qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Roland Dreier <rdreier@cisco.com>
To: Anthony Liguori <anthony@codemonkey.ws>
Cc: Dave Airlie <airlied@gmail.com>, qemu-devel@nongnu.org
Subject: [Qemu-devel] Re: [PATCH to consider for 0.12] vmware_vga: Don't crash on too-big DEFINE_CURSOR command
Date: Thu, 17 Dec 2009 14:38:08 -0800	[thread overview]
Message-ID: <adapr6d446n.fsf@roland-alpha.cisco.com> (raw)
In-Reply-To: <4B2AB1F2.3060507@codemonkey.ws> (Anthony Liguori's message of "Thu, 17 Dec 2009 16:34:26 -0600")


 > Thanks for the patch.  I'm planning on giving Dave Airlie's series a
 > try for 0.12.0.  I'm pretty comfortable with those patches (since a
 > few of them are mine :-)).  I also don't think vmware-vga is going to
 > be reliable without them so I don't think pulling in the one fix is
 > good enough.
 > 
 > His last patch has the same fix without the printf().  The printf is
 > probably something to avoid since a malicious guest could create a
 > storm of them.  Since libvirt logs stderr by default, the result could
 > be pretty nasty.

Fair enough... I just saw Dave's patches go by, and I guess we
independently fixed the cursor size thing at right around the same time.
How about the following, without the fprintf but with paranoid checks
(since a malicious guest could send a bad DEFINE_CURSOR and do who knows
what with the buffer overrun, which is even worse than spamming logs ;)

====

QEMU crashes with vmware_vga when running a Linux guest with the latest
X.org vmware video driver if QEMU is using SDL for video output.  In
this case, vmware_vga advertises cursor acceleration to the guest, and
the crash comes when the guest does a DEFINE_CURSOR command with a 64x64
32bpp cursor.  This request overruns the image[] array in struct
vmsvga_cursor_definition_s and QEMU ends up segfaulting because of
memory corruption caused by writing past the end of the array.

Fix this by enlarging the image[] array to be able to hold 4096 32-bit
pixels so we don't fail for the case of 64*64*32bpp, and also add error
checking to avoid a crash if an even bigger request is sent by a guest.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 hw/vmware_vga.c |    9 ++++++++-
 1 files changed, 8 insertions(+), 1 deletions(-)

diff --git a/hw/vmware_vga.c b/hw/vmware_vga.c
index f3e3749..75d90f2 100644
--- a/hw/vmware_vga.c
+++ b/hw/vmware_vga.c
@@ -462,7 +462,7 @@ struct vmsvga_cursor_definition_s {
     int hot_x;
     int hot_y;
     uint32_t mask[1024];
-    uint32_t image[1024];
+    uint32_t image[4096];	/* allow for 64x64 32bpp cursor */
 };
 
 #define SVGA_BITMAP_SIZE(w, h)		((((w) + 31) >> 5) * (h))
@@ -557,6 +557,13 @@ static void vmsvga_fifo_run(struct vmsvga_state_s *s)
             cursor.height = y = vmsvga_fifo_read(s);
             vmsvga_fifo_read(s);
             cursor.bpp = vmsvga_fifo_read(s);
+
+	    if (SVGA_BITMAP_SIZE(x, y) > sizeof cursor.mask ||
+		SVGA_PIXMAP_SIZE(x, y, cursor.bpp) > sizeof cursor.image) {
+		    args = SVGA_BITMAP_SIZE(x, y) + SVGA_PIXMAP_SIZE(x, y, cursor.bpp);
+		    goto badcmd;
+	    }
+
             for (args = 0; args < SVGA_BITMAP_SIZE(x, y); args ++)
                 cursor.mask[args] = vmsvga_fifo_read_raw(s);
             for (args = 0; args < SVGA_PIXMAP_SIZE(x, y, cursor.bpp); args ++)

  reply	other threads:[~2009-12-17 22:38 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-12-17 22:27 [Qemu-devel] [PATCH to consider for 0.12] vmware_vga: Don't crash on too-big DEFINE_CURSOR command Roland Dreier
2009-12-17 22:34 ` [Qemu-devel] " Anthony Liguori
2009-12-17 22:38   ` Roland Dreier [this message]
2009-12-17 22:49     ` Anthony Liguori
2009-12-17 22:41   ` Roland Dreier
2009-12-17 22:48     ` Anthony Liguori
2009-12-20 18:06       ` Roland Dreier
2010-01-06  4:43   ` [Qemu-devel] [PATCH resend] vmware_vga: Check cursor dimensions passed from guest to avoid buffer overflow Roland Dreier
2010-01-11 16:01     ` Anthony Liguori

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=adapr6d446n.fsf@roland-alpha.cisco.com \
    --to=rdreier@cisco.com \
    --cc=airlied@gmail.com \
    --cc=anthony@codemonkey.ws \
    --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).