linux-kernel - Re: [PATCH v3 5/5] perf metric: Don't compute unused events.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20201123054658.GF894261@tassilo.jf.intel.com>
Date:   Sun, 22 Nov 2020 21:46:58 -0800
From:   Andi Kleen <ak@...ux.intel.com>
To:     Ian Rogers <irogers@...gle.com>
Cc:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Mark Rutland <mark.rutland@....com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Jiri Olsa <jolsa@...hat.com>,
        Namhyung Kim <namhyung@...nel.org>,
        linux-kernel@...r.kernel.org, Jin Yao <yao.jin@...ux.intel.com>,
        John Garry <john.garry@...wei.com>,
        Paul Clarke <pc@...ibm.com>, kajoljain <kjain@...ux.ibm.com>,
        Stephane Eranian <eranian@...gle.com>,
        Sandeep Dasgupta <sdasgup@...gle.com>,
        linux-perf-users@...r.kernel.org
Subject: Re: [PATCH v3 5/5] perf metric: Don't compute unused events.

> +| expr '|' expr
> +{
> +	if (!compute_ids || (isfinite($1.val) && isfinite($3.val))) {
> +		assert($1.ids == NULL);
> +		assert($3.ids == NULL);
> +		$$.val = (long)$1.val | (long)$3.val;
> +		$$.ids = NULL;
> +	} else {
> +		/*
> +		 * LHS or RHS needs to be computed from event IDs, consequently
> +		 * so does this expression. Set val to NAN to show that the set
> +		 * of all values is possible, the events are the union of those
> +		 * on the LHS and RHS.
> +		 */
> +		$$.val = NAN;
> +		$$.ids = ids__union($1.ids, $3.ids);
> +	}


Sorry, still not a fan of the having this nan code all over. It's just ugly.

If you don't want to do the syntax change to still do it in one pass,
and given building an AST would be a little complicated.

The original parser I based this code on actually had a byte code version too
(see attachment). With that one the lazy evaluation could be done on the byte code
level. Originally I didn't include it because it wasn't really
needed for perf, but presumably if we want to do more complicated
things it might be useful. 

In theory it could speed up performance slightly when an expression needs
to be computed multiple times in interval mode.

-Andi

View attachment "code.c" of type "text/plain" (2933 bytes)

View attachment "code.h" of type "text/plain" (441 bytes)

View attachment "expr.y" of type "text/plain" (3521 bytes)