[PATCH v2] iommu/arm-smmu-v3: Limit queue allocation retry boundary to PAGE_SIZE

[PATCH v2] iommu/arm-smmu-v3: Limit queue allocation retry boundary to PAGE_SIZE

[Leo Jiang]

Stop retrying queue allocation when qsz reaches PAGE_SIZE.

Signed-off-by: Leo Jiang <leo.jiang1224@foxmail.com>
---
Changes in v2:
- Use real name "Leo Jiang".
- Remove code comments as suggested by Will.
- Limit the queue allocation retry boundary to PAGE_SIZE, per the feedback from the maintainer.
- Link to v1: https://lore.kernel.org/linux-iommu/tencent_F6E384A40D990A279B460A0CDE1927FDF509@qq.com
---
 drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
index e8d7dbe495f0..4a0b15b1d4d2 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
@@ -4418,7 +4418,7 @@ int arm_smmu_init_one_queue(struct arm_smmu_device *smmu,
 		qsz = ((1 << q->llq.max_n_shift) * dwords) << 3;
 		q->base = dmam_alloc_coherent(smmu->dev, qsz, &q->base_dma,
 					      GFP_KERNEL);
-		if (q->base || qsz < PAGE_SIZE)
+		if (q->base || qsz <= PAGE_SIZE)
 			break;
 
 		q->llq.max_n_shift--;
-- 
2.34.1

Re: [PATCH] iommu/arm-smmu-v3: Stop queue allocation retry at PAGE_SIZE

[Leo Jiang]

On Tue, Apr 21, 2026 at 05:38:40PM +0100, Will Deacon wrote:

> On Tue, Apr 21, 2026 at 04:56:47PM +0100, Robin Murphy wrote:
> > On 18/04/2026 6:31 am, leo.jiang1224@foxmail.com wrote:
> > > From: LoserJL <leo.jiang1224@foxmail.com>
> > > 
> > > In arm_smmu_init_one_queue(), the loop reduces max_n_shift if
> > > dmam_alloc_coherent() fails. However, since dmam_alloc_coherent()
> > > allocates at least PAGE_SIZE, retrying with a smaller size after
> > > a PAGE_SIZE failure is logically redundant.
> > 
> > Says who? It's certainly not a guarantee offered by the DMA API itself, and
> > indeed some allocation paths can definitely still allocate less than a page
> > - e.g. anything which hits a per-device or global coherent pool.
> > 
> > > Moreover, if a sub-page retry were to succeed due to concurrent memory
> > > release, the hardware would be configured with a smaller queue depth
> > > despite a full page being allocated. This leads to inefficient memory
> > > usage and unnecessary hardware performance limitation.
> > > 
> > > Terminate the loop once qsz reaches PAGE_SIZE to ensure logical
> > > consistency and optimal hardware configuration.
> > 
> > That's really not an argument - even if an allocator does happen to
> > over-allocate for the requested size, that is hardly the caller's concern;
> > and as far as "optimal" queue sizes go in this case, those very much depend
> > on the number of CPUs issuing commands and volume of expected stall/PRI
> > events - in many cases PAGE_SIZE would already be far too small to really
> > work well.
> > 
> > Also note that if we _were_ to fail to allocate a PAGE_SIZE or smaller
> > queue, there would be very little chance of the subsequent allocation(s) for
> > the stream table succeeding, so realistically the driver is probably going
> > to end up failing to probe in such circumstances anyway.
> 
> That's all true, but tbf I think I just fscked up the comparison in
> d25f6ead162e ("iommu/arm-smmu-v3: Increase maximum size of queues") so
> I'm not against fixing that up even though the "rationale" given by
> Loser doesn't make a whole lot of sense.

Hi Will, Robin,

Thank you both for the detailed feedback.

Robin, you are absolutely correct. After a deeper look into the source code,
I see that allocations smaller than a page are indeed possible in certain
cases. My previous assumption about the DMA API's granularity was wrong,
and I appreciate the correction.

However, as Will noted that the current logic deviates from the original
intent, I have prepared a v2 to limit the queue allocation retry boundary
to PAGE_SIZE.

In v2, I have:
 - Updated my identity to Leo Jiang.
 - Removed the code comments as suggested.

I will send the v2 as a follow-up shortly.

Best regards,
Leo Jiang

[PATCH v2] iommu/arm-smmu-v3: Limit queue allocation retry boundary to PAGE_SIZE

[Leo Jiang]

In arm_smmu_init_one_queue(), the driver attempts to allocate the largest
supported queue and retries by halving the size on failure. Currently,
the retry logic allows the allocation to fall below PAGE_SIZE if the
PAGE_SIZE attempt fails.

While dmam_alloc_coherent() can theoretically return allocations smaller
than a page in certain configurations, it is preferable to limit the
retry boundary to PAGE_SIZE for SMMUv3 queues to align with the
original design intent.

This patch ensures the retry loop terminates when qsz reaches PAGE_SIZE,
per the feedback from the maintainer.

Signed-off-by: Leo Jiang <leo.jiang1224@foxmail.com>
---
v2:
- Use real name "Leo Jiang".
- Remove code comments as suggested by Will.
- Limit the queue allocation retry boundary to PAGE_SIZE

 drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
index e8d7dbe495f0..4a0b15b1d4d2 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
@@ -4418,7 +4418,7 @@ int arm_smmu_init_one_queue(struct arm_smmu_device *smmu,
 		qsz = ((1 << q->llq.max_n_shift) * dwords) << 3;
 		q->base = dmam_alloc_coherent(smmu->dev, qsz, &q->base_dma,
 					      GFP_KERNEL);
-		if (q->base || qsz < PAGE_SIZE)
+		if (q->base || qsz <= PAGE_SIZE)
 			break;
 
 		q->llq.max_n_shift--;
-- 
2.34.1