[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-Id: <20170411090002.e74f1e80d98277b3a3f93278@kernel.org>
Date: Tue, 11 Apr 2017 09:00:02 +0900
From: Masami Hiramatsu <mhiramat@...nel.org>
To: pc@...ibm.com
Cc: LKML <linux-kernel@...r.kernel.org>,
Arnaldo Carvalho de Melo <acme@...hat.com>,
David Ahern <dsahern@...il.com>,
"linux-perf-users@...r.kernel.org" <linux-perf-users@...r.kernel.org>
Subject: Re: [PATCH v4] tools/perf: Allow user probes on versioned symbols
On Mon, 10 Apr 2017 09:53:48 -0500
Paul Clarke <pc@...ibm.com> wrote:
> >> diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h
> >> index 5245d2f..bde2100 100644
> >> --- a/tools/perf/util/symbol.h
> >> +++ b/tools/perf/util/symbol.h
> >> @@ -348,8 +348,17 @@ void arch__sym_update(struct symbol *s, GElf_Sym *sym);
> >> #define SYMBOL_A 0
> >> #define SYMBOL_B 1
> >>
> >> +int arch__compare_symbol_names(const char *namea, const char *nameb);
> >> +int arch__compare_symbol_names_n(const char *namea, const char *nameb,
> >> + unsigned int n);
> >> int arch__choose_best_symbol(struct symbol *syma, struct symbol *symb);
> >>
> >> +#define SYMBOLS__TAG_INCLUDE_NONE 0
> >> +#define SYMBOLS__TAG_INCLUDE_DEFAULT_ONLY 1
> >
> > Hmm, even if you like to pass an obvious flag, these should be
> > defined by enum, and please choose simpler and clear flag name.
> >
> > E.g.
> > enum {
> > SYMBOLS__COMP_WITH_TAG,
> > SYMBOLS__COMP_WITHOUT_TAG,
> > };
>
> Earlier in this thread, we discussed supporting non-default tagged symbols (those tagged with "@version"). To allow for that, I was thinking of these three cases:
> - SYMBOLS__TAG_INCLUDE_NONE: ignore all tagged symbols (the current behavior)
> - SYMBOLS__TAG_INCLUDE_DEFAULT_ONLY: match tagged symbols when tagged as default ("@@"; added with this patch)
> - SYMBOLS__TAG_INCLUDE_ALL: match all tagged symbols ("@" and "@@"; future)
Ah, I see now.
> For the names of the enums, I chose "SYMBOLS__TAG..." to put it first in the SYMBOLS namespace, and second in a "TAG" sub-namespace.
For the namespace, we should use same namespase as enum's name.
> I'm happy to fix the discrepancy with signed/unsigned by changing to enum.
>
> Let me know what enum names are preferred, given the above discussion, and I'll post a new patch.
OK, then what about enum symbol_tag and use "SYMBOL_TAG__" for namespace?
Thank you,
--
Masami Hiramatsu <mhiramat@...nel.org>
Powered by blists - more mailing lists