From: dahms AT ifk20 DOT mach DOT uni-karlsruhe DOT de
Subject: More bugs in b16
4 Dec 1996 18:02:20 -0800
Sender: daemon AT cygnus DOT com
Approved: cygnus DOT gnu-win32 AT cygnus DOT com
Distribution: cygnus
Message-ID: <009AC5DA.B8A1E0C0.5618.cygnus.gnu-win32@ifk20.mach.uni-karlsruhe.de>
Original-To: gnu-win32 AT cygnus DOT com
Original-CC: dahms AT ifk20 DOT mach DOT uni-karlsruhe DOT de
Original-Sender: owner-gnu-win32 AT cygnus DOT com

Hi all, I just wrote:

: I need timers urgent, too.

which reminds me about reporting some other problems I found.

I try rename() on a potentially nonexisting file and test for ENOENT, but it
does not set errno, which remains zero, so my program bombs out.

The return values from times() seems okay, but the fields tms_utime and
tms_stime of the struct tms contain garbage, since a newly started process
got a CPU usage of about 266000 seconds. Under Unix I always get near zero.
It is my understanding that the return value is only meaningful when taken
as the difference to a previously obtained value, but that should not hold
for the resource usages.

The first link attempt failed missing ntohl, ntohs, htonl and htons.
When compiling with optimization, they are handled by asm/byteorder.h,
but then floating numbers are hosed. Maybe there should be a #else clause
for the #ifdef __OPTIMIZE__. I got around it by packing __constant_ntohl
and the others wrapped as functions into an extra file.

BTW, it would be nice in gdb to be able to attach and detach to running
processes as under hp-ux, but that's low priority for me.


Bye, Heribert (dahms AT ifk20 DOT mach DOT uni-karlsruhe DOT de)
-
For help on using this list, send a message to
"gnu-win32-request AT cygnus DOT com" with one line of text: "help".