From: Matthew Dobson <colpatch@us.ibm.com>
To: linux-mm@kvack.org
Cc: Kanoj Sarcar <kanojsarcar@yahoo.com>,
Andrew Morton <akpm@zip.com.au>,
Martin Bligh <mjbligh@us.ibm.com>,
Michael Hohnbaum <hohnbaum@us.ibm.com>
Subject: Re: [patch] Useless locking in mm/numa.c
Date: Fri, 19 Jul 2002 14:45:15 -0700 [thread overview]
Message-ID: <3D38886B.4050806@us.ibm.com> (raw)
In-Reply-To: <3D387702.6010306@us.ibm.com>
[-- Attachment #1: Type: text/plain, Size: 3642 bytes --]
Whoops! wli pointed out that I forgot a reference to show_free_areas_node() in
mm.h. This patch should remedy that!
Cheers!
-Matt
Matthew Dobson wrote:
> Kanoj Sarcar wrote:
>
>> I think I put in the locks in the initial version of
>> the file becase the idea was that show_free_areas_node() could be
>> invoked from any cpu
>> in a multinode system (via the sysrq keys or other
>> intr sources), and the spin lock would provide sanity in the print out.
>
> As Bill mentioned, a grep through the source shows that
> show_free_areas_node() is never called, and since it boils down to
> *just* a call to show_free_areas_core() w/out the locking, the revised
> patch pulls it out entirely.
>
>> For nonnuma discontig machines, isn't the spin lock
>> providing protection in the pgdat list chain walking
>> in _alloc_pages()?
>
> Uhh... kinda? Since *next is static, it means that at best case, 2
> processes walking the pgdat_list chain will hip-hop over nodes... If it
> is racy code, the *best* that lock is currently doing is making it mildy
> less racy, and at worst, hiding the fact that there is a race there.
>
> I'm sure the lock was useful at some point, but it no longer is...
> Attatched is the new version, please apply..
>
> Cheers!
>
> -Matt
>
>>
>> Kanoj
>>
>> --- Matthew Dobson <colpatch@us.ibm.com> wrote:
>>
>>> There is a lock that is apparently protecting
>>> nothing. The node_lock spinlock in mm/numa.c is protecting read-only
>>> accesses to
>>> pgdat_list. Here is a patch to get rid of it.
>>>
>>> Cheers!
>>>
>>> -Matt
>>>
>>>> --- linux-2.5.26-vanilla/mm/numa.c Tue Jul 16
>>>
>>>
>>> 16:49:30 2002
>>> +++ linux-2.5.26-vanilla/mm/numa.c.fixed Thu Jul 18
>>> 17:59:35 2002
>>> @@ -44,15 +44,11 @@
>>>
>>> #define LONG_ALIGN(x)
>>> (((x)+(sizeof(long))-1)&~((sizeof(long))-1))
>>>
>>> -static spinlock_t node_lock = SPIN_LOCK_UNLOCKED;
>>> -
>>> void show_free_areas_node(pg_data_t *pgdat)
>>> {
>>> unsigned long flags;
>>>
>>> - spin_lock_irqsave(&node_lock, flags);
>>> show_free_areas_core(pgdat);
>>> - spin_unlock_irqrestore(&node_lock, flags);
>>> }
>>>
>>> /*
>>> @@ -106,11 +102,9 @@
>>> #ifdef CONFIG_NUMA
>>> temp = NODE_DATA(numa_node_id());
>>> #else
>>> - spin_lock_irqsave(&node_lock, flags);
>>> if (!next) next = pgdat_list;
>>> temp = next;
>>> next = next->node_next;
>>> - spin_unlock_irqrestore(&node_lock, flags);
>>> #endif
>>> start = temp;
>>> while (temp) {
>>>
>>
>>
>>
>> __________________________________________________
>> Do You Yahoo!?
>> Yahoo! Autos - Get free new car price quotes
>> http://autos.yahoo.com
>>
>
>
> ------------------------------------------------------------------------
>
> --- linux-2.5.26-vanilla/mm/numa.c Tue Jul 16 16:49:30 2002
> +++ linux-2.5.26-vanilla/mm/numa.c.fixed Thu Jul 18 17:59:35 2002
> @@ -43,17 +43,6 @@
> #ifdef CONFIG_DISCONTIGMEM
>
> #define LONG_ALIGN(x) (((x)+(sizeof(long))-1)&~((sizeof(long))-1))
> -
> -static spinlock_t node_lock = SPIN_LOCK_UNLOCKED;
> -
> -void show_free_areas_node(pg_data_t *pgdat)
> -{
> - unsigned long flags;
> -
> - spin_lock_irqsave(&node_lock, flags);
> - show_free_areas_core(pgdat);
> - spin_unlock_irqrestore(&node_lock, flags);
> -}
>
> /*
> * Nodes can be initialized parallely, in no particular order.
> @@ -106,11 +103,9 @@
> #ifdef CONFIG_NUMA
> temp = NODE_DATA(numa_node_id());
> #else
> - spin_lock_irqsave(&node_lock, flags);
> if (!next) next = pgdat_list;
> temp = next;
> next = next->node_next;
> - spin_unlock_irqrestore(&node_lock, flags);
> #endif
> start = temp;
> while (temp) {
[-- Attachment #2: node_lock.patch --]
[-- Type: text/plain, Size: 1266 bytes --]
--- linux-2.5.26-vanilla/mm/numa.c Tue Jul 16 16:49:30 2002
+++ linux-2.5.26-vanilla/mm/numa.c.fixed Thu Jul 18 17:59:35 2002
@@ -43,17 +43,6 @@
#ifdef CONFIG_DISCONTIGMEM
#define LONG_ALIGN(x) (((x)+(sizeof(long))-1)&~((sizeof(long))-1))
-
-static spinlock_t node_lock = SPIN_LOCK_UNLOCKED;
-
-void show_free_areas_node(pg_data_t *pgdat)
-{
- unsigned long flags;
-
- spin_lock_irqsave(&node_lock, flags);
- show_free_areas_core(pgdat);
- spin_unlock_irqrestore(&node_lock, flags);
-}
/*
* Nodes can be initialized parallely, in no particular order.
@@ -106,11 +103,9 @@
#ifdef CONFIG_NUMA
temp = NODE_DATA(numa_node_id());
#else
- spin_lock_irqsave(&node_lock, flags);
if (!next) next = pgdat_list;
temp = next;
next = next->node_next;
- spin_unlock_irqrestore(&node_lock, flags);
#endif
start = temp;
while (temp) {
--- linux-2.5.26-vanilla/include/linux/mm.h Tue Jul 16 16:49:24 2002
+++ linux-2.5.26-vanilla/include/linux/mm.h.fixed Fri Jul 19 14:40:29 2002
@@ -319,7 +319,6 @@
extern struct page *mem_map;
extern void show_free_areas(void);
-extern void show_free_areas_node(pg_data_t *pgdat);
extern int fail_writepage(struct page *);
struct page * shmem_nopage(struct vm_area_struct * vma, unsigned long address, int unused);
prev parent reply other threads:[~2002-07-19 21:45 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2002-07-19 1:03 Matthew Dobson
2002-07-19 18:36 ` Kanoj Sarcar
2002-07-19 18:38 ` William Lee Irwin III
2002-07-19 18:40 ` Kanoj Sarcar
2002-07-19 20:30 ` Matthew Dobson
2002-07-19 21:45 ` Matthew Dobson [this message]
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=3D38886B.4050806@us.ibm.com \
--to=colpatch@us.ibm.com \
--cc=akpm@zip.com.au \
--cc=hohnbaum@us.ibm.com \
--cc=kanojsarcar@yahoo.com \
--cc=linux-mm@kvack.org \
--cc=mjbligh@us.ibm.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox