Re: [patch 3/24] s390: channel path measurements.

[Andrew Morton <akpm@osdl.org>]

Martin Schwidefsky <schwidefsky@de.ibm.com> wrote:
>
> From: Cornelia Huck <cornelia.huck@de.ibm.com>
> 
> [patch 3/24] s390: channel path measurements.
> 
> Gather extended measurements for channel paths from the channel subsystem
> and expose them to userspace via a sysfs attribute.
> 
> ...
>
> +static void
> +chsc_remove_cmg_attr(struct channel_subsystem *css)
> +{
> +	int i;
> +
> +	for (i = 0; i <= __MAX_CHPID; i++) {

hm, it's somewhat unusual for MAX_FOO to be inclusive.  Usually it means
"greatest possible+1".

you have

	struct channel_subsystem {
	 	struct channel_path *chps[__MAX_CHPID + 1];

so I guess it all works..

> +static inline int
> +__chsc_do_secm(struct channel_subsystem *css, int enable, void *page)

This has two callsites.  inlining it probably deoptimises things.

> +	secm_area->request = (struct chsc_header) {
> +		.length = 0x0050,
> +		.code   = 0x0016,
> +	};

gcc tends to generate poor code for this construct.