Making sense of ACPI(CA)

rdos · Post by **rdos** » Wed Dec 14, 2011 12:46 pm

OK, so I have ACPICA working (with debug support), but the interface is cludgy and not delivering what I want.

What I basically want is:

1. A function to map a PCI address to an IRQ on IO-APIC.

2. A function which returns a list of devices, not with callbacks, but with an index #, and then allows easy retrieval of IRQ, IO-range and memory-range.

3. A function that returns a list of PCI-devices, with an index #, and that allows easy retrieval of IRQs and possibly IO and memory-ranges

4. A function that returns a description of a device

What I get is:

1. Name-space walks
2. Retrieval of resource-objects in ways that are not readable in the debugger (nor easily dereferenced in C either).

So, what I have done is:

1. Created my own object-vector with a name-space walk
2. Created my own device-vector with a name-space walk
3. Created the resource descriptions by using a number of lists of real objects instead of worthless unions.
4. Created my own PCI irq-vector with a name-space walk

What I will do shortly is:

1. Add a command-line option to display devices (once I have the RdosGetDevice), along with description, IRQ and IO/memory ranges
2. Add a command-line option to display PCI-device, along with bus, device, function, IRQ and IO/memory ranges

However, I still don't think I properly understand some central concepts in ACPICA.

For instance:

1. How do you find the PCI device(s)?

2. Which methods are typically used/required with which devices? My listing of namespace objects is huge, but how do I know that my next machine has the same ones? Especially since neither the ACPICA document, nor the ACPI document, is clear on how typical devices should be represented.

3. Is there any use in reading-out the base address for keyboard controller, PIT, PIC, PCI? Wouldn't any new design that changed these settings be so broken that nobody would want it? Isn't the chance bigger that any deviation in these values would be errors in the ACPI tables rather than a genuine difference?

rdos · Post by **rdos** » Fri Dec 23, 2011 12:36 am

I just found out that the reason one of my machines doesn't boot properly is because of bugs in ACPICA. It trashes memory by using unitialized memory when there is buggy AML-code. They simply cannot have tested it against various buggy ACPIs. I discovered this by changing the allocator to allocate one GDT per request.

Owen · Post by **Owen** » Fri Dec 23, 2011 3:10 am

rdos wrote:I just found out that the reason one of my machines doesn't boot properly is because of bugs in ACPICA. It trashes memory by using unitialized memory when there is buggy AML-code. They simply cannot have tested it against various buggy ACPIs. I discovered this by changing the allocator to allocate one GDT per request.

ACPICA is used by Linux, FreeBSD, Solaris, and various other operating systems. I know how much time the Linux developers have spent dealing with AML bugs; I somehow suspect have bugs your usage of ACPICA, or in your compiler are more probable than bugs in ACPICA itself.

rdos · Post by **rdos** » Fri Dec 23, 2011 3:38 am

Owen wrote:
rdos wrote:I just found out that the reason one of my machines doesn't boot properly is because of bugs in ACPICA. It trashes memory by using unitialized memory when there is buggy AML-code. They simply cannot have tested it against various buggy ACPIs. I discovered this by changing the allocator to allocate one GDT per request.
ACPICA is used by Linux, FreeBSD, Solaris, and various other operating systems. I know how much time the Linux developers have spent dealing with AML bugs; I somehow suspect have bugs your usage of ACPICA, or in your compiler are more probable than bugs in ACPICA itself.

It is possible it is some problem in the environment. The most probable cause of the problem is that Next pointer in the WalkState variable contains junk. It is when code follows this pointer it does random writes. Since it is this type of problem, the code sometimes actually works (when the junk is the "right" junk). When the code is run at boot-time it seems like it corrupts some important data, and thus everything simply stops.

I'm not sure if the problem would be solved if all allocated memory blocks were zeroed out, but it is one possibilty.

This description from the ACPICA reference manual about AcpiOsAllocate makes it most likely that it is the AML code itself that is at error:

This function dynamically allocates memory. The returned memory is not assumed to be initialized
to any particular value or values.

IOW, the OS should not need to zero-out allocated memory-blocks, but this is left-up to ACPICA itself to ensure.

Cognition · Post by **Cognition** » Sat Dec 24, 2011 1:45 am

ACPICA basically implements it's own calloc call by calling AcpiOsAllocate and then an internal memset. Personally I'd put a write watch on next field of the WalkState and find out what exactly is trashing it's values and go on from there.

rdos · Post by **rdos** » Sat Dec 24, 2011 5:01 pm

OK, so I have managed to collect a list of PCI devices from ACPI. For my 2-core AMD it looks like this:

Code: Select all

\_SB_.PCI0.SMB0, Bus: 0, Device: 20, Function: 0
\_SB_.PCI0.USB0, Bus: 0, Device: 19, Function: 0
\_SB_.PCI0.USB1, Bus: 0, Device: 19, Function: 1
\_SB_.PCI0.USB2, Bus: 0, Device: 19, Function: 2
\_SB_.PCI0.USB3, Bus: 0, Device: 19, Function: 3
\_SB_.PCI0.USB4, Bus: 0, Device: 19, Function: 4
\_SB_.PCI0.USB5, Bus: 0, Device: 19, Function: 5
\_SB_.PCI0.AZAL, Bus: 0, Device: 20, Function: 2
\_SB_.PCI0.LPC0, Bus: 0, Device: 20, Function: 3
\_SB_.PCI0.P2P_, Bus: 0, Device: 20, Function: 4
\_SB_.PCI0.IDE_, Bus: 0, Device: 20, Function: 1
\_SB_.PCI0.SAT0, Bus: 0, Device: 18, Function: 0
\_SB_.PCI0.PCE2, Bus: 0, Device: 2, Function: 0
\_SB_.PCI0.PCE3, Bus: 0, Device: 3, Function: 0
\_SB_.PCI0.PCE4, Bus: 0, Device: 4, Function: 0
\_SB_.PCI0.PCE5, Bus: 0, Device: 5, Function: 0
\_SB_.PCI0.PCE6, Bus: 0, Device: 6, Function: 0
\_SB_.PCI0.PCE7, Bus: 0, Device: 7, Function: 0
\_SB_.PCI0.PCE8, Bus: 0, Device: 8, Function: 0
\_SB_.PCI0.AGP_, Bus: 0, Device: 1, Function: 0

Now I only need to determined their IRQs.

rdos · Post by **rdos** » Sun Dec 25, 2011 3:21 pm

Updated list with IRQs for the 4 possible PCI pins:

Code: Select all

\_SB_.PCI0.SMB0, Bus: 0, Device: 20, Function: 0, IRQ: 3, 11, 5, 10
\_SB_.PCI0.USB0, Bus: 0, Device: 19, Function: 0, IRQ: 3, 11, 5, 10
\_SB_.PCI0.USB1, Bus: 0, Device: 19, Function: 1, IRQ: 3, 11, 5, 10
\_SB_.PCI0.USB2, Bus: 0, Device: 19, Function: 2, IRQ: 3, 11, 5, 10
\_SB_.PCI0.USB3, Bus: 0, Device: 19, Function: 3, IRQ: 3, 11, 5, 10
\_SB_.PCI0.USB4, Bus: 0, Device: 19, Function: 4, IRQ: 3, 11, 5, 10
\_SB_.PCI0.USB5, Bus: 0, Device: 19, Function: 5, IRQ: 3, 11, 5, 10
\_SB_.PCI0.AZAL, Bus: 0, Device: 20, Function: 2, IRQ: 3, 11, 5, 10
\_SB_.PCI0.LPC0, Bus: 0, Device: 20, Function: 3, IRQ: 3, 11, 5, 10
\_SB_.PCI0.P2P_, Bus: 0, Device: 20, Function: 4, IRQ: 3, 11, 5, 10
\_SB_.PCI0.IDE_, Bus: 0, Device: 20, Function: 1, IRQ: 3, 11, 5, 10
\_SB_.PCI0.SAT0, Bus: 0, Device: 18, Function: 0, IRQ: 11, 11, 11, 11
\_SB_.PCI0.PCE2, Bus: 0, Device: 2, Function: 0, IRQ: 5, 5, 5, 5
\_SB_.PCI0.PCE3, Bus: 0, Device: 3, Function: 0, IRQ: 10, 10, 10, 10
\_SB_.PCI0.PCE4, Bus: 0, Device: 4, Function: 0, IRQ: 3, 3, 3, 3
\_SB_.PCI0.PCE5, Bus: 0, Device: 5, Function: 0, IRQ: 11, 11, 11, 11
\_SB_.PCI0.PCE6, Bus: 0, Device: 6, Function: 0, IRQ: 5, 5, 5, 5
\_SB_.PCI0.PCE7, Bus: 0, Device: 7, Function: 0, IRQ: 10, 10, 10, 10
\_SB_.PCI0.PCE8, Bus: 0, Device: 8, Function: 0, IRQ: 3, 3, 3, 3
\_SB_.PCI0.AGP_, Bus: 0, Device: 1, Function: 0

However, this does not correspond to the interrupt settings in Windows XP! OTOH, they don't conflict with anything either, so maybe those are valid defaults?

My new 6-core AMD looks like this (it has two IO-APICs, but the default routing doesn't seem to use any IO-APIC specific ints either):

Code: Select all

\_SB_.PCI0.SMBS, Bus: 0, Device: 20, Function: 0, IRQ: 10, 7, 7, 3
\_SB_.PCI0.IDEC, Bus: 0, Device: 20, Function: 1, IRQ: 10, 7, 7, 3
\_SB_.PCI0.SBAZ, Bus: 0, Device: 20, Function: 2, IRQ: 10, 7, 7, 3
\_SB_.PCI0.SBRG, Bus: 0, Device: 20, Function: 3, IRQ: 10, 7, 7, 3
\_SB_.PCI0.P0PC, Bus: 0, Device: 20, Function: 4, IRQ: 10, 7, 7, 3
\_SB_.PCI0.UHC1, Bus: 0, Device: 18, Function: 0, IRQ: 7, 7 
\_SB_.PCI0.UHC2, Bus: 0, Device: 18, Function: 2, IRQ: 7, 7 
\_SB_.PCI0.USB3, Bus: 0, Device: 19, Function: 0, IRQ: 11, 10 
\_SB_.PCI0.UHC4, Bus: 0, Device: 19, Function: 2, IRQ: 11, 10 
\_SB_.PCI0.USB5, Bus: 0, Device: 22, Function: 0, IRQ: 11, 11 
\_SB_.PCI0.UHC6, Bus: 0, Device: 22, Function: 2, IRQ: 11, 11 
\_SB_.PCI0.UHC7, Bus: 0, Device: 20, Function: 5, IRQ: 10, 7, 7, 3
\_SB_.PCI0.SATA, Bus: 0, Device: 17, Function: 0, IRQ: 3
\_SB_.PCI0.PE20, Bus: 0, Device: 21, Function: 0, IRQ: 10, 7, 7, 3
\_SB_.PCI0.PE21, Bus: 0, Device: 21, Function: 1, IRQ: 10, 7, 7, 3
\_SB_.PCI0.PE22, Bus: 0, Device: 21, Function: 2, IRQ: 10, 7, 7, 3
\_SB_.PCI0.PE23, Bus: 0, Device: 21, Function: 3, IRQ: 10, 7, 7, 3
\_SB_.PCI0.IOMA, Bus: 0, Device: 0, Function: 2, IRQ: 10 
\_SB_.PCI0.RD8A, Bus: 0, Device: 0, Function: 0, IRQ: 10 
\_SB_.PCI0.PC02, Bus: 0, Device: 2, Function: 0, IRQ: 11 
\_SB_.PCI0.PC03, Bus: 0, Device: 3, Function: 0, IRQ: 11 
\_SB_.PCI0.PC04, Bus: 0, Device: 4, Function: 0, IRQ: 11 
\_SB_.PCI0.PC05, Bus: 0, Device: 5, Function: 0, IRQ: 11 
\_SB_.PCI0.PC06, Bus: 0, Device: 6, Function: 0, IRQ: 10 
\_SB_.PCI0.PC07, Bus: 0, Device: 7, Function: 0, IRQ: 10 
\_SB_.PCI0.PC09, Bus: 0, Device: 9, Function: 0, IRQ: 10 
\_SB_.PCI0.PC0A, Bus: 0, Device: 10, Function: 0, IRQ: 11 
\_SB_.PCI0.PC0B, Bus: 0, Device: 11, Function: 0, IRQ: 11 
\_SB_.PCI0.PC0C, Bus: 0, Device: 12, Function: 0, IRQ: 11 
\_SB_.PCI0.PC0D, Bus: 0, Device: 13, Function: 0, IRQ: 11

rdos · Post by **rdos** » Mon Dec 26, 2011 4:27 pm

I know why it doesn't work. In order to enable APIC-mode in ACPICA, there is a need to execute a special method just before any interrupt information is retrieved.

It looks like this:

Code: Select all

    ACPI_OBJECT_LIST        Params;
    ACPI_OBJECT             Obj;

    Params.Count = 1;
    Params.Pointer = &Obj;
    
    Obj.Type = ACPI_TYPE_INTEGER;
    Obj.Integer.Value = 1;     // 0 = PIC, 1 = APIC

    AcpiEvaluateObject(NULL, "\\_PIC", &Params, NULL);

This code could be placed in the AcpiEnableSubsystem procedure just before initializing the SCI handler.

It's really strange that the documentation says nothing about this.

Owen · Post by **Owen** » Mon Dec 26, 2011 5:23 pm

Thats not ACPICA. Thats ACPI.

Seriously. This is all documented in the (admittedly large) ACPI specification. You have to tell the firmware what mode you require before it can tell you the interrupt map you want.

rdos · Post by **rdos** » Mon Dec 26, 2011 5:40 pm

Owen wrote:Thats not ACPICA. Thats ACPI.

Seriously. This is all documented in the (admittedly large) ACPI specification. You have to tell the firmware what mode you require before it can tell you the interrupt map you want.

Well, the major problem is that you need to insert this in the middle of the AcpiEnableSubsystem, because otherwise the SCI might be initialized with a PIC setting! You cannot do it before the AcpiEnableSubsystem, because then some hardware state is uninitialized.

rdos · Post by **rdos** » Thu Dec 29, 2011 12:31 pm

Some more progress in PCI IRQ identification, but it still doesn't give the same results as Windows XP.

Code: Select all

ACPI Name                     Vendor/dev Class  Bus  Dev Func  IRQ   WinXP
\_SB_.PCI0.SMBS               1002 4385  0C05     0   20    0   16   none (SMBus)
\_SB_.PCI0.IDEC               1002 439C  0C05     0   20    1   16   none (PCI IDE)
\_SB_.PCI0.SBAZ               1002 4383  0C05     0   20    2   16   16 (UAA HD audio)
\_SB_.PCI0.SBRG               1002 439D  0C05     0   20    3   16   none (ISA bridge)
\_SB_.PCI0.P0PC               1002 4384  0C05     0   20    4   16   none (PCI-PCI bridge)
\_SB_.PCI0.UHC1               1002 4397  0C03     0   18    0   18   18 (OHCI USB)
\_SB_.PCI0.UHC2               1002 4396  0C03     0   18    2   18   17 (EHCI USB)
\_SB_.PCI0.USB3               1002 4397  0C03     0   19    0   20   20 (OHCI USB)
\_SB_.PCI0.UHC4               1002 4396  0C03     0   19    2   20   21 (EHCI USB)
\_SB_.PCI0.USB5               1002 4397  0C03     0   22    0   22   22 (OHCI USB)
\_SB_.PCI0.UHC6               1002 4396  0C03     0   22    2   22   23 (EHCI USB)
\_SB_.PCI0.UHC7               1002 4399  0C05     0   20    5   16   18 (OHCI USB)
\_SB_.PCI0.SATA               1002 4390  0101     0   17    0    0   19 (PCI IDE)
\_SB_.PCI0.RD8A               1002 5A14  FFFF     0    0    0   55   none (PCI standard host bridge)
\_SB_.PCI0.PC02               1002 5A16  0604     0    2    0   52   52 (PCI-PCI bridge)
\_SB_.PCI0.PC04               1002 5A18  0604     0    4    0   52   52 (PCI-PCI bridge)
\_SB_.PCI0.PC05               1002 5A19  0604     0    5    0   52   52 (PCI-PCI bridge)
\_SB_.PCI0.PC06               1002 5A1A  0604     0    6    0   53   53 (PCI-PCI bridge)
\_SB_.PCI0.PC07               1002 5A1B  0604     0    7    0   53   53 (PCI-PCI bridge)
\_SB_.PCI0.PC0A               1002 5A1D  0604     0   10    0   54   54 (PCI-PCI bridge)

                                                  1    0    0        24 (ATI Radeon HD 5450)
                                                  1    0    1        25 (UAA HD Audio)
                                                  2    0    0        44 (Realtek PCIe)
                                                  3    0    0        46 (XHCI USB)
                                                  4    0    0        51 (KMicron SCSI)
                                                  5    0    0        50 (XHCI USB)
                                                  6    0    0        47 (KMicron SCSI)
                                                  7    6    0        21 (1394)

First, my code only seems to identify IRQs on bus 0 (the PCI root bus). Secondly, some of the USB interrupts differ between Windows XP and my code.

There must be some additional PCI devices (probably the PCI-PCI bridges), that has additional IRQ information for other bus numbers.

When the APIC is used, there are no link-devices it seems, rather the IRQ information is directly coded in the returned IRQ routing resource. One issue is that the IRQ routings are per-device (and not function), and specifies the pin-number instead of function number. When the pin-number is read-out from various PCI-devices, it sometimes seems undefined, or has a higher than allowed value. Maybe Windows XP assigns the pin-numbers such that different functions within a devices gets different pins (all zero-functions have the correct IRQs)? That could explain why the USB-devices does not have the same IRQ in my system compared to Windows (some of the non-zero functions differ). The big question is if programming the PCI pin number setting changes the IRQ-mapping directly, unlike programing the IRQ, which has no effect. Or do Windows XP do this in some other way?

rdos · Post by **rdos** » Thu Dec 29, 2011 3:26 pm

Solved the IRQ matching issue. I read the interrupt line instead of interrupt pin, which was not a good idea. I also clobbered the PCI function in the loop, which gave the wrong class-codes.

So, here is the correct PCI mapping from ACPI:

Code: Select all

ACPI Name                     Vendor/dev Class   Bus  Dev Func IRQ
\_SB_.PCI0.SMBS               1002 4385  0C05     0   20    0    0
\_SB_.PCI0.IDEC               1002 439C  0101     0   20    1   17
\_SB_.PCI0.SBAZ               1002 4383  0403     0   20    2   16
\_SB_.PCI0.SBRG               1002 439D  0601     0   20    3    0
\_SB_.PCI0.P0PC               1002 4384  0604     0   20    4    0
\_SB_.PCI0.UHC1               1002 4397  0C03     0   18    0   18
\_SB_.PCI0.UHC2               1002 4396  0C03     0   18    2   17
\_SB_.PCI0.USB3               1002 4397  0C03     0   19    0   20
\_SB_.PCI0.UHC4               1002 4396  0C03     0   19    2   21
\_SB_.PCI0.USB5               1002 4397  0C03     0   22    0   22
\_SB_.PCI0.UHC6               1002 4396  0C03     0   22    2   23
\_SB_.PCI0.UHC7               1002 4399  0C03     0   20    5   18
\_SB_.PCI0.SATA               1002 4390  0101     0   17    0   19
\_SB_.PCI0.RD8A               1002 5A14  0600     0    0    0    0
\_SB_.PCI0.PC02               1002 5A16  0604     0    2    0   52
\_SB_.PCI0.PC04               1002 5A18  0604     0    4    0   52
\_SB_.PCI0.PC05               1002 5A19  0604     0    5    0   52
\_SB_.PCI0.PC06               1002 5A1A  0604     0    6    0   53
\_SB_.PCI0.PC07               1002 5A1B  0604     0    7    0   53
\_SB_.PCI0.PC0A               1002 5A1D  0604     0   10    0   54

Now I only need to fetch the IRQ-mappings for other buses. I think how this is done is by inspecting the class-code of each PCI-device, and if it is 0604 it is a PCI-PCI bridge, and then read the secondary bus register in PCI configuration space, and the PCI IRQ entries from ACPI, assigning them to the new bus.

rdos · Post by **rdos** » Fri Dec 30, 2011 1:46 pm

New results from the PCI command-line:

Code: Select all

ACPI Name                     Vendor/dev Class  Bus  Dev Func  IRQ
\_SB_.PCI0.SMBS               1002 4385  0C05     0   20    0     
\_SB_.PCI0.IDEC               1002 439C  0101     0   20    1   17
\_SB_.PCI0.SBAZ               1002 4383  0403     0   20    2   16
\_SB_.PCI0.SBRG               1002 439D  0601     0   20    3     
\_SB_.PCI0.P0PC               1002 4384  0604     0   20    4     
\_SB_.PCI0.UHC1               1002 4397  0C03     0   18    0   18
\_SB_.PCI0.UHC2               1002 4396  0C03     0   18    2   17
\_SB_.PCI0.USB3               1002 4397  0C03     0   19    0   20
\_SB_.PCI0.UHC4               1002 4396  0C03     0   19    2   21
\_SB_.PCI0.USB5               1002 4397  0C03     0   22    0   22
\_SB_.PCI0.UHC6               1002 4396  0C03     0   22    2   23
\_SB_.PCI0.UHC7               1002 4399  0C03     0   20    5   18
\_SB_.PCI0.SATA               1002 4390  0101     0   17    0   19
\_SB_.PCI0.RD8A               1002 5A14  0600     0    0    0     
\_SB_.PCI0.PC02               1002 5A16  0604     0    2    0   52
\_SB_.PCI0.PC04               1002 5A18  0604     0    4    0   52
\_SB_.PCI0.PC05               1002 5A19  0604     0    5    0   52
\_SB_.PCI0.PC06               1002 5A1A  0604     0    6    0   53
\_SB_.PCI0.PC07               1002 5A1B  0604     0    7    0   53
\_SB_.PCI0.PC0A               1002 5A1D  0604     0   10    0   54
\_SB_.PCI0.PC06.JMB0          197B 2362  0101     4    0    0   51
\_SB_.PCI0.PC0A.JMB0          197B 2362  0101     6    0    0   47
PCI-PCI P0PC                  1409 7168  0700     7    5    0   20
PCI-PCI P0PC                  1106 3044  0C00     7    6    0   21
PCI-PCI PC02                  1002 68F9  0300     1    0    0   24
PCI-PCI PC04                  10EC 8168  0200     2    0    0   44
PCI-PCI PC05                  1B21 1042  0C03     3    0    0   46
PCI-PCI PC07                  1B21 1042  0C03     5    0    0   50

Now the extra buses are visible (the previous method to check for PCI-PCI bridge worked). However, only the SCI devices (JMB0) are available in ACPI configuration space. The other devices are not managed by ACPI. For instance, the RTL 8168 (bus 2) is integrated behind a PCI-PCI bridge on the motherboard, and is not enumrated by ACPI. The XHCI USB is also not in ACPI. The video card could not be part of ACPI since it is an add-on board.

The devices above that are preceeded with "PCI-PCI" are not listed as ACPI devices in the name-space, but has to be added as pseudo-devices instead in order to record them. They are listed in the _PRT table for the PCI-PCI bridge, but those entries only contain a device, not a function #. There is one missing PCI function above which Windows XP has. It has bus 1, device 0, function 1. I need to add additional code to handle non-zero functions for non-ACPI listed devices. I need to check that all the PCI-devices I have found are listed.

rdos · Post by **rdos** » Fri Dec 30, 2011 3:46 pm

This is the final list for the PCI-devices, and their IRQs on the 6-core AMD:

Code: Select all

ACPI Name                     Vendor/dev Class  Bus  Dev Func  IRQ
\_SB_.PCI0.SMBS               1002 4385  0C05     0   20    0     
\_SB_.PCI0.IDEC               1002 439C  0101     0   20    1   17
\_SB_.PCI0.SBAZ               1002 4383  0403     0   20    2   16
\_SB_.PCI0.SBRG               1002 439D  0601     0   20    3     
\_SB_.PCI0.P0PC               1002 4384  0604     0   20    4     
\_SB_.PCI0.UHC1               1002 4397  0C03     0   18    0   18
\_SB_.PCI0.UHC2               1002 4396  0C03     0   18    2   17
\_SB_.PCI0.USB3               1002 4397  0C03     0   19    0   20
\_SB_.PCI0.UHC4               1002 4396  0C03     0   19    2   21
\_SB_.PCI0.USB5               1002 4397  0C03     0   22    0   22
\_SB_.PCI0.UHC6               1002 4396  0C03     0   22    2   23
\_SB_.PCI0.UHC7               1002 4399  0C03     0   20    5   18
\_SB_.PCI0.SATA               1002 4390  0101     0   17    0   19
\_SB_.PCI0.RD8A               1002 5A14  0600     0    0    0     
\_SB_.PCI0.PC02               1002 5A16  0604     0    2    0   52
\_SB_.PCI0.PC04               1002 5A18  0604     0    4    0   52
\_SB_.PCI0.PC05               1002 5A19  0604     0    5    0   52
\_SB_.PCI0.PC06               1002 5A1A  0604     0    6    0   53
\_SB_.PCI0.PC07               1002 5A1B  0604     0    7    0   53
\_SB_.PCI0.PC0A               1002 5A1D  0604     0   10    0   54
\_SB_.PCI0.PC06.JMB0          197B 2362  0101     4    0    0   51
\_SB_.PCI0.PC0A.JMB0          197B 2362  0101     6    0    0   47
PCI-PCI P0PC                  1409 7168  0700     7    5    0   20
PCI-PCI P0PC                  1409 7168  0700     7    5    1   20
PCI-PCI P0PC                  1409 7168  0700     7    5    2   20
PCI-PCI P0PC                  1409 7168  0700     7    5    3   20
PCI-PCI P0PC                  1409 7168  0700     7    5    4   20
PCI-PCI P0PC                  1409 7168  0700     7    5    5   20
PCI-PCI P0PC                  1409 7168  0700     7    5    6   20
PCI-PCI P0PC                  1409 7168  0700     7    5    7   20
PCI-PCI P0PC                  1106 3044  0C00     7    6    0   21
PCI-PCI P0PC                  1106 3044  0C00     7    6    1   21
PCI-PCI P0PC                  1106 3044  0C00     7    6    2   21
PCI-PCI P0PC                  1106 3044  0C00     7    6    3   21
PCI-PCI P0PC                  1106 3044  0C00     7    6    4   21
PCI-PCI P0PC                  1106 3044  0C00     7    6    5   21
PCI-PCI P0PC                  1106 3044  0C00     7    6    6   21
PCI-PCI P0PC                  1106 3044  0C00     7    6    7   21
PCI-PCI PC02                  1002 68F9  0300     1    0    0   24
PCI-PCI PC02                  1002 AA68  0403     1    0    1   25
PCI-PCI PC04                  10EC 8168  0200     2    0    0   44
PCI-PCI PC05                  1B21 1042  0C03     3    0    0   46
PCI-PCI PC07                  1B21 1042  0C03     5    0    0   50
                              1022 1200  0600     0   24    0     
                              1022 1201  0600     0   24    1     
                              1022 1202  0600     0   24    2     
                              1022 1203  0600     0   24    3     
                              1022 1204  0600     0   24    4

Surpsingly, the motherboard has support for 8 serial ports, and also has 8 OHCI-controllers, 7 EHCI controllers and 2 XHCI controllers.

Using this information, it should be fairly easy to implement the GetPciIrq procedure that has an PCI-address as input, and an IRQ-number as output.

Edit: Now I now why global interrupt numbers up to 53 can be present in ACPI. This machine has a primary IO-APIC with 24 channels, and a secondary IO-APIC with 32 channels, making a total of 56 channels.

OSDev.org

Making sense of ACPI(CA)

Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)

Re: Making sense of ACPI(CA)