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]
Date:   Tue, 13 Jun 2017 03:53:08 +0000
From:   Mark Santaniello <marksan@...com>
To:     Arnaldo Carvalho de Melo <acme@...nel.org>
CC:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 2/2] Support perf script -F brstackoff,dso

On 6/12/17, 4:36 PM, "Arnaldo Carvalho de Melo" <acme@...nel.org> wrote:

    Em Mon, Jun 12, 2017 at 03:51:13PM -0700, Mark Santaniello escreveu:
    > The idea here is to make AutoFDO easier in cloud environment with ASLR.
    > It's easiest to show how this is useful by example. I built a small test
    > akin to "while(1) { do_nothing(); }" where the do_nothing function is
    > loaded from a dso.  I sampled the execution with perf record -b.
    > 
    > Using the existing "brstack,dso", we get absolute addresses that are
    > affected by ASLR, and could be different on different hosts. The address
    > does not uniquely identify a branch/target in the binary:
    > $ ./perf script -F brstack,dso | sed 's/\/0 /\/0\n/g' \
    > > | grep burncpu | grep dso.so | head -n 1
    > 0x7f967139b6aa(/tmp/burncpu/dso.so)/0x4006b1(/tmp/burncpu/exe)/P/-/-/0
    > 
    > Using the existing "brstacksym,dso" is a little better, because the
    > symbol plus offset and dso name *does* uniquely identify a branch/target
    > in the binary.  Ultimately, however, AutoFDO wants a simple offset into
    > the binary, so we'd have to undo all the work perf did to symbolize in
    > the first place:
    > $ ./perf script -F brstacksym,dso | sed 's/\/0 /\/0\n/g' \
    > > | grep burncpu | grep dso.so | head -n 1
    > do_nothing+0x5(/tmp/burncpu/dso.so)/main+0x44(/tmp/burncpu/exe)/P/-/-/0
    > 
    > With the new "brstackoff,dso" we get what we need: a simple offset into a
    > specific dso/binary that uniquely identifies a branch/target:
    > $ ./perf script -F brstackoff,dso | sed 's/\/0 /\/0\n/g' \
    > > | grep burncpu | grep dso.so | head -n 1
    > 0x6aa(/tmp/burncpu/dso.so)/0x4006b1(/tmp/burncpu/exe)/P/-/-/0
    > 
    > Note this patch depends on commit 9d0ec0748cd9.
    
    Are you refererring to patch 1/2 in this series?
    
    [acme@...et linux]$ git show 9d0ec0748cd9
    fatal: ambiguous argument '9d0ec0748cd9': unknown revision or path not in the working tree.
    [acme@...et linux]$
    
    Can you please provide the test proggie source code?
    
    Seems simple enough, but to save time for people trying to test your
    patch, it would be good to have it available.
    
    - Arnaldo

Hi Arnaldo,

Sorry for the missteps – I’m new to this process.  I am indeed referring to patch 1/2 in this series.
I didn’t realize the revision hash was only meaningful locally.

Below is the little program I am using to test this.

Thanks,
Mark


~/src/burncpu$ cat burncpu.cpp
#include <dlfcn.h>

int main() {
  void* handle = dlopen("./dso.so", RTLD_LAZY);
  if (!handle) return -1;

  typedef void (*fp)();
  fp do_nothing = (fp) dlsym(handle, "do_nothing");

  while(1) {
    do_nothing();
  }
}

~/src/burncpu$ cat dso.cpp
extern "C" void do_nothing() {}

~/src/burncpu$ cat build.sh
#!/bin/bash
g++ -shared dso.cpp -o dso.so
g++ burncpu.cpp -o burncpu -ldl



     
    > Signed-off-by: Mark Santaniello <marksan@...com>
    > ---
    >  tools/perf/builtin-script.c | 56 +++++++++++++++++++++++++++++++++++++++++----
    >  1 file changed, 52 insertions(+), 4 deletions(-)
    > 
    > diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
    > index 6a7033b..1effc64 100644
    > --- a/tools/perf/builtin-script.c
    > +++ b/tools/perf/builtin-script.c
    > @@ -85,6 +85,7 @@ enum perf_output_field {
    >  	PERF_OUTPUT_INSN	    = 1U << 21,
    >  	PERF_OUTPUT_INSNLEN	    = 1U << 22,
    >  	PERF_OUTPUT_BRSTACKINSN	    = 1U << 23,
    > +	PERF_OUTPUT_BRSTACKOFF	    = 1U << 24,
    >  };
    >  
    >  struct output_option {
    > @@ -115,6 +116,7 @@ struct output_option {
    >  	{.str = "insn", .field = PERF_OUTPUT_INSN},
    >  	{.str = "insnlen", .field = PERF_OUTPUT_INSNLEN},
    >  	{.str = "brstackinsn", .field = PERF_OUTPUT_BRSTACKINSN},
    > +	{.str = "brstackoff", .field = PERF_OUTPUT_BRSTACKOFF},
    >  };
    >  
    >  /* default set to maintain compatibility with current format */
    > @@ -299,10 +301,9 @@ static int perf_evsel__check_attr(struct perf_evsel *evsel,
    >  		return -EINVAL;
    >  	}
    >  	if (PRINT_FIELD(DSO) && !PRINT_FIELD(IP) && !PRINT_FIELD(ADDR) &&
    > -	    !PRINT_FIELD(BRSTACK) && !PRINT_FIELD(BRSTACKSYM)) {
    > -		pr_err("Display of DSO requested but none of sample IP, sample address, "
    > -		       "brstack\nor brstacksym are selected. Hence, no addresses to "
    > -		       "convert to DSO.\n");
    > +	    !PRINT_FIELD(BRSTACK) && !PRINT_FIELD(BRSTACKSYM) && !PRINT_FIELD(BRSTACKOFF)) {
    > +		pr_err("Display of DSO requested but no address to convert.  Select\n"
    > +		       "sample IP, sample address, brstack, brstacksym, or brstackoff.\n");
    >  		return -EINVAL;
    >  	}
    >  	if (PRINT_FIELD(SRCLINE) && !PRINT_FIELD(IP)) {
    > @@ -606,6 +607,51 @@ static void print_sample_brstacksym(struct perf_sample *sample,
    >  	}
    >  }
    >  
    > +static void print_sample_brstackoff(struct perf_sample *sample,
    > +				    struct thread *thread,
    > +				    struct perf_event_attr *attr)
    > +{
    > +	struct branch_stack *br = sample->branch_stack;
    > +	struct addr_location alf, alt;
    > +	u64 i, from, to;
    > +
    > +	if (!(br && br->nr))
    > +		return;
    > +
    > +	for (i = 0; i < br->nr; i++) {
    > +
    > +		memset(&alf, 0, sizeof(alf));
    > +		memset(&alt, 0, sizeof(alt));
    > +		from = br->entries[i].from;
    > +		to   = br->entries[i].to;
    > +
    > +		thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, from, &alf);
    > +		if (alf.map && !alf.map->dso->adjust_symbols)
    > +			from = map__map_ip(alf.map, from);
    > +
    > +		thread__find_addr_map(thread, sample->cpumode, MAP__FUNCTION, to, &alt);
    > +		if (alt.map && !alt.map->dso->adjust_symbols)
    > +			to = map__map_ip(alt.map, to);
    > +
    > +		printf("0x%"PRIx64, from);
    > +		if (PRINT_FIELD(DSO)) {
    > +			printf("(");
    > +			map__fprintf_dsoname(alf.map, stdout);
    > +			printf(")");
    > +		}
    > +		printf("/0x%"PRIx64, to);
    > +		if (PRINT_FIELD(DSO)) {
    > +			printf("(");
    > +			map__fprintf_dsoname(alt.map, stdout);
    > +			printf(")");
    > +		}
    > +		printf("/%c/%c/%c/%d ",
    > +			mispred_str(br->entries + i),
    > +			br->entries[i].flags.in_tx ? 'X' : '-',
    > +			br->entries[i].flags.abort ? 'A' : '-',
    > +			br->entries[i].flags.cycles);
    > +	}
    > +}
    >  #define MAXBB 16384UL
    >  
    >  static int grab_bb(u8 *buffer, u64 start, u64 end,
    > @@ -1227,6 +1273,8 @@ static void process_event(struct perf_script *script,
    >  		print_sample_brstack(sample, thread, attr);
    >  	else if (PRINT_FIELD(BRSTACKSYM))
    >  		print_sample_brstacksym(sample, thread, attr);
    > +	else if (PRINT_FIELD(BRSTACKOFF))
    > +		print_sample_brstackoff(sample, thread, attr);
    >  
    >  	if (perf_evsel__is_bpf_output(evsel) && PRINT_FIELD(BPF_OUTPUT))
    >  		print_sample_bpf_output(sample);
    > -- 
    > 2.9.3
    

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ