mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* Regression?  2.6.0-test4{,-mm5} SMP interrupt distribution according to /proc/interrupts
@ 2003-09-05 11:36 Robert Davies
  0 siblings, 0 replies; only message in thread
From: Robert Davies @ 2003-09-05 11:36 UTC (permalink / raw)
  To: linux-kernel

Please personally Cc: to me, for any additional information or responses to 
this post, I am not subscribed to kernel list.

Hardware is dual Athlon-MP, AMD762 chipset for most part 2.6.0-test has 
performed well for me.  I've noticed however according to /proc/interrupts, 
CPU0 is getting all the interrupt service action (like when noapic is set), 
is this expected?

516rob@ash$ cat proc_interrupts-2.6.0-test4-mm5
+ uname -a
Linux oak 2.6.0-test4-mm5 #2 SMP Thu Sep 4 18:13:52 BST 2003 i686 AMD 
Athlon(tm) MP 2000+ AuthenticAMD GNU/Linux
+ cat /proc/interrupts
           CPU0       CPU1
  0:     354719         93    IO-APIC-edge  timer
  1:        457          1    IO-APIC-edge  i8042
  2:          0          0          XT-PIC  cascade
  8:         52          1    IO-APIC-edge  rtc
  9:          0          0   IO-APIC-level  acpi
 12:      20562          1    IO-APIC-edge  i8042
 14:       6053          1    IO-APIC-edge  ide0
 16:       7689          1   IO-APIC-level  ide2
 17:        203          1   IO-APIC-level  eth0, mga@PCI:1:5:0
 18:       2270          0   IO-APIC-level  EMU10K1
 19:          0          0   IO-APIC-level  ohci-hcd
NMI:          0          0
LOC:     354664     354663
ERR:          0
MIS:          0

The CPU1 numbers don't seem to change, except 0: and LOC:.  I have ACPI active 
in both Gentoo with vanilla 2.6.0-test, 2.6.0-test-mm5 and 2.4.22, and also 
under SuSE with their  2.4.20 kernel, 2.4 and 2.6 are consistent.

/proc/cmdline shows root=/dev/hde5 reboot=warm, using acpi=off option makes no 
change in 2.6.

Under 2.4 kernels the interrupts are distributed :

515rob@ash$ cat proc_interrupts-2.4*
           CPU0       CPU1
  0:      25148     137206    IO-APIC-edge  timer
  1:          6       1076    IO-APIC-edge  keyboard
  2:          0          0          XT-PIC  cascade
  8:         65        153    IO-APIC-edge  rtc
  9:          0          0   IO-APIC-level  acpi
 14:       6619      22404    IO-APIC-edge  ide0
 16:         12        373   IO-APIC-level  ide2
 17:     192602      49262   IO-APIC-level  eth0, eth1
 18:          0          0   IO-APIC-level  EMU10K1
 19:          0          0   IO-APIC-level  usb-ohci
NMI:          0          0
LOC:     162293     162274
ERR:          0
MIS:          0
           CPU0       CPU1
  0:      70700      62464    IO-APIC-edge  timer
  1:       2210       2328    IO-APIC-edge  keyboard
  2:          0          0          XT-PIC  cascade
  8:         92         91    IO-APIC-edge  rtc
  9:          0          0   IO-APIC-level  acpi
 14:       2652        934    IO-APIC-edge  ide0
 16:       8964       8723   IO-APIC-level  ide2
 17:       1057       1048   IO-APIC-level  eth0
NMI:          0          0
LOC:     133097     133096
ERR:          0
MIS:          0

Regards Rob


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

only message in thread, other threads:[~2003-09-05 11:36 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-09-05 11:36 Regression? 2.6.0-test4{,-mm5} SMP interrupt distribution according to /proc/interrupts Robert Davies

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

all inboxes | Powered by JetHome®