cvs commit: src/sys/amd64/amd64 machdep.c

Scott Long scottl at
Mon Nov 28 22:08:53 GMT 2005

John Baldwin wrote:
> On Monday 28 November 2005 04:51 pm, Scott Long wrote:
>>John Baldwin wrote:
>>>On Monday 28 November 2005 04:06 pm, Scott Long wrote:
>>>>John Baldwin wrote:
>>>>>On Monday 21 November 2005 01:39 pm, John Baldwin wrote:
>>>>>>jhb         2005-11-21 18:39:17 UTC
>>>>>>FreeBSD src repository
>>>>>>Modified files:
>>>>>>  sys/amd64/amd64      machdep.c
>>>>>>Expand the hack to mask the atpics if 'device atpic' is not in the
>>>>>>kernel during boot up.  Now we do a full reset of the 8259As and setup
>>>>>>a simple interrupt handler (we actually borrow the apic one that just
>>>>>>does an immediate iret) to handle any spurious interrupts triggered by
>>>>>>either chip. This should fix some folks that were getting a Trap 30
>>>>>>during bootup of certain SMP AMD systems.  This might get pushed into
>>>>>>the 6.0 branch as an errata.  For now a suitable workaround is to add
>>>>>>'device atpic' to your kernel config.
>>>>>>Tested by:      scottl
>>>>>>Helpful info from:      dillon
>>>>>>MFC after:      1 week
>>>>>Hmm, we probably still need to reprogram the ATPIC on resume as well.
>>>>>I'm not sure it's actually worth not just compiling the atpic code in on
>>>>Problems aside, what are the benefits to not having the atpic
>>>>unconditionally included on amd64?
>>>Purely space savings.  It's whatever the size of atpic.o, elcr.o, and the
>>>bits of atpic_vector.S that make it into exception.o are.
>>Ok, so it doesn't cut down on runtime overhead?  The file sizes look to
>>atpic.o     15k
>>elcr.o      2.5k
>>exception.o 200byte delta
> No, there isn't any effect on runtime.
>>If, down the road, a motherboard shows up without an atpic or one that
>>is horribly broken, would we be worse off for having the atpic code in
> Well, both i386 and amd64 assume an atpic is there.  Even if you don't include 
> 'device atpic' on amd64, we do the manual bit banging to the I/O ports that 
> assume it is there in the code I just changed.

So yeah, might as well just make it conditional then.


More information about the cvs-src mailing list