[PATCH] Please test SMP changes

John Baldwin jhb at FreeBSD.org
Wed Nov 19 10:28:30 PST 2003


On 19-Nov-2003 Peter Schultz wrote:
> John Baldwin wrote:
>> [ Bcc'd to various arch mailing lists ]
>> 
>> Please test the patch at http://www.FreeBSD.org/~jhb/patches/smp.patch
>> You can apply it to a sys/ check out by using patch -p6.  It moves
>> the setting of mp_maxid into a separate function (cpu_mp_setmaxid())
>> that is called before loader tunables so that UMA and others can use
>> it.  cpu_mp_probe() is now called at SI_SUB_CPU / SI_ORDER_SECOND like
>> it used to be.  This is necessay to re-enable the ACPI module on i386.
>> I have tested it on i386 but not on other archs.  Please be sure to
>> test that a normal SMP kernel works as well as booting an SMP kernel
>> with 'kern.smp.disabled'.  cpu_mp_probe() is no longer called if SMP
>> is disabled in the loader using the tunable.  Thanks.
>> 
> What all do I need to adjust on my end, other patches?  I took 'device 
> acpi' out of my kernel, but if I try to boot with the acpi module 
> loaded, the kernel still uses MPTable for interrupts:

No, please.  Don't disable ACPI.  This is not all the patches to do that.
Mostly I just want to mkae sure that I don't break SMP probing on all
the different architectures.  i386 seems to be working fine, I'm primarily
after testing on the other archs: alpha, amd64, ia64, and sparc64 in
particular to make sure I didn't break SMP probing on those archs.

To be clear: ACPI needs _more_ patches in addition to smp.patch
before it will work as a module again.

-- 

John Baldwin <jhb at FreeBSD.org>  <><  http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve!"  -  http://www.FreeBSD.org/


More information about the freebsd-current mailing list