Re: [PATCH v2 7/7] edac drivers: add MC owner check in init

[Kani, Toshimitsu]

On Fri, 2017-08-04 at 10:39 +0200, Borislav Petkov wrote:
> On Thu, Aug 03, 2017 at 03:57:53PM -0600, Toshi Kani wrote:
> > Change generic x86 edac drivers, which probe CPU type with
> > x86_match_cpu(), to call edac_check_mc_owner() in their
> > module init functions.ÂÂThis allows them to fail their init
> > at the beginning when ghes_edac is enabled.ÂÂSimilar change
> > can be made to other edac drivers as necessary.
> > 
> > This is an optimization and there is no functional change.
> > 
> > Signed-off-by: Toshi Kani <toshi.kani@xxxxxxx>
> > Suggested-by: Borislav Petkov <bp@xxxxxxxxx>
> > Cc: Borislav Petkov <bp@xxxxxxxxx>
> > Cc: Mauro Carvalho Chehab <mchehab@xxxxxxxxxx>
> > Cc: Tony Luck <tony.luck@xxxxxxxxx>
> > ---
> > Âdrivers/edac/amd64_edac.c |ÂÂÂÂ3 +++
> > Âdrivers/edac/pnd2_edac.cÂÂ|ÂÂÂÂ7 ++++++-
> > Âdrivers/edac/sb_edac.cÂÂÂÂ|ÂÂÂÂ7 +++++--
> > Âdrivers/edac/skx_edac.cÂÂÂ|ÂÂÂÂ6 +++++-
> > Â4 files changed, 19 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/edac/amd64_edac.c b/drivers/edac/amd64_edac.c
> > index 3aea556..cdb40d6 100644
> > --- a/drivers/edac/amd64_edac.c
> > +++ b/drivers/edac/amd64_edac.c
> > @@ -3444,6 +3444,9 @@ static int __init amd64_edac_init(void)
> > Â	if (amd_cache_northbridges() < 0)
> > Â		return -ENODEV;
> > Â
> > +	if (!edac_check_mc_owner(EDAC_MOD_STR))
> > +		return -EBUSY;
> > +
> 
> That needs to happen first in the init function.

Not sure if anyone cares, but I thought it should return with -ENODEV
when this modules found no target, and -EBUSY when it found a target
but it's busy.  Hence, this ordering.

> > Â	opstate_init();
> > Â
> > Â	err = -ENOMEM;
> > diff --git a/drivers/edac/pnd2_edac.c b/drivers/edac/pnd2_edac.c
> > index 8e59949..a5b7855 100644
> > --- a/drivers/edac/pnd2_edac.c
> > +++ b/drivers/edac/pnd2_edac.c
> > @@ -45,6 +45,8 @@
> > Â#include "edac_module.h"
> > Â#include "pnd2_edac.h"
> > Â
> > +#define PND2_MOD_NAME		"pnd2_edac.c"
> 
> EDAC_MOD_STR and look how the other drivers define it, i.e., without
> the ".c"

OK, I will change to use EDAC_MOD_STR and remove ".c".

> > +
> > Â#define APL_NUM_CHANNELS	4
> > Â#define DNV_NUM_CHANNELS	2
> > Â#define DNV_MAX_DIMMS		2 /* Max DIMMs per channel */
> > @@ -1313,7 +1315,7 @@ static int pnd2_register_mci(struct
> > mem_ctl_info **ppmci)
> > Â	pvt = mci->pvt_info;
> > Â	memset(pvt, 0, sizeof(*pvt));
> > Â
> > -	mci->mod_name = "pnd2_edac.c";
> > +	mci->mod_name = PND2_MOD_NAME;
> > Â	mci->dev_name = ops->name;
> > Â	mci->ctl_name = "Pondicherry2";
> > Â
> > @@ -1513,6 +1515,9 @@ static int __init pnd2_init(void)
> > Â	if (!id)
> > Â		return -ENODEV;
> > Â
> > +	if (!edac_check_mc_owner(PND2_MOD_NAME))
> > +		return -EBUSY;
> 
> Also first thing to do in the function.
>
> > +
> > Â	ops = (struct dunit_ops *)id->driver_data;
> > Â
> > Â	/* Ensure that the OPSTATE is set correctly for POLL or
> > NMI */
> > diff --git a/drivers/edac/sb_edac.c b/drivers/edac/sb_edac.c
> > index 80d860c..71bd66e 100644
> > --- a/drivers/edac/sb_edac.c
> > +++ b/drivers/edac/sb_edac.c
> > @@ -36,7 +36,7 @@ static LIST_HEAD(sbridge_edac_list);
> > Â * Alter this version for the module when modifications are made
> > Â */
> > Â#define SBRIDGE_REVISIONÂÂÂÂ" Ver: 1.1.2 "
> > -#define EDAC_MOD_STRÂÂÂÂÂÂ"sbridge_edac"
> > +#define SBRIDGE_MOD_NAMEÂÂÂÂ"sb_edac.c"
> 
> Why? EDAC_MOD_STR is just fine.

I simply kept what it has today.  Will remove ".c".

Thanks,
-Toshi