Re: [PATCH] KVM: use modern cpumask primitives, no cpumask_t on stack

Re: [PATCH] KVM: use modern cpumask primitives, no cpumask_t on stack

[Hollis Blanchard]

On Mon, 2008-12-08 at 16:09 +0000, Avi Kivity wrote:
> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> index ba4275d..2d6ca79 100644
> --- a/virt/kvm/kvm_main.c
> +++ b/virt/kvm/kvm_main.c
> @@ -568,14 +570,17 @@ static bool make_all_cpus_request(struct kvm *kvm, unsigned int req)
>  		if (test_and_set_bit(req, &vcpu->requests))
>  			continue;
>  		cpu = vcpu->cpu;
> -		if (cpu != -1 && cpu != me)
> -			cpu_set(cpu, cpus);
> -	}
> -	if (!cpus_empty(cpus)) {
> -		smp_call_function_mask(cpus, ack_flush, NULL, 1);
> -		called = true;
> +		if (cpus != NULL && cpu != -1 && cpu != me)
> +			cpumask_set_cpu(cpu, cpus);
>  	}
> +	if (unlikely(cpus == NULL))
> +		smp_call_function_many(cpu_online_mask, ack_flush, NULL, 1);
> +	else if (!cpumask_empty(cpus))
> +		smp_call_function_many(cpus, ack_flush, NULL, 1);
> +	else
> +		called = false;
>  	put_cpu();
> +	free_cpumask_var(cpus);
>  	return called;
>  }

This patch breaks uniprocessor builds, because smp_call_function_many()
is only defined for CONFIG_SMP.

Avi, I think you should be able to build a PowerPC KVM kernel at this
point? That would have caught this error. Rusty, could you ack the
following:



cpumask: define smp_call_function_many() for non-SMP builds

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

diff --git a/include/linux/smp.h b/include/linux/smp.h
--- a/include/linux/smp.h
+++ b/include/linux/smp.h
@@ -67,13 +67,6 @@ int smp_call_function(void(*func)(void *
 /* Deprecated: use smp_call_function_many() which uses a cpumask ptr. */
 int smp_call_function_mask(cpumask_t mask, void(*func)(void *info), void *info,
                                int wait);
-
-static inline void smp_call_function_many(const struct cpumask *mask,
-                                         void (*func)(void *info), void *info,
-                                         int wait)
-{
-       smp_call_function_mask(*mask, func, info, wait);
-}
 
 int smp_call_function_single(int cpuid, void (*func) (void *info), void *info,
                                int wait);
@@ -151,6 +144,13 @@ static inline void init_call_single_data
 }
 #endif /* !SMP */
 
+static inline void smp_call_function_many(const struct cpumask *mask,
+                                         void (*func)(void *info), void *info,
+                                         int wait)
+{
+       smp_call_function_mask(*mask, func, info, wait);
+}
+
 /*
  * smp_processor_id(): get the current CPU ID.
  *

-- 
Hollis Blanchard
IBM Linux Technology Center

Re: [PATCH] KVM: use modern cpumask primitives, no cpumask_t on stack

[Rusty Russell]

On Friday 12 December 2008 03:21:24 Hollis Blanchard wrote:
> This patch breaks uniprocessor builds, because smp_call_function_many()
> is only defined for CONFIG_SMP.

Good catch.  I missed it because it's fixed as a side-effect of a
later patch in my series (before I convert users).

Linus, can you please apply this fix?

Thanks,
Rusty.

Subject: Define smp_call_function_many for UP

Otherwise those using it in transition patches (eg. kvm) can't compile
with CONFIG_SMP=n:

arch/x86/kvm/../../../virt/kvm/kvm_main.c: In function 'make_all_cpus_request':
arch/x86/kvm/../../../virt/kvm/kvm_main.c:380: error: implicit declaration of function 'smp_call_function_many'

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

diff --git a/include/linux/smp.h b/include/linux/smp.h
--- a/include/linux/smp.h
+++ b/include/linux/smp.h
@@ -146,6 +147,8 @@ static inline void smp_send_reschedule(i
 })
 #define smp_call_function_mask(mask, func, info, wait) \
 			(up_smp_call_function(func, info))
+#define smp_call_function_many(mask, func, info, wait) \
+			(up_smp_call_function(func, info))
 static inline void init_call_single_data(void)
 {
 }

[PATCH] Define smp_call_function_many for UP

[Rusty Russell]

Linus, please apply.  Otherwise those using it in transition patches (eg. kvm)
can't compile with CONFIG_SMP=n:

arch/x86/kvm/../../../virt/kvm/kvm_main.c: In function 'make_all_cpus_request':
arch/x86/kvm/../../../virt/kvm/kvm_main.c:380: error: implicit declaration of function 'smp_call_function_many'

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
---
 include/linux/smp.h |    2 ++
 1 file changed, 2 insertions(+)

diff --git a/include/linux/smp.h b/include/linux/smp.h
--- a/include/linux/smp.h
+++ b/include/linux/smp.h
@@ -146,6 +147,8 @@ static inline void smp_send_reschedule(i
 })
 #define smp_call_function_mask(mask, func, info, wait) \
 			(up_smp_call_function(func, info))
+#define smp_call_function_many(mask, func, info, wait) \
+			(up_smp_call_function(func, info))
 static inline void init_call_single_data(void)
 {
 }

Re: [PATCH] Define smp_call_function_many for UP

[Linus Torvalds]

On Fri, 19 Dec 2008, Rusty Russell wrote:
>
> Linus, please apply.  Otherwise those using it in transition patches (eg. kvm)
> can't compile with CONFIG_SMP=n:

Umm. I already applied this a long time ago. It's commit 
d2ff911882b6bc693d86ca9566daac70aacbb2b3

		Linus

Re: [PATCH] Define smp_call_function_many for UP

[Rusty Russell]

On Friday 19 December 2008 08:46:37 Linus Torvalds wrote:
> 
> On Fri, 19 Dec 2008, Rusty Russell wrote:
> >
> > Linus, please apply.  Otherwise those using it in transition patches (eg. kvm)
> > can't compile with CONFIG_SMP=n:
> 
> Umm. I already applied this a long time ago. It's commit 
> d2ff911882b6bc693d86ca9566daac70aacbb2b3

Wow, it's like Santa came early!

Thanks,
Rusty.

Re: [PATCH] Define smp_call_function_many for UP

[Linus Torvalds]

On Fri, 19 Dec 2008, Rusty Russell wrote:
>
> On Friday 19 December 2008 08:46:37 Linus Torvalds wrote:
> > 
> > Umm. I already applied this a long time ago. It's commit 
> > d2ff911882b6bc693d86ca9566daac70aacbb2b3
> 
> Wow, it's like Santa came early!

Christmas at Rusty's house:

Dreary.

			Linus

Re: [PATCH] Define smp_call_function_many for UP

[Rusty Russell]

On Saturday 20 December 2008 03:33:26 Linus Torvalds wrote:
> Christmas at Rusty's house:
> 
> Dreary.

Far from!  Sure, it's Arabella's first Xmas, but a Very Kernel Christmas can
be lots of fun.  Why just this morning, we did this together!  See if you
can tell which bits are hers...

Subject: Decorate lib/rbtree.c
From: Rusty and Arabella Russell <rustabella@rustcorp.com.au>

A slight enhancement to the comments in lib/rbtree.c.

Signed-off-by: Rusty and Arabella Russell <rustabella@rustcorp.com.au>
---
 lib/rbtree.c |    4 +++-
 1 files changed, 3 insertions(+), 1 deletions(-)

diff --git a/lib/rbtree.c b/lib/rbtree.c
index 48499c2..7848186 100644
--- a/lib/rbtree.c
+++ b/lib/rbtree.c
@@ -17,7 +17,8 @@
   along with this program; if not, write to the Free Software
   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 
-  linux/lib/rbtree.c
+  An excellent description of this algorithm can be found at:
+	http://en.wikipedia.org/wiki/Red-black_tree
 */
 
 #include <linux/rbtree.h>
@@ -291,6 +292,7 @@ EXPORT_SYMBOL(rb_erase);
 
 /*
  * This function returns the first node (in sort order) of the tree.
+ * ssdddsddddddddddddd
  */
 struct rb_node *rb_first(struct rb_root *root)
 {