linux-kernel - Re: [PATCH v6 10/11] perf callchain: Support x86 target platform

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

Message-ID: <574C1CE0.1050309@huawei.com>
Date:	Mon, 30 May 2016 18:58:40 +0800
From:	Hekuang <hekuang@...wei.com>
To:	Jiri Olsa <jolsa@...hat.com>
CC:	<peterz@...radead.org>, <mingo@...hat.com>, <acme@...nel.org>,
	<alexander.shishkin@...ux.intel.com>, <wangnan0@...wei.com>,
	<jpoimboe@...hat.com>, <ak@...ux.intel.com>, <eranian@...gle.com>,
	<namhyung@...nel.org>, <adrian.hunter@...el.com>,
	<sukadev@...ux.vnet.ibm.com>, <masami.hiramatsu.pt@...achi.com>,
	<tumanova@...ux.vnet.ibm.com>, <kan.liang@...el.com>,
	<penberg@...nel.org>, <dsahern@...il.com>,
	<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v6 10/11] perf callchain: Support x86 target platform



在 2016/5/30 17:30, Jiri Olsa 写道:
> On Mon, May 30, 2016 at 05:11:35PM +0800, Hekuang wrote:
>> hi
>>
>> 在 2016/5/30 16:53, Jiri Olsa 写道:
>>> On Sat, May 28, 2016 at 11:59:59AM +0000, He Kuang wrote:
>>>> Support x86(32-bit) cross platform callchain unwind.
>>>>
>>>> Signed-off-by: He Kuang <hekuang@...wei.com>
>>>> ---
>>>>    tools/perf/arch/Build                              |  1 +
>>>>    tools/perf/arch/x86/util/unwind-libunwind.c        |  7 ++++---
>>>>    tools/perf/arch/x86/util/unwind-libunwind_x86_32.c | 21 +++++++++++++++++++++
>>>>    tools/perf/util/unwind-libunwind-local.c           |  4 ++++
>>>>    tools/perf/util/unwind-libunwind.c                 | 19 +++++++++++++------
>>>>    tools/perf/util/unwind.h                           | 10 ++++++++++
>>>>    6 files changed, 53 insertions(+), 9 deletions(-)
>>>>    create mode 100644 tools/perf/arch/x86/util/unwind-libunwind_x86_32.c
>>>>
>>>> diff --git a/tools/perf/arch/Build b/tools/perf/arch/Build
>>>> index 109eb75..3fc4af1 100644
>>>> --- a/tools/perf/arch/Build
>>>> +++ b/tools/perf/arch/Build
>>>> @@ -1,2 +1,3 @@
>>>>    libperf-y += common.o
>>>>    libperf-y += $(ARCH)/
>>>> +libperf-$(CONFIG_LIBUNWIND_X86)      += x86/util/unwind-libunwind_x86_32.o
>>> we have Build file directly in arch/x86/util/
>>>
>>> if you do it like this to include generic file easily
>>> we better fix the include then
>> This is because "libperf-y += $(ARCH)" will only sink into $(ARCH) folder,
>> for example on x86_64, only tools/perf/arch/x86 will be built. But for
>>   remote libunwind, we also need
>>   'tools/perf/arch/arm64/util/unwind-libunwind.o', while arm64 folder is
>> not added to libperf-y. Is there a gracefull to deal with this?
> you just need to include the file, right?
>
> I think it's ok to include arch/arm/....c
> from arch/x86/util/unwind-libunwind-arm64.c
>
> jirka

By following your advise, if ARCH=x86, the file tree will
be like this:

arch/x86
-    arch/x86/util/unwind-libunwind-arm64.c
-    arch/x86/util/unwind-libunwind-x86_32.c
-    arch/x86/util/unwind-libunwind-x86_64.c
-    arch/x86/util/unwind-libunwind-arm.c

And for ARCH=arm （host machine is arm, it should be considered)
arch/arm
-    arch/arm/util/unwind-libunwind-arm64.c
-    arch/arm/util/unwind-libunwind-x86_32.c
-    arch/arm/util/unwind-libunwind-x86_64.c
-    arch/arm/util/unwind-libunwind-arm.c

For arm64:
arch/arm64
-    arch/arm64/util/unwind-libunwind-arm64.c
-    arch/arm64/util/unwind-libunwind-x86_32.c
-    arch/arm64/util/unwind-libunwind-x86_64.c
-    arch/arm64/util/unwind-libunwind-arm.c

But in my patch, the file tree is like this:

arch
-    arch/arm64/util/unwind-libunwind-arm64.c
-    arch/x86/util/unwind-libunwind-x86_64.c
-    arch/x86/util/unwind-libunwind-x86_32.c
-    arch/arm/util/unwind-libunwind-arm.c

I admit that

+libperf-$(CONFIG_LIBUNWIND_X86)      += x86/util/unwind-libunwind_x86_32.o

is not so good, but do you think the above file tree is
too redunctant?

Thank you.
>