linux-next: tip tree build warning

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 431 bytes --]

Hi all,

Today's linux-next build (powerpc allnoconfig) produced this warning:

kernel/sched.c:122: warning: 'double_rq_lock' declared 'static' but never defined

Introduced by commit 18a3885fc1ffa92c2212ff0afdf033403d5b0fa0 ("sched:
Remove reciprocal for cpu_power").  This is a build with CONFIG_SMP
undefined.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: tip tree build warning

[Geert Uytterhoeven]

On Fri, Sep 11, 2009 at 10:56, Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> Hi all,
>
> Today's linux-next build (powerpc allnoconfig) produced this warning:
>
> kernel/sched.c:122: warning: 'double_rq_lock' declared 'static' but never defined
>
> Introduced by commit 18a3885fc1ffa92c2212ff0afdf033403d5b0fa0 ("sched:
> Remove reciprocal for cpu_power").  This is a build with CONFIG_SMP
> undefined.

Yep. same here.

It's even more fishy:
  - double_rq_lock() is useded inside #ifdef CONFIG_PREEMPT
  - double_rq_lock() is defined inside #ifdef CONFIG_SMP
  - double_rq_lock() is used inside #ifdef CONFIG_HOTPLUG_CPU

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 377 bytes --]

Hi all,

Today's linux-next build (powerpc allyesconfig) produced this warning:

kernel/sched.c:1636: warning: 'update_shares_locked' defined but not used

Introduced by commit f492e12ef050e02bf0185b6b57874992591b9be1 ("sched:
Remove load_balance_newidle()").

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 414 bytes --]

Hi all,

Today's linux-next build (powerpc allnoconfig) produced this warning:

kernel/sched.c: In function 'wake_up_new_task':
kernel/sched.c:2631: warning: unused variable 'cpu'

Introduced by commit fabf318e5e4bda0aca2b0d617b191884fda62703 ("sched:
Fix fork vs hotplug vs cpuset namespaces").

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 494 bytes --]

Hi all,

Today's (and the previous few) linux-next build (powerpc allyesconfig)
produced this warning:

lib/dma-debug.c: In function 'dma_debug_device_change':
lib/dma-debug.c:680: warning: 'return' with no value, in function returning non-void

Introduced by commit f797d9881b62c2ddb1d2e7bd80d87141949c84aa
("dma-debug: Do not add notifier when dma debugging is disabled").

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 624 bytes --]

Hi all,

Today's linux-next build (powerpc allyesconfig) produced this warning:

In file included from kernel/trace/trace.h:14,
                 from kernel/trace/trace_selftest_dynamic.c:1:
include/linux/hw_breakpoint.h: In function 'modify_user_hw_breakpoint':
include/linux/hw_breakpoint.h:96: warning: return makes integer from pointer without a cast

Introduced by commit 44234adcdce38f83c56e05f808ce656175b4beeb
("hw-breakpoints: Modify breakpoints without unregistering them") from
the tip tree.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 421 bytes --]

Hi all,

Today's linux-next build (powerpc ppc64_defconfig) produced this warning:

kernel/perf_event.c:4306: warning: 'bp_perf_event_destroy' defined but not used

Introduced by commit 24f1e32c60c45c89a997c73395b69c8af6f0a84e
("hw-breakpoints: Rewrite the hw-breakpoints layer on top of perf
events").

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: tip tree build warning

[Frederic Weisbecker]

On Wed, Nov 25, 2009 at 09:53:38PM +1100, Stephen Rothwell wrote:
> Hi all,
> 
> Today's linux-next build (powerpc ppc64_defconfig) produced this warning:
> 
> kernel/perf_event.c:4306: warning: 'bp_perf_event_destroy' defined but not used
> 
> Introduced by commit 24f1e32c60c45c89a997c73395b69c8af6f0a84e
> ("hw-breakpoints: Rewrite the hw-breakpoints layer on top of perf
> events").



Oh I see, it's not used in the off-case. I'll fix that soon.
Thanks for the report!

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 489 bytes --]

Hi all,

Today's linux-next build (x86_64 allmodconfig) produced this warning:

arch/x86/kernel/apic/apic.c: In function 'calibrate_APIC_clock':
arch/x86/kernel/apic/apic.c:650: warning: format '%ld' expects type 'long int', but argument 2 has type 'u32'

Introduced by commit 23af368e9a904f59256c27d371ce223d6cee0430
("clockevents: Use u32 for mult and shift factors").

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: tip tree build warning

[Randy Dunlap]

On Mon, 16 Nov 2009 16:25:03 +1100 Stephen Rothwell wrote:

> Hi all,
> 
> Today's linux-next build (x86_64 allmodconfig) produced this warning:
> 
> arch/x86/kernel/apic/apic.c: In function 'calibrate_APIC_clock':
> arch/x86/kernel/apic/apic.c:650: warning: format '%ld' expects type 'long int', but argument 2 has type 'u32'
> 
> Introduced by commit 23af368e9a904f59256c27d371ce223d6cee0430
> ("clockevents: Use u32 for mult and shift factors").


also this warning:

arch/x86/kernel/vmiclock_32.c:230: warning: format '%lu' expects type 'long unsigned int', but argument 3 has type 'u32'

(i386 build)

---
~Randy

Re: linux-next: tip tree build warning

[Randy Dunlap]

On Mon, 16 Nov 2009 09:05:07 -0800 Randy Dunlap wrote:

> On Mon, 16 Nov 2009 16:25:03 +1100 Stephen Rothwell wrote:
> 
> > Hi all,
> > 
> > Today's linux-next build (x86_64 allmodconfig) produced this warning:
> > 
> > arch/x86/kernel/apic/apic.c: In function 'calibrate_APIC_clock':
> > arch/x86/kernel/apic/apic.c:650: warning: format '%ld' expects type 'long int', but argument 2 has type 'u32'
> > 
> > Introduced by commit 23af368e9a904f59256c27d371ce223d6cee0430
> > ("clockevents: Use u32 for mult and shift factors").
> 
> 
> also this warning:
> 
> arch/x86/kernel/vmiclock_32.c:230: warning: format '%lu' expects type 'long unsigned int', but argument 3 has type 'u32'
> 
> (i386 build)


Thomas posted a patch for this, but I'm still seeing this warning (20091123),
so I guess his patch hasn't been merged/pushed yet?

thanks,
---
~Randy

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1018 bytes --]

Hi all,

Today's linux-next build (x86_64 allmodconfig) produced this warning:

In file included from arch/x86/include/asm/smp.h:13,
                 from arch/x86/include/asm/mmzone_64.h:12,
                 from arch/x86/include/asm/mmzone.h:4,
                 from include/linux/mmzone.h:783,
                 from include/linux/gfp.h:4,
                 from include/linux/kmod.h:22,
                 from include/linux/module.h:13,
                 from arch/x86/kernel/apic/apic_noop.c:14:
arch/x86/include/asm/apic.h: In function 'default_apicid_to_cpu_present':
arch/x86/include/asm/apic.h:591: warning: the frame size of 8192 bytes is larger than 2048 bytes

It may not have been caused by the tip tree, but I can't find what
changed to cause this and a commit from the tip tree has exposed it
(9844ab11c763bfed9f054c82366b19dcda66aca9 "x86, apic: Introduce the NOOP
apic driver").
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: tip tree build warning

[Ingo Molnar]

* Stephen Rothwell <sfr@canb.auug.org.au> wrote:

> Hi all,
> 
> Today's linux-next build (x86_64 allmodconfig) produced this warning:
> 
> In file included from arch/x86/include/asm/smp.h:13,
>                  from arch/x86/include/asm/mmzone_64.h:12,
>                  from arch/x86/include/asm/mmzone.h:4,
>                  from include/linux/mmzone.h:783,
>                  from include/linux/gfp.h:4,
>                  from include/linux/kmod.h:22,
>                  from include/linux/module.h:13,
>                  from arch/x86/kernel/apic/apic_noop.c:14:
> arch/x86/include/asm/apic.h: In function 'default_apicid_to_cpu_present':
> arch/x86/include/asm/apic.h:591: warning: the frame size of 8192 bytes is larger than 2048 bytes
> 
> It may not have been caused by the tip tree, but I can't find what 
> changed to cause this and a commit from the tip tree has exposed it 
> (9844ab11c763bfed9f054c82366b19dcda66aca9 "x86, apic: Introduce the 
> NOOP apic driver").

That commit is very simple. Are you sure it's not GCC bogosity?

	Ingo

Re: linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 760 bytes --]

Hi Ingo,

On Wed, 28 Oct 2009 08:31:45 +0100 Ingo Molnar <mingo@elte.hu> wrote:
>
> 
> That commit is very simple. Are you sure it's not GCC bogosity?

That commit just produces the warning (since it introduced the
apic_noop.c file) and probably has nothing directly to do with the
problem (as I said).

The relevant part of apic.h is:

static inline physid_mask_t default_apicid_to_cpu_present(int phys_apicid)
{
        return physid_mask_of_physid(phys_apicid);
}

Where physid_mask_of_physid() manipulates a physid_mask_t which contains
an array of unsigned longs that is BITS_TO_LONGS(MAX_APICS) long and
MAX_APICS is 32768.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 607 bytes --]

Hi all,

On Wed, 28 Oct 2009 18:41:26 +1100 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
>
> static inline physid_mask_t default_apicid_to_cpu_present(int phys_apicid)
> {
>         return physid_mask_of_physid(phys_apicid);
> }

I just noticed that this function (default_apicid_to_cpu_present) is
declared "static inline in a header" but looks like it is only used by
assigning its address to a function pointer.  Its only use for x86_64 is
in arch/x86/kernel/apic/apic_noop.c ...

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: tip tree build warning

[Ingo Molnar]

* Stephen Rothwell <sfr@canb.auug.org.au> wrote:

> Hi all,
> 
> On Wed, 28 Oct 2009 18:41:26 +1100 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> >
> > static inline physid_mask_t default_apicid_to_cpu_present(int phys_apicid)
> > {
> >         return physid_mask_of_physid(phys_apicid);
> > }
> 
> I just noticed that this function (default_apicid_to_cpu_present) is 
> declared "static inline in a header" but looks like it is only used by 
> assigning its address to a function pointer.  Its only use for x86_64 
> is in arch/x86/kernel/apic/apic_noop.c ...

yes, that might be a real problem - returning the mask like that is 
messy. Thanks, will check.

	Ingo

Re: linux-next: tip tree build warning

[Cyrill Gorcunov]

[Ingo Molnar - Wed, Oct 28, 2009 at 08:50:12AM +0100]
| 
| * Stephen Rothwell <sfr@canb.auug.org.au> wrote:
| 
| > Hi all,
| > 
| > On Wed, 28 Oct 2009 18:41:26 +1100 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
| > >
| > > static inline physid_mask_t default_apicid_to_cpu_present(int phys_apicid)
| > > {
| > >         return physid_mask_of_physid(phys_apicid);
| > > }
| > 
| > I just noticed that this function (default_apicid_to_cpu_present) is 
| > declared "static inline in a header" but looks like it is only used by 
| > assigning its address to a function pointer.  Its only use for x86_64 
| > is in arch/x86/kernel/apic/apic_noop.c ...
| 
| yes, that might be a real problem - returning the mask like that is 
| messy. Thanks, will check.
| 
| 	Ingo
| 

Darn, my fault sorry! Here is an update which fixes the issue.
(Btw, Stephen could you CC me next time if you get commit id
 with me in authors, so I wouldn't miss problem).

Please review, comments/complains are quite welcome!

	-- Cyrill
---
x86,apic: Get rid of apicid_to_cpu_present assign on X86-64

In fact it's never get used on x86-64 (for 64 bit platform
we use differ technique to enumerate io-units).

Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
---
 arch/x86/kernel/apic/apic_noop.c |    5 +++++
 1 file changed, 5 insertions(+)

Index: linux-2.6.git/arch/x86/kernel/apic/apic_noop.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/apic_noop.c
+++ linux-2.6.git/arch/x86/kernel/apic/apic_noop.c
@@ -162,7 +162,12 @@ struct apic apic_noop = {
 
 	.cpu_to_logical_apicid		= noop_cpu_to_logical_apicid,
 	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
+
+#ifdef CONFIG_X86_32
 	.apicid_to_cpu_present		= default_apicid_to_cpu_present,
+#else
+	.apicid_to_cpu_present		= NULL,
+#endif
 
 	.setup_portio_remap		= NULL,
 	.check_phys_apicid_present	= default_check_phys_apicid_present,

Re: linux-next: tip tree build warning

[Ingo Molnar]

* Cyrill Gorcunov <gorcunov@openvz.org> wrote:

> [Ingo Molnar - Wed, Oct 28, 2009 at 08:50:12AM +0100]
> | 
> | * Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> | 
> | > Hi all,
> | > 
> | > On Wed, 28 Oct 2009 18:41:26 +1100 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> | > >
> | > > static inline physid_mask_t default_apicid_to_cpu_present(int phys_apicid)
> | > > {
> | > >         return physid_mask_of_physid(phys_apicid);
> | > > }
> | > 
> | > I just noticed that this function (default_apicid_to_cpu_present) is 
> | > declared "static inline in a header" but looks like it is only used by 
> | > assigning its address to a function pointer.  Its only use for x86_64 
> | > is in arch/x86/kernel/apic/apic_noop.c ...
> | 
> | yes, that might be a real problem - returning the mask like that is 
> | messy. Thanks, will check.
> | 
> | 	Ingo
> | 
> 
> Darn, my fault sorry! Here is an update which fixes the issue.
> (Btw, Stephen could you CC me next time if you get commit id
>  with me in authors, so I wouldn't miss problem).
> 
> Please review, comments/complains are quite welcome!
> 
> 	-- Cyrill
> ---
> x86,apic: Get rid of apicid_to_cpu_present assign on X86-64
> 
> In fact it's never get used on x86-64 (for 64 bit platform
> we use differ technique to enumerate io-units).
> 
> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
> ---
>  arch/x86/kernel/apic/apic_noop.c |    5 +++++
>  1 file changed, 5 insertions(+)
> 
> Index: linux-2.6.git/arch/x86/kernel/apic/apic_noop.c
> =====================================================================
> --- linux-2.6.git.orig/arch/x86/kernel/apic/apic_noop.c
> +++ linux-2.6.git/arch/x86/kernel/apic/apic_noop.c
> @@ -162,7 +162,12 @@ struct apic apic_noop = {
>  
>  	.cpu_to_logical_apicid		= noop_cpu_to_logical_apicid,
>  	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
> +
> +#ifdef CONFIG_X86_32
>  	.apicid_to_cpu_present		= default_apicid_to_cpu_present,
> +#else
> +	.apicid_to_cpu_present		= NULL,
> +#endif

would be better to unify this instead ...

	Ingo

Re: linux-next: tip tree build warning

[Cyrill Gorcunov]

[Ingo Molnar - Sun, Nov 08, 2009 at 02:32:47PM +0100]
...
| 
| > +
| > +#ifdef CONFIG_X86_32
| >  	.apicid_to_cpu_present		= default_apicid_to_cpu_present,
| > +#else
| > +	.apicid_to_cpu_present		= NULL,
| > +#endif
| 
| would be better to unify this instead ...
| 
| 	Ingo
| 

OK, wait a bit.

	-- Cyrill

Re: linux-next: tip tree build warning

[Cyrill Gorcunov]

[Ingo Molnar - Sun, Nov 08, 2009 at 02:32:47PM +0100]
| 
...
| > +
| > +#ifdef CONFIG_X86_32
| >  	.apicid_to_cpu_present		= default_apicid_to_cpu_present,
| > +#else
| > +	.apicid_to_cpu_present		= NULL,
| > +#endif
| 
| would be better to unify this instead ...
| 
| 	Ingo
| 

[not for inclusion]

Here is what I'm going to implement (it is not finished yet,
but just to show the idea -- ie to get rid of physid_mask_t
passed as an argument at all but use pointers instead since
callers already have bitmask allocated).

And physid_set_mask_of_physid already do the work for us.
Hmm?

(Ingo, I think you may apply the former patch just to
 have the issue fixed this way temporary)

	-- Cyrill
---
 arch/x86/include/asm/apic.h      |    7 +------
 arch/x86/kernel/apic/apic_noop.c |    2 +-
 arch/x86/kernel/apic/bigsmp_32.c |    7 +------
 arch/x86/kernel/apic/probe_32.c  |    2 +-
 4 files changed, 4 insertions(+), 14 deletions(-)

Index: linux-2.6.git/arch/x86/include/asm/apic.h
=====================================================================
--- linux-2.6.git.orig/arch/x86/include/asm/apic.h
+++ linux-2.6.git/arch/x86/include/asm/apic.h
@@ -310,7 +310,7 @@ struct apic {
 	int (*apicid_to_node)(int logical_apicid);
 	int (*cpu_to_logical_apicid)(int cpu);
 	int (*cpu_present_to_apicid)(int mps_cpu);
-	physid_mask_t (*apicid_to_cpu_present)(int phys_apicid);
+	void (*apicid_to_cpu_present)(int phys_apicid, physid_mask_t *bitmap);
 	void (*setup_portio_remap)(void);
 	int (*check_phys_apicid_present)(int phys_apicid);
 	void (*enable_apic_mode)(void);
@@ -585,11 +585,6 @@ extern int default_cpu_present_to_apicid
 extern int default_check_phys_apicid_present(int phys_apicid);
 #endif
 
-static inline physid_mask_t default_apicid_to_cpu_present(int phys_apicid)
-{
-	return physid_mask_of_physid(phys_apicid);
-}
-
 #endif /* CONFIG_X86_LOCAL_APIC */
 
 #ifdef CONFIG_X86_32
Index: linux-2.6.git/arch/x86/kernel/apic/apic_noop.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/apic_noop.c
+++ linux-2.6.git/arch/x86/kernel/apic/apic_noop.c
@@ -162,7 +162,7 @@ struct apic apic_noop = {
 
 	.cpu_to_logical_apicid		= noop_cpu_to_logical_apicid,
 	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
-	.apicid_to_cpu_present		= default_apicid_to_cpu_present,
+	.apicid_to_cpu_present		= physid_set_mask_of_physid,
 
 	.setup_portio_remap		= NULL,
 	.check_phys_apicid_present	= default_check_phys_apicid_present,
Index: linux-2.6.git/arch/x86/kernel/apic/bigsmp_32.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/bigsmp_32.c
+++ linux-2.6.git/arch/x86/kernel/apic/bigsmp_32.c
@@ -93,11 +93,6 @@ static int bigsmp_cpu_present_to_apicid(
 	return BAD_APICID;
 }
 
-static physid_mask_t bigsmp_apicid_to_cpu_present(int phys_apicid)
-{
-	return physid_mask_of_physid(phys_apicid);
-}
-
 /* Mapping from cpu number to logical apicid */
 static inline int bigsmp_cpu_to_logical_apicid(int cpu)
 {
@@ -230,7 +225,7 @@ struct apic apic_bigsmp = {
 	.apicid_to_node			= bigsmp_apicid_to_node,
 	.cpu_to_logical_apicid		= bigsmp_cpu_to_logical_apicid,
 	.cpu_present_to_apicid		= bigsmp_cpu_present_to_apicid,
-	.apicid_to_cpu_present		= bigsmp_apicid_to_cpu_present,
+	.apicid_to_cpu_present		= physid_set_mask_of_physid,
 	.setup_portio_remap		= NULL,
 	.check_phys_apicid_present	= bigsmp_check_phys_apicid_present,
 	.enable_apic_mode		= NULL,
Index: linux-2.6.git/arch/x86/kernel/apic/probe_32.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/probe_32.c
+++ linux-2.6.git/arch/x86/kernel/apic/probe_32.c
@@ -108,7 +108,7 @@ struct apic apic_default = {
 	.apicid_to_node			= default_apicid_to_node,
 	.cpu_to_logical_apicid		= default_cpu_to_logical_apicid,
 	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
-	.apicid_to_cpu_present		= default_apicid_to_cpu_present,
+	.apicid_to_cpu_present		= physid_set_mask_of_physid,
 	.setup_portio_remap		= NULL,
 	.check_phys_apicid_present	= default_check_phys_apicid_present,
 	.enable_apic_mode		= NULL,

Re: linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 471 bytes --]

Hi Cyril,

On Sun, 8 Nov 2009 16:16:45 +0300 Cyrill Gorcunov <gorcunov@openvz.org> wrote:
>
> (Btw, Stephen could you CC me next time if you get commit id
>  with me in authors, so I wouldn't miss problem).

Yeah, sorry about that - at the time I didn't know if this was caused
directly by your commit or was just a side effect of some some other tree.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: linux-next: tip tree build warning

[Cyrill Gorcunov]

[Ingo Molnar - Wed, Oct 28, 2009 at 08:50:12AM +0100]
| 
| * Stephen Rothwell <sfr@canb.auug.org.au> wrote:
| 
| > Hi all,
| > 
| > On Wed, 28 Oct 2009 18:41:26 +1100 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
| > >
| > > static inline physid_mask_t default_apicid_to_cpu_present(int phys_apicid)
| > > {
| > >         return physid_mask_of_physid(phys_apicid);
| > > }
| > 
| > I just noticed that this function (default_apicid_to_cpu_present) is 
| > declared "static inline in a header" but looks like it is only used by 
| > assigning its address to a function pointer.  Its only use for x86_64 
| > is in arch/x86/kernel/apic/apic_noop.c ...
| 
| yes, that might be a real problem - returning the mask like that is 
| messy. Thanks, will check.
| 
| 	Ingo
|

ok, here is what I've just cooked. Please review, I've CC'ed a few
"knowing-apic-code-quite-well" persons just to be sure.

Note that we still have a few items in "struct apic" which
operate with physid_mask_t on stack.

I think perhaps it's a good idea to touch this data via pointers
povided by caller:

	physid_mask_t (*ioapic_phys_id_map)(physid_mask_t map);

I didn't manage to cover it today -- will do tomorrow if patch
approach would be approved.

Also, it's just warning (yet) since we don't use those routines
in x86-64 so it doesn't harm.

Anyway, please review, comment, complain and etc... would appreciate.

	-- Cyrill

p.s.: i don't have inet access in office so will be able to reply
at tomorrow evening only.
---
x86,apic: Do not use stack'ed physid_mask_t in apicid_to_cpu_present

Stephen Rothwell pointed out that apic-noop (when gets compiled
in x86-84 environment) potentially may consume too much stack space.

|
| Hi all,
|
| Today's linux-next build (x86_64 allmodconfig) produced this warning:
|
| In file included from arch/x86/include/asm/smp.h:13,
|                 from arch/x86/include/asm/mmzone_64.h:12,
|                 from arch/x86/include/asm/mmzone.h:4,
|                 from include/linux/mmzone.h:783,
|                 from include/linux/gfp.h:4,
|                 from include/linux/kmod.h:22,
|                 from include/linux/module.h:13,
|                 from arch/x86/kernel/apic/apic_noop.c:14:
| arch/x86/include/asm/apic.h: In function 'default_apicid_to_cpu_present':
| arch/x86/include/asm/apic.h:591: warning: the frame size of 8192 bytes is larger than 2048 bytes
|
| It may not have been caused by the tip tree, but I can't find what
| changed to cause this and a commit from the tip tree has exposed it
| (9844ab11c763bfed9f054c82366b19dcda66aca9 "x86, apic: Introduce the NOOP
| apic driver").
|

So I would say this is a bug in apic-noop (in fact we don't use
default_apicid_to_cpu_present if operate in 64bit mode but it's a sign
that something is wrong with code design). The key problem is that
physid_mask_t is an array with a size depending on MAX_APICS, which
in turn is big enough on x86-64 to trigger compiler warning.

So to prevent such a situation in future we should use physid_mask_t
pointer leaving apic driver with a task to operate over data but not
allocate it. Caller should instead.

This allow us throw out some code as well since physid_set_mask_of_physid
already implement the functionality we need.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
---
 arch/x86/include/asm/apic.h      |    7 +------
 arch/x86/kernel/apic/apic_noop.c |    2 +-
 arch/x86/kernel/apic/bigsmp_32.c |    7 +------
 arch/x86/kernel/apic/es7000_32.c |    8 ++------
 arch/x86/kernel/apic/io_apic.c   |    4 ++--
 arch/x86/kernel/apic/numaq_32.c  |    4 ++--
 arch/x86/kernel/apic/probe_32.c  |    2 +-
 arch/x86/kernel/apic/summit_32.c |    4 ++--
 arch/x86/kernel/visws_quirks.c   |    2 +-
 9 files changed, 13 insertions(+), 27 deletions(-)

Index: linux-2.6.git/arch/x86/include/asm/apic.h
=====================================================================
--- linux-2.6.git.orig/arch/x86/include/asm/apic.h
+++ linux-2.6.git/arch/x86/include/asm/apic.h
@@ -310,7 +310,7 @@ struct apic {
 	int (*apicid_to_node)(int logical_apicid);
 	int (*cpu_to_logical_apicid)(int cpu);
 	int (*cpu_present_to_apicid)(int mps_cpu);
-	physid_mask_t (*apicid_to_cpu_present)(int phys_apicid);
+	void (*apicid_to_cpu_present)(int phys_apicid, physid_mask_t *map);
 	void (*setup_portio_remap)(void);
 	int (*check_phys_apicid_present)(int phys_apicid);
 	void (*enable_apic_mode)(void);
@@ -585,11 +585,6 @@ extern int default_cpu_present_to_apicid
 extern int default_check_phys_apicid_present(int phys_apicid);
 #endif
 
-static inline physid_mask_t default_apicid_to_cpu_present(int phys_apicid)
-{
-	return physid_mask_of_physid(phys_apicid);
-}
-
 #endif /* CONFIG_X86_LOCAL_APIC */
 
 #ifdef CONFIG_X86_32
Index: linux-2.6.git/arch/x86/kernel/apic/apic_noop.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/apic_noop.c
+++ linux-2.6.git/arch/x86/kernel/apic/apic_noop.c
@@ -162,7 +162,7 @@ struct apic apic_noop = {
 
 	.cpu_to_logical_apicid		= noop_cpu_to_logical_apicid,
 	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
-	.apicid_to_cpu_present		= default_apicid_to_cpu_present,
+	.apicid_to_cpu_present		= physid_set_mask_of_physid,
 
 	.setup_portio_remap		= NULL,
 	.check_phys_apicid_present	= default_check_phys_apicid_present,
Index: linux-2.6.git/arch/x86/kernel/apic/bigsmp_32.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/bigsmp_32.c
+++ linux-2.6.git/arch/x86/kernel/apic/bigsmp_32.c
@@ -93,11 +93,6 @@ static int bigsmp_cpu_present_to_apicid(
 	return BAD_APICID;
 }
 
-static physid_mask_t bigsmp_apicid_to_cpu_present(int phys_apicid)
-{
-	return physid_mask_of_physid(phys_apicid);
-}
-
 /* Mapping from cpu number to logical apicid */
 static inline int bigsmp_cpu_to_logical_apicid(int cpu)
 {
@@ -230,7 +225,7 @@ struct apic apic_bigsmp = {
 	.apicid_to_node			= bigsmp_apicid_to_node,
 	.cpu_to_logical_apicid		= bigsmp_cpu_to_logical_apicid,
 	.cpu_present_to_apicid		= bigsmp_cpu_present_to_apicid,
-	.apicid_to_cpu_present		= bigsmp_apicid_to_cpu_present,
+	.apicid_to_cpu_present		= physid_set_mask_of_physid,
 	.setup_portio_remap		= NULL,
 	.check_phys_apicid_present	= bigsmp_check_phys_apicid_present,
 	.enable_apic_mode		= NULL,
Index: linux-2.6.git/arch/x86/kernel/apic/es7000_32.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/es7000_32.c
+++ linux-2.6.git/arch/x86/kernel/apic/es7000_32.c
@@ -539,14 +539,10 @@ static int es7000_cpu_present_to_apicid(
 
 static int cpu_id;
 
-static physid_mask_t es7000_apicid_to_cpu_present(int phys_apicid)
+static void es7000_apicid_to_cpu_present(int phys_apicid, physid_mask_t *map)
 {
-	physid_mask_t mask;
-
-	mask = physid_mask_of_physid(cpu_id);
+	physid_set_mask_of_physid(cpu_id, map);
 	++cpu_id;
-
-	return mask;
 }
 
 /* Mapping from cpu number to logical apicid */
Index: linux-2.6.git/arch/x86/kernel/apic/io_apic.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/io_apic.c
+++ linux-2.6.git/arch/x86/kernel/apic/io_apic.c
@@ -2073,7 +2073,7 @@ void __init setup_ioapic_ids_from_mpc(vo
 			mp_ioapics[apic_id].apicid = i;
 		} else {
 			physid_mask_t tmp;
-			tmp = apic->apicid_to_cpu_present(mp_ioapics[apic_id].apicid);
+			apic->apicid_to_cpu_present(mp_ioapics[apic_id].apicid, &tmp);
 			apic_printk(APIC_VERBOSE, "Setting %d in the "
 					"phys_id_present_map\n",
 					mp_ioapics[apic_id].apicid);
@@ -3969,7 +3969,7 @@ int __init io_apic_get_unique_id(int ioa
 		apic_id = i;
 	}
 
-	tmp = apic->apicid_to_cpu_present(apic_id);
+	apic->apicid_to_cpu_present(apic_id, &tmp);
 	physids_or(apic_id_map, apic_id_map, tmp);
 
 	if (reg_00.bits.ID != apic_id) {
Index: linux-2.6.git/arch/x86/kernel/apic/numaq_32.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/numaq_32.c
+++ linux-2.6.git/arch/x86/kernel/apic/numaq_32.c
@@ -402,12 +402,12 @@ static inline int numaq_apicid_to_node(i
 	return logical_apicid >> 4;
 }
 
-static inline physid_mask_t numaq_apicid_to_cpu_present(int logical_apicid)
+static void numaq_apicid_to_cpu_present(int logical_apicid, physid_mask_t *map)
 {
 	int node = numaq_apicid_to_node(logical_apicid);
 	int cpu = __ffs(logical_apicid & 0xf);
 
-	return physid_mask_of_physid(cpu + 4*node);
+	physid_set_mask_of_physid(cpu + 4*node, map);
 }
 
 /* Where the IO area was mapped on multiquad, always 0 otherwise */
Index: linux-2.6.git/arch/x86/kernel/apic/probe_32.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/probe_32.c
+++ linux-2.6.git/arch/x86/kernel/apic/probe_32.c
@@ -108,7 +108,7 @@ struct apic apic_default = {
 	.apicid_to_node			= default_apicid_to_node,
 	.cpu_to_logical_apicid		= default_cpu_to_logical_apicid,
 	.cpu_present_to_apicid		= default_cpu_present_to_apicid,
-	.apicid_to_cpu_present		= default_apicid_to_cpu_present,
+	.apicid_to_cpu_present		= physid_set_mask_of_physid,
 	.setup_portio_remap		= NULL,
 	.check_phys_apicid_present	= default_check_phys_apicid_present,
 	.enable_apic_mode		= NULL,
Index: linux-2.6.git/arch/x86/kernel/apic/summit_32.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic/summit_32.c
+++ linux-2.6.git/arch/x86/kernel/apic/summit_32.c
@@ -267,9 +267,9 @@ static physid_mask_t summit_ioapic_phys_
 	return physids_promote(0x0F);
 }
 
-static physid_mask_t summit_apicid_to_cpu_present(int apicid)
+static void summit_apicid_to_cpu_present(int apicid, physid_mask_t *map)
 {
-	return physid_mask_of_physid(0);
+	physid_set_mask_of_physid(0, map);
 }
 
 static int summit_check_phys_apicid_present(int physical_apicid)
Index: linux-2.6.git/arch/x86/kernel/visws_quirks.c
=====================================================================
--- linux-2.6.git.orig/arch/x86/kernel/visws_quirks.c
+++ linux-2.6.git/arch/x86/kernel/visws_quirks.c
@@ -183,7 +183,7 @@ static void __init MP_processor_info(str
 		return;
 	}
 
-	apic_cpus = apic->apicid_to_cpu_present(m->apicid);
+	apic->apicid_to_cpu_present(m->apicid, &apic_cpus);
 	physids_or(phys_cpu_present_map, phys_cpu_present_map, apic_cpus);
 	/*
 	 * Validate version

Re: linux-next: tip tree build warning

[Ingo Molnar]

* Cyrill Gorcunov <gorcunov@gmail.com> wrote:

> +++ linux-2.6.git/arch/x86/include/asm/apic.h
> @@ -310,7 +310,7 @@ struct apic {
>  	int (*apicid_to_node)(int logical_apicid);
>  	int (*cpu_to_logical_apicid)(int cpu);
>  	int (*cpu_present_to_apicid)(int mps_cpu);
> -	physid_mask_t (*apicid_to_cpu_present)(int phys_apicid);
> +	void (*apicid_to_cpu_present)(int phys_apicid, physid_mask_t *map);

Yep, passing masks by reference is unconditonal goodness - that's the 
cleanup we want.

Mind sending a delta patch against tip:master?

	Ingo

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 503 bytes --]

Hi all,

Today's linux-next build (x86_64 allmodconfig) produced this warning:

arch/x86/kernel/cpu/perf_event.c: In function 'intel_get_event_idx':
arch/x86/kernel/cpu/perf_event.c:1445: warning: 'event_constraint' is used uninitialized in this function

Introduced by commit b690081d4d3f6a23541493f1682835c3cd5c54a1
("perf_events: Add event constraints support for Intel processors").
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

linux-next: tip tree build warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 720 bytes --]

Hi all,

Today's linux-next build (powerpc ppc64_defconfig) produced these warnings:

kernel/trace/ring_buffer.c: In function 'rb_head_page_set':
kernel/trace/ring_buffer.c:704: warning: initialization makes pointer from integer without a cast
kernel/trace/ring_buffer.c:704: warning: initialization makes pointer from integer without a cast
kernel/trace/ring_buffer.c: In function 'rb_head_page_replace':
kernel/trace/ring_buffer.c:797: warning: initialization makes integer from pointer without a cast

Introduced by commit 77ae365eca895061c8bf2b2e3ae1d9ea62869739
("ring-buffer: make lockless").

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: linux-next: tip tree build warning

[Steven Rostedt]

On Tue, 2009-08-04 at 16:16 +1000, Stephen Rothwell wrote:
> Hi all,
> 
> Today's linux-next build (powerpc ppc64_defconfig) produced these warnings:
> 
> kernel/trace/ring_buffer.c: In function 'rb_head_page_set':
> kernel/trace/ring_buffer.c:704: warning: initialization makes pointer from integer without a cast
> kernel/trace/ring_buffer.c:704: warning: initialization makes pointer from integer without a cast
> kernel/trace/ring_buffer.c: In function 'rb_head_page_replace':
> kernel/trace/ring_buffer.c:797: warning: initialization makes integer from pointer without a cast
> 
> Introduced by commit 77ae365eca895061c8bf2b2e3ae1d9ea62869739
> ("ring-buffer: make lockless").
> 

Thanks, I'll take a look at it.

-- Steve