Re: buffer overflow in /proc/sys/sunrpc/transports

["Vegard Nossum" <vegard.nossum@gmail.com>]

On Sat, Aug 30, 2008 at 9:56 PM, Cyrill Gorcunov <gorcunov@gmail.com> wrote:
> | BTW, look at this:
> |
> |     $ od -A x -t x1z /proc/sys/sunrpc/transports
> |     000000 74 63 70 20 31 30 34 38 35 37 36 0a 75 64 70 20  >tcp 1048576.udp <
> |     000010 33 32 37 36 38 0a 00 00 00 00 00 00 00 00 00 00  >32768...........<
> |     000020 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  >................<
> |     *
> |     0003e0 00 00 00 00 00 00 00 00 00 00                    >..........<
> |     0003ea
> |
> | ...and:
> |
> |     $ strace -e trace=read cat /proc/sys/sunrpc/transports > /dev/null
> |     read(3, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0@G\316E4\0\0\0"...,
> | 512) = 512
> |     read(3, "tcp 1048576\nudp 32768\n\0\0\0\0\0\0\0\0\0\0"..., 4096) = 4074
> |     read(3, "", 4096)                       = 0
> |
> | ...why does it have a huge return value? The output is only about 40
> | bytes... why add all the \0? Would your patch also fix this?
>
> I think it's from strace side - it pass 4096 zero'ed buffer.

"cat" passed buffer of size 4096, yes. But read() still returned 4074.
It should have returned 38 or so.

> At least I don't see additional issues from kernel side in buffer
> filling - except from svc_print_xprts() which walk over list.
> But I think sunpc guys should know details :)
> Will send short-fix patch soon :)

It looks like it's returning (sizeof(buffer) - x) where it really
should be returning x. Maybe it's this one that should be different?

        *lenp -= len;


Vegard

-- 
"The animistic metaphor of the bug that maliciously sneaked in while
the programmer was not looking is intellectually dishonest as it
disguises that the error is the programmer's own creation."
	-- E. W. Dijkstra, EWD1036