Re: HEADSUP: acpi mpsafe committed

From: Robert Watson <rwatson_at_freebsd.org>
Date: Mon, 16 Aug 2004 14:33:30 -0400 (EDT)
On Mon, 16 Aug 2004, Nate Lawson wrote:

> Stefan Farfeleder wrote:
> > On Thu, Aug 12, 2004 at 11:35:48PM -0700, Nate Lawson wrote:
> > 
> >>Let me know if there are any problems.
> > 
> > 
> > I'm now getting a panic if I want to suspend my Thinkpad R32 via Fn-F4
> > (manually transcribed):
> > 
> > panic: mutex Giant not owned at /usr/src/sys/net/if.c:1874
> > 
> > db> trace
> > kdb_enter
> > panic
> > _mtx_assert
> > if_start
> > ieee80211_mgmt_output
> > ieee80211_send_mgmt
> > ieee80211_newstate

wi_newstate() should probably call NET_LOCK_GIANT() before entering the
802.11 framework, and call NET_UNLOCK_GIANT() on its return.  This will
cause it to conditionally acquire and release Giant based on
debug_mpsafenet. 

> I can't see how the acpi commit affects this.  The assertion is in the
> if code so it's likely that wi(4) is not setting the right flag to
> acquire Giant before if_start. 

Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
robert_at_fledge.watson.org      Principal Research Scientist, McAfee Research
Received on Mon Aug 16 2004 - 16:35:36 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:06 UTC