ID:               22657
 Updated by:       [EMAIL PROTECTED]
 Reported By:      vesely at tana dot it
 Status:           Bogus
 Bug Type:         HTTP related
 Operating System: Any
 PHP Version:      4.3.0
 New Comment:

It must not be tested for not being NULL because
it is only NULL while self->bytes_in_buffer is
0. There will be no action on it until bytes_in_buffer
becomes > 0. This only happens after the buf begin
is set to the buffer again...



Previous Comments:
------------------------------------------------------------------------

[2003-03-13 01:13:01] vesely at tana dot it

> buf_begin can be NULL, no DOS possible -> BOGUS1

it is not initialized to NULL, it is never tested for not
being NULL, and around line 232, function fill_buffer

  if (self->bytes_in_buffer > 0 && self->buf_begin != self->buffer) {
    memmove(self->buffer, self->buf_begin, self->bytes_in_buffer);
        }

should grant the job will be done. Why do you say
it can be NULL? If I have some more time next week
I'll try and prepare a proof of concept.

>  searching for ',' is correct this works around a bug
>  in some IE version -> BOGUS2

Ha ha! I should have guessed it... :-)

------------------------------------------------------------------------

[2003-03-12 16:06:33] [EMAIL PROTECTED]

buf_begin can be NULL, no DOS possible -> BOGUS1

searching for ',' is correct this works around a bug
in some IE version -> BOGUS2

------------------------------------------------------------------------

[2003-03-12 03:44:37] vesely at tana dot it

Hi,
watch out rfc1867.c around line 342, in function
next_line() there is (was?) the following code:

        if (ptr) {
                /* ... */       
        } else {        
                /* ... */

                line[self->bufsize] = 0;
                self->buf_begin = ptr;    /* <=== */
                self->bytes_in_buffer = 0;
        }

ptr is obviously NULL, buf_begin should never be NULL
or the program may crash. So this is a potential
vulnerability for DOS attackers who submit long lines.

Since you're there, would you mind to check why at line
721, in the rfc1867_post_handler function, there is

                boundary_end = strchr(boundary, ',');

Shouldn't it be ';' (semicolon) rather than ',' (comma)?
(Just wandering)

------------------------------------------------------------------------


-- 
Edit this bug report at http://bugs.php.net/?id=22657&edit=1

Reply via email to