1 Jul 2011 10:06
1 Jul 2011 13:41
Re: kdump: even more fancy
Otto Moerbeek <otto <at> drijf.net>
2011-07-01 11:41:06 GMT
2011-07-01 11:41:06 GMT
On Mon, Jun 27, 2011 at 08:39:19AM +0200, Otto Moerbeek wrote:
> On Tue, Jun 21, 2011 at 11:49:37AM +0200, Otto Moerbeek wrote:
>
> > Hi,
> >
> > with kdump, file offsets are printed as two ints on 32-bit systems.
> > This diff should fix that and print a single long long.
> >
> > Tested on vax and hppa. Strange things is that hppa prints a trailing
> > int for lseek(2). Have to dig into that. This has nothing to do with
> > fancy mode, since it is also printed with -n.
> >
> > Here a sample of the test I used:
> >
> > #include <sys/mman.h>
> > #include <unistd.h>
> >
> > main()
> > {
> > mmap(1,2,3,4,5,0x8000000090000000LL);
> > lseek(3,0x12345678abcdef12LL,1);
> > }
> >
> > Hope I got the tricky sign-extends all covered.
>
> Did anybody test or review this? I like to move on...
OK, I decide to move on anyway.
(Continue reading)
2 Jul 2011 00:08
Re: disable altq functions
Claudio Jeker <claudio <at> openbsd.org>
2011-07-01 22:08:37 GMT
2011-07-01 22:08:37 GMT
On Wed, Jun 29, 2011 at 03:00:09PM -0400, Ted Unangst wrote: > On Wed, 22 Jun 2011, Ted Unangst wrote: > > > Does anybody use altq cdnr or rio? Do they work? Why do we have them, > > but not enabled? > > So it seems like cdnr and rio are dead. Last chance, would anyone like to > object before I start my chainsaw? I would prefer if someone would actually do real work on altq and not just chainsawing the mostly dead branches of the tree. Wishful think I know. Did you check the other BSD if they finished these two modes? -- -- :wq Claudio
2 Jul 2011 01:07
Re: disable altq functions
Henning Brauer <lists-openbsdtech <at> bsws.de>
2011-07-01 23:07:42 GMT
2011-07-01 23:07:42 GMT
* Claudio Jeker <claudio <at> openbsd.org> [2011-07-02 00:11]: > On Wed, Jun 29, 2011 at 03:00:09PM -0400, Ted Unangst wrote: > > On Wed, 22 Jun 2011, Ted Unangst wrote: > > > > > Does anybody use altq cdnr or rio? Do they work? Why do we have them, > > > but not enabled? > > > > So it seems like cdnr and rio are dead. Last chance, would anyone like to > > object before I start my chainsaw? > > I would prefer if someone would actually do real work on altq and not > just chainsawing the mostly dead branches of the tree. Wishful think I > know. Did you check the other BSD if they finished these two modes? what i have in mind in that ara only involved prio and hfsc (the latter just because ppl will want it :(( ) - I think cdnr and rio belong to the attic, so kill kill kill. if anyone wants to revive them they're still available in Attic. -- -- Henning Brauer, hb <at> bsws.de, henning <at> openbsd.org BS Web Services, http://bsws.de Full-Service ISP - Secure Hosting, Mail and DNS Services Dedicated Servers, Rootservers, Application Hosting
2 Jul 2011 03:04
hostid 64-bit support
Ted Unangst <tedu <at> tedunangst.com>
2011-07-02 01:04:04 GMT
2011-07-02 01:04:04 GMT
hostid is a long, and that ship has sailed, but still 64-bit archs should
not truncate the value.
I would like somebody to test this on a 64-bit machine. Even try
setting a 64-bit hostid maybe.
Index: kern_sysctl.c
===================================================================
RCS file: /home/tedu/cvs/src/sys/kern/kern_sysctl.c,v
retrieving revision 1.205
diff -u -r1.205 kern_sysctl.c
--- kern_sysctl.c 27 Jun 2011 03:40:35 -0000 1.205
+++ kern_sysctl.c 2 Jul 2011 00:59:56 -0000
<at> <at> -349,9 +349,18 <at> <at>
domainnamelen = newlen;
return (error);
case KERN_HOSTID:
- inthostid = hostid; /* XXX assumes sizeof long <= sizeof int */
- error = sysctl_int(oldp, oldlenp, newp, newlen, &inthostid);
- hostid = inthostid;
+#ifdef LP64
+ if (*oldlenp == sizeof(int64_t) || newlen == sizeof(int64_t)) {
+ error = sysctl_quad(oldp, oldlenp, newp, newlen,
+ &hostid);
+ } else
+#endif
+ {
+ inthostid = hostid;
+ error = sysctl_int(oldp, oldlenp, newp, newlen,
+ &inthostid);
(Continue reading)
2 Jul 2011 07:20
Re: hostid 64-bit support
Brynet <brynet <at> gmail.com>
2011-07-02 05:20:23 GMT
2011-07-02 05:20:23 GMT
This might be useful.. but POSIX defines a hostid as 32-bit, however the use of 'long' makes it ambiguous. http://pubs.opengroup.org/onlinepubs/9699919799/functions/gethostid.html Not sure if anything actually would break if you bumped this, do you know if anything relies on it? or was it mainly intended for humans. Does it default to '0' on other systems as well? -Bryan.
RSS Feed