linux-kernel - [PATCH v2 RESEND 3] video: fbdev: cirrusfb: check pixclock to avoid divide by zero

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 2 Feb 2022 13:22:54 -0500
From:   George Kennedy <george.kennedy@...cle.com>
To:     Andrew Morton <akpm@...ux-foundation.org>
Cc:     Geert Uytterhoeven <geert@...ux-m68k.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Dan Carpenter <dan.carpenter@...cle.com>
Subject: [PATCH v2 RESEND 3] video: fbdev: cirrusfb: check pixclock to avoid
 divide by zero

Do a sanity check on pixclock value to avoid divide by zero.

If the pixclock value is zero, the cirrusfb driver will round up
pixclock to get the derived frequency as close to maxclock as
possible.

Syzkaller reported a divide error in cirrusfb_check_pixclock.

divide error: 0000 [#1] SMP KASAN PTI
CPU: 0 PID: 14938 Comm: cirrusfb_test Not tainted 5.15.0-rc6 #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2
RIP: 0010:cirrusfb_check_var+0x6f1/0x1260

Call Trace:
  fb_set_var+0x398/0xf90
  do_fb_ioctl+0x4b8/0x6f0
  fb_ioctl+0xeb/0x130
  __x64_sys_ioctl+0x19d/0x220
  do_syscall_64+0x3a/0x80
  entry_SYSCALL_64_after_hwframe+0x44/0xae

Cc: <stable@...r.kernel.org> # 5.4.y
Reported-by: syzkaller <syzkaller@...glegroups.com>
Signed-off-by: George Kennedy<george.kennedy@...cle.com>
Reviewed-by: Geert Uytterhoeven <geert@...ux-m68k.org>
---
Hi Andrew, can you pick this patch up?  Some of the fbdev drivers have
maintainers but the cirrusfb.c is basically orphaned.  I sent this patch
in Oct and again in Dec and no one picked it up.  Geert has reviewed it.
Obviously the original code just crashed so this shouldn't be
controversial.

v2: Geert says that instead of just returning an error, this is supposed
to find a valid pixclock and use that.

RESEND: Added the Reviewed by: and Cc: <stable@...r.kernel.org>

Thank you,
George

  drivers/video/fbdev/cirrusfb.c | 16 ++++++++--------
  1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/video/fbdev/cirrusfb.c 
b/drivers/video/fbdev/cirrusfb.c
index 93802ab..3d47c34 100644
--- a/drivers/video/fbdev/cirrusfb.c
+++ b/drivers/video/fbdev/cirrusfb.c
@@ -469,7 +469,7 @@ static int cirrusfb_check_mclk(struct fb_info *info, 
long freq)
      return 0;
  }

-static int cirrusfb_check_pixclock(const struct fb_var_screeninfo *var,
+static int cirrusfb_check_pixclock(struct fb_var_screeninfo *var,
                     struct fb_info *info)
  {
      long freq;
@@ -478,9 +478,7 @@ static int cirrusfb_check_pixclock(const struct 
fb_var_screeninfo *var,
      unsigned maxclockidx = var->bits_per_pixel >> 3;

      /* convert from ps to kHz */
-    freq = PICOS2KHZ(var->pixclock);
-
-    dev_dbg(info->device, "desired pixclock: %ld kHz\n", freq);
+    freq = PICOS2KHZ(var->pixclock ? : 1);

      maxclock = cirrusfb_board_info[cinfo->btype].maxclock[maxclockidx];
      cinfo->multiplexing = 0;
@@ -488,11 +486,13 @@ static int cirrusfb_check_pixclock(const struct 
fb_var_screeninfo *var,
      /* If the frequency is greater than we can support, we might be able
       * to use multiplexing for the video mode */
      if (freq > maxclock) {
-        dev_err(info->device,
-            "Frequency greater than maxclock (%ld kHz)\n",
-            maxclock);
-        return -EINVAL;
+        var->pixclock = KHZ2PICOS(maxclock);
+
+        while ((freq = PICOS2KHZ(var->pixclock)) > maxclock)
+            var->pixclock++;
      }
+    dev_dbg(info->device, "desired pixclock: %ld kHz\n", freq);
+
      /*
       * Additional constraint: 8bpp uses DAC clock doubling to allow 
maximum
       * pixel clock
-- 1.8.3.1