Date: Sat, 7 Oct 2000 16:21:31 +0400 (MSD) From: Igor Timkin <ivt@logger.gamma.ru> To: Justin Gibbs <gibbs@plutotech.com> Cc: Jordan Hubbard <jkh@winston.osd.bsdi.com>, gibbs@FreeBSD.ORG, cvs-committers@FreeBSD.ORG, cvs-all@FreeBSD.ORG Subject: Re: cvs commit: src/sys/dev/aic7xxx aic7xxx.c aic7xxx_freebsd.c Message-ID: <200010071221.QAA60436@logger.gamma.ru> In-Reply-To: <200010061441.IAA58551@pluto.plutotech.com> "from Justin Gibbs at Oct 6, 2000 08:41:08 am"
next in thread | previous in thread | raw e-mail | index | archive | help
I have problem with this commit. During 2-3 minutes systep panic: ahc2:A:11 ahc intr referenced scb not valid during seqint 0x71 scb(8) ahc2: warning no commend for scb 8 (cndcmplt) QUOTPOS=34 IP=0xc012fc24 (0xc012fb58 ahc_busy_tcl) System check out at 2 Oct panic avery 2 days (no crash dump due problem with adaptect, system haven't DDB). > > This is sort of why I'd really be happier if you held back a bit on > > MFCs in the future - the commit to current and then -stable > > immediately after it was probably not all that prudent. Thanks. > > > > - Jordan > > This is another instance where you've drawn poor assumptions. The > bug actually was introduced to -stable on Sept 16th. after a MFC that > occurred after code was in -current for some time. Unfortunately, > SMPNG and the disabling of spls() masked the problem and no amount > of time in current would have shown it up. When I became aware of > the issue, I corrected it in -current and immediately merged it > back to -stable because it was a necessary bug fix. Unforunately I > took patches from the wrong source tree so only half of the fix > was effected with the first commit. > > Would any amount of review by a third party have found this problem? > Probably not. I found it by inspection before any reports of failure > from -stable but only because I am still actively working on the code. > > I have been reading the discussions on how to deal with -stable with > some interest, but so far I think most comments miss an important > point. So long as -current and -stable are different, any MFC > will contain some risk. This will mean temporary but inevitable > short term instability in -stable as we strive to improve its quality. > The only way to improve -stable releases in my mind is to more proactively > plan when those releases will occur and provide similar shakeout and > beta peroids to releases made off of the mainline. For those tracking > -stable, they should take their decission to sync to any change with > caution, because no change is risk free. > > -- > Justin > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe cvs-all" in the body of the message > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200010071221.QAA60436>