public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* 2.4.23, 2.4.24-pre: Interrupt balancing problem
@ 2003-12-23 12:48 Andreas Haumer
  0 siblings, 0 replies; only message in thread
From: Andreas Haumer @ 2003-12-23 12:48 UTC (permalink / raw)
  To: linux-kernel@vger.kernel.org

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi!

Interrupt balancing doesn't work anymore on an Asus Dual
Xeon Server (AP1700-S5, ASUS PR-DLS533 MB, Dual Xeon HT).

Last kernel where this worked was 2.4.21-ac:

 cat /proc/interrupts
           CPU0       CPU1       CPU2       CPU3
  0:   32414607   32353905   32486435   32470963    IO-APIC-edge  timer
  1:          1          0          0          1    IO-APIC-edge  keyboard
  2:          0          0          0          0          XT-PIC  cascade
 15:          1          0          0          1    IO-APIC-edge  ide1
 18:   22702419   22631022   22668864   22843971   IO-APIC-level  eth0
 22:   15021656   15223388   15112963   14929184   IO-APIC-level  ioc0
 23:          9          7         15         11   IO-APIC-level  ioc1
NMI:          0          0          0          0
LOC:  129727116  129727115  129727114  129727114
ERR:          0
MIS:          0

Note that on this particular hardware, I have to boot with
"acpi=off" or "pci=noacpi" in order to get the Fusion MPT
hardware recognized (see my other bug report a few minutes ago)

I didn't check 2.4.22, but with 2.4.23 and 2.4.24-pre2
I get something like the following:

Booted with "acpi=off":
root@tolstoi:~ {502} $ cat /proc/interrupts
           CPU0       CPU1       CPU2       CPU3
  0:      41424          0          0          0    IO-APIC-edge  timer
  1:          2          0          0          0    IO-APIC-edge  keyboard
  2:          0          0          0          0          XT-PIC  cascade
 11:          0          0          0          0   IO-APIC-level  usb-ohci
 18:       8235          0          0          0   IO-APIC-level  eth0
 22:      34704          0          0          0   IO-APIC-level  ioc0
 23:         42          0          0          0   IO-APIC-level  ioc1
NMI:          0          0          0          0
LOC:      41271      41270      41270      41269
ERR:          0
MIS:          0


Booted with "pci=noacpi":
root@tolstoi:~ {502} $ cat /proc/interrupts
           CPU0       CPU1       CPU2       CPU3
  0:      89806          0          0          0    IO-APIC-edge  timer
  1:          2          0          0          0    IO-APIC-edge  keyboard
  2:          0          0          0          0          XT-PIC  cascade
  9:          0          0          0          0          XT-PIC  acpi
 11:          0          0          0          0   IO-APIC-level  usb-ohci
 18:      12481          0          0          0   IO-APIC-level  eth0
 22:      12518          0          0          0   IO-APIC-level  ioc0
 23:         42          0          0          0   IO-APIC-level  ioc1
NMI:          0          0          0          0
LOC:      89651      89650      89650      89650
ERR:          0
MIS:          0


I also tried boot options "noirqbalance" and "acpi_irq_balance",
but interrupts were still served on CPU0 only (SMP is working,
though: I can build the kernel with "make -j4" and as "top" tells
all four logical CPU's are used perfectly well)

Bug or feature?

- - andreas

- --
Andreas Haumer                     | mailto:andreas@xss.co.at
*x Software + Systeme              | http://www.xss.co.at/
Karmarschgasse 51/2/20             | Tel: +43-1-6060114-0
A-1100 Vienna, Austria             | Fax: +43-1-6060114-71
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQE/6DmtxJmyeGcXPhERAtRnAJ0aJwcH0BL2yY2o9hz1rNU4Nmh9DgCfeUHc
LRlp1wMHcRRPd25q0LiQk2c=
=13fn
-----END PGP SIGNATURE-----


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2003-12-23 12:48 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-12-23 12:48 2.4.23, 2.4.24-pre: Interrupt balancing problem Andreas Haumer

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox