Re: [PATCH] Fix off-by-one buffer overflow in getdelim

[Rich Felker <dalias@libc.org>]

On Sat, Oct 24, 2015 at 07:35:15PM -0400, Rich Felker wrote:
> On Sun, Oct 25, 2015 at 12:25:52AM +0200, Felix Janda wrote:
> > Rich Felker wrote:
> > > On Sat, Oct 24, 2015 at 10:43:39PM +0200, Felix Janda wrote:
> > > > when deciding whether to resize the buffer, the terminating null byte
> > > > was not taken into account
> > > > ---
> > > >  src/stdio/getdelim.c | 2 +-
> > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > > 
> > > > diff --git a/src/stdio/getdelim.c b/src/stdio/getdelim.c
> > > > index a88c393..3077490 100644
> > > > --- a/src/stdio/getdelim.c
> > > > +++ b/src/stdio/getdelim.c
> > > > @@ -27,7 +27,7 @@ ssize_t getdelim(char **restrict s, size_t *restrict n, int delim, FILE *restric
> > > >  	for (;;) {
> > > >  		z = memchr(f->rpos, delim, f->rend - f->rpos);
> > > >  		k = z ? z - f->rpos + 1 : f->rend - f->rpos;
> > > > -		if (i+k >= *n) {
> > > > +		if (i+k+1 >= *n) {
> > > >  			if (k >= SIZE_MAX/2-i) goto oom;
> > > >  			*n = i+k+2;
> > > >  			if (*n < SIZE_MAX/4) *n *= 2;
> > > > -- 
> > > > 2.4.9
> > > 
> > > I think you're mistaken. i+k is the space needed so far in the buffer
> > > (not counting the terminating null byte) and *n is the usable space.
> > > The equality case of the i+k >= *n conditional covers the need to
> > > expand the buffer when the new content of length k would exactly fit
> > > but would not leave room for null termination.
> > > 
> > > Just to make sure I wrote a quick test program, which I've attached,
> > > that should crash in free if the overflow occurs. It does not crash
> > > and the output demonstrates correct resizing.
> > 
> > Thanks for the test program!
> > 
> > I did not see the 'if (z) break;'. The off-by-one should only occur
> > when memchr returns 0 but the byte from getc_unlocked is the delimiter.
> > (This makes it not so easy to observe the bug.)
> 
> Are you saying you still think there is a bug, that's only triggered
> when the byte from getc_unlocked causes the loop to break? I'll have
> to review that but it seems plausible. Do you have any ideas for
> adapting the test program to check this case?

Never mind; I can produce the expected crash just by adding
setbuf(f,0) right after the file is opened.

Rich