qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Stefan Hajnoczi <stefanha@redhat.com>
To: Peter Maydell <peter.maydell@linaro.org>
Cc: qemu-devel@nongnu.org, qemu-block@nongnu.org,
	Hanna Reitz <hreitz@redhat.com>, Kevin Wolf <kwolf@redhat.com>
Subject: Re: [PATCH] hw/block/dataplane/virtio-block: Avoid dynamic stack allocation
Date: Thu, 24 Aug 2023 13:15:12 -0400	[thread overview]
Message-ID: <20230824171512.GA1690425@fedora> (raw)
In-Reply-To: <20230824165740.2653919-1-peter.maydell@linaro.org>

[-- Attachment #1: Type: text/plain, Size: 2897 bytes --]

On Thu, Aug 24, 2023 at 05:57:40PM +0100, Peter Maydell wrote:
> Instead of using a variable length array in notify_guest_bh(), always
> use a fixed sized bitmap (this will be 128 bytes).  This means we
> need to avoid assuming that bitmap and the s->batch_notify_vqs bitmap
> are the same size; the neatest way to do this is to switch to using
> bitmap.h APIs to declare, copy and clear, because then we can specify
> the length in bits, exactly as we did when creating
> s->batch_notify_vqs with bitmap_new().
> 
> The codebase has very few VLAs, and if we can get rid of them all we
> can make the compiler error on new additions.  This is a defensive
> measure against security bugs where an on-stack dynamic allocation
> isn't correctly size-checked (e.g.  CVE-2021-3527).
> 
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> In discussion on Philippe's attempt at getting rid of this VLA:
> https://patchew.org/QEMU/20210505211047.1496765-1-philmd@redhat.com/20210505211047.1496765-7-philmd@redhat.com/
> Stefan suggested getting rid of the local bitmap array entirely.
> But I don't know this code at all and have no idea of the
> implications (presumably there is a reason we have the local
> array rather than iterating directly on batch_notify_vqs),
> so I have opted for the more minimal change.
> 
> Usual disclaimer: tested only with "make check" and
> "make check-avocado".
> ---
>  hw/block/dataplane/virtio-blk.c | 11 ++++++++---
>  1 file changed, 8 insertions(+), 3 deletions(-)

Hi Peter,
I recently sent a patch series that removes notify_guest_bh() completely:
https://lore.kernel.org/qemu-devel/20230817155847.3605115-5-stefanha@redhat.com/

If it's urgent we can merge your patch immediately, though I hope my
series will be merged soon anyway:
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>

> diff --git a/hw/block/dataplane/virtio-blk.c b/hw/block/dataplane/virtio-blk.c
> index da36fcfd0b5..f31ec79d0b2 100644
> --- a/hw/block/dataplane/virtio-blk.c
> +++ b/hw/block/dataplane/virtio-blk.c
> @@ -59,11 +59,16 @@ static void notify_guest_bh(void *opaque)
>  {
>      VirtIOBlockDataPlane *s = opaque;
>      unsigned nvqs = s->conf->num_queues;
> -    unsigned long bitmap[BITS_TO_LONGS(nvqs)];
> +    DECLARE_BITMAP(bitmap, VIRTIO_QUEUE_MAX);
>      unsigned j;
>  
> -    memcpy(bitmap, s->batch_notify_vqs, sizeof(bitmap));
> -    memset(s->batch_notify_vqs, 0, sizeof(bitmap));
> +    /*
> +     * Note that our local 'bitmap' is declared at a fixed
> +     * worst case size, but s->batch_notify_vqs has only
> +     * nvqs bits in it.
> +     */
> +    bitmap_copy(bitmap, s->batch_notify_vqs, nvqs);
> +    bitmap_zero(s->batch_notify_vqs, nvqs);
>  
>      for (j = 0; j < nvqs; j += BITS_PER_LONG) {
>          unsigned long bits = bitmap[j / BITS_PER_LONG];
> -- 
> 2.34.1
> 

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

  reply	other threads:[~2023-08-24 17:16 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-08-24 16:57 [PATCH] hw/block/dataplane/virtio-block: Avoid dynamic stack allocation Peter Maydell
2023-08-24 17:15 ` Stefan Hajnoczi [this message]
2023-08-29  9:52   ` Peter Maydell

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=20230824171512.GA1690425@fedora \
    --to=stefanha@redhat.com \
    --cc=hreitz@redhat.com \
    --cc=kwolf@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.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;
as well as URLs for NNTP newsgroup(s).