From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@redhat.com>, Anton Blanchard <anton@au1.ibm.com>,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH] perf/powerpc: Cache the DWARF debug info
Date: Tue, 21 Oct 2014 16:08:39 -0300 [thread overview]
Message-ID: <20141021190839.GA4160@kernel.org> (raw)
In-Reply-To: <20141021185610.GA18366@us.ibm.com>
Em Tue, Oct 21, 2014 at 11:56:10AM -0700, Sukadev Bhattiprolu escreveu:
> >From 773a3608a0cd2daf02e244cb9ffbf5bb6a0e724e Mon Sep 17 00:00:00 2001
> From: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
> Date: Tue, 21 Oct 2014 13:20:22 -0500
> Subject: [PATCH 1/1] perf/powerpc: Cache DWARF debug info
Jiri, isn't his related to that dsos caching thing you created?
Anyway, there is a per machine struct (struct machine) where things like
this should be put, please do not add globals or static variables in
functions.
- Arnaldo
> Cache the DWARF debug info for DSO so we don't have to rebuild it for
> each address in the DSO (duh!).
>
> $ time /tmp/perf.orig report -g > /tmp/report.1
>
> real 0m1.845s
> user 0m0.963s
> sys 0m0.879s
>
> $ time /tmp/perf report -g > /tmp/report.2
>
> real 0m0.089s
> user 0m0.082s
> sys 0m0.006s
>
> $ diff /tmp/report.1 /tmp/report.2
> $
>
> Reported-by: Anton Blanchard <anton@samba.org>
> Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
> ---
> tools/perf/arch/powerpc/util/skip-callchain-idx.c | 91 ++++++++++++++++++++---
> 1 file changed, 82 insertions(+), 9 deletions(-)
>
> diff --git a/tools/perf/arch/powerpc/util/skip-callchain-idx.c b/tools/perf/arch/powerpc/util/skip-callchain-idx.c
> index d73ef8b..bfe254d 100644
> --- a/tools/perf/arch/powerpc/util/skip-callchain-idx.c
> +++ b/tools/perf/arch/powerpc/util/skip-callchain-idx.c
> @@ -38,6 +38,63 @@ static const Dwfl_Callbacks offline_callbacks = {
> .section_address = dwfl_offline_section_address,
> };
>
> +struct list_head perf_dwfl_files;
> +
> +struct perf_dwfl_node {
> + struct list_head node;
> + char *exec_file;
> + Dwfl *dwfl;
> +};
> +
> +static void perf_init_dwfl(void)
> +{
> + static int dwfl_inited;
> +
> + if (!dwfl_inited) {
> + dwfl_inited = 1;
> + INIT_LIST_HEAD(&perf_dwfl_files);
> + }
> +}
> +
> +static Dwfl *perf_dwfl_find(const char *file_name)
> +{
> + struct list_head *pos;
> + struct perf_dwfl_node *node;
> +
> + perf_init_dwfl();
> +
> + list_for_each(pos, &perf_dwfl_files) {
> + node = list_entry(pos, struct perf_dwfl_node, node);
> + if (!strcmp(node->exec_file, file_name))
> + return node->dwfl;
> + }
> +
> + return NULL;
> +}
> +
> +
> +/*
> + * Return 1 if were able to cache the DWARF debug info. 0 otherwise.
> + */
> +static int perf_dwfl_cache(const char *file_name, Dwfl *dwfl)
> +{
> + struct perf_dwfl_node *dwfl_node;
> +
> + dwfl_node = malloc(sizeof(struct perf_dwfl_node));
> +
> + if (!dwfl_node) {
> + pr_debug("%s(): Unable to alloc memory\n", __func__);
> + return 0;
> + }
> + INIT_LIST_HEAD(&dwfl_node->node);
> + dwfl_node->dwfl = dwfl;
> + dwfl_node->exec_file = strdup(file_name);
> +
> + list_add(&dwfl_node->node, &perf_dwfl_files);
> +
> + return 1;
> +}
> +
>
> /*
> * Use the DWARF expression for the Call-frame-address and determine
> @@ -155,16 +212,26 @@ static int check_return_addr(const char *exec_file, Dwarf_Addr pc)
> Dwarf_Addr start = pc;
> Dwarf_Addr end = pc;
> bool signalp;
> + int cached;
>
> - dwfl = dwfl_begin(&offline_callbacks);
> + cached = 1;
> + dwfl = perf_dwfl_find(exec_file);
> if (!dwfl) {
> - pr_debug("dwfl_begin() failed: %s\n", dwarf_errmsg(-1));
> - return -1;
> - }
> -
> - if (dwfl_report_offline(dwfl, "", exec_file, -1) == NULL) {
> - pr_debug("dwfl_report_offline() failed %s\n", dwarf_errmsg(-1));
> - goto out;
> + dwfl = dwfl_begin(&offline_callbacks);
> + if (!dwfl) {
> + pr_debug("dwfl_begin() failed: %s\n", dwarf_errmsg(-1));
> + return -1;
> + }
> +
> + if (dwfl_report_offline(dwfl, "", exec_file, -1) == NULL) {
> + pr_debug("dwfl_report_offline() failed %s\n",
> + dwarf_errmsg(-1));
> + goto out;
> + }
> + /*
> + * If we fail to alloc memory, we lose the benefit of caching
> + */
> + cached = perf_dwfl_cache(exec_file, dwfl);
> }
>
> mod = dwfl_addrmodule(dwfl, pc);
> @@ -194,7 +261,13 @@ static int check_return_addr(const char *exec_file, Dwarf_Addr pc)
> rc = check_return_reg(ra_regno, frame);
>
> out:
> - dwfl_end(dwfl);
> + /*
> + * In the unlikely event that we were unable to cache the debug
> + * info, release it so we don't leak fds.
> + */
> + if (!cached)
> + dwfl_end(dwfl);
> +
> return rc;
> }
>
> --
> 1.8.4.2
next prev parent reply other threads:[~2014-10-21 19:08 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-10-21 18:56 [PATCH] perf/powerpc: Cache the DWARF debug info Sukadev Bhattiprolu
2014-10-21 19:08 ` Arnaldo Carvalho de Melo [this message]
2014-10-22 0:09 ` Sukadev Bhattiprolu
2014-10-22 8:17 ` Jiri Olsa
2014-10-22 17:46 ` Sukadev Bhattiprolu
2014-10-23 13:37 ` Arnaldo Carvalho de Melo
2014-10-23 14:12 ` Jiri Olsa
2014-10-23 14:26 ` Arnaldo Carvalho de Melo
2014-10-23 15:13 ` Jiri Olsa
2014-10-23 15:33 ` Arnaldo Carvalho de Melo
2014-10-23 15:39 ` Jiri Olsa
2014-10-30 6:42 ` [tip:perf/core] perf tools powerpc: " tip-bot for Sukadev Bhattiprolu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20141021190839.GA4160@kernel.org \
--to=acme@kernel.org \
--cc=anton@au1.ibm.com \
--cc=jolsa@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=sukadev@linux.vnet.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.