linux-kernel - Re: [Internet]Re: [PATCH] fbcon: Out-Of-Bounds write in sys

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]

Message-ID: <28F2D8E8-B519-40F6-B6CD-98A0FAD67CD7@tencent.com>
Date:   Tue, 27 Jul 2021 01:53:13 +0000
From:   tcs_kernel(腾讯云内核开发者) 
        <tcs_kernel@...cent.com>
To:     Sam Ravnborg <sam@...nborg.org>
CC:     "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
        "daniel.vetter@...ll.ch" <daniel.vetter@...ll.ch>,
        "yepeilin.cs@...il.com" <yepeilin.cs@...il.com>,
        "penguin-kernel@...ove.SAKURA.ne.jp" 
        <penguin-kernel@...ove.sakura.ne.jp>,
        "tzimmermann@...e.de" <tzimmermann@...e.de>,
        "george.kennedy@...cle.com" <george.kennedy@...cle.com>,
        "ducheng2@...il.com" <ducheng2@...il.com>,
        "dri-devel@...ts.freedesktop.org" <dri-devel@...ts.freedesktop.org>,
        "linux-fbdev@...r.kernel.org" <linux-fbdev@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [Internet]Re: [PATCH] fbcon: Out-Of-Bounds write in
 sys_imageblit, add range check

yres and vyres can be controlled by user mode paramaters, and cause p->vrows to become a negative value. While this value be passed to real_y function, the ypos will be out of screen range.
This is an out-of-bounds write bug.
I think updatescrollmode is the right place to validate values supplied by a user ioctl, because only here makes --operation,and 0 is a legal value before that.

Signed-off-by: Tencent Cloud System tcs_kernel@...cent.com

diff --git a/drivers/video/fbdev/core/fbcon.c b/drivers/video/fbdev/core/fbcon.c
index 22bb3892f6bd..0970de46782f 100644
--- a/drivers/video/fbdev/core/fbcon.c
+++ b/drivers/video/fbdev/core/fbcon.c
@@ -1956,11 +1956,12 @@ static void updatescrollmode(struct fbcon_display *p,
        int yres = FBCON_SWAP(ops->rotate, info->var.yres, info->var.xres);
        int vyres = FBCON_SWAP(ops->rotate, info->var.yres_virtual,
                                   info->var.xres_virtual);
+       int rows = vc->vc_rows;
 
        p->vrows = vyres/fh;
-       if (yres > (fh * (vc->vc_rows + 1)))
-               p->vrows -= (yres - (fh * vc->vc_rows)) / fh;
-       if ((yres % fh) && (vyres % fh < yres % fh))
+       if ((yres > (fh * (rows + 1))) && (vyres >= (yres - (fh * rows))) && p->vrows)
+               p->vrows -= (yres - (fh * rows)) / fh;
+       if ((yres % fh) && (vyres % fh < yres % fh) && p->vrows)
                p->vrows--;
 }

在 2021/7/26 21:45，“Sam Ravnborg”<sam@...nborg.org> 写入:

    Hi,
    On Mon, Jul 26, 2021 at 11:32:37AM +0000, tcs_kernel(腾讯云内核开发者) wrote:
    > yres and vyres can be controlled by user mode paramaters, and cause p->vrows to become a negative value. While this value be passed to real_y function, the ypos will be out of screen range.
    > This is an out-of-bounds write bug.

    Please investigate if you can validate the user-supplied values for yres
    and vyres earlier so the code never reaches the below statements.
    This would also make it much more explicit what is going on.

    	Sam

    > 
    > 
    > diff --git a/drivers/video/fbdev/core/fbcon.c b/drivers/video/fbdev/core/fbcon.c
    > index 22bb3892f6bd..0970de46782f 100644
    > --- a/drivers/video/fbdev/core/fbcon.c
    > +++ b/drivers/video/fbdev/core/fbcon.c
    > @@ -1956,11 +1956,12 @@ static void updatescrollmode(struct fbcon_display *p,
    >         int yres = FBCON_SWAP(ops->rotate, info->var.yres, info->var.xres);
    >         int vyres = FBCON_SWAP(ops->rotate, info->var.yres_virtual,
    >                                    info->var.xres_virtual);
    > +       int rows = vc->vc_rows;
    >  
    >         p->vrows = vyres/fh;
    > -       if (yres > (fh * (vc->vc_rows + 1)))
    > -               p->vrows -= (yres - (fh * vc->vc_rows)) / fh;
    > -       if ((yres % fh) && (vyres % fh < yres % fh))
    > +       if ((yres > (fh * (rows + 1))) && (vyres >= (yres - (fh * rows))) && p->vrows)
    > +               p->vrows -= (yres - (fh * rows)) / fh;
    > +       if ((yres % fh) && (vyres % fh < yres % fh) && p->vrows)
    >                 p->vrows--;
    >  }
    >