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: <793C2320-255A-4894-AA07-70EDBB1DDDA5@iki.fi>
Date:	Thu, 12 Apr 2012 17:47:41 +0300
From:	Jouni Siren <jouni.siren@....fi>
To:	linux-ext4@...r.kernel.org
Subject: Bug: Large writes can fail on ext4 if the write buffer is not empty

Hi,

I recently ran into problems when writing large blocks of data (more than about 2 GB) with a single call, if there is already some data in the write buffer. The problem seems to be specific to ext4, or at least it does not happen when writing to nfs on the same system. Also, the problem does not happen, if the write buffer is flushed before the large write.

The following C++ program should write a total of 4294967304 bytes, but I end up with a file of size 2147483664.

#include <fstream>

int
main(int argc, char** argv)
{
  std::streamsize data_size = (std::streamsize)1 << 31;
  char* data = new char[data_size];

  std::ofstream output("test.dat", std::ios_base::binary);
  output.write(data, 8);
  output.write(data, data_size);
  output.write(data, data_size);
  output.close();

  delete[] data;
  return 0;
}


The relevant part of strace is the following:

open("test.dat", O_WRONLY|O_CREAT|O_TRUNC, 0666) = 3
writev(3, [{"\0\0\0\0\0\0\0\0", 8}, {"", 2147483648}], 2) = -2147483640
writev(3, [{0xffffffff80c6d258, 2147483648}, {"", 2147483648}], 2) = -1 EFAULT (Bad address)
write(3, "\0\0\0\0\0\0\0\0", 8)         = 8
close(3)                                = 0


The first two writes are combined into a single writev call that reports having written -2147483640 bytes. This is the same as 8 + 2147483648, when interpreted as a signed 32-bit integer. After the first call, everything more or less fails. This happens on a Linux system, where uname -a returns

Linux alm01 2.6.32-220.7.1.el6.x86_64 #1 SMP Tue Mar 6 15:45:33 CST 2012 x86_64 x86_64 x86_64 GNU/Linux


I believe that the bug can be found in file.c, function ext4_file_write, where variable ret has type int. Function generic_file_aio_write returns the number of bytes written as a ssize_t, and the returned value is stored in ret and eventually returned by ext4_file_write. If the number of bytes written is more than INT_MAX, the value returned by ext4_file_write will be incorrect.

If you need more information on the problem, I will be happy to provide it.

-- 
Jouni Siren - jouni.siren@....fi - http://iki.fi/jouni.siren/




--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ