linux-mips
[Top] [All Lists]

Sign extension problem in llseek(2)

To: linux-arch@vger.kernel.org
Subject: Sign extension problem in llseek(2)
From: Ralf Baechle <ralf@linux-mips.org>
Date: Tue, 24 Mar 2009 19:24:18 +0100
Cc: dann frazier <dannf@dannf.org>, linux-mips@linux-mips.org
Original-recipient: rfc822;linux-mips@linux-mips.org
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.5.18 (2008-05-17)
In fs/read_write.c:

SYSCALL_DEFINE5(llseek, unsigned int, fd, unsigned long, offset_high,
                unsigned long, offset_low, loff_t __user *, result,
                unsigned int, origin)
...
        offset = vfs_llseek(file, ((loff_t) offset_high << 32) | offset_low,
                        origin);
 
On a 64-bit system that define CONFIG_HAVE_SYSCALL_WRAPPERS SYSCALL_DEFINEx
will truncate long arguments to 32-bit and on some architectures such as
MIPS sign-extended to 64-bit again.  On such architectures passing a
value with bit 31 in offset_low set will result in a huge 64-bit offset
being passed to vfs_llseek() and it failiing with EINVAL.

MIPS is affected by this issue.  Other 64-bit architectures which also
set CONFIG_HAVE_SYSCALL_WRAPPERS and __ARCH_WANT_SYS_LLSEEK are PowerPC,
S390 and sparc.

The issue was discovered on Debian's MIPS infrastructure machines running
e2fsck:

[...]
                          This was noticed on one of the Debian
infrastructure machines where, after an upgrade, e2fsck began failing
with errors like:

  Error reading block 524290 (Invalid argument) while getting next inode
  from scan.  Ignore error<y>?
[...]

  Ralf

<Prev in Thread] Current Thread [Next in Thread>
  • Sign extension problem in llseek(2), Ralf Baechle <=