Use cpu ops to deal with the per cpu data instead of a local_t. Reduces memory
requirements, cache footprint and decreases cycle counts.
The this_cpu_xx operations are also used for !SMP mode. Otherwise we could
not drop the use of __module_ref_addr() which would make per cpu data handling
complicated. this_cpu_xx operations have their own fallback for !SMP.
The last hold out of users of local_t is the tracing ringbuffer after this patch
has been applied.
Signed-off-by: Christoph Lameter <[email protected]>
---
include/linux/module.h | 38 ++++++++++++++------------------------
kernel/module.c | 30 ++++++++++++++++--------------
kernel/trace/ring_buffer.c | 1 +
3 files changed, 31 insertions(+), 38 deletions(-)
Index: linux-2.6/include/linux/module.h
===================================================================
--- linux-2.6.orig/include/linux/module.h 2009-12-18 13:13:24.000000000 -0600
+++ linux-2.6/include/linux/module.h 2009-12-18 14:16:17.000000000 -0600
@@ -16,8 +16,7 @@
#include <linux/kobject.h>
#include <linux/moduleparam.h>
#include <linux/tracepoint.h>
-
-#include <asm/local.h>
+#include <linux/percpu.h>
#include <asm/module.h>
#include <trace/events/module.h>
@@ -363,11 +362,9 @@ struct module
/* Destruction function. */
void (*exit)(void);
-#ifdef CONFIG_SMP
- char *refptr;
-#else
- local_t ref;
-#endif
+ struct module_ref {
+ int count;
+ } *refptr;
#endif
#ifdef CONFIG_CONSTRUCTORS
@@ -454,25 +451,16 @@ void __symbol_put(const char *symbol);
#define symbol_put(x) __symbol_put(MODULE_SYMBOL_PREFIX #x)
void symbol_put_addr(void *addr);
-static inline local_t *__module_ref_addr(struct module *mod, int cpu)
-{
-#ifdef CONFIG_SMP
- return (local_t *) (mod->refptr + per_cpu_offset(cpu));
-#else
- return &mod->ref;
-#endif
-}
-
/* Sometimes we know we already have a refcount, and it's easier not
to handle the error case (which only happens with rmmod --wait). */
static inline void __module_get(struct module *module)
{
if (module) {
- unsigned int cpu = get_cpu();
- local_inc(__module_ref_addr(module, cpu));
+ preempt_disable();
+ __this_cpu_inc(module->refptr->count);
trace_module_get(module, _THIS_IP_,
- local_read(__module_ref_addr(module, cpu)));
- put_cpu();
+ __this_cpu_read(module->refptr->count));
+ preempt_enable();
}
}
@@ -481,15 +469,17 @@ static inline int try_module_get(struct
int ret = 1;
if (module) {
- unsigned int cpu = get_cpu();
+ preempt_disable();
+
if (likely(module_is_live(module))) {
- local_inc(__module_ref_addr(module, cpu));
+ __this_cpu_inc(module->refptr->count);
trace_module_get(module, _THIS_IP_,
- local_read(__module_ref_addr(module, cpu)));
+ __this_cpu_read(module->refptr->count));
}
else
ret = 0;
- put_cpu();
+
+ preempt_enable();
}
return ret;
}
Index: linux-2.6/kernel/module.c
===================================================================
--- linux-2.6.orig/kernel/module.c 2009-12-18 13:13:24.000000000 -0600
+++ linux-2.6/kernel/module.c 2009-12-18 14:15:57.000000000 -0600
@@ -474,9 +474,10 @@ static void module_unload_init(struct mo
INIT_LIST_HEAD(&mod->modules_which_use_me);
for_each_possible_cpu(cpu)
- local_set(__module_ref_addr(mod, cpu), 0);
+ per_cpu_ptr(mod->refptr, cpu)->count = 0;
+
/* Hold reference count during initialization. */
- local_set(__module_ref_addr(mod, raw_smp_processor_id()), 1);
+ __this_cpu_write(mod->refptr->count, 1);
/* Backwards compatibility macros put refcount during init. */
mod->waiter = current;
}
@@ -555,6 +556,7 @@ static void module_unload_free(struct mo
kfree(use);
sysfs_remove_link(i->holders_dir, mod->name);
/* There can be at most one match. */
+ free_percpu(i->refptr);
break;
}
}
@@ -619,7 +621,7 @@ unsigned int module_refcount(struct modu
int cpu;
for_each_possible_cpu(cpu)
- total += local_read(__module_ref_addr(mod, cpu));
+ total += per_cpu_ptr(mod->refptr, cpu)->count;
return total;
}
EXPORT_SYMBOL(module_refcount);
@@ -796,14 +798,15 @@ static struct module_attribute refcnt =
void module_put(struct module *module)
{
if (module) {
- unsigned int cpu = get_cpu();
- local_dec(__module_ref_addr(module, cpu));
+ preempt_disable();
+ __this_cpu_dec(module->refptr->count);
+
trace_module_put(module, _RET_IP_,
- local_read(__module_ref_addr(module, cpu)));
+ __this_cpu_read(module->refptr->count));
/* Maybe they're waiting for us to drop reference? */
if (unlikely(!module_is_live(module)))
wake_up_process(module->waiter);
- put_cpu();
+ preempt_enable();
}
}
EXPORT_SYMBOL(module_put);
@@ -1394,9 +1397,9 @@ static void free_module(struct module *m
kfree(mod->args);
if (mod->percpu)
percpu_modfree(mod->percpu);
-#if defined(CONFIG_MODULE_UNLOAD) && defined(CONFIG_SMP)
+#if defined(CONFIG_MODULE_UNLOAD)
if (mod->refptr)
- percpu_modfree(mod->refptr);
+ free_percpu(mod->refptr);
#endif
/* Free lock-classes: */
lockdep_free_key_range(mod->module_core, mod->core_size);
@@ -2159,9 +2162,8 @@ static noinline struct module *load_modu
mod = (void *)sechdrs[modindex].sh_addr;
kmemleak_load_module(mod, hdr, sechdrs, secstrings);
-#if defined(CONFIG_MODULE_UNLOAD) && defined(CONFIG_SMP)
- mod->refptr = percpu_modalloc(sizeof(local_t), __alignof__(local_t),
- mod->name);
+#if defined(CONFIG_MODULE_UNLOAD)
+ mod->refptr = alloc_percpu(struct module_ref);
if (!mod->refptr) {
err = -ENOMEM;
goto free_init;
@@ -2393,8 +2395,8 @@ static noinline struct module *load_modu
kobject_put(&mod->mkobj.kobj);
free_unload:
module_unload_free(mod);
-#if defined(CONFIG_MODULE_UNLOAD) && defined(CONFIG_SMP)
- percpu_modfree(mod->refptr);
+#if defined(CONFIG_MODULE_UNLOAD)
+ free_percpu(mod->refptr);
free_init:
#endif
module_free(mod, mod->module_init);
Index: linux-2.6/kernel/trace/ring_buffer.c
===================================================================
--- linux-2.6.orig/kernel/trace/ring_buffer.c 2009-12-18 13:13:24.000000000 -0600
+++ linux-2.6/kernel/trace/ring_buffer.c 2009-12-18 14:15:57.000000000 -0600
@@ -12,6 +12,7 @@
#include <linux/hardirq.h>
#include <linux/kmemcheck.h>
#include <linux/module.h>
+#include <asm/local.h>
#include <linux/percpu.h>
#include <linux/mutex.h>
#include <linux/init.h>
--
Hello,
On 12/19/2009 07:26 AM, Christoph Lameter wrote:
> Use cpu ops to deal with the per cpu data instead of a local_t. Reduces memory
> requirements, cache footprint and decreases cycle counts.
>
> The this_cpu_xx operations are also used for !SMP mode. Otherwise we could
> not drop the use of __module_ref_addr() which would make per cpu data handling
> complicated. this_cpu_xx operations have their own fallback for !SMP.
>
> The last hold out of users of local_t is the tracing ringbuffer after this patch
> has been applied.
>
> Signed-off-by: Christoph Lameter <[email protected]>
Please keep Rusty Russell cc'd on module changes.
> Index: linux-2.6/kernel/trace/ring_buffer.c
> ===================================================================
> --- linux-2.6.orig/kernel/trace/ring_buffer.c 2009-12-18 13:13:24.000000000 -0600
> +++ linux-2.6/kernel/trace/ring_buffer.c 2009-12-18 14:15:57.000000000 -0600
> @@ -12,6 +12,7 @@
> #include <linux/hardirq.h>
> #include <linux/kmemcheck.h>
> #include <linux/module.h>
> +#include <asm/local.h>
This doesn't belong to this patch, right? I stripped this part out,
added Cc: to Rusty and applied 1, 2, 7 and 8 to percpu branch. I'll
post the updated patch here.
Thanks.
--
tejun
Here's the version I committed to percpu branch.
Thanks.
>From 8af47ddd01364ae3c663c0bc92415c06fe887ba1 Mon Sep 17 00:00:00 2001
From: Christoph Lameter <[email protected]>
Date: Fri, 18 Dec 2009 16:26:24 -0600
Subject: [PATCH] module: Use this_cpu_xx to dynamically allocate counters
Use cpu ops to deal with the per cpu data instead of a local_t. Reduces memory
requirements, cache footprint and decreases cycle counts.
The this_cpu_xx operations are also used for !SMP mode. Otherwise we could
not drop the use of __module_ref_addr() which would make per cpu data handling
complicated. this_cpu_xx operations have their own fallback for !SMP.
The last hold out of users of local_t is the tracing ringbuffer after this patch
has been applied.
Signed-off-by: Christoph Lameter <[email protected]>
Cc: Rusty Russell <[email protected]>
Signed-off-by: Tejun Heo <[email protected]>
---
include/linux/module.h | 38 ++++++++++++++------------------------
kernel/module.c | 30 ++++++++++++++++--------------
2 files changed, 30 insertions(+), 38 deletions(-)
diff --git a/include/linux/module.h b/include/linux/module.h
index 482efc8..9350577 100644
--- a/include/linux/module.h
+++ b/include/linux/module.h
@@ -16,8 +16,7 @@
#include <linux/kobject.h>
#include <linux/moduleparam.h>
#include <linux/tracepoint.h>
-
-#include <asm/local.h>
+#include <linux/percpu.h>
#include <asm/module.h>
#include <trace/events/module.h>
@@ -361,11 +360,9 @@ struct module
/* Destruction function. */
void (*exit)(void);
-#ifdef CONFIG_SMP
- char *refptr;
-#else
- local_t ref;
-#endif
+ struct module_ref {
+ int count;
+ } *refptr;
#endif
#ifdef CONFIG_CONSTRUCTORS
@@ -452,25 +449,16 @@ void __symbol_put(const char *symbol);
#define symbol_put(x) __symbol_put(MODULE_SYMBOL_PREFIX #x)
void symbol_put_addr(void *addr);
-static inline local_t *__module_ref_addr(struct module *mod, int cpu)
-{
-#ifdef CONFIG_SMP
- return (local_t *) (mod->refptr + per_cpu_offset(cpu));
-#else
- return &mod->ref;
-#endif
-}
-
/* Sometimes we know we already have a refcount, and it's easier not
to handle the error case (which only happens with rmmod --wait). */
static inline void __module_get(struct module *module)
{
if (module) {
- unsigned int cpu = get_cpu();
- local_inc(__module_ref_addr(module, cpu));
+ preempt_disable();
+ __this_cpu_inc(module->refptr->count);
trace_module_get(module, _THIS_IP_,
- local_read(__module_ref_addr(module, cpu)));
- put_cpu();
+ __this_cpu_read(module->refptr->count));
+ preempt_enable();
}
}
@@ -479,15 +467,17 @@ static inline int try_module_get(struct module *module)
int ret = 1;
if (module) {
- unsigned int cpu = get_cpu();
+ preempt_disable();
+
if (likely(module_is_live(module))) {
- local_inc(__module_ref_addr(module, cpu));
+ __this_cpu_inc(module->refptr->count);
trace_module_get(module, _THIS_IP_,
- local_read(__module_ref_addr(module, cpu)));
+ __this_cpu_read(module->refptr->count));
}
else
ret = 0;
- put_cpu();
+
+ preempt_enable();
}
return ret;
}
diff --git a/kernel/module.c b/kernel/module.c
index 64787cd..9bc04de 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -474,9 +474,10 @@ static void module_unload_init(struct module *mod)
INIT_LIST_HEAD(&mod->modules_which_use_me);
for_each_possible_cpu(cpu)
- local_set(__module_ref_addr(mod, cpu), 0);
+ per_cpu_ptr(mod->refptr, cpu)->count = 0;
+
/* Hold reference count during initialization. */
- local_set(__module_ref_addr(mod, raw_smp_processor_id()), 1);
+ __this_cpu_write(mod->refptr->count, 1);
/* Backwards compatibility macros put refcount during init. */
mod->waiter = current;
}
@@ -555,6 +556,7 @@ static void module_unload_free(struct module *mod)
kfree(use);
sysfs_remove_link(i->holders_dir, mod->name);
/* There can be at most one match. */
+ free_percpu(i->refptr);
break;
}
}
@@ -619,7 +621,7 @@ unsigned int module_refcount(struct module *mod)
int cpu;
for_each_possible_cpu(cpu)
- total += local_read(__module_ref_addr(mod, cpu));
+ total += per_cpu_ptr(mod->refptr, cpu)->count;
return total;
}
EXPORT_SYMBOL(module_refcount);
@@ -796,14 +798,15 @@ static struct module_attribute refcnt = {
void module_put(struct module *module)
{
if (module) {
- unsigned int cpu = get_cpu();
- local_dec(__module_ref_addr(module, cpu));
+ preempt_disable();
+ __this_cpu_dec(module->refptr->count);
+
trace_module_put(module, _RET_IP_,
- local_read(__module_ref_addr(module, cpu)));
+ __this_cpu_read(module->refptr->count));
/* Maybe they're waiting for us to drop reference? */
if (unlikely(!module_is_live(module)))
wake_up_process(module->waiter);
- put_cpu();
+ preempt_enable();
}
}
EXPORT_SYMBOL(module_put);
@@ -1377,9 +1380,9 @@ static void free_module(struct module *mod)
kfree(mod->args);
if (mod->percpu)
percpu_modfree(mod->percpu);
-#if defined(CONFIG_MODULE_UNLOAD) && defined(CONFIG_SMP)
+#if defined(CONFIG_MODULE_UNLOAD)
if (mod->refptr)
- percpu_modfree(mod->refptr);
+ free_percpu(mod->refptr);
#endif
/* Free lock-classes: */
lockdep_free_key_range(mod->module_core, mod->core_size);
@@ -2145,9 +2148,8 @@ static noinline struct module *load_module(void __user *umod,
mod = (void *)sechdrs[modindex].sh_addr;
kmemleak_load_module(mod, hdr, sechdrs, secstrings);
-#if defined(CONFIG_MODULE_UNLOAD) && defined(CONFIG_SMP)
- mod->refptr = percpu_modalloc(sizeof(local_t), __alignof__(local_t),
- mod->name);
+#if defined(CONFIG_MODULE_UNLOAD)
+ mod->refptr = alloc_percpu(struct module_ref);
if (!mod->refptr) {
err = -ENOMEM;
goto free_init;
@@ -2373,8 +2375,8 @@ static noinline struct module *load_module(void __user *umod,
kobject_put(&mod->mkobj.kobj);
free_unload:
module_unload_free(mod);
-#if defined(CONFIG_MODULE_UNLOAD) && defined(CONFIG_SMP)
- percpu_modfree(mod->refptr);
+#if defined(CONFIG_MODULE_UNLOAD)
+ free_percpu(mod->refptr);
free_init:
#endif
module_free(mod, mod->module_init);
--
1.6.4.2
On 12/21/2009 04:59 PM, Tejun Heo wrote:
> Here's the version I committed to percpu branch.
>
> Thanks.
>
> From 8af47ddd01364ae3c663c0bc92415c06fe887ba1 Mon Sep 17 00:00:00 2001
> From: Christoph Lameter <[email protected]>
> Date: Fri, 18 Dec 2009 16:26:24 -0600
> Subject: [PATCH] module: Use this_cpu_xx to dynamically allocate counters
>
> Use cpu ops to deal with the per cpu data instead of a local_t. Reduces memory
> requirements, cache footprint and decreases cycle counts.
>
> The this_cpu_xx operations are also used for !SMP mode. Otherwise we could
> not drop the use of __module_ref_addr() which would make per cpu data handling
> complicated. this_cpu_xx operations have their own fallback for !SMP.
>
> The last hold out of users of local_t is the tracing ringbuffer after this patch
> has been applied.
>
> Signed-off-by: Christoph Lameter <[email protected]>
> Cc: Rusty Russell <[email protected]>
This was changed to Acked-by as Rusty acked on the previous thread.
Thanks.
--
tejun
On Mon, 21 Dec 2009 06:29:36 pm Tejun Heo wrote:
> @@ -555,6 +556,7 @@ static void module_unload_free(struct module *mod)
> kfree(use);
> sysfs_remove_link(i->holders_dir, mod->name);
> /* There can be at most one match. */
> + free_percpu(i->refptr);
> break;
> }
> }
This looks very wrong.
Rusty
On 12/22/2009 08:28 AM, Rusty Russell wrote:
> On Mon, 21 Dec 2009 06:29:36 pm Tejun Heo wrote:
>> @@ -555,6 +556,7 @@ static void module_unload_free(struct module *mod)
>> kfree(use);
>> sysfs_remove_link(i->holders_dir, mod->name);
>> /* There can be at most one match. */
>> + free_percpu(i->refptr);
>> break;
>> }
>> }
>
> This looks very wrong.
Indeed, thanks for spotting it. Christoph, I'm rolling back all
patches from this series. Please re-post with updates.
Thanks.
--
tejun
On Mon, 21 Dec 2009, Tejun Heo wrote:
>
> > Index: linux-2.6/kernel/trace/ring_buffer.c
> > ===================================================================
> > --- linux-2.6.orig/kernel/trace/ring_buffer.c 2009-12-18 13:13:24.000000000 -0600
> > +++ linux-2.6/kernel/trace/ring_buffer.c 2009-12-18 14:15:57.000000000 -0600
> > @@ -12,6 +12,7 @@
> > #include <linux/hardirq.h>
> > #include <linux/kmemcheck.h>
> > #include <linux/module.h>
> > +#include <asm/local.h>
>
> This doesn't belong to this patch, right? I stripped this part out,
> added Cc: to Rusty and applied 1, 2, 7 and 8 to percpu branch. I'll
> post the updated patch here. Thanks.
If you strip this part out then ringbuffer.c will no longer compile
since it relies on the "#include <local.h>" that is removed by this patch
from the module.h file.
On Mon, 21 Dec 2009, Tejun Heo wrote:
> index 482efc8..9350577 100644
> --- a/include/linux/module.h
> +++ b/include/linux/module.h
> @@ -16,8 +16,7 @@
> #include <linux/kobject.h>
> #include <linux/moduleparam.h>
> #include <linux/tracepoint.h>
> -
> -#include <asm/local.h>
> +#include <linux/percpu.h>
This is going to break ringbuffer.c.
On Tue, 22 Dec 2009, Tejun Heo wrote:
> On 12/22/2009 08:28 AM, Rusty Russell wrote:
> > On Mon, 21 Dec 2009 06:29:36 pm Tejun Heo wrote:
> >> @@ -555,6 +556,7 @@ static void module_unload_free(struct module *mod)
> >> kfree(use);
> >> sysfs_remove_link(i->holders_dir, mod->name);
> >> /* There can be at most one match. */
> >> + free_percpu(i->refptr);
> >> break;
> >> }
> >> }
> >
> > This looks very wrong.
>
> Indeed, thanks for spotting it. Christoph, I'm rolling back all
> patches from this series. Please re-post with updates.
Simply drop the chunk?
On 12/23/2009 12:57 AM, Christoph Lameter wrote:
> On Mon, 21 Dec 2009, Tejun Heo wrote:
>
>>
>>> Index: linux-2.6/kernel/trace/ring_buffer.c
>>> ===================================================================
>>> --- linux-2.6.orig/kernel/trace/ring_buffer.c 2009-12-18 13:13:24.000000000 -0600
>>> +++ linux-2.6/kernel/trace/ring_buffer.c 2009-12-18 14:15:57.000000000 -0600
>>> @@ -12,6 +12,7 @@
>>> #include <linux/hardirq.h>
>>> #include <linux/kmemcheck.h>
>>> #include <linux/module.h>
>>> +#include <asm/local.h>
>>
>> This doesn't belong to this patch, right? I stripped this part out,
>> added Cc: to Rusty and applied 1, 2, 7 and 8 to percpu branch. I'll
>> post the updated patch here. Thanks.
>
> If you strip this part out then ringbuffer.c will no longer compile
> since it relies on the "#include <local.h>" that is removed by this patch
> from the module.h file.
Oh... alright. I'll add that comment and drop the offending chunk and
recommit.
Thanks.
--
tejun
On Wed, 23 Dec 2009, Tejun Heo wrote:
> On 12/23/2009 12:57 AM, Christoph Lameter wrote:
> > On Mon, 21 Dec 2009, Tejun Heo wrote:
> >
> >>
> >>> Index: linux-2.6/kernel/trace/ring_buffer.c
> >>> ===================================================================
> >>> --- linux-2.6.orig/kernel/trace/ring_buffer.c 2009-12-18 13:13:24.000000000 -0600
> >>> +++ linux-2.6/kernel/trace/ring_buffer.c 2009-12-18 14:15:57.000000000 -0600
> >>> @@ -12,6 +12,7 @@
> >>> #include <linux/hardirq.h>
> >>> #include <linux/kmemcheck.h>
> >>> #include <linux/module.h>
> >>> +#include <asm/local.h>
> >>
> >> This doesn't belong to this patch, right? I stripped this part out,
> >> added Cc: to Rusty and applied 1, 2, 7 and 8 to percpu branch. I'll
> >> post the updated patch here. Thanks.
> >
> > If you strip this part out then ringbuffer.c will no longer compile
> > since it relies on the "#include <local.h>" that is removed by this patch
> > from the module.h file.
>
> Oh... alright. I'll add that comment and drop the offending chunk and
> recommit.
So we need a separate patch to deal with remval of the #include
<asm/local.h> from modules.h?
* Christoph Lameter ([email protected]) wrote:
> On Wed, 23 Dec 2009, Tejun Heo wrote:
>
> > On 12/23/2009 12:57 AM, Christoph Lameter wrote:
> > > On Mon, 21 Dec 2009, Tejun Heo wrote:
> > >
> > >>
> > >>> Index: linux-2.6/kernel/trace/ring_buffer.c
> > >>> ===================================================================
> > >>> --- linux-2.6.orig/kernel/trace/ring_buffer.c 2009-12-18 13:13:24.000000000 -0600
> > >>> +++ linux-2.6/kernel/trace/ring_buffer.c 2009-12-18 14:15:57.000000000 -0600
> > >>> @@ -12,6 +12,7 @@
> > >>> #include <linux/hardirq.h>
> > >>> #include <linux/kmemcheck.h>
> > >>> #include <linux/module.h>
> > >>> +#include <asm/local.h>
> > >>
> > >> This doesn't belong to this patch, right? I stripped this part out,
> > >> added Cc: to Rusty and applied 1, 2, 7 and 8 to percpu branch. I'll
> > >> post the updated patch here. Thanks.
> > >
> > > If you strip this part out then ringbuffer.c will no longer compile
> > > since it relies on the "#include <local.h>" that is removed by this patch
> > > from the module.h file.
> >
> > Oh... alright. I'll add that comment and drop the offending chunk and
> > recommit.
>
> So we need a separate patch to deal with remval of the #include
> <asm/local.h> from modules.h?
>
I think this would make sense, yes. This would be a patch specific to
the ring-buffer code that would go through (or be acked by) Steven.
Mathieu
--
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68