2005-03-03 21:23:37

by Marcelo Tosatti

[permalink] [raw]
Subject: [[email protected]: Kernel 2.4.28 can't boot into OS without noapic]

Hi Thomas,

I'm forwarding your message to Mikael and Len, who have knowledge
on the IOAPIC infrastructure.

----- Forwarded message from [email protected] -----

From: [email protected]
Date: Wed, 2 Mar 2005 13:37:03 +0800
To: [email protected]
Subject: Kernel 2.4.28 can't boot into OS without noapic

Hi,

Sorry to interrupt you. I'm BIOS engineer from Taiwan. I found an kernel IO-APIC
bug in my current project.
The BUS id report in MPTable was miss interpreter by kernel. For more detail
information please reference below description.

System architecture

+-----------+ HT LINK +-------+ HT LINK +-------+
| nvidia +---------------+ CPU0 +-----------+ CPU1 |
| CK804 | (bus 80) +---+---+ +-------+
+-----------+ | (HT LINK)
+---+------+
| AMD 8131 | (bus 0)
+---+------+
|
+---+------+
| AMD 8111 | (bus 1)
+----------+


Kernel 2.4.xx will parsing MPTable and contruct a table as below.
It is ok for signal PCI bus system. But in Multiple PCI bus system BIOS
will using actual PCI bus number as the bus ID. In this case bus number
and array index not equivalent. There is no problem to create this array.
But when boot with SMP kernel IO-APIC code 'pin_2_irq' will using bus ID
as the array index to reference 'mp_bus_id_to_type' array. It will
get incorrect bus type thus irq pin will not programming and system may go
suspend by peripheral driver never get it interrupt signal.
This problem can easily solve by increase 'MAX_MP_BUSSES' but it will waste
kernel memory size. Or we can correct 'smp_read_mpc' to support
multiple PCI bus ID assignment.


+---------------+
| |
+---------------+
| |
: :
| |
+---------------+ <- mp_bus_id_to_type, bus_data
| bus0 |
| bus1 |
| bus2 | ? mp_bus_id_to_type[80]
| bus3 |
| bus80 |
| bus81 |
+---------------+ <- mp_bus_id_to_node
| |
| |
| |
+---------------+ <- mp_bus_id_to_local
| |
| |
| |
+---------------+ <- mp_bus_id_to_pci_bus
| |
| |
| |
+---------------+ <- mp_irqs
| |
| |
| |
+---------------+


Here is MP spec about multiple PCI bus section.
D.2 Bus Entries in Systems with More Than One PCI Bus
To accommodate systems with more than one PCI bus within the confines of version
1.1 of this
specification, construction of the bus entries on the MP configuration table
must be handled in a
very particular sequence:
1. Begin with bus entries for the PCI buses. Start at bus zero, using the actual
PCI bus number as
the bus ID for the bus entry.
2. Add entries for other buses. These entries can use bus ID numbers left vacant
by the PCI bus
entries.
This sequence implies that bus ID numbers do not have to increase sequentially
by increments of
one; the requirement is that they must appear in ascending order by bus ID
number. This specific
interpretation of the information presented in Table 4-7 ensures consistency
between the
information in the MP configuration table and the model for systems with
multiple PCI buses that
is presented in the formal PCI specification, which allows for more flexibility
in bus numbering.
This numbering scheme requires bus entries in the MP configuration table to be
sorted
appropriately. For example, bus entries should appear in the order PCI (0), EISA
(1), and PCI(4)
in a system with three buses, two PCI buses numbered 0 and 4, and a single EISA
bus numbered
as 1.


Best Regard
Thomas CJ Chang
Office - 8691-2319
Mobile - 0933825636

----- End forwarded message -----


2005-03-03 22:19:54

by Brown, Len

[permalink] [raw]
Subject: Re: [[email protected]: Kernel 2.4.28 can't boot into OS without noapic]

On Thu, 2005-03-03 at 11:27, Marcelo Tosatti wrote:
> Hi Thomas,
>
> I'm forwarding your message to Mikael and Len, who have knowledge
> on the IOAPIC infrastructure.
>
> ----- Forwarded message from [email protected] -----
>
> From: [email protected]
> Date: Wed, 2 Mar 2005 13:37:03 +0800
> To: [email protected]
> Subject: Kernel 2.4.28 can't boot into OS without noapic
>
> Hi,
>
> Sorry to interrupt you. I'm BIOS engineer from Taiwan. I found an
> kernel IO-APIC bug in my current project...

Thomas,
Is this specific to 2.4, or does 2.6 fail also?
Is this specific to MPS mode, or does ACPI mode fail also?

If this is a publically available platform, can you identify it so maybe
we can corrolate this with other reports?

thanks,
-Len