Uwe Bonnes <[EMAIL PROTECTED]> writes:
> int MSVCRT_fclose(MSVCRT_FILE* file)
> {
> int r;
> r=_close(file->_file);
> TRACE("r 0x%08x flag 0x%08x %s\n",r, file->_flag,
> (file->_flag & MSVCRT__IOERR)?"MSVCRT__IOERR":"");
> return ((r==MSVCRT_EOF) || (file->_flag & MSVCRT__IOERR) ? MSVCRT_EOF :0);
> }
>
> While the trace in line 321037 shows that MSVCRT__IOERR is not set, the
> return value implies that MSVCRT__IOERR is set.
>
> I run on a dual celeron. Is msvcrt perhaps not thread safe?
Well, seeing that there are no locks at all in the above function I
think the answer is clear... and yes that should probably be fixed.
--
Alexandre Julliard
[EMAIL PROTECTED]
