How to fix MPS 1.4 + ACPI behaviour ?

Hi,

(kernel: 2.5.69)

The conclusion of bug 699 is that some / all i386 SMP systems that use MPS 1.4 (and higher ? or all MPS versions ?), should boot with the "pci=noacpi" parameter to prevent IRQ problems.

What exactly happens: The MPS 1.4 interpreter causes PCI IRQs to be remapped to IRQ 16 and higher, which is the desired behaviour. The ACPI interpreter comes in and finds no MADT table, for the Multiprocessor info is stored as MPS table. No MADT table, so ACPI sets up the APIC in PIC mode (which I wonder wether correct, but ok). As a result, the kernels pci_dev table tells us that the IRQs have not been remapped (i.e. all values less than 16), while the IRQs are actually mapped above 16.

All drivers of PCI cards claim the wrong IRQ line, and the end of story is timeouts while waiting for an IRQ that never comes.

Remark: I think it is strange, that the kernel actually says: "ACPI: Using PIC for interrupt routing", but it doesn't set up the PIC correctly (otherwise the APIC rerouting table would be reset or something).

Now, my big question his: how to fix this. It is possible to have some code in the kernel that does the same as "pci=noacpi", but what and where do I have to do the check, with what condition ?

1) In the ACPI code, when MADT is not present ? Problem here is that the MPS parser comes after the ACPI parser, so it isn't known yet that the MPS table is present.

2) In the MPS parser ? As soon as an I/O APIC is detected by MPS, tell ACPI not to touch the APIC ? You get acpi related code in non-acpi procedures then...

3) Somewhere else ? How early in the kernel boot process should this option be set ?

And an additional question: is "pci=noapic" the correct way to fix this ? It runs fine here, but maybe we should only touch the IRQ related part ? If so, how to do that ?

Please shoot... I found the problem, but this doesn't mean I understand the kernel :)

Jos

Hi,

If no Multiple APIC Description Table (MADT) is described, in this case the _PIC method can be used to tell the bios to return the right table (PIC or APIC routing table).

In this case, if the MPS table describes matches the ACPI APIC table (this is the case, because the ACPI APIC table is built from the MPS table), you do not need to remap all IRQs.

I am really new in Kernel so I can't help you too much, but I can help you on ACPI (I worked 5 years in Bios).

Hope it helps,

Yann

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Jos Hulzink
Sent: lundi 12 mai 2003 21:36
To: linux-kernel
Cc: Zwane Mwaikambo
Subject: [RFC] How to fix MPS 1.4 + ACPI behaviour ?

Hi,

(kernel: 2.5.69)

The conclusion of bug 699 is that some / all i386 SMP systems that use MPS 1.4
(and higher ? or all MPS versions ?), should boot with the "pci=noacpi" parameter to prevent IRQ problems.

What exactly happens: The MPS 1.4 interpreter causes PCI IRQs to be remapped
to IRQ 16 and higher, which is the desired behaviour. The ACPI interpreter
comes in and finds no MADT table, for the Multiprocessor info is stored as
MPS table. No MADT table, so ACPI sets up the APIC in PIC mode (which I wonder wether correct, but ok). As a result, the kernels pci_dev table tells
us that the IRQs have not been remapped (i.e. all values less than 16), while
the IRQs are actually mapped above 16.

All drivers of PCI cards claim the wrong IRQ line, and the end of story is
timeouts while waiting for an IRQ that never comes.

Remark: I think it is strange, that the kernel actually says: "ACPI:
Using PIC
for interrupt routing", but it doesn't set up the PIC correctly (otherwise
the APIC rerouting table would be reset or something).

Now, my big question his: how to fix this. It is possible to have some code in
the kernel that does the same as "pci=noacpi", but what and where do I have
to do the check, with what condition ?

1) In the ACPI code, when MADT is not present ? Problem here is that the MPS
parser comes after the ACPI parser, so it isn't known yet that the MPS table
is present.

2) In the MPS parser ? As soon as an I/O APIC is detected by MPS, tell ACPI
not to touch the APIC ? You get acpi related code in non-acpi procedures

then...

3) Somewhere else ? How early in the kernel boot process should this option be
set ?

And an additional question: is "pci=noapic" the correct way to fix this ? It
runs fine here, but maybe we should only touch the IRQ related part ? If so,
how to do that ?

Please shoot... I found the problem, but this doesn't mean I understand the
kernel :)

Jos
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

So, it's more or less a bug in the ACPI code that should do some things when no MADT is dectected ? Or do I understand you wrong ?

Jos

For me the linux kernel should invoke the _PIC method with the right parameter.

Look at the specification Section

<< ================ Start ================
5.8.1 _PIC Method
The \_PIC optional method is to report to the BIOS the current interrupt model used by the OS. This
control method returns nothing. The argument passed into the method signifies the interrupt model OSPM
has chosen, PIC mode, APIC mode, or SAPIC mode. Notice that calling this method is optional for OSPM.
If the method is never called, the BIOS must assume PIC mode. It is important that the BIOS save the value
passed in by OSPM for later use during wake operations. _PIC(x):
_PIC(0) => PIC Mode
_PIC(1) => APIC Mode
_PIC(2) => SAPIC Mode
_PIC(3-n) => Reserved

==================== End ====================>

==> No MADT table, so ACPI sets up the APIC in PIC mode (which I wonder wether correct, but ok)
For me the kernel should invoke the _PIC method with the right parameter, in this case the ACPI module will receive the right table during the _PRT

Bios ASL code:

                Method(_PRT)
                {
                If(\PICF) <== internal variable set by _PIC
                {   
                	==> Returning APIC Mode
                	Return(APIC)
                }
                Else 
                {
                	==> Returning PIC Mode
                	Return(PICM)
                }
			}

Regards,

Yann

-----Original Message-----
From: Jos Hulzink [mailto:[email protected]]
Sent: lundi 12 mai 2003 22:51
To: STK; 'linux-kernel'
Cc: 'Zwane Mwaikambo'
Subject: Re: [RFC] How to fix MPS 1.4 + ACPI behaviour ?

So, it's more or less a bug in the ACPI code that should do some things when
no MADT is dectected ? Or do I understand you wrong ?

Jos

Jos, if you switch to IOAPIC mode after failing you're going to have to reprogram the IOAPIC again as specified in the mptables, setup_IOAPIC_irqs does this. As well as do some other setup which ACPI let you get past in arch/i386/kernel/mpparse.c. In a nutshell it'll get very convoluted quickly.

Zwane

function.linuxpower.ca

Hi,

As you can see below, current ACPI implementation concludes it has to use the PIC, before it finds the PRT. No occurences of PRT in dmesg before this one.

vvvv
ACPI: Using PIC for interrupt routing
^^^^
ACPI: PCI Interrupt Link [LNKA] (IRQs 3 4 *5 6 7 9 10 11 12 14 15)
ACPI: PCI Interrupt Link [LNKB] (IRQs 3 4 5 6 *7 9 10 11 12 14 15)
ACPI: PCI Interrupt Link [LNKC] (IRQs 3 4 5 6 7 9 *10 11 12 14 15)
ACPI: PCI Interrupt Link [LNKD] (IRQs 3 4 5 6 7 9 10 *11 12 14 15)
ACPI: PCI Root Bridge [PCI0] (00:00)
PCI: Probing PCI hardware (bus 00)
vvvv
ACPI: PCI Interrupt Routing Table [\_SB_.PCI0._PRT] ^^^^

Jos

From: Jos Hulzink <[email protected]>
Date: Mon, May 12, 2003 at 09:35:53PM +0200

Is this with or without IOAPIC? I got some problems with MPS 1.4, acpi and the local ioapic on a uniprocessor system, see bugzilla 678. I think it's a different problem, though.

Jurriaan

Evayne's smile turned sharply inward, although it remained on her face,
changed in tone and texture. "It loses none of its power," she whispered,
"and all of its romance."
	Michelle West - Hunter's Death
Debian (Unstable) GNU/Linux 2.5.69 4112 bogomips load av: 0.64 0.16 0.05

Your problem looks the same, though isn't, for your kernel finds a MADT:

ACPI: RSDP (v000 KT400 ) @ 0x000f74a0
ACPI: RSDT (v001 KT400 AWRDACPI 16944.11825) @ 0x3fff3000
ACPI: FADT (v001 KT400 AWRDACPI 16944.11825) @ 0x3fff3040 vvvvv
ACPI: MADT (v001 KT400 AWRDACPI 16944.11825) @ 0x3fff71c0 ^^^^^
ACPI: DSDT (v001 KT400 AWRDACPI 00000.04096) @ 0x00000000
ACPI: BIOS passes blacklist

And ACPI uses the IOAPIC, as expected:

ACPI: Interpreter enabled
ACPI: Using IOAPIC for interrupt routing

Unfortunately, we're talking two different bugs :(

Jos