svn commit: r201999 - head/lib/libc/stdio

[Colin Percival]

Andrey Chernov wrote:
> On Sun, Jan 10, 2010 at 02:30:30PM +0000, Colin Percival wrote:
>> +	if (((count | size) > 0xFFFF) &&
>> +	    (count > SIZE_MAX / size)) {
>> +		errno = EINVAL;
>> +		fp->_flags |= __SERR;
>> +		return (0);
>> +	}
> 
> 1) I don't think that this is good place of exact constants like 0xFFFF, 
> usually we don't use such things in overflow checks (see all other ones).
> fread/fwrite are already slow as designed, so optimizing one time argument 
> check looks strange.

On my laptop, this optimization speeds up 1-byte fwrite calls by 27% compared
to always doing the division.  You can argue that people who care about code
performance shouldn't be making millions of 1-byte fwrite calls, but I think
it's a simple enough optimization and has enough of an impact to be worthwhile.

> 2) fp->_flags |= __SERR;
> This flag is for errors in the file stream, not for errors in 
> the arguments. Please back that line out.

Quoting fread(3):
     The function fread() does not distinguish between end-of-file and error,
     and callers must use feof(3) and ferror(3) to determine which occurred.
This would seem to imply that for any failed request, either feof or ferror
should return a non-zero value.  Removing this line would break the common
	while (fread(ptr, reclen, 1, f) == 1) {
		/* Do stuff here. */
	}
	if (ferror(f)) {
		/* Die due to read error */
	}
	/* No error?  Ok, we must have hit EOF */
idiom.

> 3) errno should be EOVERFLOW, see other owerflow checks in the stdio.

I picked EINVAL because this is the code used by read(2) and write(2) if they
are passed nbytes > INT_MAX.  It would seem odd to use one error code for a
number of bytes between INT_MAX and SIZE_MAX and then switch to a different
error code for > SIZE_MAX bytes.

-- 
Colin Percival
Security Officer, FreeBSD | freebsd.org | The power to serve
Founder / author, Tarsnap | tarsnap.com | Online backups for the truly paranoid