I have been one of these people who have been having to boot with
pci=noacpi to get up with much of my hardware initialized. My system is
now working without it. It isn't getting shutoff on irq storms or
anything.
My only possible problem is this:
13:59:40 up 8 min, 3 users, load average: 0.86, 0.81, 0.36
CPU0
0: 516847 XT-PIC timer
I am not sure how fast the irq's for the timer should be going up. So,
that may be an issue.
I am aware some of these changes are causing problems for others, but
they fixed the mess for me.
Thank you.
Trever
--
"He who chops his own wood, is warm twice." -- Abraham Lincoln
On Mon, 25 Aug 2003, Trever L. Adams wrote:
> I have been one of these people who have been having to boot with
> pci=noacpi to get up with much of my hardware initialized. My system is
> now working without it. It isn't getting shutoff on irq storms or
> anything.
>
> My only possible problem is this:
>
> 13:59:40 up 8 min, 3 users, load average: 0.86, 0.81, 0.36
> CPU0
> 0: 516847 XT-PIC timer
>
8 mins ~ 8 * 60 = 480 seconds
516847 / 480 = 1076...
Looks like your HZ value is 1024. If so, everything is fine.
Cheers,
Dick Johnson
Penguin : Linux version 2.4.20 on an i686 machine (797.90 BogoMips).
Note 96.31% of all statistics are fiction.
On Monday 25 August 2003 19:00, Trever L. Adams wrote:
> I have been one of these people who have been having to boot with
> pci=noacpi to get up with much of my hardware initialized. My system is
> now working without it. It isn't getting shutoff on irq storms or
> anything.
Likewise, my EPoX 8RDA+ board is working 100% perfectly since the nforce2-apic
fixes were merged in -mm. No spurious interrupts, no weird ACPI glitches,
everything from power management to PCI IRQ routing is just fine.
I'm still not sure I understand the local apic lockups experienced by others,
but anybody considering the purchase of an nForce 2 board can probably put
their mind to rest.
>
> My only possible problem is this:
>
> 13:59:40 up 8 min, 3 users, load average: 0.86, 0.81, 0.36
> CPU0
> 0: 516847 XT-PIC timer
With the 1000Hz timer in linux 2.6, I'd guess your PC had been up for
516847/1000/60 = 8.61 minutes.. Oh, look!
>
> I am not sure how fast the irq's for the timer should be going up. So,
> that may be an issue.
>
Evidently not.
Cheers,
Alistair.
Instead of turning the apic off in bios like I'm doing, either smp code
or today's patch to turn apic off before on might do it for you. I have
to do something with apic to get my pci netcard to work and not get "irq
disabled". pci=noacpi wasn't it for me. Turning apic off in bios or
using smp code on uniprocessor is not all, I disable everything to get
errors and lockups to stop. Power management works. UDMA6 and irq
unmasking and ide options work.
Alistair J Strachan wrote:
>On Monday 25 August 2003 19:00, Trever L. Adams wrote:
>
>
>>I have been one of these people who have been having to boot with
>>pci=noacpi to get up with much of my hardware initialized. My system is
>>now working without it. It isn't getting shutoff on irq storms or
>>anything.
>>
>>
>Likewise, my EPoX 8RDA+ board is working 100% perfectly since the nforce2-apic
>fixes were merged in -mm. No spurious interrupts, no weird ACPI glitches,
>everything from power management to PCI IRQ routing is just fine.
>
>BD> I have to turn the APIC off in BIOS, then linux turns it on and my rt8139c pci netcard will work. If I turn off parallel and serial ports and usb as well, and remove cd drives, I don't get lockups or see netdev watchdog failure to transmit and disable irq7 along with netcard failure. Needless to say the onboard ethernet controller and audio don't work. But I have another reliable textmode file server with an MSI nforce2 board. There are a few "correctable errors" trapped with slightly underclocked overtested csl2 pc3200 ram running at 2 channel 166 = 333. Altogether a very fine agp8 textmode linux box. Ecstatically, I can copy and paste script text with a mouse, but of course it's on a ps2 port, not usb. USB code, when not logging errors, doesn't recognize usb mouse or keyboard either booted or hot-plugged. I can hibernate with textmode agp8, writing scripts, while somebody hacks nforce2 and nvidia geforce 4200ti--no, I will not dump 2.6 for a 2.4 kernel patched with nvidia driver--more important things don't work in 2.4 than 2.6! 6-channel surround silence and frozen pizza...with the cd's pulled out md might not lockup when copies overflow the drive buffer cache!
>