New errors when booting current

[Dana H. Myers]

Nate Lawson wrote:
> Kevin Oberman wrote:
>> I am seeing large numbers of "bad write" and "bad read" errors from acpi
>> since my last update of the OS. I'm getting these from both desktops with
>> minimal ACPI capability and my laptop.
>>
>> I don't see any problems, so I suspect they are cosmetic, but I am
>> curious as to what changed to trigger them and whether they might be
>> eliminated.
>>
>> Let me know if you want to see my ASL, but it seems to be happening on
>> all of my current systems.
>>
>> Here is an excerpt from my dmesg on the desktop:
>> acpi: bad write to port 0x070 (8), val 0x26
>> acpi: bad read from port 0x071 (8)
>> acpi: bad write to port 0x070 (8), val 0x26
>> acpi: bad read from port 0x071 (8)
>> acpi: bad write to port 0x070 (8), val 0x26
> 
> Your ASL is writing to the RTC hardware directly.  This is not allowed
> by Windows XP and newer systems for obvious reasons.  We added the check
> recently but do not block the accesses yet.  I don't think we use the
> RTC much so we're just lucky there's no collision.  I'll try to add a
> patch soon that only prints the first 5 errors or so but can do more if
> we want to use it for debugging.
> 
> http://www.microsoft.com/whdc/system/pnppwr/powermgmt/BIOSAML.mspx

Windows apparently does more than block the accesses; it does to some
effort to make them work safely.  I eventually gave up on most I/O
port access control in the Solaris ACPI CA OSL; we'd been running for
several years (first with our homegrown interpreter, and then with
the ACPI CA port) without port checks and we didn't have a single
report of trouble as a result; when I added the checks, I immediately
started getting reports of trouble.

Dana