Re: [PATCH] oom killer (Core)

[Andrea Arcangeli]

On Sat, Dec 04, 2004 at 08:00:04AM +0100, Voluspa wrote:
> 
> On 2004-12-03 23:08:55 Andrea Arcangeli wrote:
> 
> >You mean my patch is preventing your machine to boot? Then you're doing
> >something else wrong because it's impossible my patch is preventing 
> >your machine to boot.
> 
> Same experience as Thomas here. Full stop like his first log (no errors)
> . PIII (Celeron) 900@1 gig, 256 meg mem, 1 gig swap, preempt enabled.

The places I modified are never invoked during boot (except the below
one). It works like a charm here. I tried again building on top of
2.6.10-rc3 and it works again just fine here (previously I was on top of
the kernel CVS out of sync currently). No idea what's preventing you to
boot, but it's very hard to believe that my patch is to blame for it.

The only modified piece of code that may run during boot is this:

-       might_sleep_if(wait);
+       if (wait)
+               cond_resched();


You can try to put back a might_slee_if(wait), but if it deadlocks with
that change sure it's not a bug in my patch, it's instead a bug
somewhere else that calls alloc_pages w/o GFP_ATOMIC. Ingo's
lowlatency patch would expose the same bug too since they're aliasing
the might_sleep to cond_resched.

But I can't reproduce anything wrong here.

Enabling unmasked SIMD FPU exception support... done.
Checking 'hlt' instruction... OK.
 tbxface-0118 [02] acpi_load_tables      : ACPI Tables successfully
acquired
Parsing all Control
Methods:...........................................................................................................................................
Table [DSDT](id F004) - 502 Objects with 45 Devices 139 Methods 29
Regions
ACPI Namespace successfully loaded at root c0503d40
evxfevnt-0094 [03] acpi_enable           : Transition to ACPI mode
successful
CPU0: Intel(R) XEON(TM) CPU 2.40GHz stepping 04
per-CPU timeslice cutoff: 1462.70 usecs.
[..]

machine boots and run just fine and the patch is definitely applied.

PREEMPT=n but it can't be a bug in my patch even if PREEMPT=y breaks.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/