AIC7XXX for SMP kernel

[Doug Ledford]

Dave Perks wrote:
> 
> > "Robert L. Brown Jr." wrote:
> > > Can anybody tell me what version of Linux has an aic7xxx driver
> > > that will work in an SMP compiled kernel?
> > >
> > > I compiled 2.2.13, 2.2.5-12, 2.2.12-20 and both have bad aic7xxx drivers
> > > for SMP compiled kernels. I keep getting reset and timout loops on my
> > > 2940UW Adaptec card everytime during boot.
> > >
> > > Has anyone with an SMP board and an Adaptec 2940 successfully compiled a
> > > Linux kernel and the aic7xxx driver that work?
> 
> > No, of course not!  The aic7xxx driver and SMP kernels never have and never
> > will work!!
> >
> > (sometimes I honestly wonder just how many SMP systems there are out there
> > running the aic7xxx driver right now....)
> 
> >   Doug Ledford   <dledford at redhat.com>
> 
> *RedHat* Linux 6.1 installs an aic7xxx driver on an SMP PII system. Now I know why it doesn't work all the time (timeouts, lockups).
> 
> Why not, when I wrote a Linux video driver I just protected everything shared between h/w and s/w with a spinlock so only one s/w element at a time would touch it. Okay, also interrupt s/w and non interrupt s/w, but you get the idea...


OK, for all those people lacking the somewhat elusive sarcasm gene, the above
comments from me where satirical in nature.  The aic7xxx driver is actually
100% SMP safe.  As long ago as the 1998 Linux Expo in Durham, North Carolina,
Linus Torvalds ranked the aic7xxx driver as one of the three best SMP drivers
in the entire linux kernel.  Those things that people have reported regarding
aic7xxx and SMP have been related to issues other than aic7xxx SMP locking
(such as interrupt delivery in the core interrupt dispatch code or sequencer
related resets that aren't actually SMP specific but can be triggered easier
under SMP kernels due to interrupt latency and timing issues).  Those items
that seem related to SMP in particular are typically related to linux kernel
core SMP code and would effect other drivers for other cards if they were
plugged into the same PCI slots and treated the same by the motherboard BIOS
and MP tables.


-- 
  Doug Ledford   <dledford at redhat.com>
   Opinions expressed are my own, but
      they should be everybody's.


To Unsubscribe: send mail to majordomo at FreeBSD.org
with "unsubscribe aic7xxx" in the body of the message