The Linux Kernel Mailing List
 help / color / mirror / Atom feed
From: Jason Gunthorpe <jgg@ziepe.ca>
To: Rik van Riel <riel@surriel.com>
Cc: linux-kernel@vger.kernel.org, robin.murphy@arm.com,
	joro@8bytes.org, will@kernel.org, iommu@lists.linux.dev,
	kyle@mcmartin.ca, kernel-team@meta.com,
	Rik van Riel <riel@meta.com>
Subject: Re: [PATCH 5/5] iova: add KUnit test suite
Date: Fri, 15 May 2026 19:43:17 -0300	[thread overview]
Message-ID: <20260515224317.GM7702@ziepe.ca> (raw)
In-Reply-To: <20260513020304.1528751-6-riel@surriel.com>

On Tue, May 12, 2026 at 10:00:22PM -0400, Rik van Riel wrote:
> From: Rik van Riel <riel@meta.com>
> 
> Add a kunit suite for the augmented-rbtree IOVA allocator, plus an
> iova_domain_verify_invariants() helper (compiled only when the test
> config is enabled) that walks the tree and confirms every node's
> gap_to_prev, clamped_gap32, __subtree_max_gap, and __subtree_max_gap32
> match what recomputation from scratch yields.
> 
> Test cases:
>   - test_init_destroy: domain lifecycle, no leaks.
>   - test_basic_alloc_free: single alloc/free roundtrip, top-down reuse.
>   - test_size_aligned: alignment of size_aligned allocs across orders 0..7.
>   - test_top_down_preference: sequential allocs decrease in pfn_lo.
>   - test_limit_pfn_respected: 100 allocs all stay <= limit_pfn.
>   - test_reserve_iova: allocs avoid the reserved range.
>   - test_find_iova: lookup by pfn returns the right iova.
>   - test_32bit_in_64bit_domain: 1000 64-bit allocs followed by a 32-bit
>     alloc must still find a slot below DMA_BIT_MASK(32) -- exercises
>     the __subtree_max_gap32 augmentation.
>   - test_two_phase_alignment: pack size-2 size_aligned allocs, free
>     every other; subsequent size-2 alloc must succeed via the phase-2
>     fallback search since phase-1's S+A-1 threshold prunes the size-2
>     gaps.
>   - test_pci_32bit_workaround_pattern: alternate 32-bit-first allocation
>     attempts with 64-bit fallback, mirroring dma-iommu.c.
>   - test_stress_random: 2048 random alloc/free operations with mixed
>     sizes, alignments, and 32/64-bit limits, with periodic invariant
>     checks.
> 
> Each test verifies the augmented invariants both during and after the
> test run so that any sequencing bug in insert / erase / rotate /
> propagate is caught at the operation that introduced it.
> 
> Tested by: building drivers/iommu/iova.o and drivers/iommu/iova-kunit.o
> (no warnings); runtime execution requires booting a kernel with
> CONFIG_IOMMU_IOVA_KUNIT_TEST=y under qemu-system-x86_64 (not available
> on this devvm).
    ^^^^^^^^^^^^^^^

Heh, you should still read the patches when using claude :)

Can you add a .kunitconfig please? The tests should run with a command like:

    tools/testing/kunit/kunit.py run --build_dir build_kunit_x86_64 --arch x86_64 --kunitconfig ./drivers/iommu/.kunitconfig

Also it is worth while to carefully read each test because claude will
happily write nonsense tests that avoid actual bugs so they succeed...

> +static struct iova_test_ctx *iova_test_init_ctx(struct kunit *test)
> +{
> +	struct iova_test_ctx *ctx;
> +	int ret;
> +
> +	ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
> +	KUNIT_ASSERT_NOT_NULL(test, ctx);
> +
> +	ret = iova_cache_get();
> +	KUNIT_ASSERT_EQ(test, ret, 0);
> +
> +	init_iova_domain(&ctx->iovad, TEST_GRANULE, 1);
> +	ret = iova_domain_init_rcaches(&ctx->iovad);
> +	KUNIT_ASSERT_EQ(test, ret, 0);
> +	ctx->initialized = true;
> +	KUNIT_ASSERT_TRUE(test, iova_domain_verify_invariants(&ctx->iovad));
> +	return ctx;
> +}
> +
> +static void iova_test_cleanup(struct kunit *test, struct iova_test_ctx *ctx)
> +{
> +	if (ctx->initialized) {
> +		KUNIT_EXPECT_TRUE(test, iova_domain_verify_invariants(&ctx->iovad));
> +		put_iova_domain(&ctx->iovad);
> +		iova_cache_put();
> +		ctx->initialized = false;
> +	}
> +}
> +
> +static void test_init_destroy(struct kunit *test)
> +{
> +	struct iova_test_ctx *ctx = iova_test_init_ctx(test);
> +
> +	iova_test_cleanup(test, ctx);
> +}

This iova_test_init_ctx()/iova_test_cleanup() stuff is not the right
way to use kunit. Add a fixture and remove it from every test.

> +bool iova_domain_verify_invariants(struct iova_domain *iovad)
> +{
> +	bool ok = true;
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&iovad->iova_rbtree_lock, flags);
> +	iova_walk_verify(iovad->rbroot.rb_node, iovad, &ok);
> +	spin_unlock_irqrestore(&iovad->iova_rbtree_lock, flags);
> +	return ok;
> +}
> +EXPORT_SYMBOL_GPL(iova_domain_verify_invariants);

EXPORT_SYMBOL_IF_KUNIT

>  int iova_domain_init_rcaches(struct iova_domain *iovad);
>  struct iova *find_iova(struct iova_domain *iovad, unsigned long pfn);
>  void put_iova_domain(struct iova_domain *iovad);
> +#if IS_ENABLED(CONFIG_IOMMU_IOVA_KUNIT_TEST)
> +bool iova_domain_verify_invariants(struct iova_domain *iovad);
> +#endif

Don't really need this #ifdef

Jason

  reply	other threads:[~2026-05-15 22:43 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-13  2:00 [PATCH 0/5] iova augmented rbtree O(log n) alloc_iova Rik van Riel
2026-05-13  2:00 ` [PATCH 1/5] iova: switch to augmented rbtree for log(n) allocation Rik van Riel
2026-05-13  2:00 ` [PATCH 2/5] iova: drop dead cached_node / cached32_node infrastructure Rik van Riel
2026-05-13  2:00 ` [PATCH 3/5] iova: limit_pfn-aware augmentation for log(n) 32-bit alloc Rik van Riel
2026-05-13  2:00 ` [PATCH 4/5] iova: alignment-aware two-phase search for log(n) aligned alloc Rik van Riel
2026-05-13  2:00 ` [PATCH 5/5] iova: add KUnit test suite Rik van Riel
2026-05-15 22:43   ` Jason Gunthorpe [this message]
2026-05-16  3:20     ` Rik van Riel

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260515224317.GM7702@ziepe.ca \
    --to=jgg@ziepe.ca \
    --cc=iommu@lists.linux.dev \
    --cc=joro@8bytes.org \
    --cc=kernel-team@meta.com \
    --cc=kyle@mcmartin.ca \
    --cc=linux-kernel@vger.kernel.org \
    --cc=riel@meta.com \
    --cc=riel@surriel.com \
    --cc=robin.murphy@arm.com \
    --cc=will@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox