linux-kernel - Re: [PATCH V2 5/5] powerpc, perf: Enable branch stack sampling framework support with BHRB

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] [day] [month] [year] [list]

Message-ID: <516F6165.9070205@linux.vnet.ibm.com>
Date:	Thu, 18 Apr 2013 08:28:45 +0530
From:	Anshuman Khandual <khandual@...ux.vnet.ibm.com>
To:	Michael Ellerman <michael@...erman.id.au>
CC:	linuxppc-dev@...abs.org, mikey@...ling.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH V2 5/5] powerpc, perf: Enable branch stack sampling framework
 support with BHRB

On 04/17/2013 05:37 PM, Anshuman Khandual wrote:
> On 04/17/2013 12:38 PM, Michael Ellerman wrote:
>> On Tue, Apr 16, 2013 at 09:24:10PM +0530, Anshuman Khandual wrote:
>>> This patch provides basic enablement for perf branch stack sampling framework
>>> on POWER8 processor with a new PMU feature called BHRB (Branch History Rolling
>>> Buffer).
>>>
>>> Signed-off-by: Anshuman Khandual <khandual@...ux.vnet.ibm.com>
>>> ---
>>>  arch/powerpc/perf/core-book3s.c     | 96 +++++++++++++++++++++++++++++++++++--
>>>  arch/powerpc/perf/perf_event_bhrb.c | 85 ++++++++++++++++++++++++++++++++
>>>  2 files changed, 178 insertions(+), 3 deletions(-)
>>>  create mode 100644 arch/powerpc/perf/perf_event_bhrb.c
>>>
>>> diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c
>>> index 4ac6e64..f4d1347 100644
>>> --- a/arch/powerpc/perf/core-book3s.c
>>> +++ b/arch/powerpc/perf/core-book3s.c
>>> @@ -19,6 +19,8 @@
>>>  #include <asm/firmware.h>
>>>  #include <asm/ptrace.h>
>>>  
>>> +#define BHRB_MAX_ENTRIES	32
>>> +
>>>  struct cpu_hw_events {
>>>  	int n_events;
>>>  	int n_percpu;
>>> @@ -38,11 +40,21 @@ struct cpu_hw_events {
>>>  
>>>  	unsigned int group_flag;
>>>  	int n_txn_start;
>>> +
>>> +	/* BHRB bits */
>>> +	u64				bhrb_filter;	/* BHRB HW branch filter */
>>> +	int				bhrb_users;
>>> +	void				*bhrb_context;
>>> +	struct	perf_branch_stack	bhrb_stack;
>>> +	struct	perf_branch_entry	bhrb_entries[BHRB_MAX_ENTRIES];
>>>  };
>>> +
>>>  DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
>>>  
>>>  struct power_pmu *ppmu;
>>>  
>>> +#include "perf_event_bhrb.c"
>>> +
>>
>> Um, why are you doing that?
>>
> 
> There was no specific reason for that.
> 

Ahh, I remember it now. The function in the new file uses "cpu_hw_events" structure
which is passed during "record_and_restart" data capture phase. Right now cpu_hw_events is
not defined in the header file but inside core-book3s.c itself. Solution to this problem
could be any of these.

(0) Move all the code from the new file perf_event_bhrb.c into core-book3s.c
(1) Move cpu_hw_events structure to perf_event_server.h
(2) Create additional BHRB processing function inside struct power_pmu and
    define it for P8 inside power8_pmu.c

Regards
Anshuman

--
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/