Linux RDMA and InfiniBand development
 help / color / mirror / Atom feed
From: David Matlack <dmatlack@google.com>
To: Jason Gunthorpe <jgg@nvidia.com>
Cc: Alex Williamson <alex@shazbot.org>,
	kvm@vger.kernel.org, Leon Romanovsky <leon@kernel.org>,
	linux-kselftest@vger.kernel.org, linux-rdma@vger.kernel.org,
	Mark Bloch <mbloch@nvidia.com>,
	netdev@vger.kernel.org, Saeed Mahameed <saeedm@nvidia.com>,
	Shuah Khan <shuah@kernel.org>, Tariq Toukan <tariqt@nvidia.com>,
	patches@lists.linux.dev
Subject: Re: [PATCH v4 06/10] vfio: selftests: Allow drivers to specify required region size
Date: Wed, 12 Aug 2026 22:11:30 +0000	[thread overview]
Message-ID: <anzvkibgzTae9eV9@google.com> (raw)
In-Reply-To: <6-v4-021df3fb5a3f+98e-mlx5st_jgg@nvidia.com>

On 2026-08-12 11:59 AM, Jason Gunthorpe wrote:
> Add a region_size field to struct vfio_pci_driver_ops so drivers can
> declare how much DMA-mapped region they need. The mlx5 driver will need
> ~18MB for firmware pages. Existing drivers pass in the sizeof their state
> struct. Since drivers all use only their declared state we can use the
> value directly and remove the SZ_2M hard coding.
> 
> Assisted-by: Claude:claude-opus-4.6
> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

...

> diff --git a/tools/testing/selftests/vfio/lib/iova_allocator.c b/tools/testing/selftests/vfio/lib/iova_allocator.c
> index 4a660f636f4972..c8075575e9131d 100644
> --- a/tools/testing/selftests/vfio/lib/iova_allocator.c
> +++ b/tools/testing/selftests/vfio/lib/iova_allocator.c
> @@ -13,8 +13,10 @@
>  
>  #include <linux/iommufd.h>
>  #include <linux/limits.h>
> +#include <linux/log2.h>
>  #include <linux/mman.h>
>  #include <linux/overflow.h>
> +#include <linux/sizes.h>
>  #include <linux/types.h>
>  #include <linux/vfio.h>
>  
> @@ -50,7 +52,8 @@ void iova_allocator_cleanup(struct iova_allocator *allocator)
>  iova_t iova_allocator_alloc(struct iova_allocator *allocator, size_t size)
>  {
>  	VFIO_ASSERT_GT(size, 0, "Invalid size arg, zero\n");
> -	VFIO_ASSERT_EQ(size & (size - 1), 0, "Invalid size arg, non-power-of-2\n");
> +	/* Adjust the size to the allocator requirements */
> +	size = max_t(u64, roundup_pow_of_two(size), getpagesize());

...

> diff --git a/tools/testing/selftests/vfio/lib/vfio_pci_driver.c b/tools/testing/selftests/vfio/lib/vfio_pci_driver.c
> index 5e65434d2318b2..52b9c2a2bbb9b6 100644
> --- a/tools/testing/selftests/vfio/lib/vfio_pci_driver.c
> +++ b/tools/testing/selftests/vfio/lib/vfio_pci_driver.c
> @@ -32,6 +32,9 @@ void vfio_pci_driver_probe(struct vfio_pci_device *device)
>  			continue;
>  
>  		device->driver.ops = ops;
> +
> +		VFIO_ASSERT_NE(ops->region_size, 0);
> +		device->driver.region.size = ops->region_size;

I'm seeing failures with the IGB driver with this series. I think it's
coming from here. The region size needs to be page-aligned so that the
IOMMU map/unmap calls can work?

The size of struct igb is 0x20080 bytes which is not page-aligned. The
size of struct mlx5st_device is 0x200D000, which is page-aligned, so
that's probably why your Claude did not notice.

# #  RUN           vfio_pci_driver_test.iommufd.send_msi ...
# /usr/local/google/home/dmatlack/kernel/trees/review/tools/testing/selftests/vfio/lib/include/libvfio/iommu.h:42:
# Assertion Failure
#
#   Expression: __iommu_map(iommu, region) == 0
#   Observed: 0xffffffffffffffea == 0
#   [errno: 22 - Invalid argument]
#
# # send_msi: Test failed
# #          FAIL  vfio_pci_driver_test.iommufd.send_msi
# not ok 33 vfio_pci_driver_test.iommufd.send_msi

Applying this on top fixes things and also avoids a potential overflow
when doing the roundup_pow_of_two() in the allocator.

diff --git a/tools/testing/selftests/vfio/lib/iova_allocator.c b/tools/testing/selftests/vfio/lib/iova_allocator.c
index c8075575e913..66a160d4ff9d 100644
--- a/tools/testing/selftests/vfio/lib/iova_allocator.c
+++ b/tools/testing/selftests/vfio/lib/iova_allocator.c
@@ -52,8 +52,11 @@ void iova_allocator_cleanup(struct iova_allocator *allocator)
 iova_t iova_allocator_alloc(struct iova_allocator *allocator, size_t size)
 {
 	VFIO_ASSERT_GT(size, 0, "Invalid size arg, zero\n");
-	/* Adjust the size to the allocator requirements */
-	size = max_t(u64, roundup_pow_of_two(size), getpagesize());
+	VFIO_ASSERT_LT(size, rounddown_pow_of_two(SIZE_MAX),
+		       "Invalid size arg, too large (%lu)\n", size);
+
+	size = roundup_pow_of_two(size);

 	for (;;) {
 		struct iommu_iova_range *range;
diff --git a/tools/testing/selftests/vfio/lib/vfio_pci_driver.c b/tools/testing/selftests/vfio/lib/vfio_pci_driver.c
index 96a940ea59b5..b31abfe2c7ff 100644
--- a/tools/testing/selftests/vfio/lib/vfio_pci_driver.c
+++ b/tools/testing/selftests/vfio/lib/vfio_pci_driver.c
@@ -36,7 +36,8 @@ void vfio_pci_driver_probe(struct vfio_pci_device *device)
 		device->driver.ops = ops;

 		VFIO_ASSERT_NE(ops->region_size, 0);
-		device->driver.region.size = ops->region_size;
+		device->driver.region.size =
+			round_up(ops->region_size, getpagesize());
 	}
 }

  reply	other threads:[~2026-08-12 22:11 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-08-12 14:58 [PATCH v4 00/10] mlx5 support for VFIO self test Jason Gunthorpe
2026-08-12 14:58 ` [PATCH v4 01/10] net/mlx5: Add IFC structures for CQE and WQE Jason Gunthorpe
2026-08-12 14:59 ` [PATCH v4 02/10] net/mlx5: Move HW constant groups from device.h/cq.h to mlx5_ifc.h Jason Gunthorpe
2026-08-12 14:59 ` [PATCH v4 03/10] net/mlx5: Extract MLX5_SET/GET macros into mlx5_ifc_macros.h Jason Gunthorpe
2026-08-12 14:59 ` [PATCH v4 04/10] net/mlx5: Add ONCE and MMIO accessor variants to mlx5_ifc_macros.h Jason Gunthorpe
2026-08-12 14:59 ` [PATCH v4 05/10] selftests: Add additional kernel functions to tools/include/ Jason Gunthorpe
2026-08-12 14:59 ` [PATCH v4 06/10] vfio: selftests: Allow drivers to specify required region size Jason Gunthorpe
2026-08-12 22:11   ` David Matlack [this message]
2026-08-12 22:13     ` David Matlack
2026-08-12 23:44     ` Jason Gunthorpe
2026-08-12 14:59 ` [PATCH v4 07/10] vfio: selftests: Add dev_dbg Jason Gunthorpe
2026-08-12 14:59 ` [PATCH v4 08/10] vfio: selftests: Add mlx5 driver - HW init and command interface Jason Gunthorpe
2026-08-12 21:41   ` David Matlack
2026-08-12 23:32     ` Jason Gunthorpe
2026-08-12 14:59 ` [PATCH v4 09/10] vfio: selftests: Add mlx5 driver - data path and memcpy ops Jason Gunthorpe
2026-08-12 14:59 ` [PATCH v4 10/10] vfio: selftests: mlx5 driver - add send_msi support Jason Gunthorpe

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=anzvkibgzTae9eV9@google.com \
    --to=dmatlack@google.com \
    --cc=alex@shazbot.org \
    --cc=jgg@nvidia.com \
    --cc=kvm@vger.kernel.org \
    --cc=leon@kernel.org \
    --cc=linux-kselftest@vger.kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=mbloch@nvidia.com \
    --cc=netdev@vger.kernel.org \
    --cc=patches@lists.linux.dev \
    --cc=saeedm@nvidia.com \
    --cc=shuah@kernel.org \
    --cc=tariqt@nvidia.com \
    /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