Re: [PATCH 2/3] perf alias: Rebuild alias expression string to make it comparable

From: Jiri Olsa
Date: Fri Jun 15 2018 - 05:27:07 EST


On Fri, Jun 15, 2018 at 11:09:05AM +0200, Thomas-Mich Richter wrote:
> On 06/15/2018 10:12 AM, Jiri Olsa wrote:
> > On Thu, Jun 14, 2018 at 08:53:14AM -0500, Paul Clarke wrote:
> >
> > SNIP
> >
> >>> + if (ret)
> >>> + ret += scnprintf(newval + ret, sizeof(newval) - ret,
> >>> + ",");
> >>> + if (term->type_val == PARSE_EVENTS__TERM_TYPE_NUM)
> >>> + ret += scnprintf(newval + ret, sizeof(newval) - ret,
> >>> + "%s=%#x", term->config, term->val.num);
> >>> + else if (term->type_val == PARSE_EVENTS__TERM_TYPE_STR)
> >>> + ret += scnprintf(newval + ret, sizeof(newval) - ret,
> >>> + "%s=%s", term->config, term->val.str);
> >>
> >> If we exceed 256, we just suddenly terminate the rebuilding without reporting any issues.
> >>
> >>> + }
> >>> +
> >>> alias->name = strdup(name);
> >>> if (dir) {
> >>> /*
> >>> @@ -285,7 +308,7 @@ static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
> >>> snprintf(alias->unit, sizeof(alias->unit), "%s", unit);
> >>> }
> >>> alias->per_pkg = perpkg && sscanf(perpkg, "%d", &num) == 1 && num == 1;
> >>> - alias->str = strdup(val);
> >>> + alias->str = strdup(newval);
> >
> > hum, how is newval different from val? AFAICS it's the same
> >
>
> Not really, depends on the platform, here is some debug output from s390:
> root@s35lp76 perf]# ./perf stat -e tx_nc_tend -- ~/mytesttx 1 >/tmp/111
>
> Performance counter stats for '/root/mytesttx 1':
>
> 1 tx_nc_tend
>
> 0.001050150 seconds time elapsed
>
> [root@s35lp76 perf]# fgrep -i tx_nc_tend /tmp/111
> __perf_pmu__new_alias TX_NC_TEND val:event=0x008d newval:event=0x8d
> __perf_pmu__new_alias tx_nc_tend val:event=0x8d newval:event=0x8d
> TX_NC_TEND 1 rc 8
> [root@s35lp76 perf]#
>
> On s390 the events in the PMU sysfs file are printed with leading zeroes.
> This means the strcmp() of alias->str differs, but the contents is logically
> identical. (Same of some files contains spaces).
>
> Thats why I do the rewrite of val into newval.
>
> The alias name does not match too, but we use strcasecmp() to ignore case.
>
> Hope this helps.

yep.. should have read the full change log ;-) thanks

jirka