EDAC: Don't add devices under /sys/bus/edac

EDAC: Don't add devices under /sys/bus/edac

[Mauro Carvalho Chehab]

Em Mon, 1 Oct 2018 15:43:13 -0700
"Luck, Tony" <tony.luck@intel.com> escreveu:

> Nobody(*) uses them.  Dropping this will allow us to make the total
> number of memory controllers configurable (as we won't have to
> worry about duplicated device names under this directory).
> 
> (*) https://marc.info/?l=linux-edac&m=153809709903987&w=2
> 
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
> 
> Boris: Apply this, then your earlier patch to get rid of the
> hard coded limit on the number of memory controllers:
>   https://marc.info/?l=linux-edac&m=153797567628947&w=2
> the combination works on my 4 socket machine. Perhaps HPE
> can test on their superdome.
> 

For both this and the referred patch:

Acked-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

>  drivers/edac/edac_mc_sysfs.c | 3 ---
>  1 file changed, 3 deletions(-)
> 
> diff --git a/drivers/edac/edac_mc_sysfs.c b/drivers/edac/edac_mc_sysfs.c
> index 20374b8248f0..4c1bee59c2e6 100644
> --- a/drivers/edac/edac_mc_sysfs.c
> +++ b/drivers/edac/edac_mc_sysfs.c
> @@ -405,7 +405,6 @@ static int edac_create_csrow_object(struct mem_ctl_info *mci,
>  				    struct csrow_info *csrow, int index)
>  {
>  	csrow->dev.type = &csrow_attr_type;
> -	csrow->dev.bus = mci->bus;
>  	csrow->dev.groups = csrow_dev_groups;
>  	device_initialize(&csrow->dev);
>  	csrow->dev.parent = &mci->dev;
> @@ -636,7 +635,6 @@ static int edac_create_dimm_object(struct mem_ctl_info *mci,
>  	dimm->mci = mci;
>  
>  	dimm->dev.type = &dimm_attr_type;
> -	dimm->dev.bus = mci->bus;
>  	device_initialize(&dimm->dev);
>  
>  	dimm->dev.parent = &mci->dev;
> @@ -940,7 +938,6 @@ int edac_create_sysfs_mci_device(struct mem_ctl_info *mci,
>  	device_initialize(&mci->dev);
>  
>  	mci->dev.parent = mci_pdev;
> -	mci->dev.bus = mci->bus;
>  	mci->dev.groups = groups;
>  	dev_set_name(&mci->dev, "mc%d", mci->mc_idx);
>  	dev_set_drvdata(&mci->dev, mci);



Thanks,
Mauro

EDAC: Don't add devices under /sys/bus/edac

[Borislav Petkov]

On Tue, Nov 13, 2018 at 07:09:24PM +0000, Ernst, Justin wrote:
> Looks good on a 32 socket system. All 64 memory controllers show up
> and I'm able to see the same sysfs diff. Thanks

Thanks for testing, much appreciated!

I'm queuing the stuff for 4.21.

EDAC: Don't add devices under /sys/bus/edac

[Justin Ernst]

Looks good on a 32 socket system. All 64 memory controllers show up and I'm able to see the same sysfs diff.
Thanks
-Justin

> -----Original Message-----
> From: Borislav Petkov [mailto:bp@alien8.de]
> Sent: Tuesday, November 6, 2018 8:46 AM
> To: Ernst, Justin <justin.ernst@hpe.com>; Luck, Tony <tony.luck@intel.com>
> Cc: Greg KH <gregkh@linuxfoundation.org>; Anderson, Russ
> <russ.anderson@hpe.com>; Mauro Carvalho Chehab
> <mchehab@kernel.org>; linux-edac@vger.kernel.org; linux-
> kernel@vger.kernel.org; Aristeu Rozanski Filho <arozansk@redhat.com>
> Subject: Re: [PATCH] EDAC: Don't add devices under /sys/bus/edac
> 
> On Tue, Oct 02, 2018 at 06:26:08PM +0200, Borislav Petkov wrote:
> > On Tue, Oct 02, 2018 at 03:51:41PM +0000, Ernst, Justin wrote:
> > > The combined patches work on a 20 socket system.
> > > Thanks!
> >
> > Cool, thanks for testing.
> >
> > Nevertheless, I'll queue them for 4.21 so that we have a full cycle of
> > testing before we really kill the bus thing.
> 
> Ok, I've pushed the two patches ontop of EDAC's for-next branch, here:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/bp/bp.git/log/?h=edac-for-
> 4.21-bus
> 
> and I'd appreciate testing them one more time on the big boxes you guys
> have.
> 
> Diffing sysfs here shows this:
> 
> --- edac.before	2018-11-06 13:37:37.925448609 +0100
> +++ edac.after	2018-11-06 15:36:11.229497795 +0100
> @@ -37,7 +37,6 @@
>  /sys/devices/system/edac/mc/mc0/dimm3/power/control
>  /sys/devices/system/edac/mc/mc0/dimm3/dimm_dev_type
>  /sys/devices/system/edac/mc/mc0/dimm3/size
> -/sys/devices/system/edac/mc/mc0/dimm3/subsystem
>  /sys/devices/system/edac/mc/mc0/dimm3/dimm_ce_count
>  /sys/devices/system/edac/mc/mc0/dimm3/dimm_label
>  /sys/devices/system/edac/mc/mc0/dimm3/dimm_location
> 
> which is the bus symlink:
> 
> subsystem -> ../../../../../../bus/mc0
> 
> and I think that's ok as nothing should be using it.
> 
> Thx.
> 
> --
> Regards/Gruss,
>     Boris.
> 
> Good mailing practices for 400: avoid top-posting and trim the reply.

EDAC: Don't add devices under /sys/bus/edac

[Borislav Petkov]

On Tue, Oct 02, 2018 at 06:26:08PM +0200, Borislav Petkov wrote:
> On Tue, Oct 02, 2018 at 03:51:41PM +0000, Ernst, Justin wrote:
> > The combined patches work on a 20 socket system. 
> > Thanks!
> 
> Cool, thanks for testing.
> 
> Nevertheless, I'll queue them for 4.21 so that we have a full cycle of
> testing before we really kill the bus thing.

Ok, I've pushed the two patches ontop of EDAC's for-next branch, here:

https://git.kernel.org/pub/scm/linux/kernel/git/bp/bp.git/log/?h=edac-for-4.21-bus

and I'd appreciate testing them one more time on the big boxes you guys have.

Diffing sysfs here shows this:


which is the bus symlink:

subsystem -> ../../../../../../bus/mc0

and I think that's ok as nothing should be using it.

Thx.

--- edac.before	2018-11-06 13:37:37.925448609 +0100
+++ edac.after	2018-11-06 15:36:11.229497795 +0100
@@ -37,7 +37,6 @@
 /sys/devices/system/edac/mc/mc0/dimm3/power/control
 /sys/devices/system/edac/mc/mc0/dimm3/dimm_dev_type
 /sys/devices/system/edac/mc/mc0/dimm3/size
-/sys/devices/system/edac/mc/mc0/dimm3/subsystem
 /sys/devices/system/edac/mc/mc0/dimm3/dimm_ce_count
 /sys/devices/system/edac/mc/mc0/dimm3/dimm_label
 /sys/devices/system/edac/mc/mc0/dimm3/dimm_location

EDAC: Don't add devices under /sys/bus/edac

[Borislav Petkov]

On Tue, Oct 02, 2018 at 03:51:41PM +0000, Ernst, Justin wrote:
> The combined patches work on a 20 socket system. 
> Thanks!

Cool, thanks for testing.

Nevertheless, I'll queue them for 4.21 so that we have a full cycle of
testing before we really kill the bus thing.

EDAC: Don't add devices under /sys/bus/edac

[Justin Ernst]

The combined patches work on a 20 socket system. 
Thanks!
-Justin

> -----Original Message-----
> From: Mauro Carvalho Chehab [mailto:mchehab+samsung@kernel.org]
> Sent: Monday, October 1, 2018 8:23 PM
> To: Luck, Tony <tony.luck@intel.com>
> Cc: Borislav Petkov <bp@alien8.de>; Anderson, Russ
> <russ.anderson@hpe.com>; Greg KH <gregkh@linuxfoundation.org>; Ernst,
> Justin <justin.ernst@hpe.com>; Anderson, Russ <russ.anderson@hpe.com>;
> Mauro Carvalho Chehab <mchehab@kernel.org>; linux-
> edac@vger.kernel.org; linux-kernel@vger.kernel.org; Aristeu Rozanski Filho
> <arozansk@redhat.com>
> Subject: Re: [PATCH] EDAC: Don't add devices under /sys/bus/edac
> 
> Em Mon, 1 Oct 2018 15:43:13 -0700
> "Luck, Tony" <tony.luck@intel.com> escreveu:
> 
> > Nobody(*) uses them.  Dropping this will allow us to make the total
> > number of memory controllers configurable (as we won't have to
> > worry about duplicated device names under this directory).
> >
> > (*) https://marc.info/?l=linux-edac&m=153809709903987&w=2
> >
> > Signed-off-by: Tony Luck <tony.luck@intel.com>
> > ---
> >
> > Boris: Apply this, then your earlier patch to get rid of the
> > hard coded limit on the number of memory controllers:
> >   https://marc.info/?l=linux-edac&m=153797567628947&w=2
> > the combination works on my 4 socket machine. Perhaps HPE
> > can test on their superdome.
> >
> 
> For both this and the referred patch:
> 
> Acked-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
> 
> >  drivers/edac/edac_mc_sysfs.c | 3 ---
> >  1 file changed, 3 deletions(-)
> >
> > diff --git a/drivers/edac/edac_mc_sysfs.c b/drivers/edac/edac_mc_sysfs.c
> > index 20374b8248f0..4c1bee59c2e6 100644
> > --- a/drivers/edac/edac_mc_sysfs.c
> > +++ b/drivers/edac/edac_mc_sysfs.c
> > @@ -405,7 +405,6 @@ static int edac_create_csrow_object(struct
> mem_ctl_info *mci,
> >  				    struct csrow_info *csrow, int index)
> >  {
> >  	csrow->dev.type = &csrow_attr_type;
> > -	csrow->dev.bus = mci->bus;
> >  	csrow->dev.groups = csrow_dev_groups;
> >  	device_initialize(&csrow->dev);
> >  	csrow->dev.parent = &mci->dev;
> > @@ -636,7 +635,6 @@ static int edac_create_dimm_object(struct
> mem_ctl_info *mci,
> >  	dimm->mci = mci;
> >
> >  	dimm->dev.type = &dimm_attr_type;
> > -	dimm->dev.bus = mci->bus;
> >  	device_initialize(&dimm->dev);
> >
> >  	dimm->dev.parent = &mci->dev;
> > @@ -940,7 +938,6 @@ int edac_create_sysfs_mci_device(struct
> mem_ctl_info *mci,
> >  	device_initialize(&mci->dev);
> >
> >  	mci->dev.parent = mci_pdev;
> > -	mci->dev.bus = mci->bus;
> >  	mci->dev.groups = groups;
> >  	dev_set_name(&mci->dev, "mc%d", mci->mc_idx);
> >  	dev_set_drvdata(&mci->dev, mci);
> 
> 
> 
> Thanks,
> Mauro

EDAC: Don't add devices under /sys/bus/edac

[Luck, Tony]

Nobody(*) uses them.  Dropping this will allow us to make the total
number of memory controllers configurable (as we won't have to
worry about duplicated device names under this directory).

(*) https://marc.info/?l=linux-edac&m=153809709903987&w=2

Signed-off-by: Tony Luck <tony.luck@intel.com>
---

Boris: Apply this, then your earlier patch to get rid of the
hard coded limit on the number of memory controllers:
  https://marc.info/?l=linux-edac&m=153797567628947&w=2
the combination works on my 4 socket machine. Perhaps HPE
can test on their superdome.

 drivers/edac/edac_mc_sysfs.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/edac/edac_mc_sysfs.c b/drivers/edac/edac_mc_sysfs.c
index 20374b8248f0..4c1bee59c2e6 100644
--- a/drivers/edac/edac_mc_sysfs.c
+++ b/drivers/edac/edac_mc_sysfs.c
@@ -405,7 +405,6 @@ static int edac_create_csrow_object(struct mem_ctl_info *mci,
 				    struct csrow_info *csrow, int index)
 {
 	csrow->dev.type = &csrow_attr_type;
-	csrow->dev.bus = mci->bus;
 	csrow->dev.groups = csrow_dev_groups;
 	device_initialize(&csrow->dev);
 	csrow->dev.parent = &mci->dev;
@@ -636,7 +635,6 @@ static int edac_create_dimm_object(struct mem_ctl_info *mci,
 	dimm->mci = mci;
 
 	dimm->dev.type = &dimm_attr_type;
-	dimm->dev.bus = mci->bus;
 	device_initialize(&dimm->dev);
 
 	dimm->dev.parent = &mci->dev;
@@ -940,7 +938,6 @@ int edac_create_sysfs_mci_device(struct mem_ctl_info *mci,
 	device_initialize(&mci->dev);
 
 	mci->dev.parent = mci_pdev;
-	mci->dev.bus = mci->bus;
 	mci->dev.groups = groups;
 	dev_set_name(&mci->dev, "mc%d", mci->mc_idx);
 	dev_set_drvdata(&mci->dev, mci);