 0)
 #define MI_CONDITIONAL_BATCH_BUFFER_END MI_INSTR(0x36, 0)
+#define  MI_DO_COMPARE		REG_BIT(21)
 
 #define STATE_BASE_ADDRESS \
 	((0x3 << 29) | (0x0 << 27) | (0x1 << 24) | (0x1 << 16))
diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c
index f0dbfc434e0773..001a7ec5b86182 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt.c
+++ b/drivers/gpu/drm/i915/gt/intel_gt.c
@@ -1205,3 +1205,7 @@  void intel_gt_invalidate_tlb(struct intel_gt *gt, u32 seqno)
 		mutex_unlock(&gt->tlb.invalidate_lock);
 	}
 }
+
+#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
+#include "selftest_tlb.c"
+#endif
diff --git a/drivers/gpu/drm/i915/gt/selftest_tlb.c b/drivers/gpu/drm/i915/gt/selftest_tlb.c
new file mode 100644
index 00000000000000..166d18a614d51d
--- /dev/null
+++ b/drivers/gpu/drm/i915/gt/selftest_tlb.c
@@ -0,0 +1,379 @@ 
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright © 2022 Intel Corporation
+ */
+
+#include "i915_selftest.h"
+
+#include "gem/i915_gem_internal.h"
+#include "gem/i915_gem_region.h"
+
+#include "gen8_engine_cs.h"
+#include "i915_gem_ww.h"
+#include "intel_engine_regs.h"
+#include "intel_gpu_commands.h"
+#include "intel_context.h"
+#include "intel_gt.h"
+#include "intel_ring.h"
+
+#include "selftests/igt_flush_test.h"
+#include "selftests/i915_random.h"
+
+static void vma_set_qw(struct i915_vma *vma, u64 addr, u64 val)
+{
+	GEM_BUG_ON(addr < i915_vma_offset(vma));
+	GEM_BUG_ON(addr >= i915_vma_offset(vma) + i915_vma_size(vma) + sizeof(val));
+	memset64(page_mask_bits(vma->obj->mm.mapping) +
+		 (addr - i915_vma_offset(vma)), val, 1);
+}
+
+static int
+pte_tlbinv(struct intel_context *ce,
+	   struct i915_vma *va,
+	   struct i915_vma *vb,
+	   u64 align,
+	   void (*tlbinv)(struct i915_address_space *vm, u64 addr, u64 length),
+	   u64 length,
+	   struct rnd_state *prng)
+{
+	struct drm_i915_gem_object *batch;
+	struct i915_request *rq;
+	struct i915_vma *vma;
+	u64 addr;
+	int err;
+	u32 *cs;
+
+	batch = i915_gem_object_create_internal(ce->vm->i915, 4096);
+	if (IS_ERR(batch))
+		return PTR_ERR(batch);
+
+	vma = i915_vma_instance(batch, ce->vm, NULL);
+	if (IS_ERR(vma)) {
+		err = PTR_ERR(vma);
+		goto out;
+	}
+
+	err = i915_vma_pin(vma, 0, 0, PIN_USER);
+	if (err)
+		goto out;
+
+	/* Pin va at random but aligned offset after vma */
+	addr = round_up(vma->node.start + vma->node.size, align);
+	/* MI_CONDITIONAL_BATCH_BUFFER_END limits address to 48b */
+	addr = igt_random_offset(prng, addr, min(ce->vm->total, BIT_ULL(48)),
+				 va->size, align);
+	err = i915_vma_pin(va,  0, 0, addr | PIN_OFFSET_FIXED | PIN_USER);
+	if (err) {
+		pr_err("Cannot pin at %llx+%llx\n", addr, va->size);
+		goto out;
+	}
+	GEM_BUG_ON(i915_vma_offset(va) != addr);
+	vb->node = va->node; /* overwrites the _same_ PTE  */
+
+	/*
+	 * Now choose random dword at the 1st pinned page.
+	 *
+	 * SZ_64K pages on dg1 require that the whole PT be marked
+	 * containing 64KiB entries. So we make sure that vma
+	 * covers the whole PT, despite being randomly aligned to 64KiB
+	 * and restrict our sampling to the 2MiB PT within where
+	 * we know that we will be using 64KiB pages.
+	 */
+	if (align == SZ_64K)
+		addr = round_up(addr, SZ_2M);
+	addr = igt_random_offset(prng, addr, addr + align, 8, 8);
+
+	pr_info("%s(%s): Sampling %llx, with alignment %llx, using PTE size %x (phys %x, sg %x), invalidate:%llx+%llx\n",
+		ce->engine->name, va->obj->mm.region->name ?: "smem",
+		addr, align, va->resource->page_sizes_gtt, va->page_sizes.phys,
+		va->page_sizes.sg, addr & -length, length);
+
+	cs = i915_gem_object_pin_map_unlocked(batch, I915_MAP_WC);
+	*cs++ = MI_NOOP; /* for later termination */
+	/*
+	 * Sample the target to see if we spot an incorrect page.
+	 * Gen8 VCS compares immediate value with bitwise-and of two
+	 * consecutive DWORDS pointed by addr, other gen/engines compare value
+	 * with DWORD pointed by addr. Moreover we want to exercise DWORD size
+	 * invalidations. To fulfill all these requirements below values
+	 * has been chosen.
+	 */
+	*cs++ = MI_CONDITIONAL_BATCH_BUFFER_END | MI_DO_COMPARE | 2;
+	*cs++ = 0; /* break if *addr == 0 */
+	*cs++ = lower_32_bits(addr);
+	*cs++ = upper_32_bits(addr);
+	vma_set_qw(va, addr, -1);
+	vma_set_qw(vb, addr, 0);
+
+	/* Keep sampling until we get bored */
+	*cs++ = MI_BATCH_BUFFER_START | BIT(8) | 1;
+	*cs++ = lower_32_bits(i915_vma_offset(vma));
+	*cs++ = upper_32_bits(i915_vma_offset(vma));
+
+	i915_gem_object_flush_map(batch);
+
+	rq = i915_request_create(ce);
+	if (IS_ERR(rq)) {
+		err = PTR_ERR(rq);
+		goto out_va;
+	}
+
+	err = rq->engine->emit_bb_start(rq, i915_vma_offset(vma), 0, 0);
+	if (err) {
+		i915_request_add(rq);
+		goto out_va;
+	}
+
+	i915_request_get(rq);
+	i915_request_add(rq);
+
+	/* Short sleep to sanitycheck the batch is spinning before we begin */
+	msleep(10);
+	if (va == vb) {
+		if (!i915_request_completed(rq)) {
+			pr_err("Semaphore sanitycheck failed\n");
+			err = -EIO;
+		}
+	} else if (!i915_request_completed(rq)) {
+		struct i915_vma_resource vb_res = {
+			.bi.pages = vb->obj->mm.pages,
+			.bi.page_sizes = vb->obj->mm.page_sizes,
+			.start = i915_vma_offset(vb),
+			.vma_size = i915_vma_size(vb)
+		};
+		unsigned int pte_flags = 0;
+
+		/* Flip the PTE between A and B */
+		if (i915_gem_object_is_lmem(vb->obj))
+			pte_flags |= PTE_LM;
+		ce->vm->insert_entries(ce->vm, &vb_res, 0, pte_flags);
+
+		/* Flush the PTE update to concurrent HW */
+		tlbinv(ce->vm, addr & -length, length);
+
+		if (wait_for(i915_request_completed(rq), HZ / 2)) {
+			pr_err("%s: Request did not complete; the COND_BBE did not read the updated PTE\n",
+			       ce->engine->name);
+			err = -EINVAL;
+		}
+	} else {
+		pr_err("Spinner ended unexpectedly\n");
+		err = -EIO;
+	}
+	i915_request_put(rq);
+
+	cs = page_mask_bits(batch->mm.mapping);
+	*cs = MI_BATCH_BUFFER_END;
+	wmb();
+
+out_va:
+	if (vb != va)
+		memset(&vb->node, 0, sizeof(vb->node));
+	i915_vma_unpin(va);
+	if (i915_vma_unbind_unlocked(va))
+		err = -EIO;
+out:
+	i915_gem_object_put(batch);
+	return err;
+}
+
+static struct drm_i915_gem_object *create_lmem(struct intel_gt *gt)
+{
+	/*
+	 * Allocation of largest possible page size allows to test all types
+	 * of pages.
+	 */
+	return i915_gem_object_create_lmem(gt->i915, SZ_1G, I915_BO_ALLOC_CONTIGUOUS);
+}
+
+static struct drm_i915_gem_object *create_smem(struct intel_gt *gt)
+{
+	/*
+	 * SZ_64K pages require covering the whole 2M PT (gen8 to tgl/dg1).
+	 * While that does not require the whole 2M block to be contiguous
+	 * it is easier to make it so, since we need that for SZ_2M pagees.
+	 * Since we randomly offset the start of the vma, we need a 4M object
+	 * so that there is a 2M range within it is suitable for SZ_64K PTE.
+	 */
+	return i915_gem_object_create_internal(gt->i915, SZ_4M);
+}
+
+static int
+mem_tlbinv(struct intel_gt *gt,
+	   struct drm_i915_gem_object *(*create_fn)(struct intel_gt *),
+	   void (*tlbinv)(struct i915_address_space *vm, u64 addr, u64 length))
+{
+	unsigned int ppgtt_size = RUNTIME_INFO(gt->i915)->ppgtt_size;
+	struct intel_engine_cs *engine;
+	struct drm_i915_gem_object *A, *B;
+	struct i915_ppgtt *ppgtt;
+	struct i915_vma *va, *vb;
+	enum intel_engine_id id;
+	I915_RND_STATE(prng);
+	void *vaddr;
+	int err;
+
+	/*
+	 * Check that the TLB invalidate is able to revoke an active
+	 * page. We load a page into a spinning COND_BBE loop and then
+	 * remap that page to a new physical address. The old address, and
+	 * so the loop keeps spinning, is retained in the TLB cache until
+	 * we issue an invalidate.
+	 */
+
+	A = create_fn(gt);
+	if (IS_ERR(A))
+		return PTR_ERR(A);
+
+	vaddr = i915_gem_object_pin_map_unlocked(A, I915_MAP_WC);
+	if (IS_ERR(vaddr)) {
+		err = PTR_ERR(vaddr);
+		goto out_a;
+	}
+
+	B = create_fn(gt);
+	if (IS_ERR(B)) {
+		err = PTR_ERR(B);
+		goto out_a;
+	}
+
+	vaddr = i915_gem_object_pin_map_unlocked(B, I915_MAP_WC);
+	if (IS_ERR(vaddr)) {
+		err = PTR_ERR(vaddr);
+		goto out_b;
+	}
+
+	GEM_BUG_ON(A->base.size != B->base.size);
+	if ((A->mm.page_sizes.phys | B->mm.page_sizes.phys) & (A->base.size - 1))
+		pr_warn("Failed to allocate contiguous pages for size %zx\n",
+			A->base.size);
+
+	ppgtt = i915_ppgtt_create(gt, 0);
+	if (IS_ERR(ppgtt)) {
+		err = PTR_ERR(ppgtt);
+		goto out_b;
+	}
+
+	va = i915_vma_instance(A, &ppgtt->vm, NULL);
+	if (IS_ERR(va)) {
+		err = PTR_ERR(va);
+		goto out_vm;
+	}
+
+	vb = i915_vma_instance(B, &ppgtt->vm, NULL);
+	if (IS_ERR(vb)) {
+		err = PTR_ERR(vb);
+		goto out_vm;
+	}
+
+	err = 0;
+	for_each_engine(engine, gt, id) {
+		struct i915_gem_ww_ctx ww;
+		struct intel_context *ce;
+		int bit;
+
+		ce = intel_context_create(engine);
+		if (IS_ERR(ce)) {
+			err = PTR_ERR(ce);
+			break;
+		}
+
+		i915_vm_put(ce->vm);
+		ce->vm = i915_vm_get(&ppgtt->vm);
+
+		for_i915_gem_ww(&ww, err, true)
+			err = intel_context_pin_ww(ce, &ww);
+		if (err)
+			goto err_put;
+
+		for_each_set_bit(bit,
+				 (unsigned long *)&RUNTIME_INFO(gt->i915)->page_sizes,
+				 BITS_PER_TYPE(RUNTIME_INFO(gt->i915)->page_sizes)) {
+			unsigned int len;
+
+			if (BIT_ULL(bit) < i915_vm_obj_min_alignment(va->vm, va->obj))
+				continue;
+
+			/* sanitycheck the semaphore wake up */
+			err = pte_tlbinv(ce, va, va,
+					 BIT_ULL(bit),
+					 NULL, SZ_4K,
+					 &prng);
+			if (err)
+				goto err_unpin;
+
+			for (len = 2; len <= ppgtt_size; len = min(2 * len, ppgtt_size)) {
+				err = pte_tlbinv(ce, va, vb,
+						BIT_ULL(bit),
+						tlbinv,
+						BIT_ULL(len),
+						&prng);
+				if (err)
+					goto err_unpin;
+				if (len == ppgtt_size)
+					break;
+			}
+		}
+err_unpin:
+		intel_context_unpin(ce);
+err_put:
+		intel_context_put(ce);
+		if (err)
+			break;
+	}
+
+	if (igt_flush_test(gt->i915))
+		err = -EIO;
+
+out_vm:
+	i915_vm_put(&ppgtt->vm);
+out_b:
+	i915_gem_object_put(B);
+out_a:
+	i915_gem_object_put(A);
+	return err;
+}
+
+static void tlbinv_full(struct i915_address_space *vm, u64 addr, u64 length)
+{
+	intel_gt_invalidate_tlb(vm->gt, intel_gt_tlb_seqno(vm->gt) | 1);
+}
+
+static int invalidate_full(void *arg)
+{
+	struct intel_gt *gt = arg;
+	int err;
+
+	if (GRAPHICS_VER(gt->i915) < 8)
+		return 0; /* TLB invalidate not implemented */
+
+	err = mem_tlbinv(gt, create_smem, tlbinv_full);
+	if (err == 0)
+		err = mem_tlbinv(gt, create_lmem, tlbinv_full);
+	if (err == -ENODEV || err == -ENXIO)
+		err = 0;
+
+	return err;
+}
+
+int intel_tlb_live_selftests(struct drm_i915_private *i915)
+{
+	static const struct i915_subtest tests[] = {
+		SUBTEST(invalidate_full),
+	};
+	struct intel_gt *gt;
+	unsigned int i;
+
+	for_each_gt(gt, i915, i) {
+		int err;
+
+		if (intel_gt_is_wedged(gt))
+			continue;
+
+		err = intel_gt_live_subtests(tests, gt);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
diff --git a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
index aaf8a380e5c789..5aee6c9a8295ce 100644
--- a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
+++ b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
@@ -25,6 +25,7 @@  selftest(gt_lrc, intel_lrc_live_selftests)
 selftest(gt_mocs, intel_mocs_live_selftests)
 selftest(gt_pm, intel_gt_pm_live_selftests)
 selftest(gt_heartbeat, intel_heartbeat_live_selftests)
+selftest(gt_tlb, intel_tlb_live_selftests)
 selftest(requests, i915_request_live_selftests)
 selftest(migrate, intel_migrate_live_selftests)
 selftest(active, i915_active_live_selftests)