The patch below does not apply to the 5.4-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <[email protected]>.
Thanks,
Sasha
------------------ original commit in Linus's tree ------------------
From f2d5dcb48f7ba9e3ff249d58fc1fa963d374e66a Mon Sep 17 00:00:00 2001
From: "Matthew Wilcox (Oracle)" <[email protected]>
Date: Tue, 10 Oct 2023 15:55:49 +0100
Subject: [PATCH] bounds: support non-power-of-two CONFIG_NR_CPUS
ilog2() rounds down, so for example when PowerPC 85xx sets CONFIG_NR_CPUS
to 24, we will only allocate 4 bits to store the number of CPUs instead of
5. Use bits_per() instead, which rounds up. Found by code inspection.
The effect of this would probably be a misaccounting when doing NUMA
balancing, so to a user, it would only be a performance penalty. The
effects may be more wide-spread; it's hard to tell.
Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Matthew Wilcox (Oracle) <[email protected]>
Fixes: 90572890d202 ("mm: numa: Change page last {nid,pid} into {cpu,pid}")
Reviewed-by: Rik van Riel <[email protected]>
Acked-by: Mel Gorman <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
---
kernel/bounds.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/bounds.c b/kernel/bounds.c
index b529182e8b04f..c5a9fcd2d6228 100644
--- a/kernel/bounds.c
+++ b/kernel/bounds.c
@@ -19,7 +19,7 @@ int main(void)
DEFINE(NR_PAGEFLAGS, __NR_PAGEFLAGS);
DEFINE(MAX_NR_ZONES, __MAX_NR_ZONES);
#ifdef CONFIG_SMP
- DEFINE(NR_CPUS_BITS, ilog2(CONFIG_NR_CPUS));
+ DEFINE(NR_CPUS_BITS, bits_per(CONFIG_NR_CPUS));
#endif
DEFINE(SPINLOCK_SIZE, sizeof(spinlock_t));
#ifdef CONFIG_LRU_GEN
--
2.43.0
On Wed, Mar 27, 2024 at 08:21:25AM -0400, Sasha Levin wrote:
> The patch below does not apply to the 5.4-stable tree.
> If someone wants it applied there, or to any other stable or longterm
> tree, then please email the backport, including the original git commit
> id to <[email protected]>.
Looks like you just need a little more fuzz on the patch.
diff --git a/kernel/bounds.c b/kernel/bounds.c
index 9795d75b09b2..a94e3769347e 100644
--- a/kernel/bounds.c
+++ b/kernel/bounds.c
@@ -19,7 +19,7 @@ int main(void)
DEFINE(NR_PAGEFLAGS, __NR_PAGEFLAGS);
DEFINE(MAX_NR_ZONES, __MAX_NR_ZONES);
#ifdef CONFIG_SMP
- DEFINE(NR_CPUS_BITS, ilog2(CONFIG_NR_CPUS));
+ DEFINE(NR_CPUS_BITS, bits_per(CONFIG_NR_CPUS));
#endif
DEFINE(SPINLOCK_SIZE, sizeof(spinlock_t));
/* End of constants */
> Thanks,
> Sasha
>
> ------------------ original commit in Linus's tree ------------------
>
> >From f2d5dcb48f7ba9e3ff249d58fc1fa963d374e66a Mon Sep 17 00:00:00 2001
> From: "Matthew Wilcox (Oracle)" <[email protected]>
> Date: Tue, 10 Oct 2023 15:55:49 +0100
> Subject: [PATCH] bounds: support non-power-of-two CONFIG_NR_CPUS
>
> ilog2() rounds down, so for example when PowerPC 85xx sets CONFIG_NR_CPUS
> to 24, we will only allocate 4 bits to store the number of CPUs instead of
> 5. Use bits_per() instead, which rounds up. Found by code inspection.
> The effect of this would probably be a misaccounting when doing NUMA
> balancing, so to a user, it would only be a performance penalty. The
> effects may be more wide-spread; it's hard to tell.
>
> Link: https://lkml.kernel.org/r/[email protected]
> Signed-off-by: Matthew Wilcox (Oracle) <[email protected]>
> Fixes: 90572890d202 ("mm: numa: Change page last {nid,pid} into {cpu,pid}")
> Reviewed-by: Rik van Riel <[email protected]>
> Acked-by: Mel Gorman <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Ingo Molnar <[email protected]>
> Cc: <[email protected]>
> Signed-off-by: Andrew Morton <[email protected]>
> ---
> kernel/bounds.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/bounds.c b/kernel/bounds.c
> index b529182e8b04f..c5a9fcd2d6228 100644
> --- a/kernel/bounds.c
> +++ b/kernel/bounds.c
> @@ -19,7 +19,7 @@ int main(void)
> DEFINE(NR_PAGEFLAGS, __NR_PAGEFLAGS);
> DEFINE(MAX_NR_ZONES, __MAX_NR_ZONES);
> #ifdef CONFIG_SMP
> - DEFINE(NR_CPUS_BITS, ilog2(CONFIG_NR_CPUS));
> + DEFINE(NR_CPUS_BITS, bits_per(CONFIG_NR_CPUS));
> #endif
> DEFINE(SPINLOCK_SIZE, sizeof(spinlock_t));
> #ifdef CONFIG_LRU_GEN
> --
> 2.43.0
>
>
>
>
On Wed, Mar 27, 2024 at 02:10:10PM +0000, Matthew Wilcox wrote:
> On Wed, Mar 27, 2024 at 08:21:25AM -0400, Sasha Levin wrote:
> > The patch below does not apply to the 5.4-stable tree.
> > If someone wants it applied there, or to any other stable or longterm
> > tree, then please email the backport, including the original git commit
> > id to <[email protected]>.
>
> Looks like you just need a little more fuzz on the patch.
>
> diff --git a/kernel/bounds.c b/kernel/bounds.c
> index 9795d75b09b2..a94e3769347e 100644
> --- a/kernel/bounds.c
> +++ b/kernel/bounds.c
> @@ -19,7 +19,7 @@ int main(void)
> DEFINE(NR_PAGEFLAGS, __NR_PAGEFLAGS);
> DEFINE(MAX_NR_ZONES, __MAX_NR_ZONES);
> #ifdef CONFIG_SMP
> - DEFINE(NR_CPUS_BITS, ilog2(CONFIG_NR_CPUS));
> + DEFINE(NR_CPUS_BITS, bits_per(CONFIG_NR_CPUS));
> #endif
> DEFINE(SPINLOCK_SIZE, sizeof(spinlock_t));
> /* End of constants */
Now fuzzed, thanks.
greg k-h
On Fri, Mar 29, 2024 at 02:34:43PM +0100, Greg KH wrote:
> On Wed, Mar 27, 2024 at 02:10:10PM +0000, Matthew Wilcox wrote:
> > On Wed, Mar 27, 2024 at 08:21:25AM -0400, Sasha Levin wrote:
> > > The patch below does not apply to the 5.4-stable tree.
> > > If someone wants it applied there, or to any other stable or longterm
> > > tree, then please email the backport, including the original git commit
> > > id to <[email protected]>.
> >
> > Looks like you just need a little more fuzz on the patch.
> >
> > diff --git a/kernel/bounds.c b/kernel/bounds.c
> > index 9795d75b09b2..a94e3769347e 100644
> > --- a/kernel/bounds.c
> > +++ b/kernel/bounds.c
> > @@ -19,7 +19,7 @@ int main(void)
> > DEFINE(NR_PAGEFLAGS, __NR_PAGEFLAGS);
> > DEFINE(MAX_NR_ZONES, __MAX_NR_ZONES);
> > #ifdef CONFIG_SMP
> > - DEFINE(NR_CPUS_BITS, ilog2(CONFIG_NR_CPUS));
> > + DEFINE(NR_CPUS_BITS, bits_per(CONFIG_NR_CPUS));
> > #endif
> > DEFINE(SPINLOCK_SIZE, sizeof(spinlock_t));
> > /* End of constants */
>
> Now fuzzed, thanks.
But it breaks the build on 4.19.y, so I'll go drop it from there. If
you want it added there, please provide a working fix.
thanks,
greg k-h
On Fri, Mar 29, 2024 at 06:05:20PM +0100, Greg KH wrote:
> On Fri, Mar 29, 2024 at 02:34:43PM +0100, Greg KH wrote:
> > On Wed, Mar 27, 2024 at 02:10:10PM +0000, Matthew Wilcox wrote:
> > > On Wed, Mar 27, 2024 at 08:21:25AM -0400, Sasha Levin wrote:
> > > > The patch below does not apply to the 5.4-stable tree.
> > > > If someone wants it applied there, or to any other stable or longterm
> > > > tree, then please email the backport, including the original git commit
> > > > id to <[email protected]>.
> > >
> > > Looks like you just need a little more fuzz on the patch.
> > >
> > > diff --git a/kernel/bounds.c b/kernel/bounds.c
> > > index 9795d75b09b2..a94e3769347e 100644
> > > --- a/kernel/bounds.c
> > > +++ b/kernel/bounds.c
> > > @@ -19,7 +19,7 @@ int main(void)
> > > DEFINE(NR_PAGEFLAGS, __NR_PAGEFLAGS);
> > > DEFINE(MAX_NR_ZONES, __MAX_NR_ZONES);
> > > #ifdef CONFIG_SMP
> > > - DEFINE(NR_CPUS_BITS, ilog2(CONFIG_NR_CPUS));
> > > + DEFINE(NR_CPUS_BITS, bits_per(CONFIG_NR_CPUS));
> > > #endif
> > > DEFINE(SPINLOCK_SIZE, sizeof(spinlock_t));
> > > /* End of constants */
> >
> > Now fuzzed, thanks.
>
> But it breaks the build on 4.19.y, so I'll go drop it from there. If
> you want it added there, please provide a working fix.
Looks like bits_per() didn't exist in 4.19. It was added as part of commit 69842cba9ace84849bb9b8edcdf2cefccd97901c
Author: Patrick Bellasi <[email protected]>
Date: Fri Jun 21 09:42:02 2019 +0100
sched/uclamp: Add CPU's clamp buckets refcounting
Up to you; I can provide bits_per() to 4.19 which will aid backporting
other fixes (we currently have 17 uses of bits_per() in 6.9), or we can
just drop this whole thing for 4.19.
On Tue, Apr 02, 2024 at 06:33:55AM +0100, Matthew Wilcox wrote:
> On Fri, Mar 29, 2024 at 06:05:20PM +0100, Greg KH wrote:
> > On Fri, Mar 29, 2024 at 02:34:43PM +0100, Greg KH wrote:
> > > On Wed, Mar 27, 2024 at 02:10:10PM +0000, Matthew Wilcox wrote:
> > > > On Wed, Mar 27, 2024 at 08:21:25AM -0400, Sasha Levin wrote:
> > > > > The patch below does not apply to the 5.4-stable tree.
> > > > > If someone wants it applied there, or to any other stable or longterm
> > > > > tree, then please email the backport, including the original git commit
> > > > > id to <[email protected]>.
> > > >
> > > > Looks like you just need a little more fuzz on the patch.
> > > >
> > > > diff --git a/kernel/bounds.c b/kernel/bounds.c
> > > > index 9795d75b09b2..a94e3769347e 100644
> > > > --- a/kernel/bounds.c
> > > > +++ b/kernel/bounds.c
> > > > @@ -19,7 +19,7 @@ int main(void)
> > > > DEFINE(NR_PAGEFLAGS, __NR_PAGEFLAGS);
> > > > DEFINE(MAX_NR_ZONES, __MAX_NR_ZONES);
> > > > #ifdef CONFIG_SMP
> > > > - DEFINE(NR_CPUS_BITS, ilog2(CONFIG_NR_CPUS));
> > > > + DEFINE(NR_CPUS_BITS, bits_per(CONFIG_NR_CPUS));
> > > > #endif
> > > > DEFINE(SPINLOCK_SIZE, sizeof(spinlock_t));
> > > > /* End of constants */
> > >
> > > Now fuzzed, thanks.
> >
> > But it breaks the build on 4.19.y, so I'll go drop it from there. If
> > you want it added there, please provide a working fix.
>
> Looks like bits_per() didn't exist in 4.19. It was added as part of commit 69842cba9ace84849bb9b8edcdf2cefccd97901c
> Author: Patrick Bellasi <[email protected]>
> Date: Fri Jun 21 09:42:02 2019 +0100
>
> sched/uclamp: Add CPU's clamp buckets refcounting
>
> Up to you; I can provide bits_per() to 4.19 which will aid backporting
> other fixes (we currently have 17 uses of bits_per() in 6.9), or we can
> just drop this whole thing for 4.19.
>
At this point in time for 4.19, unless it's really criticial, I would
just say drop it.
thanks,
greg k-h