Hi all,

After merging the cgroup tree, today's linux-next build (x86_64
allmodconfig) failed like this:

kernel/cgroup/cgroup.c:5275:26: error: 'CFTYPE_PRESSURE' undeclared here (not in a function)
 5275 |                 .flags = CFTYPE_PRESSURE,
      |                          ^~~~~~~~~~~~~~~

Caused by commit

  8a693f7766f9 ("cgroup: Remove CFTYPE_PRESSURE")

inteacting with commits

  52b1364ba0b1 ("sched/psi: Add PSI_IRQ to track IRQ/SOFTIRQ pressure")
  34f26a15611a ("sched/psi: Per-cgroup PSI accounting disable/re-enable interface")

from the tip tree.

I have applied the following merge fix patch.

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Mon, 12 Sep 2022 16:15:33 +1000
Subject: [PATCH] sched/psi: fix up for "cgroup: Remove CFTYPE_PRESSURE"

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
---
 kernel/cgroup/cgroup.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
index 974ca46c6d7b..829aa42e773e 100644
--- a/kernel/cgroup/cgroup.c
+++ b/kernel/cgroup/cgroup.c
@@ -5272,7 +5272,6 @@ static struct cftype cgroup_psi_files[] = {
 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
 	{
 		.name = "irq.pressure",
-		.flags = CFTYPE_PRESSURE,
 		.file_offset = offsetof(struct cgroup, psi_files[PSI_IRQ]),
 		.seq_show = cgroup_irq_pressure_show,
 		.write = cgroup_irq_pressure_write,
@@ -5282,7 +5281,6 @@ static struct cftype cgroup_psi_files[] = {
 #endif
 	{
 		.name = "cgroup.pressure",
-		.flags = CFTYPE_PRESSURE,
 		.seq_show = cgroup_pressure_show,
 		.write = cgroup_pressure_write,
 	},
-- 
2.35.1

-- 
Cheers,
Stephen Rothwell

Hi all,

On Mon, 12 Sep 2022 16:18:12 +1000 Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> 
> After merging the cgroup tree, today's linux-next build (x86_64
> allmodconfig) failed like this:
> 
> kernel/cgroup/cgroup.c:5275:26: error: 'CFTYPE_PRESSURE' undeclared here (not in a function)
>  5275 |                 .flags = CFTYPE_PRESSURE,
>       |                          ^~~~~~~~~~~~~~~
> 
> Caused by commit
> 
>   8a693f7766f9 ("cgroup: Remove CFTYPE_PRESSURE")
> 
> inteacting with commits
> 
>   52b1364ba0b1 ("sched/psi: Add PSI_IRQ to track IRQ/SOFTIRQ pressure")
>   34f26a15611a ("sched/psi: Per-cgroup PSI accounting disable/re-enable interface")
> 
> from the tip tree.
> 
> I have applied the following merge fix patch.
> 
> From: Stephen Rothwell <sfr@canb.auug.org.au>
> Date: Mon, 12 Sep 2022 16:15:33 +1000
> Subject: [PATCH] sched/psi: fix up for "cgroup: Remove CFTYPE_PRESSURE"
> 
> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
> ---
>  kernel/cgroup/cgroup.c | 2 --
>  1 file changed, 2 deletions(-)
> 
> diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
> index 974ca46c6d7b..829aa42e773e 100644
> --- a/kernel/cgroup/cgroup.c
> +++ b/kernel/cgroup/cgroup.c
> @@ -5272,7 +5272,6 @@ static struct cftype cgroup_psi_files[] = {
>  #ifdef CONFIG_IRQ_TIME_ACCOUNTING
>  	{
>  		.name = "irq.pressure",
> -		.flags = CFTYPE_PRESSURE,
>  		.file_offset = offsetof(struct cgroup, psi_files[PSI_IRQ]),
>  		.seq_show = cgroup_irq_pressure_show,
>  		.write = cgroup_irq_pressure_write,
> @@ -5282,7 +5281,6 @@ static struct cftype cgroup_psi_files[] = {
>  #endif
>  	{
>  		.name = "cgroup.pressure",
> -		.flags = CFTYPE_PRESSURE,
>  		.seq_show = cgroup_pressure_show,
>  		.write = cgroup_pressure_write,
>  	},
> -- 
> 2.35.1

This is now a conflict between the tip tree and Linus' tree.

-- 
Cheers,
Stephen Rothwell

On Tue, Oct 11, 2022 at 07:57:48AM +1100, Stephen Rothwell wrote:
> This is now a conflict between the tip tree and Linus' tree.

The conflict and resolution were noted in the pull request. This should be
fine.

Thanks.

-- 
tejun

On Mon, Sep 12, 2022 at 04:18:12PM +1000, Stephen Rothwell wrote:
> Hi all,
> 
> After merging the cgroup tree, today's linux-next build (x86_64
> allmodconfig) failed like this:

Hmm,. TJ should I base sched/psi on top of drivers-core-next and your
cgroup tree?

On Mon, Sep 12, 2022 at 09:38:58AM +0200, Peter Zijlstra wrote:
> On Mon, Sep 12, 2022 at 04:18:12PM +1000, Stephen Rothwell wrote:
> > Hi all,
> > 
> > After merging the cgroup tree, today's linux-next build (x86_64
> > allmodconfig) failed like this:
> 
> Hmm,. TJ should I base sched/psi on top of drivers-core-next and your
> cgroup tree?

Yeah, this is kinda nasty. Lemme just pull drivers-core-next into
cgorup/for-6.1 so that folks don't have to worry about this and you can just
pull that one.

Thanks.

-- 
tejun

Hello,

On Mon, Sep 12, 2022 at 08:44:48AM -1000, Tejun Heo wrote:
> On Mon, Sep 12, 2022 at 09:38:58AM +0200, Peter Zijlstra wrote:
> > On Mon, Sep 12, 2022 at 04:18:12PM +1000, Stephen Rothwell wrote:
> > > Hi all,
> > > 
> > > After merging the cgroup tree, today's linux-next build (x86_64
> > > allmodconfig) failed like this:
> > 
> > Hmm,. TJ should I base sched/psi on top of drivers-core-next and your
> > cgroup tree?
> 
> Yeah, this is kinda nasty. Lemme just pull drivers-core-next into
> cgorup/for-6.1 so that folks don't have to worry about this and you can just
> pull that one.

I initially thought that driver-core-next was conflicting with
cgroup/for-6.1 but this is simpler. It's a straight forward conflict between
the PSI patchset in the tip tree and changes in cgroup/for-6.1 which are
cleaning up some cftype handling, and the patch that Stephen applied is
correct. So, we can do one of the following two:

1. Leave it as-is. Stephen's patch is correct and we can just let Linus know
   how to resolve it on the pull requests.

2. Pull cgroup/for-6.1 into the tip tree and preemptively resolve the
   conflict there.

I'm leaning towards #1 but #2 is fine too. Peter, what do you wanna do?

Thanks.

-- 
tejun

On Mon, Sep 12, 2022 at 12:16:32PM -1000, Tejun Heo wrote:

> 1. Leave it as-is. Stephen's patch is correct and we can just let Linus know
>    how to resolve it on the pull requests.

> I'm leaning towards #1 but #2 is fine too. Peter, what do you wanna do?

Works for me; I'll make a note for the pull req.