Andrey Chernov wrote: > On Wed, Mar 28, 2007 at 10:47:19AM +0200, Andre Oppermann wrote: >> Andrey Chernov wrote: >>> On Tue, Mar 27, 2007 at 02:18:46PM +0200, Andre Oppermann wrote: >>> >>>> Could you revert sys/netinet/tcp_input.c back to rev. 1.327 while leaving >>>> all others at HEAD and look if the bug can be reproduced? >>> >>> Yes! Reverting to 1.327 really helps, lockup is gone! Thanx! >> You tried with a fully updated system before the backout as well? >> There was a time window where TCP was indeed broken. It panic'ed >> though. > > Yes. My case is not related to SACK options KASSERT - my kernel even not > go so far to get that panic and I get lockup even when disable SACK by > sysctl. The problem is fixed in rev. 1.330 of tcp_input.c. Thanks to Robert for finding the bug. It was a INP_INFO_LOCK leak in the blackhole case that I missed during the tcp_input() split. There was a slight change in semantics. Sorry for the trouble. -- AndreReceived on Wed Mar 28 2007 - 11:00:42 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:07 UTC