From: Ingo Molnar <mingo@elte.hu>
To: Tejun Heo <tj@kernel.org>
Cc: rusty@rustcorp.com.au, tglx@linutronix.de, x86@kernel.org,
linux-kernel@vger.kernel.org, hpa@zytor.com, jeremy@goop.org,
cpw@sgi.com, nickpiggin@yahoo.com.au, ink@jurassic.park.msu.ru
Subject: Re: [PATCHSET x86/core/percpu] improve the first percpu chunk allocation
Date: Tue, 24 Feb 2009 15:12:17 +0100 [thread overview]
Message-ID: <20090224141217.GA17287@elte.hu> (raw)
In-Reply-To: <49A3F5C5.4060107@kernel.org>
* Tejun Heo <tj@kernel.org> wrote:
> What's missing is unification of static and dynamic accessors
> and thus the faster accessors - percpu_read() and friends -
> for dynamic ones. This will be the next round of patches.
Ok, good - we are in agreement then and i'll wait for those
patches.
And i think i finally decoded the real source of the disconnect
:-)
It's still about this restriction:
+ /*
+ * If large page isn't supported, there's no benefit in doing
+ * this. Also, embedding allocation doesn't play well with
+ * NUMA.
+ */
+ if (!cpu_has_pse || pcpu_need_numa())
+ return -EINVAL;
This is what makes no sense (why force the static percpu area
into 4K mappings on NUMA).
You do it because i think you misunderstood my original 2MB TLB
static area suggestion. setup_pcpu_embed() does this now:
+ pcpue_ptr = pcpu_alloc_bootmem(0, num_possible_cpus() * pcpue_unit_size,
+ PAGE_SIZE);
That is not NUMA-friendly indeed.
What should be done instead is to up the unit size to 2MB as i
suggested, and to allocate 2MB sized and 2MB aligned
numa-correct area for each CPU, via bootmem.
To quote my original mail:
> > - allocate the static percpu area using bootmem-alloc, but
> > using a 2MB alignment parameter and a 2MB aligned size. Then
> > we can remap it to some convenient and undisturbed virtual
> > memory area, using 2MB TLBs. [*]
I.e. each individual 2MB allocated largepage can then be
remapped as a 2MB TLB to the high (vmalloc) area. Followed by
ordinary 4K mappings for regular percpu_alloc() pages.
( and the partial, unused pages within this initial chunk are
returned to bootmem. )
That will be NUMA-friendly and i suspect we should also use it
on SMP just to get that aspect of the code tested better.
Do _not_ allocate the units together in one bootmem allocation
because that's not NUMA-friendly.
Ok?
Ingo
next prev parent reply other threads:[~2009-02-24 14:12 UTC|newest]
Thread overview: 45+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-02-24 3:11 [PATCHSET x86/core/percpu] improve the first percpu chunk allocation Tejun Heo
2009-02-24 3:11 ` [PATCH 01/10] percpu: fix pcpu_chunk_struct_size Tejun Heo
2009-02-24 3:11 ` [PATCH 02/10] bootmem: clean up arch-specific bootmem wrapping Tejun Heo
2009-02-24 11:30 ` Johannes Weiner
2009-02-24 11:39 ` Tejun Heo
2009-02-24 3:11 ` [PATCH 03/10] bootmem: reorder interface functions and add a missing one Tejun Heo
2009-02-24 3:11 ` [PATCH 04/10] vmalloc: add @align to vm_area_register_early() Tejun Heo
2009-02-24 3:11 ` [PATCH 05/10] x86: update populate_extra_pte() and add populate_extra_pmd() Tejun Heo
2009-02-24 3:11 ` [PATCH 06/10] percpu: remove unit_size power-of-2 restriction Tejun Heo
2009-02-24 3:11 ` [PATCH 07/10] percpu: give more latitude to arch specific first chunk initialization Tejun Heo
2009-02-24 3:11 ` [PATCH 08/10] x86: separate out setup_pcpu_4k() from setup_per_cpu_areas() Tejun Heo
2009-02-24 3:11 ` [PATCH 09/10] x86: add embedding percpu first chunk allocator Tejun Heo
2009-02-24 3:11 ` [PATCH 10/10] x86: add remapping " Tejun Heo
2009-02-24 9:57 ` [PATCHSET x86/core/percpu] improve the first percpu chunk allocation Ingo Molnar
2009-02-24 11:48 ` Tejun Heo
2009-02-24 12:40 ` Ingo Molnar
2009-02-24 13:27 ` Tejun Heo
2009-02-24 14:12 ` Ingo Molnar [this message]
2009-02-24 14:37 ` Tejun Heo
2009-02-24 15:15 ` Ingo Molnar
2009-02-24 23:33 ` Tejun Heo
2009-03-04 0:03 ` Rusty Russell
2009-03-04 0:15 ` H. Peter Anvin
2009-03-04 0:50 ` Ingo Molnar
2009-02-24 12:51 ` Ingo Molnar
2009-02-24 14:47 ` Tejun Heo
2009-02-24 15:19 ` Ingo Molnar
2009-02-24 15:30 ` Nick Piggin
2009-02-24 13:02 ` Ingo Molnar
2009-02-24 14:40 ` Tejun Heo
2009-02-24 20:17 ` Ingo Molnar
2009-02-24 20:51 ` Ingo Molnar
2009-02-24 21:02 ` Yinghai Lu
2009-02-24 21:12 ` [PATCH] x86: check range in reserve_early() -v2 Yinghai Lu
2009-02-24 21:16 ` [PATCHSET x86/core/percpu] improve the first percpu chunk allocation Ingo Molnar
2009-02-25 2:09 ` [PATCH x86/core/percpu 1/2] x86, percpu: fix minor bugs in setup_percpu.c Tejun Heo
2009-02-25 2:10 ` [PATCH x86/core/percpu 2/2] x86: convert cacheflush macros inline functions Tejun Heo
2009-02-25 2:23 ` [PATCHSET x86/core/percpu] improve the first percpu chunk allocation Tejun Heo
2009-02-25 2:56 ` Tejun Heo
2009-02-25 12:59 ` Ingo Molnar
2009-02-25 13:43 ` WARNING: at include/linux/percpu.h:159 __create_workqueue_key+0x1f6/0x220() Ingo Molnar
2009-02-26 2:03 ` [PATCH core/percpu] percpu: fix too low alignment restriction on UP Tejun Heo
2009-02-26 3:26 ` Ingo Molnar
2009-02-25 6:40 ` [PATCHSET x86/core/percpu] improve the first percpu chunk allocation Rusty Russell
2009-02-25 12:54 ` Ingo Molnar
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20090224141217.GA17287@elte.hu \
--to=mingo@elte.hu \
--cc=cpw@sgi.com \
--cc=hpa@zytor.com \
--cc=ink@jurassic.park.msu.ru \
--cc=jeremy@goop.org \
--cc=linux-kernel@vger.kernel.org \
--cc=nickpiggin@yahoo.com.au \
--cc=rusty@rustcorp.com.au \
--cc=tglx@linutronix.de \
--cc=tj@kernel.org \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.