Re: panics on 6.3-RELEASE in IP stack
Robert Watson wrote: On Fri, 18 Apr 2008, Bruce M. Simpson wrote: I concur, this fix should really be MFCed. The problem went away in 7.x due to a total rewrite. I am distracted by other stuff at the moment, so, starter's orders... This just bit me. The fix is in RELENG_6, but it is not present after the -p1 tag. So updating to STABLE should fix the problem for Petr. Whilst 6 is no longer the STABLE branch, I think this really should go onto 6 in case any other releases happen from that branch. There are (tentative) plans to release a 6.4-RELEASE, but I don't think there's a schedule yet. Developers should plan on continuing to merge changes, especially bug fixes, to RELENG_6. Robert N M Watson Computer Laboratory University of Cambridge ___ Just a note that I'd really like it for a 6.4-RELEASE to be made available if possible. I still have some hosts on 6-STABLE for a variety of reasons (including one very nasty problem with 7-STABLE, pf, and IPSec with NAT traversal that I'm trying to troubleshoot in preparation for a possible PR, once I find out exactly where the problem is). Thanks, Proto ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: panics on 6.3-RELEASE in IP stack
On Mon, 2008-04-21 at 11:27 -0400, Michael Proto wrote: Robert Watson wrote: On Fri, 18 Apr 2008, Bruce M. Simpson wrote: I concur, this fix should really be MFCed. The problem went away in 7.x due to a total rewrite. I am distracted by other stuff at the moment, so, starter's orders... This just bit me. The fix is in RELENG_6, but it is not present after the -p1 tag. So updating to STABLE should fix the problem for Petr. Whilst 6 is no longer the STABLE branch, I think this really should go onto 6 in case any other releases happen from that branch. There are (tentative) plans to release a 6.4-RELEASE, but I don't think there's a schedule yet. Developers should plan on continuing to merge changes, especially bug fixes, to RELENG_6. Robert N M Watson Computer Laboratory University of Cambridge ___ Just a note that I'd really like it for a 6.4-RELEASE to be made available if possible. I still have some hosts on 6-STABLE for a variety of reasons (including one very nasty problem with 7-STABLE, pf, and IPSec with NAT traversal that I'm trying to troubleshoot in preparation for a possible PR, once I find out exactly where the problem is). Thanks, Proto Remember that 6.3-RELEASE is an extended support release, and will be maintained by the Security Team until (estimated) Jan 31, 2010 - as will 6-STABLE. http://www.freebsd.org/security/ Cheers Tom signature.asc Description: This is a digitally signed message part
Re: panics on 6.3-RELEASE in IP stack
On Fri, 18 Apr 2008, Bruce M. Simpson wrote: I concur, this fix should really be MFCed. The problem went away in 7.x due to a total rewrite. I am distracted by other stuff at the moment, so, starter's orders... This just bit me. The fix is in RELENG_6, but it is not present after the -p1 tag. So updating to STABLE should fix the problem for Petr. Whilst 6 is no longer the STABLE branch, I think this really should go onto 6 in case any other releases happen from that branch. There are (tentative) plans to release a 6.4-RELEASE, but I don't think there's a schedule yet. Developers should plan on continuing to merge changes, especially bug fixes, to RELENG_6. Robert N M Watson Computer Laboratory University of Cambridge ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: panics on 6.3-RELEASE in IP stack
If memory serves me right, John Baldwin wrote: Given how simple the patch is and that if fixes a known panic this might be worthy of an errata notice or errata candidate. (At least a note in the errata pointing to the 1.85.2.10 commit if not an actual patch to RELENG_6_3.) I added an item to the post-release errata for this. Bruce. signature.asc Description: OpenPGP digital signature
Re: panics on 6.3-RELEASE in IP stack
Bruce M. Simpson wrote: I started to play with RAT application (ports: mbone/rat + an SVN version) and it seems to crash my 6.3-RELEASE-p1 box in rather deterministic way. Crash details are shown below. Has anyone seen a problem like this? Yes, there's an off-by-one reference count bug in the multicast stuff. I concur, this fix should really be MFCed. The problem went away in 7.x due to a total rewrite. I am distracted by other stuff at the moment, so, starter's orders... This just bit me. The fix is in RELENG_6, but it is not present after the -p1 tag. So updating to STABLE should fix the problem for Petr. Whilst 6 is no longer the STABLE branch, I think this really should go onto 6 in case any other releases happen from that branch. cheers BMS ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: panics on 6.3-RELEASE in IP stack
On Monday 07 April 2008 12:16:53 pm Petr Holub wrote:
Hi all,
I started to play with RAT application (ports: mbone/rat + an SVN version)
and
it seems to crash my 6.3-RELEASE-p1 box in rather deterministic way. Crash
details are shown below. Has anyone seen a problem like this?
Yes, there's an off-by-one reference count bug in the multicast stuff. You
need 1.85.2.10 of sys/netinet/in.c:
Index: in.c
===
RCS file: /host/cvs/usr/cvs/src/sys/netinet/in.c,v
retrieving revision 1.85.2.9
retrieving revision 1.85.2.10
diff -u -r1.85.2.9 -r1.85.2.10
--- in.c7 Jul 2007 00:54:46 - 1.85.2.9
+++ in.c22 Feb 2008 19:13:56 - 1.85.2.10
@@ -991,7 +991,6 @@
* a new record. Otherwise, we are done.
*/
if (ifma-ifma_protospec != NULL) {
- if_delmulti_ent(ifma); /* We don't need another reference */
IN_MULTI_UNLOCK();
IFF_UNLOCKGIANT(ifp);
return ifma-ifma_protospec;
Given how simple the patch is and that if fixes a known panic this might be
worthy of an errata notice or errata candidate. (At least a note in the
errata pointing to the 1.85.2.10 commit if not an actual patch to
RELENG_6_3.)
--
John Baldwin
___
freebsd-stable@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to [EMAIL PROTECTED]
RE: panics on 6.3-RELEASE in IP stack
Yes, there's an off-by-one reference count bug in the multicast stuff. You need 1.85.2.10 of sys/netinet/in.c: I can confirm that RAT can be restarted many times now with this patch applied (at least when RAT doesn't touch audio devices as I may be also reporting some sound related panic soon ;-) ). I'd also propose to release errata on this. Thanks Petr ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: panics on 6.3-RELEASE in IP stack
John Baldwin wrote: On Monday 07 April 2008 12:16:53 pm Petr Holub wrote: Hi all, I started to play with RAT application (ports: mbone/rat + an SVN version) and it seems to crash my 6.3-RELEASE-p1 box in rather deterministic way. Crash details are shown below. Has anyone seen a problem like this? Yes, there's an off-by-one reference count bug in the multicast stuff. I concur, this fix should really be MFCed. The problem went away in 7.x due to a total rewrite. I am distracted by other stuff at the moment, so, starter's orders... cheers BMS ___ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to [EMAIL PROTECTED]
