aboutsummaryrefslogtreecommitdiffstats
path: root/Documentation/i386/IO-APIC.txt
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@ppc970.osdl.org>2005-04-16 18:20:36 -0400
committerLinus Torvalds <torvalds@ppc970.osdl.org>2005-04-16 18:20:36 -0400
commit1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 (patch)
tree0bba044c4ce775e45a88a51686b5d9f90697ea9d /Documentation/i386/IO-APIC.txt
Linux-2.6.12-rc2v2.6.12-rc2
Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
Diffstat (limited to 'Documentation/i386/IO-APIC.txt')
-rw-r--r--Documentation/i386/IO-APIC.txt117
1 files changed, 117 insertions, 0 deletions
diff --git a/Documentation/i386/IO-APIC.txt b/Documentation/i386/IO-APIC.txt
new file mode 100644
index 000000000000..435e69e6e9aa
--- /dev/null
+++ b/Documentation/i386/IO-APIC.txt
@@ -0,0 +1,117 @@
1Most (all) Intel-MP compliant SMP boards have the so-called 'IO-APIC',
2which is an enhanced interrupt controller, it enables us to route
3hardware interrupts to multiple CPUs, or to CPU groups.
4
5Linux supports all variants of compliant SMP boards, including ones with
6multiple IO-APICs. (multiple IO-APICs are used in high-end servers to
7distribute IRQ load further).
8
9There are (a few) known breakages in certain older boards, which bugs are
10usually worked around by the kernel. If your MP-compliant SMP board does
11not boot Linux, then consult the linux-smp mailing list archives first.
12
13If your box boots fine with enabled IO-APIC IRQs, then your
14/proc/interrupts will look like this one:
15
16 ---------------------------->
17 hell:~> cat /proc/interrupts
18 CPU0
19 0: 1360293 IO-APIC-edge timer
20 1: 4 IO-APIC-edge keyboard
21 2: 0 XT-PIC cascade
22 13: 1 XT-PIC fpu
23 14: 1448 IO-APIC-edge ide0
24 16: 28232 IO-APIC-level Intel EtherExpress Pro 10/100 Ethernet
25 17: 51304 IO-APIC-level eth0
26 NMI: 0
27 ERR: 0
28 hell:~>
29 <----------------------------
30
31some interrupts are still listed as 'XT PIC', but this is not a problem,
32none of those IRQ sources is performance-critical.
33
34
35in the unlikely case that your board does not create a working mp-table,
36you can use the pirq= boot parameter to 'hand-construct' IRQ entries. This
37is nontrivial though and cannot be automated. One sample /etc/lilo.conf
38entry:
39
40 append="pirq=15,11,10"
41
42the actual numbers depend on your system, on your PCI cards and on their
43PCI slot position. Usually PCI slots are 'daisy chained' before they are
44connected to the PCI chipset IRQ routing facility (the incoming PIRQ1-4
45lines):
46
47 ,-. ,-. ,-. ,-. ,-.
48 PIRQ4 ----| |-. ,-| |-. ,-| |-. ,-| |--------| |
49 |S| \ / |S| \ / |S| \ / |S| |S|
50 PIRQ3 ----|l|-. `/---|l|-. `/---|l|-. `/---|l|--------|l|
51 |o| \/ |o| \/ |o| \/ |o| |o|
52 PIRQ2 ----|t|-./`----|t|-./`----|t|-./`----|t|--------|t|
53 |1| /\ |2| /\ |3| /\ |4| |5|
54 PIRQ1 ----| |- `----| |- `----| |- `----| |--------| |
55 `-' `-' `-' `-' `-'
56
57every PCI card emits a PCI IRQ, which can be INTA,INTB,INTC,INTD:
58
59 ,-.
60 INTD--| |
61 |S|
62 INTC--|l|
63 |o|
64 INTB--|t|
65 |x|
66 INTA--| |
67 `-'
68
69These INTA-D PCI IRQs are always 'local to the card', their real meaning
70depends on which slot they are in. If you look at the daisy chaining diagram,
71a card in slot4, issuing INTA IRQ, it will end up as a signal on PIRQ2 of
72the PCI chipset. Most cards issue INTA, this creates optimal distribution
73between the PIRQ lines. (distributing IRQ sources properly is not a
74necessity, PCI IRQs can be shared at will, but it's a good for performance
75to have non shared interrupts). Slot5 should be used for videocards, they
76do not use interrupts normally, thus they are not daisy chained either.
77
78so if you have your SCSI card (IRQ11) in Slot1, Tulip card (IRQ9) in
79Slot2, then you'll have to specify this pirq= line:
80
81 append="pirq=11,9"
82
83the following script tries to figure out such a default pirq= line from
84your PCI configuration:
85
86 echo -n pirq=; echo `scanpci | grep T_L | cut -c56-` | sed 's/ /,/g'
87
88note that this script wont work if you have skipped a few slots or if your
89board does not do default daisy-chaining. (or the IO-APIC has the PIRQ pins
90connected in some strange way). E.g. if in the above case you have your SCSI
91card (IRQ11) in Slot3, and have Slot1 empty:
92
93 append="pirq=0,9,11"
94
95[value '0' is a generic 'placeholder', reserved for empty (or non-IRQ emitting)
96slots.]
97
98generally, it's always possible to find out the correct pirq= settings, just
99permute all IRQ numbers properly ... it will take some time though. An
100'incorrect' pirq line will cause the booting process to hang, or a device
101won't function properly (if it's inserted as eg. a module).
102
103If you have 2 PCI buses, then you can use up to 8 pirq values. Although such
104boards tend to have a good configuration.
105
106Be prepared that it might happen that you need some strange pirq line:
107
108 append="pirq=0,0,0,0,0,0,9,11"
109
110use smart try-and-err techniques to find out the correct pirq line ...
111
112good luck and mail to linux-smp@vger.kernel.org or
113linux-kernel@vger.kernel.org if you have any problems that are not covered
114by this document.
115
116-- mingo
117