Date: Mon, 1 Apr 2013 11:54:09 -0400 From: John Baldwin <jhb@freebsd.org> To: "Alexander V. Chernikov" <melifaro@freebsd.org> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r248705 - head/sys/dev/ipmi Message-ID: <201304011154.09795.jhb@freebsd.org> In-Reply-To: <201303251430.r2PEUYcx045864@svn.freebsd.org> References: <201303251430.r2PEUYcx045864@svn.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday, March 25, 2013 10:30:34 am Alexander V. Chernikov wrote: > Author: melifaro > Date: Mon Mar 25 14:30:34 2013 > New Revision: 248705 > URL: http://svnweb.freebsd.org/changeset/base/248705 > > Log: > Unlock IPMI sc while performing requests via KCS and SMIC interfaces. > It is already done in SSIF interface code. > This reduces contention/spinning reported by many users. Eh, this was on purpose to prevent concurrent access to the hardware. SSIF doesn't do this because the smbus driver itself does locking internally. There are no followups in the PR to state how this patch helps (or if it was tested). OTOH, we should probably make KCS and SMIC perform their requests synchronously rather than kicking them over to a worker thread and only use a worker thread for SSIF. This has the advantage that you could make it interruptible so you could Ctrl-C ipmiutil and have it do something useful. Also, the current KCS/SMIC code has a timeout, it shouldn't spin forever. The fact that it is spinning forever is a different bug (possibly recently fixed by making 'ticks' volatile). That different bug is probably what should be fixed instead. -- John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201304011154.09795.jhb>