| 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, 6 Feb 2020 15:01:52 +0000
From: Mike Leach <mike.leach@...aro.org>
To: Leo Yan <leo.yan@...aro.org>
Cc: Arnaldo Carvalho de Melo <acme@...nel.org>,
Mathieu Poirier <mathieu.poirier@...aro.org>,
Suzuki K Poulose <suzuki.poulose@....com>,
Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...hat.com>,
Mark Rutland <mark.rutland@....com>,
Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
Jiri Olsa <jolsa@...hat.com>,
Namhyung Kim <namhyung@...nel.org>,
linux-arm-kernel <linux-arm-kernel@...ts.infradead.org>,
linux-kernel@...r.kernel.org,
Robert Walker <robert.walker@....com>,
Coresight ML <coresight@...ts.linaro.org>
Subject: Re: [PATCH v4 5/5] perf cs-etm: Synchronize instruction sample with
the thread stack
Hi Leo,
On Mon, 3 Feb 2020 at 02:08, Leo Yan <leo.yan@...aro.org> wrote:
>
> The synthesized flow use 'tidq->packet' for instruction samples; on the
> other hand, 'tidp->prev_packet' is used to generate the thread stack and
> the branch samples, this results in the instruction samples using one
> packet ahead than thread stack and branch samples ('tidp->prev_packet'
> vs 'tidq->packet').
>
> This leads to an instruction's callchain error as shows in below
> example:
>
> main 1579 100 instructions:
> ffff000010214854 perf_event_update_userpage+0x4c ([kernel.kallsyms])
> ffff000010214850 perf_event_update_userpage+0x48 ([kernel.kallsyms])
> ffff000010219360 perf_swevent_add+0x88 ([kernel.kallsyms])
> ffff0000102135f4 event_sched_in.isra.57+0xbc ([kernel.kallsyms])
> ffff0000102137a0 group_sched_in+0x60 ([kernel.kallsyms])
> ffff000010213b84 flexible_sched_in+0xfc ([kernel.kallsyms])
> ffff00001020c0b4 visit_groups_merge+0x12c ([kernel.kallsyms])
>
> In the callchain log, for the two continuous lines the up line contains
> one child function info and the followed line contains the caller
> function info, and so forth. So the first two lines are:
>
> perf_event_update_userpage+0x4c => the sampled instruction
> perf_event_update_userpage+0x48 => the parent function's calling
>
> The child function and parent function both are the same function
> perf_event_update_userpage(), but this isn't a recursive function, thus
> the sequence for perf_event_update_userpage() calling itself shouldn't
> never happen. This callchain error is caused by the instruction sample
> using an ahead packet than the thread stack, the thread stack is deferred
> to process the new packet and misses to pop stack if it is just a return
> packet.
>
> To fix this issue, we can simply change to use 'tidq->prev_packet' to
> generate the instruction samples, this allows the thread stack to push
> and pop synchronously with instruction sample. Finally, the callchain
> can be displayed correctly as below:
>
> main 1579 100 instructions:
> ffff000010214854 perf_event_update_userpage+0x4c ([kernel.kallsyms])
> ffff000010219360 perf_swevent_add+0x88 ([kernel.kallsyms])
> ffff0000102135f4 event_sched_in.isra.57+0xbc ([kernel.kallsyms])
> ffff0000102137a0 group_sched_in+0x60 ([kernel.kallsyms])
> ffff000010213b84 flexible_sched_in+0xfc ([kernel.kallsyms])
> ffff00001020c0b4 visit_groups_merge+0x12c ([kernel.kallsyms])
>
> Signed-off-by: Leo Yan <leo.yan@...aro.org>
> ---
> tools/perf/util/cs-etm.c | 8 +++++---
> 1 file changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/tools/perf/util/cs-etm.c b/tools/perf/util/cs-etm.c
> index 8f805657658d..410e40ce19f2 100644
> --- a/tools/perf/util/cs-etm.c
> +++ b/tools/perf/util/cs-etm.c
> @@ -1414,7 +1414,7 @@ static int cs_etm__sample(struct cs_etm_queue *etmq,
> struct cs_etm_packet *tmp;
> int ret;
> u8 trace_chan_id = tidq->trace_chan_id;
> - u64 instrs_executed = tidq->packet->instr_count;
> + u64 instrs_executed = tidq->prev_packet->instr_count;
>
> tidq->period_instructions += instrs_executed;
>
> @@ -1505,7 +1505,8 @@ static int cs_etm__sample(struct cs_etm_queue *etmq,
> * instruction)
> */
> addr = cs_etm__instr_addr(etmq, trace_chan_id,
> - tidq->packet, offset - 1);
> + tidq->prev_packet,
> + offset - 1);
> ret = cs_etm__synth_instruction_sample(
> etmq, tidq, addr,
> etm->instructions_sample_period);
> @@ -1525,7 +1526,8 @@ static int cs_etm__sample(struct cs_etm_queue *etmq,
> * instruction)
> */
> addr = cs_etm__instr_addr(etmq, trace_chan_id,
> - tidq->packet, offset - 1);
> + tidq->prev_packet,
> + offset - 1);
> ret = cs_etm__synth_instruction_sample(
> etmq, tidq, addr,
> etm->instructions_sample_period);
> --
> 2.17.1
>
I am really not convinced that this is a correct solution.
Consider a set of trace range packet inputs:
current: 0x3000-0x3050
prev: 0x2000-0x2100
prev-1: 0x1020-0x1080
Before your modification.....
cs_etm__sample() processes the current packet....
On entry, the branch stack will contain:0x1080=>0x2000;
We add to this from the current packet to get: 0x1080=>0x2000; 0x2100=>0x3000;
This is then copied by cs_etm__copy_last_branch_rb()
We find the instruction sample address in the range 0x3000 to 0x3050,
e.g. 0x3010.
cs_etm__synth_instruction_sample() will then generate a sample with values
sample.ip = 0x3010
sample.branch_stack = 0x1080=>0x2000; 0x2100=>0x3000;
to be passed to the perf session / injected as required.
This sample has the correct branch context for the sampled address -
i.e. how the code arrived @0x3010
After the modification.....
The branch stack will be the same, but the sample address will be from
the range 0x2000-0x2010, e.g. 0x2008 to give a sample in
cs_etm__synth_instruction_sample() of
sample.ip = 0x2008
sample.branch_stack = 0x1080=>0x2000; 0x2100=>0x3000;
This really does not make much sense - the branch stack no longer
relates to the sample.ip.
Further - cs_etm__synth_instruction_sample() calls cs_etm__copy_insn()
using the _current_ packet and sample.ip. This is a clear mismatch.
I don't know what is causing the apparent error in the callchain, but
given that the previous features added in this set, work without this
alteration, I feel there must be another solution.
Regards
Mike
--
Mike Leach
Principal Engineer, ARM Ltd.
Manchester Design Centre. UK
Powered by blists - more mailing lists