From: Jason Gunthorpe <jgg@ziepe.ca>
To: Alex Williamson <alex.williamson@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>,
kvm@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] vfio: align capability structures
Date: Tue, 8 Aug 2023 20:29:25 -0300 [thread overview]
Message-ID: <ZNLP1dU1Ijzm/NPE@ziepe.ca> (raw)
In-Reply-To: <20230803151823.4e5943e6.alex.williamson@redhat.com>
On Thu, Aug 03, 2023 at 03:18:23PM -0600, Alex Williamson wrote:
> diff --git a/drivers/vfio/vfio_main.c b/drivers/vfio/vfio_main.c
> index 902f06e52c48..2d074cbd371d 100644
> --- a/drivers/vfio/vfio_main.c
> +++ b/drivers/vfio/vfio_main.c
> @@ -1362,6 +1362,8 @@ struct vfio_info_cap_header *vfio_info_cap_add(struct vfio_info_cap *caps,
> void *buf;
> struct vfio_info_cap_header *header, *tmp;
>
> + size = ALIGN(size, sizeof(u64));
> +
> buf = krealloc(caps->buf, caps->size + size, GFP_KERNEL);
> if (!buf) {
> kfree(caps->buf);
> @@ -1395,6 +1397,8 @@ void vfio_info_cap_shift(struct vfio_info_cap *caps, size_t offset)
> struct vfio_info_cap_header *tmp;
> void *buf = (void *)caps->buf;
>
> + WARN_ON(!IS_ALIGNED(offset, sizeof(u64)));
> +
> for (tmp = buf; tmp->next; tmp = buf + tmp->next - offset)
> tmp->next += offset;
> }
> diff --git a/include/uapi/linux/vfio.h b/include/uapi/linux/vfio.h
> index fa06e3eb4955..fd2761841ffe 100644
> --- a/include/uapi/linux/vfio.h
> +++ b/include/uapi/linux/vfio.h
> @@ -217,6 +217,7 @@ struct vfio_device_info {
> __u32 num_regions; /* Max region index + 1 */
> __u32 num_irqs; /* Max IRQ index + 1 */
> __u32 cap_offset; /* Offset within info struct of first cap */
> + __u32 pad; /* Size must be aligned for caps */
> };
> #define VFIO_DEVICE_GET_INFO _IO(VFIO_TYPE, VFIO_BASE + 7)
>
> @@ -1444,6 +1445,7 @@ struct vfio_iommu_type1_info {
> #define VFIO_IOMMU_INFO_CAPS (1 << 1) /* Info supports caps */
> __u64 iova_pgsizes; /* Bitmap of supported page sizes */
> __u32 cap_offset; /* Offset within info struct of first cap */
> + __u32 pad; /* Size must be aligned for caps */
> };
IMHO this is partially being caused by not using __aligned_u64 for the
other __u64's in the same struct..
Both of these structs have u64s in them and many arches will
automatically add the above padding. __aligned_u64 will force the
reset to do it, and then making padding explicit as you have done will
make it really true.
This is a subtle x64/x32 compatability issue also. It is probably best
just to do the change across the whole header file.
Please also include the matching hunk for iommufd:
--- a/drivers/iommu/iommufd/vfio_compat.c
+++ b/drivers/iommu/iommufd/vfio_compat.c
@@ -483,6 +483,8 @@ static int iommufd_vfio_iommu_get_info(struct iommufd_ctx *ictx,
rc = cap_size;
goto out_put;
}
+ cap_size = ALIGN(cap_size, sizeof(u64));
+
if (last_cap && info.argsz >= total_cap_size &&
put_user(total_cap_size, &last_cap->next)) {
rc = -EFAULT;
Thanks,
Jason
next prev parent reply other threads:[~2023-08-08 23:29 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-08-03 14:41 [PATCH] vfio: align capability structures Stefan Hajnoczi
2023-08-03 21:18 ` Alex Williamson
2023-08-04 13:33 ` Stefan Hajnoczi
2023-08-08 23:29 ` Jason Gunthorpe [this message]
2023-08-09 20:24 ` Stefan Hajnoczi
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=ZNLP1dU1Ijzm/NPE@ziepe.ca \
--to=jgg@ziepe.ca \
--cc=alex.williamson@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=stefanha@redhat.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