RE: [PATCH V5 1/1] misc: NOC Probe Counters

From: N, Pandith
Date: Thu Jul 15 2021 - 03:00:54 EST


Hi Greg KH,

> -----Original Message-----
> From: Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx>
> Sent: Tuesday, July 13, 2021 12:28 PM
> To: N, Pandith <pandith.n@xxxxxxxxx>
> Cc: arnd@xxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Zhou, Furong
> <furong.zhou@xxxxxxxxx>; mgross@xxxxxxxxxxxxxxx; Sangannavar,
> Mallikarjunappa <mallikarjunappa.sangannavar@xxxxxxxxx>;
> andriy.shevchenko@xxxxxxxxxxxxxxx; Raja Subramanian, Lakshmi Bai
> <lakshmi.bai.raja.subramanian@xxxxxxxxx>
> Subject: Re: [PATCH V5 1/1] misc: NOC Probe Counters
>
> On Tue, Jul 13, 2021 at 12:23:47PM +0530, pandith.n@xxxxxxxxx wrote:
> > --- /dev/null
> > +++ b/drivers/misc/noc/Kconfig
> > @@ -0,0 +1,16 @@
> > +# Copyright (C) 2019 Intel Corporation # SPDX-License-Identifier:
> > +GPL-2.0-only
> > +
> > +
> > +config NOC_DSS
> > + tristate "Enable DSS NOC probing"
>
> Please be more descriptive of what exactly this is, it does not make sense to
> anyone who does not know what DSS or NOC means.
>
Will rephrase and add more description about DSS, NOC in help section

tristate "Enable DDR profiling using NOC for Intel Keem Bay"
depends on ARCH_KEEMBAY || COMPILE_TEST
help
Enable this option for DDR bandwidth measurements using NOC

Add support for Network on chip (NOC) in DDR Subsystem(DSS).
DSS NOC has capabilities to enable and get statistics profiling.
NOC driver enables features to configure and capture NOC probe
counters, needed for DSS bandwidth measurement.
Say Y if using a processor that includes the Intel VPU such as
Keem Bay. If unsure, say N.

> > + depends on ARCH_KEEMBAY
> > + default y
>
> You only use 'y' if the machine can not boot without the option.
> Otherwise it is left alone.
>
Have removed default 'y'

> And what about CONFIG_TEST ability to test build this code? Without that, you
> will get almost no one every building this code to see if anything breaks over
> time.
>
Have added COMPILE_TEST option as below
depends on ARCH_KEEMBAY || COMPILE_TEST

> thanks,
>
> greg k-h