Re: Possible bug in 2.6.34 slub

From: Randy Dunlap
Date: Tue Jun 01 2010 - 15:40:18 EST


On Tue, 1 Jun 2010 21:17:40 +0200 Giangiacomo Mariotti wrote:

> On Tue, Jun 1, 2010 at 5:22 AM, Bill Davidsen <davidsen@xxxxxxx> wrote:
> > Randy Dunlap wrote:
> >>
> >> On Tue, 1 Jun 2010 01:39:43 +0200 Giangiacomo Mariotti wrote:
> >>
> >>> Hi, I've recently noticed this line on the dmesg output(kernel 2.6.34):
> >>> [    0.000000] SLUB: Genslabs=14, HWalign=64, Order=0-3, MinObjects=0,
> >>> CPUs=16, Nodes=1
> >>>
> >>> My cpu is an I7 920, so it has 4 cores and there's hyperthreading
> >>> enabled, so there are 8 logical cpus. Is this a bug?
> >>
> >>
> >> No, it's just some boot/init time information.
> >>
> > I would consider it a bug to claim CPUs=xx when xx is something other than
> > the number of cores or the number of SMT threads supported by the processor.
> > Of course if /proc/cpuinfo shows four siblings per core or something
> > exciting, then it's right and you have a CPU you can sell to gizmodo and
> > tell them a drunk left on the bar.
> >
> So....is it a bug or not?

Sorry, I think that I misread your report.
It does look like misinformation.
Let's cc Christoph Lameter & Pekka.


> The point is, I guess(didn't actually look at the code), if that's
> just the count of MAX number of cpus supported, which is a config time
> define and then the actual count gets refined afterwards by slub
> too(because I know that the rest of the kernel knows I've got 4
> cores/8 logical cpus) or not. Is that it? If this is not the case(that
> is, it's not a static define used as a MAX value), then I can't see
> what kind of boot/init time info it is. If it's a boot-time info, it
> just means it's a _wrong_ boot-time info.



---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
--
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/