| 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
| ||
|
Date: Thu, 28 Jan 2010 15:59:38 -0500
From: Masami Hiramatsu <mhiramat@...hat.com>
To: Arnaldo Carvalho de Melo <acme@...radead.org>
CC: Ingo Molnar <mingo@...e.hu>, linux-kernel@...r.kernel.org,
Frédéric Weisbecker <fweisbec@...il.com>,
Mike Galbraith <efault@....de>,
Peter Zijlstra <a.p.zijlstra@...llo.nl>,
Paul Mackerras <paulus@...ba.org>
Subject: Re: [PATCH 4/9] perf session: Create kernel maps in the constructor
Arnaldo Carvalho de Melo wrote:
> Em Thu, Jan 28, 2010 at 11:28:29AM -0500, Masami Hiramatsu escreveu:
>
>> Yeah, and if you can separate an interface only for handling symbols
>> from debug/elf binaries, it will be helpful for me too.
>
> It was done on this same series, now you can do as the patch at the end
> of this message, that will be on my next series, does.
Great! thank you!
>
> Testing with this patch:
>
> [root@...pio ~]# perf probe --list
> [root@...pio ~]# perf probe schedule
> Added new event:
> probe:schedule (on schedule+0)
>
> You can now use it on all perf tools, such as:
>
> perf record -e probe:schedule -a sleep 1
>
> [root@...pio ~]# perf record -e probe:schedule -a sleep 1
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.158 MB perf.data (~6923 samples) ]
> [root@...pio ~]# perf report | head -9
> # Samples: 4
> #
> # Overhead Command Shared Object Symbol
> # ........ ............... ..................... ......
> #
> 25.00% perf libc-2.10.2.so [.] __GI___libc_read
> 25.00% kondemand/1 [kernel.kallsyms] [k] kernel_thread_helper
> 25.00% kondemand/0 [kernel.kallsyms] [k] kernel_thread_helper
> [root@...pio ~]#
>
> And what about support for modules? You have everything you need there
> as well :-)
>
> If you do:
>
> struct map *mod;
> struct symbol *sym;
>
> mod = map_groups__find_by_name(&session.kmap_groups,
> MAP__FUNCTION, "[ipv6]");
>
> sym = map__find_symbol_by_name(module, "ip6_xmit");
>
> you'll get ipv6.ko located and loaded, mod->dso->long_name thus will
> have the file to use with libdwarf and sym->start will have the
> unrelocated address, and mod->start will have where ipv6 is loaded in
> kernel space
Oh, nice :-) Is that available for the modules which aren't loaded?
And yeah, I'd like to support modules and it will requires some
enhancement in kprobe-tracer too.
.
> Also you should be able to find variables by name or address too, just
> use MAP__VARIABLE where above you see MAP__FUNCTION.
>
> - Arnaldo
>
> From 387ab05301cce32c8b091ba9feb94c906124e8a8 Mon Sep 17 00:00:00 2001
> From: Arnaldo Carvalho de Melo <acme@...hat.com>
> Date: Thu, 28 Jan 2010 16:12:31 -0200
> Subject: [PATCH 1/1] perf probe: Don't use a perf_session instance just to resolve symbols
> MIME-Version: 1.0
> Content-Type: text/plain; charset=utf-8
> Content-Transfer-Encoding: 8bit
>
> With the recent modifications done to untie the session and symbol
> layers, 'perf probe' now can use just the symbols layer.
Could you remove session.kmap too?
other parts look good to me.
Thank you,
>
> Cc: Frédéric Weisbecker <fweisbec@...il.com>
> Cc: Masami Hiramatsu <mhiramat@...hat.com>
> Cc: Mike Galbraith <efault@....de>
> Cc: Peter Zijlstra <a.p.zijlstra@...llo.nl>
> Cc: Paul Mackerras <paulus@...ba.org>
> Signed-off-by: Arnaldo Carvalho de Melo <acme@...hat.com>
> ---
> tools/perf/builtin-probe.c | 15 +++++++++------
> 1 files changed, 9 insertions(+), 6 deletions(-)
>
> diff --git a/tools/perf/builtin-probe.c b/tools/perf/builtin-probe.c
> index 4fa73ec..cd45e02 100644
> --- a/tools/perf/builtin-probe.c
> +++ b/tools/perf/builtin-probe.c
> @@ -41,7 +41,6 @@
> #include "util/debugfs.h"
> #include "util/symbol.h"
> #include "util/thread.h"
> -#include "util/session.h"
> #include "util/parse-options.h"
> #include "util/parse-events.h" /* For debugfs_path */
> #include "util/probe-finder.h"
> @@ -59,7 +58,8 @@ static struct {
> int nr_probe;
> struct probe_point probes[MAX_PROBES];
> struct strlist *dellist;
> - struct perf_session *psession;
> + struct map_groups kmap_groups;
> + struct map *kmaps[MAP__NR_TYPES];
> struct map *kmap;
> struct line_range line_range;
> } session;
> @@ -212,10 +212,13 @@ static void init_vmlinux(void)
> pr_debug("Use vmlinux: %s\n", symbol_conf.vmlinux_name);
> if (symbol__init() < 0)
> die("Failed to init symbol map.");
> - session.psession = perf_session__new(NULL, O_WRONLY, false);
> - if (session.psession == NULL)
> - die("Failed to init perf_session.");
> - session.kmap = session.psession->vmlinux_maps[MAP__FUNCTION];
> +
> + map_groups__init(&session.kmap_groups);
> + if (map_groups__create_kernel_maps(&session.kmap_groups,
> + session.kmaps) < 0)
> + die("Failed to create kernel maps.");
> +
> + session.kmap = session.kmaps[MAP__FUNCTION];
> if (!session.kmap)
> die("Could not find kernel map.\n");
> }
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division
e-mail: mhiramat@...hat.com
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists