Re: [PATCH v3] vc_screen: reload load of struct vc_data pointer in vcs_write() to avoid UAF

[linux]

May 12, 2023 13:10:48 George Kennedy <george.kennedy@xxxxxxxxxx>:

> After a call to console_unlock() in vcs_write() the vc_data struct can be
> freed by vc_port_destruct(). Because of that, the struct vc_data pointer
> must be reloaded in the while loop in vcs_write() after console_lock() to
> avoid a UAF when vcs_size() is called.
>
> Syzkaller reported a UAF in vcs_size().
>
> BUG: KASAN: slab-use-after-free in vcs_size (drivers/tty/vt/vc_screen.c:215)
> Read of size 4 at addr ffff8880beab89a8 by task repro_vcs_size/4119
>
> Call Trace:
> <TASK>
> __asan_report_load4_noabort (mm/kasan/report_generic.c:380)
> vcs_size (drivers/tty/vt/vc_screen.c:215)
> vcs_write (drivers/tty/vt/vc_screen.c:664)
> vfs_write (fs/read_write.c:582 fs/read_write.c:564)
> ...
> <TASK>
>
> Allocated by task 1213:
> kmalloc_trace (mm/slab_common.c:1064)
> vc_allocate (./include/linux/slab.h:559 ./include/linux/slab.h:680
>     drivers/tty/vt/vt.c:1078 drivers/tty/vt/vt.c:1058)
> con_install (drivers/tty/vt/vt.c:3334)
> tty_init_dev (drivers/tty/tty_io.c:1303 drivers/tty/tty_io.c:1415
>     drivers/tty/tty_io.c:1392)
> tty_open (drivers/tty/tty_io.c:2082 drivers/tty/tty_io.c:2128)
> chrdev_open (fs/char_dev.c:415)
> do_dentry_open (fs/open.c:921)
> vfs_open (fs/open.c:1052)
> ...
>
> Freed by task 4116:
> kfree (mm/slab_common.c:1016)
> vc_port_destruct (drivers/tty/vt/vt.c:1044)
> tty_port_destructor (drivers/tty/tty_port.c:296)
> tty_port_put (drivers/tty/tty_port.c:312)
> vt_disallocate_all (drivers/tty/vt/vt_ioctl.c:662 (discriminator 2))
> vt_ioctl (drivers/tty/vt/vt_ioctl.c:903)
> tty_ioctl (drivers/tty/tty_io.c:2778)
> ...
>
> The buggy address belongs to the object at ffff8880beab8800
> which belongs to the cache kmalloc-1k of size 1024
> The buggy address is located 424 bytes inside of
> freed 1024-byte region [ffff8880beab8800, ffff8880beab8c00)
>
> The buggy address belongs to the physical page:
> page:00000000afc77580 refcount:1 mapcount:0 mapping:0000000000000000
>     index:0x0 pfn:0xbeab8
> head:00000000afc77580 order:3 entire_mapcount:0 nr_pages_mapped:0
>     pincount:0
> flags: 0xfffffc0010200(slab|head|node=0|zone=1|lastcpupid=0x1fffff)
> page_type: 0xffffffff()
> raw: 000fffffc0010200 ffff888100042dc0 ffffea000426de00 dead000000000002
> raw: 0000000000000000 0000000000100010 00000001ffffffff 0000000000000000
> page dumped because: kasan: bad access detected
>
> Memory state around the buggy address:
> ffff8880beab8880: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> ffff8880beab8900: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>> ffff8880beab8980: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>                                   ^
> ffff8880beab8a00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> ffff8880beab8a80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> ==================================================================
> Disabling lock debugging due to kernel taint
>
> Fixes: ac751efa6a0d ("console: rename acquire/release_console_sem() to console_lock/unlock()")
> Reported-by: syzkaller <syzkaller@xxxxxxxxxxxxxxxx>
> Signed-off-by: George Kennedy <george.kennedy@xxxxxxxxxx>

Reviewed-by: Thomas Weißschuh <linux@xxxxxxxxxxxxxx>

> ---
> v3: changed to "ret = -ENXIO;"
>
> v2: added "ret = -EINVAL;"
>
> v1: This patch is similar to the recent UAF vcs_read() patches (226fae124b2d & 46d733d0efc7).
> vcs_write() should have been checked for the same UAF issue at the same time the fixes went
> into vcs_read().
>
> drivers/tty/vt/vc_screen.c | 11 +++++++++--
> 1 file changed, 9 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/tty/vt/vc_screen.c b/drivers/tty/vt/vc_screen.c
> index 498ba9c0ee93..829c4be66f3b 100644
> --- a/drivers/tty/vt/vc_screen.c
> +++ b/drivers/tty/vt/vc_screen.c
> @@ -656,10 +656,17 @@ vcs_write(struct file *file, const char __user *buf, size_t count, loff_t *ppos)
>             }
>         }
>
> -       /* The vcs_size might have changed while we slept to grab
> -        * the user buffer, so recheck.
> +       /* The vc might have been freed or vcs_size might have changed
> +        * while we slept to grab the user buffer, so recheck.
>          * Return data written up to now on failure.
>          */
> +       vc = vcs_vc(inode, &viewed);
> +       if (!vc) {
> +           if (written)
> +               break;
> +           ret = -ENXIO;
> +           goto unlock_out;
> +       }
>         size = vcs_size(vc, attr, false);
>         if (size < 0) {
>             if (written)
> -- > 2.31.1