Re: [PATCH v3 05/10] EDAC/mc: Create new function edac_inc_csrow()

[Robert Richter]

On 13.02.20 13:47:08, Borislav Petkov wrote:
> On Thu, Jan 23, 2020 at 09:02:58AM +0000, Robert Richter wrote:
> > Have a separate function to count errors in csrow/channel. This better
> > separates code and reduces the indentation level. No functional
> > changes.
> > 
> > Signed-off-by: Robert Richter <rrichter@xxxxxxxxxxx>
> > Reviewed-by: Mauro Carvalho Chehab <mchehab@xxxxxxxxxx>
> > Acked-by: Aristeu Rozanski <aris@xxxxxxxxxx>
> > ---
> >  drivers/edac/edac_mc.c | 40 +++++++++++++++++++++++++---------------
> >  1 file changed, 25 insertions(+), 15 deletions(-)
> > 
> > diff --git a/drivers/edac/edac_mc.c b/drivers/edac/edac_mc.c
> > index 3c00c046acc9..e75cb7a9c454 100644
> > --- a/drivers/edac/edac_mc.c
> > +++ b/drivers/edac/edac_mc.c
> > @@ -1091,6 +1091,26 @@ static void edac_ue_error(struct mem_ctl_info *mci,
> >  	edac_inc_ue_error(mci, enable_per_layer_report, pos, error_count);
> >  }
> >  
> > +static void edac_inc_csrow(struct edac_raw_error_desc *e, int row, int chan)
> > +{
> > +	struct mem_ctl_info *mci = error_desc_to_mci(e);
> > +	u16 count = e->error_count;
> > +	enum hw_event_mc_err_type type = e->type;
> 
> Please sort function local variables declaration in a reverse christmas
> tree order:
> 
> 	<type A> longest_variable_name;
> 	<type B> shorter_var_name;
> 	<type C> even_shorter;
> 	<type D> i;

I can change this. Does variable name length include the assignment?

Generally I prefer to sort it by size_of() to avoid holes due to
padding, though the compiler uses probably registers here anyway.
It's just a flavor.

If it's just this change, could you edit the patch to avoid respin?

> 
> > +
> > +	if (row < 0)
> > +		return;
> > +
> > +	edac_dbg(4, "csrow/channel to increment: (%d,%d)\n", row, chan);
> > +
> > +	if (type == HW_EVENT_ERR_CORRECTED) {
> > +		mci->csrows[row]->ce_count += count;
> > +		if (chan >= 0)
> > +			mci->csrows[row]->channels[chan]->ce_count += count;
> > +	} else {
> > +		mci->csrows[row]->ue_count += count;
> > +	}
> > +}
> > +
> >  void edac_raw_mc_handle_error(struct edac_raw_error_desc *e)
> >  {
> >  	struct mem_ctl_info *mci = error_desc_to_mci(e);
> > @@ -1258,22 +1278,12 @@ void edac_mc_handle_error(const enum hw_event_mc_err_type type,
> >  			chan = -2;
> >  	}
> >  
> > -	if (!e->enable_per_layer_report) {
> > +	if (!e->enable_per_layer_report)
> >  		strcpy(e->label, "any memory");
> > -	} else {
> > -		edac_dbg(4, "csrow/channel to increment: (%d,%d)\n", row, chan);
> > -		if (p == e->label)
> > -			strcpy(e->label, "unknown memory");
> > -		if (type == HW_EVENT_ERR_CORRECTED) {
> > -			if (row >= 0) {
> > -				mci->csrows[row]->ce_count += error_count;
> > -				if (chan >= 0)
> > -					mci->csrows[row]->channels[chan]->ce_count += error_count;
> > -			}
> > -		} else
> > -			if (row >= 0)
> > -				mci->csrows[row]->ue_count += error_count;
> > -	}
> > +	else if (!*e->label)
> > +		strcpy(e->label, "unknown memory");
> > +
> > +	edac_inc_csrow(e, row, chan);
> 
> Err, but this has functional changes: the !e->enable_per_layer_report
> case sets only the e->label and the else branch only does increment
> ->ce_count.
> 
> Your change gets ->ce_count incremented in both cases.

No, there is a check in edac_inc_csrow(): if (row < 0) ... In the case
of "any memory", row is also < 0, so nothing is counted. This is
reasonable since no dimm is found and row/channel is still set to the
setup value of -1.

-Robert