Re: [Timers SMP] can this machine be helped?

[Guennadi Liakhovetski <g.liakhovetski@gmx.de>]

On Mon, 3 Dec 2007, Pavel Machek wrote:

> On Mon 2007-12-03 22:45:06, Guennadi Liakhovetski wrote:
> > On Sun, 2 Dec 2007, Pavel Machek wrote:
> > 
> > > > I compiled a .24-ish kernel for it with CONFIG_NO_HZ and 
> > > > CONFIG_HIGH_RES_TIMERS. To get the system boot at least sometimes I have 
> > > > to specify nohz=off. Then I get
> > > 
> > > Try highres=off, too... Hehe, and even idle=poll might help.
> > 
> > Ok, for now I've compiled a kernel with all "advanced" features off like 
> > hrt, nohz. Will see how it behaves. But thanks for the hints.
> 
> Let us know...

Ok, it cold-booted (after a power-off, which is usually the most 
difficult case for this machine) once fine. On another boot it was 
veeeeery slow. I have a dmesg from that boot, here're some of the first 
differences with a subsequent normal boot:

-apm: BIOS version 1.2 Flags 0x03 (Driver version 1.16ac)
+apm: BIOS version 1.2 Flags 0x0b (Driver version 1.16ac)

...

 Serial: 8250/16550 driver $Revision: 1.90 $ 2 ports, IRQ sharing disabled
+serial 00:08: activated
 00:08: ttyS0 at I/O 0x3e8 (irq = 4) is a 16550A
+serial 00:09: activated
 00:09: ttyS1 at I/O 0x2e8 (irq = 3) is a NS16550A

...

 ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
 PIIX4: IDE controller (0x8086:0x7111 rev 0x01) at  PCI slot 0000:00:14.1
 PIIX4: not 100% native mode: will probe irqs later
-PIIX4: IDE port disabled
+    ide0: BM-DMA at 0x58c0-0x58c7, BIOS settings: hda:pio, hdb:pio
     ide1: BM-DMA at 0x58c8-0x58cf, BIOS settings: hdc:pio, hdd:pio
+Probing IDE interface ide0...
 Probing IDE interface ide1...
 Probing IDE interface ide0...
 Probing IDE interface ide1...

"+" is the slow boot. A reboot helped. And I've got two soft lockups now 
on that new boot each for 13s (hm, it's always 13s...:-() and this time I 
didn't specify "clocksource=tsc" as the last time.

On Wed, 5 Dec 2007, Robert Hancock wrote:

> On a machine that old, you shouldn't need ACPI to detect both CPUs, it should
> be able to use MPS..

Don't know the details, but remember it didn't work without ACPI. The only 
solution was acpi=noirq.

Thanks
Guennadi
---
Guennadi Liakhovetski