n Node-based order.
command:
%echo Z > /proc/sys/vm/numa_zonelist_order
Will rebuild zonelist in Zone-based order.
Tested on ia64 2-Node NUMA. works well.
Thanks to Lee Schermerhorn, he gives me much help and codes.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
---
include/linux
Hi, this is zonelist-order-fix patch version 5.
against 2.6.21-mm1. works well in my ia64/NUMA environment.
ChangeLog V4 -> V5
- separated 'doc' patch and rewrote it.
- more clean ups.
- sysctl/boot option params are simplified.
ChangeLog V2 -> V4
- automatic configuration is added.
- automatic
On Mon, 07 May 2007 19:10:05 +0900
Satoru Takeuchi <[EMAIL PROTECTED]> wrote:
> kstopmachine is created, bound to the CPU1, and woken up here, but
> this process can't start to run because reschedule doesn't occur on
> CPU1. Hence CPU0 also be able to run because it's waiting completion
> of CPU1
Node (A)'s ZONE_DMA/DMA32 occupies 60% of Node(A)'s memory.
otherwise, ZONE_ORDER_ZONE is selected.
Note: a user can specifiy this ordering from boot option.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
---
mm/page_alloc.c | 44 +
ing order(new style, ZONE order)
Node(0)'s NORMAL -> Node(1)'s NORMAL -> Node(0)'s DMA.
means put more priority on zone_type.
And you can specify this option as boot param.
Because autoconfig function does nothing. Default is "Node" order.
Tested on ia64 2-Node NU
Hi, this is version 4. including Lee Schermerhon's good rework.
and automatic configuration at boot time.
(This patch is reworked from V2, so skip V3 changelog.)
ChangeLog V2 -> V4
- automatic configuration is added.
- automatic configuration is now default.
- relaxed_zone_order is renamed to be
On Thu, 26 Apr 2007 18:25:10 -0700 (PDT)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Fri, 27 Apr 2007, KAMEZAWA Hiroyuki wrote:
>
> > > DMA memory.
> > >
> > It seems a bit complicated. If we do so, following can occur,
> >
> > Node
On Thu, 26 Apr 2007 17:57:40 -0400
Lee Schermerhorn <[EMAIL PROTECTED]> wrote:
> On Thu, 2007-04-26 at 18:34 +0900, KAMEZAWA Hiroyuki wrote:
> > Changelog from V1 -> V2
> > - sysctl name is changed to be relaxed_zone_order
> > - NORMAL->NORMAL->->DM
On Thu, 26 Apr 2007 08:48:19 -0700 (PDT)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Thu, 26 Apr 2007, KAMEZAWA Hiroyuki wrote:
>
> > (1)Use new zonelist ordering always and move init_task's tied cpu to a
> > cpu on the best node.
> > Child process
changes *default* zone order to
Node(0)'s NORMAL -> Node(1)'s NORMAL -> Node(0)'s DMA.
tested ia64 2-Node NUMA. works well.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: linux-2.6.21-rc7-mm2/mm/page_alloc.c
===
On Thu, 26 Apr 2007 11:47:44 +0200
Andi Kleen <[EMAIL PROTECTED]> wrote:
> On Thursday 26 April 2007 11:34:17 KAMEZAWA Hiroyuki wrote:
> >
> > Changelog from V1 -> V2
> > - sysctl name is changed to be relaxed_zone_order
> > - NORMAL->NORMAL->-
pported by arch.
But this style zonelist can easily cause OOM-Kill because of ZONE_DMA
exhaition. be careful.
command:
echo 0 > /proc/sys/vm/relaxed_zone_order
will rebuild zonelist as
Node(0)'s NORMAL -> Node(1)'s NORMAL -> Node(0)'s DM
On Thu, 26 Apr 2007 09:31:12 +0900
KAMEZAWA Hiroyuki <[EMAIL PROTECTED]> wrote:
> >
> > So a IA64 platform with i386 sicknesses? And pretty bad case of it since I
> > assume that the memory sizes per node are equal. Your solution of taking
> > 4G off node 0 and t
On Wed, 25 Apr 2007 12:17:15 -0700 (PDT)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Wed, 25 Apr 2007, KAMEZAWA Hiroyuki wrote:
>
> > Make zonelist policy selectable from sysctl.
> >
> > Assume 2 node NUMA, only node(0) has ZONE_DMA (ZONE_DMA32).
> >
On Wed, 25 Apr 2007 00:42:14 -0700
Andrew Morton <[EMAIL PROTECTED]> wrote:
> On Wed, 25 Apr 2007 12:19:46 +0900 KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
> wrote:
>
> > Make zonelist policy selectable from sysctl.
> >
> > Assume 2 node NUMA, only node(0)
useful in some users with heavy memory pressure and mlocks.
Tested under ia64 2 node NUMA against 2.6.21-rc7.. works well.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: linux-2.6.21-rc7/kernel/sysctl.c
===
-
On Tue, 24 Apr 2007 10:54:27 +0900
Hisashi Hifumi <[EMAIL PROTECTED]> wrote:
> In the case that changing the same bit concurrently, lock prefix or other
> spinlock is needed. But, I think that concurrent bit operation on different
> bits
> is just like OR operation , so lock prefix is not needed.
On Mon, 23 Apr 2007 19:32:46 +0100
[EMAIL PROTECTED] (Mel Gorman) wrote:
> > > I wasn't even aware of this kernelcore thing. It's pretty nasty-looking.
> > > yet another reminder that this code hasn't been properly reviewed in the
> > > past year or three.
> >
> > Just now, I'm making memory-un
ce we initialize nodemask in constrained_alloc().
>
thank you for catching bug.
Acked-by: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
> Signed-off-by: Hugh Dickins <[EMAIL PROTECTED]>
> ---
> Perhaps appropriate for 2.6.20-stable too - regression since 2.6.19.
>
>
On Thu, 22 Mar 2007 21:56:03 -0700
"Tony Luck" <[EMAIL PROTECTED]> wrote:
> On 3/22/07, KAMEZAWA Hiroyuki <[EMAIL PROTECTED]> wrote:
> > I hear some people says that "When I set stack-size-limit to 32M,
> > I want to use 32M of memory stack..." an
r-backing store cannot be expanded because the
memory stack uses the whole stack".
How about this ?
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: linux-2.6.21-rc4/arch/ia64/mm/init.c
===
--- linux-2.6.21-rc4.orig/
by adjusting register-stack.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: linux-2.6.21-rc4/arch/ia64/mm/init.c
===
--- linux-2.6.21-rc4.orig/arch/ia64/mm/init.c
+++ linux-2.6.21-rc4/arch/ia64/mm/init.c
@@ -155,7
On Thu, 15 Mar 2007 11:06:21 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> > On Tue, 13 Mar 2007 18:33:20 +0200 Dan Aloni <[EMAIL PROTECTED]> wrote:
> > Hello,
> >
> > This question is relevent to 2.6.20.
> >
> > I noticed that if the RSS for the stack size is say, 8MB, running
> > a single-t
plz allow me to explain more.
"Why register-stack/memory-stack upside down is bad" is a bit complicated.
So...this is a test and result for explaining bug.
This is a sample code and its result on 2.6.21-rc3.
Note: base address of memory'stack can be randomly change.
== sample code ==
[EMAIL PRO
On Fri, 16 Mar 2007 06:20:47 +0900
KAMEZAWA Hiroyuki <[EMAIL PROTECTED]> wrote:
> On Thu, 15 Mar 2007 09:57:28 -0600
> "David Mosberger-Tang" <[EMAIL PROTECTED]> wrote:
>
> > But aren't you going to be limited to less than a page worth of
> > reg
y if I don't catch your point.
--Kame
> --david
>
> On 3/15/07, KAMEZAWA Hiroyuki <[EMAIL PROTECTED]> wrote:
> > This patch fixes ia64's bug in ulimit -s handling. against 2.6.21-rc3.
> >
> > At first,the address of register stack is defined by
27;t handle this case. In this case,
register stack expansion causes SEGV.
This means that the user program can use only 1 page for its register stack.
This patch fixes the above case by moving register stack to suitable place.
Note) fixing page fault handler seems to be another way...but a bit
On Mon, 12 Mar 2007 19:57:58 +0100
Michal Hocko <[EMAIL PROTECTED]> wrote:
> What do you think about that. Is this way correct?
>
If you are sure that your "original" pages is never freed while you are
migrating it.maybe.
-Kame
-
To unsubscribe from this list: send the line "unsubscribe li
On Tue, 6 Mar 2007 22:57:10 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> On Wed, 7 Mar 2007 15:23:17 +0900 KAMEZAWA Hiroyuki <[EMAIL PROTECTED]> wrote:
>
> > This patch fixes boot failure because irq_desc->mask() is NULL.
> >
> > - Added mask/unmas
This patch fixes boot failure because irq_desc->mask() is NULL.
- Added mask/unmask functions to ia64's irq desc function table.
But I'm not sure this fix is correct or not. please review.
- rename hw_interrupt_type to irq_chip. hw_interrupt_type is old name.
Signed-Off-By: KAME
On Tue, 6 Mar 2007 03:09:27 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> > ==
> >
> > Is "mask" always valid pointer ?
>
> I can only find two `struct irq_chip's in arch/ia64 and they both have a
> .mask. And a .unmask. So perhaps that is a misreading of what oopsed.
>
> There are no cha
zones are overkill
> there and anti-fragmentation on its own is good enough). Pages hot-added
> to ZONE_MOVABLE will always be reclaimable or migratable in the case of
> mlock(). Kamezawa Hiroyuki has indicated that his hot-remove patches also
> do something like ZONE_MOVABLE. I would hope t
On Thu, 1 Mar 2007 21:11:58 -0800 (PST)
Linus Torvalds <[EMAIL PROTECTED]> wrote:
> The whole DRAM power story is a bedtime story for gullible children. Don't
> fall for it. It's not realistic. The hardware support for it DOES NOT
> EXIST today, and probably won't for several years. And the real
On Thu, 1 Mar 2007 16:09:15 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> On Thu, 1 Mar 2007 10:12:50 +
> [EMAIL PROTECTED] (Mel Gorman) wrote:
>
> > Any opinion on merging these patches into -mm
> > for wider testing?
>
> I'm a little reluctant to make changes to -mm's core mm unless tho
On Tue, 27 Feb 2007 09:50:16 +0900
Tomoki Sekiyama <[EMAIL PROTECTED]> wrote:
> Hi Kamezawa-san,
>
> thanks for your reply.
>
> KAMEZAWA Hiroyuki wrote:
> > Interesting, but how about adjust this parameter like below instead of
> > adding new control knob ?(thi
On Thu, 22 Feb 2007 10:42:23 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> > > G. Slab merging
> > >
> > >We often have slab caches with similar parameters. SLUB detects those
> > >on bootup and merges them into the corresponding general caches. This
> > >leads to more ef
On Fri, 23 Feb 2007 21:03:37 +0900
Tomoki Sekiyama <[EMAIL PROTECTED]> wrote:
> Hi,
>
> I have observed a problem that write(2) can be blocked for a long time
> if a system has several disks and is under heavy I/O pressure. This
> patchset is to avoid the problem.
>
> Example of the probrem:
>
On Fri, 23 Feb 2007 16:37:43 +0100
Nick Piggin <[EMAIL PROTECTED]> wrote:
> +static void dcache_hash_resize(unsigned int new_shift);
> +static void mod_nr_dentry(int mod)
> +{
> + unsigned long dentry_size;
> +
> + dentry_stat.nr_dentry += mod;
> +
> + dentry_size = 1 << dentry_hash->s
On Tue, 20 Feb 2007 15:10:07 -0800 (PST)
Roland McGrath <[EMAIL PROTECTED]> wrote:
> I'm usually the stickler for anal POSIX compliance, but this is one thing
> that I did notice a while ago, realized Linux had never done it, and
> decided I didn't care.
>
Okay, I don't think this is a big troubl
On Tue, 20 Feb 2007 20:20:49 +0300
Oleg Nesterov <[EMAIL PROTECTED]> wrote:
> > > > + clear_stale_sigchild(current, retval);
> > >
> > > But we are not checking that SIGCHLD is blocked?
> > >
> > I'm sorry if I don't read SUSv3 correctly. SUSv3 doesn't define how we
> > should
> >
On Tue, 20 Feb 2007 17:22:57 +0300
Oleg Nesterov <[EMAIL PROTECTED]> wrote:
>
> I'd suggest to make a separate function, but not complicate collect_signal().
>
okay. I'll try again if people admit me to go ahead.
> > --- linux-2.6.20-devel.orig/kernel/exit.c
> > +++ linux-2.6.20-devel/kernel/exi
reaped
process.)
please review...
works well on 2.6.20/ia64/NUMA environment and passed my easy test.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: linux-2.6.20-devel/kernel/signal.c
===
--- linux-2.6.20-devel.orig/
On Mon, 19 Feb 2007 12:20:42 +0530
Balbir Singh <[EMAIL PROTECTED]> wrote:
> +int memctlr_mm_overlimit(struct mm_struct *mm, void *sc_cont)
> +{
> + struct container *cont;
> + struct memctlr *mem;
> + long usage, limit;
> + int ret = 1;
> +
> + if (!sc_cont)
> + go
On Mon, 19 Feb 2007 10:22:52 +0200 (EET)
Pekka J Enberg <[EMAIL PROTECTED]> wrote:
> @@ -2987,6 +2987,14 @@
> slabp = list_entry(entry, struct slab, list);
> check_slabp(cachep, slabp);
> check_spinlock_acquired(cachep);
> +
> + /*
> +
Hi,
>From SUSv3, I expected SIGCHLD from dead processes (already reaped by wait(2))
should be cleared. But it seems that such situation is not handled in Linux.
Here is a test program. set sigchld handler and call waitpid() in main().
==
#include
#include
#include
#include
int sigchld_handl
log: v2 -> v3
- removed ambiguous void *pointer usage.
- fixed warnings...misuse of PTR_ERR.
Changelog: v1 -> v2
- avoid extra pgdat scanningit is not necessary.
tested on ia64/NUMA with memory-less-node.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: li
On Tue, 13 Feb 2007 10:50:53 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Tue, 13 Feb 2007, Martin J. Bligh wrote:
>
> > What's wrong with just setting the existing counters like
> > node_spanned_pages / node_present_pages to zero?
>
> Will this fix the breakage that Kame-san sa
On Tue, 13 Feb 2007 09:25:00 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Tue, 13 Feb 2007, KAMEZAWA Hiroyuki wrote:
>
> > NOD_DATA(nid) is always valid pointer if a node is online.
> > NODE_DATA(nid)->present_pages can be 0 even if a node is online,
On Tue, 13 Feb 2007 09:29:49 +0100
Andi Kleen <[EMAIL PROTECTED]> wrote:
>
> > In my understanding, a "node" is a block of cpu, memory, devices.
> > and there could be cpu-only-node, memory-only-node, device-only-node...
>
> The trouble with this is that you'll need to harden large parts
> of co
continue;
x
This patch adds a new node mask "node_memory_online_map" for nodes
which have memory.
for_each_node_mask(nid, node_memory_online_map) walks all memory-ready-nodes.
This mask is updated at node-hotplug ops.
Signed-Off-By: KAMEZAWA H
On Thu, 8 Feb 2007 11:28:30 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> > @@ -193,9 +197,11 @@
> > break;
> > case MPOL_BIND:
> > policy->v.zonelist = bind_zonelist(nodes);
> > - if (policy->v.zonelist == NULL) {
> > + if (IS_ERR(policy
of zonelist is zero, just returns -EINVAL.
Changelog: v2 -> v3
- changed handling of void *pointer
- fixed warnings...misuse of PTR_ERR.
Changelog: v1 -> v2
- avoid extra pgdat scanningit is not necessary.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: lin
On Thu, 8 Feb 2007 08:49:41 +0100
Andi Kleen <[EMAIL PROTECTED]> wrote:
>
> > This panic(hang) was found by a numa test-set on a system with 3 nodes,
> > where
> > node(2) was memory-less-node.
>
> I still think it's the wrong fix -- just get rid of the memory less node.
> I expect you'll likel
-EINVAL.
Changelog: v1 -> v2
- avoid extra pgdat scanningit is not necessary.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: linux-2.6.20/mm/mempolicy.c
===
--- linux-2.6.20.orig/mm/mempolicy.c2007-02
On Wed, 7 Feb 2007 09:43:44 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> > > and to
> > > accurately present the machine's topology to the user without us having to
> > > go adding falsehoods like this?
> >
> > a node is a piece of memory. Without memory it doesn't make sense.
>
> Who said?
On Wed, 7 Feb 2007 06:05:56 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Wed, 7 Feb 2007, KAMEZAWA Hiroyuki wrote:
>
> > > IMHO there shouldn't be any memory less nodes. The architecture code
> > > should not create them. The CPU
On Wed, 7 Feb 2007 12:32:36 +0100
Andi Kleen <[EMAIL PROTECTED]> wrote:
>
> > How for_each_online_node(nid) works ? it can handle alias-nid ?
> >
> > ==
> > for_each_online_node(nid) {
> > pgdat = NODE_DATA(nid);
> > ==
> > This code never accesses pgdat_for_A twice ?
>
> It wou
On Wed, 7 Feb 2007 11:41:25 +0100
Andi Kleen <[EMAIL PROTECTED]> wrote:
> On Wednesday 07 February 2007 11:37, KAMEZAWA Hiroyuki wrote:
> > On Wed, 7 Feb 2007 11:19:02 +0100
> > Andi Kleen <[EMAIL PROTECTED]> wrote:
> > > You can also alias node numbers to
On Wed, 7 Feb 2007 11:19:02 +0100
Andi Kleen <[EMAIL PROTECTED]> wrote:
>
> > AFAIK, ia64 creates nodes just depends on SRAT's possible resource
> > information.
> > Then, ia64 can create cpu-memory-less-node(node with no available
> > resource.).
> > (*)I don't like this.
> >
> > If we don't
On 07 Feb 2007 11:20:06 +0100
Andi Kleen <[EMAIL PROTECTED]> wrote:
> KAMEZAWA Hiroyuki <[EMAIL PROTECTED]> writes:
>
> > current mempolicy just checks whether a node is online or not.
> > If there is memory-less-node, mempolicy's target node can be
>
On Wed, 7 Feb 2007 00:04:41 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Wed, 7 Feb 2007, KAMEZAWA Hiroyuki wrote:
>
> > > Hmmm... Remove the node from the node_online_map instead?
> > >
> > Changing defintion of node_online_map is har
On Tue, 6 Feb 2007 09:26:53 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Tue, 6 Feb 2007, KAMEZAWA Hiroyuki wrote:
>
> > This means an access to NULL,here.
> > ==
> > unsigned slab_node(struct mempolicy *policy)
&g
node.
*/
return zone_to_nid(policy->v.zonelist->zones[0]);
}
==
length of this zonelist was 0.
It seems fixing a NULL access here is also O.K.
This patch is just an idea.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: linux-2.6.
On Fri, 26 Jan 2007 10:10:27 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> On Fri, 26 Jan 2007 07:44:42 -0800 (PST)
> Christoph Lameter <[EMAIL PROTECTED]> wrote:
>
> > On Fri, 26 Jan 2007, Andrew Morton wrote:
> >
> > > > > > Track mlocked pages via a ZVC
> > >
> > > Why?
> >
> > Large amo
On Fri, 26 Jan 2007 02:29:55 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> On Wed, 24 Jan 2007 14:15:10 +0900
> KAMEZAWA Hiroyuki <[EMAIL PROTECTED]> wrote:
>
> > - One for stability
> > When a customer constructs their detabase(Oracle), the system ofte
On Thu, 25 Jan 2007 00:40:54 -0500
Rik van Riel <[EMAIL PROTECTED]> wrote:
> KAMEZAWA Hiroyuki wrote:
> > On Wed, 24 Jan 2007 23:28:15 -0500
> > Rik van Riel <[EMAIL PROTECTED]> wrote:
> >
> >> KAMEZAWA Hiroyuki wrote:
> > I always says Linux
On Wed, 24 Jan 2007 23:28:15 -0500
Rik van Riel <[EMAIL PROTECTED]> wrote:
> KAMEZAWA Hiroyuki wrote:
>
> > FYI:
> > Because some customers are migrated from mainframes, they want to control
> > almost all features in OS, IOW, designing memory usages.
>
>
On Wed, 24 Jan 2007 18:41:27 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> > But I can't think of the way to show that.
> > ==
> > [EMAIL PROTECTED] src]$ free
> > total used free sharedbuffers cached
> > Mem:741604 724628 16976
On Wed, 24 Jan 2007 14:15:10 +0900
KAMEZAWA Hiroyuki <[EMAIL PROTECTED]> wrote:
> And...some customers want to keep memory Free as much as possible.
> 99% memory usage makes insecure them ;)
>
If there is a way that the "free" command can show "never used&q
On Tue, 23 Jan 2007 20:30:16 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Wed, 24 Jan 2007, KAMEZAWA Hiroyuki wrote:
>
> > I don't prefer to cause zone fallback by this.
> > This may use ZONE_DMA before exhausing ZONE_NORMAL (ia64),
>
> H
one more thing...
On Tue, 23 Jan 2007 16:49:55 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> @@ -1168,6 +1170,11 @@ zonelist_scan:
> !cpuset_zone_allowed_softwall(zone, gfp_mask))
> goto try_next_zone;
>
> + if ((gfp_
On Tue, 23 Jan 2007 16:49:55 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> If we enter reclaim and the number of page cache pages
> is too high then we switch off swapping during reclaim
> to avoid touching anonymous pages.
In general, I like this (kind of) feature.
> + /*
> +
n
zonelist[].
contstrained_alloc() should get memory_less_node into count.
Otherwise, it always thinks 'oom is from mempolicy'.
This means that current process dies at any time. This patch fix it.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
mm/oom_kill.c |7 ++-
1 files c
On Thu, 21 Dec 2006 21:18:12 -0800
Paul Jackson <[EMAIL PROTECTED]> wrote:
> KAMEZAWA-san wrote:
> > But there is memory-less-node. contstrained_alloc() should get
> > memory_less_node into count.
>
> This patch looks ok to me.
>
> One line in the patch comment seems backward:
>
> If zone_lis
time. This patch fix it.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
mm/oom_kill.c |7 ++-
1 files changed, 6 insertions(+), 1 deletion(-)
Index: devel-2.6.20-rc1-mm1/mm/oom_kill.c
===
--- devel-2.6.20-rc1-mm
On some system, there are memory-less-nodes. (IOW, cpu-only-node)
Then,there are online nodes which has no memory.
Now, below code is used to detect the context where oom happens.
===
static inline int constrained_alloc(struct zonelist *zonelist, gfp_t gfp_mask)
{
#ifdef CONFIG_NUMA
struc
On Wed, 20 Dec 2006 15:06:28 -0500
"Bob Picco" <[EMAIL PROTECTED]> wrote:
> Sorry I was looking for AIM VII and/or reaim which are multiuser loads.
> The results (2.6.20-rc1-mm1) for EXTREME, SPARSEMEM+VMEMMAP and
> SPARSEMEM+VMEMMAP+your+patch are below. Note SPARSEMEM+VMEMMAP AIM VII
> wasn't be
On Mon, 18 Dec 2006 15:16:20 -0800
Dave Hansen <[EMAIL PROTECTED]> wrote:
> enum context
> {
> EARLY,
> HOTPLUG
> };
I like this :)
Thanks,
-Kame
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo
On Sat, 16 Dec 2006 10:38:53 -0800 (PST)
Christoph Lameter <[EMAIL PROTECTED]> wrote:
> On Sat, 16 Dec 2006, KAMEZAWA Hiroyuki wrote:
>
> > By this, we'll not access mem_section[] in usual ops.
>
> Why do we need mem_section? We have a page table that fulfills the s
On Fri, 15 Dec 2006 11:45:36 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> Perhaps if the function's role in the world was commented it would be clearer.
>
How about patch like this ? (this one is not tested.)
Already-exisiting-more-generic-flag is available ?
-Kame
==
include/linux/memory
ia64 support for sparsemem vmem_map optimize pfn_valid() patch.
Because ia64 has its own virtual mem_map, we can reuse the same code.
So this patch is simple.
To support optimized pfn_valid() in other arch, you (may) have to modify fault
handler in kernel address space.
Signed-Off-By: KAMEZAWA
m_map range.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
include/linux/mmzone.h | 10 ++
mm/Kconfig |4
mm/sparse.c|7 +++
3 files changed, 21 insertions(+)
Index: devel-2.6.20-rc1-mm1/include/lin
This patch implements pfn_valid() micro optimization.
This uses ia64_pfn_valid() idea to check mem_map is valid or not instead of
sparsemem's logic.
By this, we'll not access mem_section[] in usual ops.
I attaches my easy test result with *micro* benchmark on SMP system.
I'm glad if you give me
On Mon, 11 Dec 2006 22:06:17 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> > When I use ftp on 2.6.19-mm1, transfered file is always broken.
> > like this:
> > ==
> > [EMAIL PROTECTED] ~]$ file ./linux-2.6.19.tar.bz2 (got on 2.6.19-mm1)
> > ./linux-2.6.19.tar.bz2: data
> > (I confirmed original
On Mon, 11 Dec 2006 00:58:07 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
>
> Temporarily at
>
> http://userweb.kernel.org/~akpm/2.6.19-mm1/
>
> Will appear later at
>
>
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.19/2.6.19-mm1/
>
When I use ftp on 2.6.
On Sat, 9 Dec 2006 22:17:00 +0900
KAMEZAWA Hiroyuki <[EMAIL PROTECTED]> wrote:
> I'll renew this in the next week.
>
Hi, this is a fix patch. Sorry for my carelessness.
I'll post next add-on patch against the next -mm which will be shipped.
What I have now are
- pf
On Sat, 9 Dec 2006 12:51:37 +0100
Heiko Carstens <[EMAIL PROTECTED]> wrote:
> > Virtual mem_map is not useful for 32bit archs. This uses huge virtual
> > address range.
>
> Why? The s390 vmem_map implementation which I sent last week to linux-mm
> is merged in the meantime. It supports both 32 an
On Sat, 9 Dec 2006 13:05:47 +0100
Heiko Carstens <[EMAIL PROTECTED]> wrote:
> > +#ifdef CONFIG_SPARSEMEM_VMEMMAP
> > +#if (((BITS_PER_LONG/4) * PAGES_PER_SECTION) % PAGE_SIZE) != 0
> > +#error "PAGE_SIZE/SECTION_SIZE relationship is not suitable for vmem_map"
> > +#endif
>
> Why the BITS_PER_LONG
I tested ia64 with this patch under
- DISCONTIGMEM + VIRTUAL_MEM_MAP
- SPARSEMEM
- SPARSEMEM_VMEMMAP
on SMP with tiger4_defconfig.
Fix typo for DISCONTIGMEM
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: devel-2.6.19/include/asm-ia64/pgt
This removes implicit default actions in map_generic_kernel() call.
Also changes comments in vmalloc.h
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: devel-2.6.19/include/linux/vmalloc.h
===
--- devel-2.6.1
for avoiding complex inclusion of headr file in the middle of another header
file.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: devel-2.6.19/include/linux/mmzone.h
===
--- devel-2.6.19.orig/include/linux/mm
On Fri, 8 Dec 2006 19:33:23 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> >
> > > Would prefer to unconditionally include the header file - conditional
> > > inclusions
> > > like this can cause compile failures when someone changes a config
> > > option. They
> > > generally raise the comp
On Fri, 8 Dec 2006 16:30:20 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> > +#ifdef CONFIG_SPARSEMEM_VMEMMAP_STATIC
> > +#include
> > +extern struct page mem_map[];
> > +#else
> > extern struct page* mem_map;
> > #endif
> > +#endif
>
> This looks rather unpleasant - what went wrong here?
>
On Fri, 8 Dec 2006 16:28:19 -0800
Andrew Morton <[EMAIL PROTECTED]> wrote:
> Generally we prefer to simply *require* that the function vector be filled
> in appropriately. So if the caller has no special needs, the caller will
> set their gen_map_kern_ops.k_pte_alloc to point at pte_alloc_kernel(
ia64 support for sparsemem/vmem_map.
* defines mem_map[] and set its value (by static way).
* changes definitions of VMALLOC_START.
* adds CONFIGS.
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: devel-2.6.19/arch/ia64/K
This patch adds support for statically allocated virtual mem_map.
(means virtual address of mem_map array is defined statically.)
This removes reference to *(&mem_map).
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
Index: devel-2.6.19/include/linu
This patch implements of virtual mem_map on sparsemem.
This includes only arch independent part and depends on
generic map/unmap in the kernel function in this patch series.
Usual sparsemem(_extreme) have to do global table look up in
pfn_to_page()/page_to_pfn(), this seems a bit costly.
If an ar
e their virtual/physical space by themselves.
Because it's complex and danger to manage virtual address space by
each function's own code, it's better to use fixed address.
Note: My first purpose is supporting virtual mem_map both at boot/hotplug
sharing the same logic.
Signed-Off
Hi, virtual mem_map on sparsemem/generic patch version 3.
I myself likes this patch.
But someone may feels this patch is intrusive and scattered.
please pointing out.
Changes v2 -> v3
- make map/unmap function for general purpose. (for my purpose ;)
- drop memory hotplug support. will be posted a
Hi, your plan looks good to me.
some comments.
On Mon, 4 Dec 2006 23:45:32 + (GMT)
Mel Gorman <[EMAIL PROTECTED]> wrote:
> 1. Use lumpy-reclaim to intelligently reclaim contigous pages. The same
> logic can be used to reclaim within a PFN range
> 2. Merge anti-frag to help high-order allo
601 - 700 of 709 matches
Mail list logo
