linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Martin Liška" <mliska@suse.cz>
To: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: linux-perf-users@vger.kernel.org,
	lkml <linux-kernel@vger.kernel.org>, Jiri Olsa <jolsa@redhat.com>
Subject: Re: [PATCH] Properly interpret indirect call in perf annotate.
Date: Mon, 27 Aug 2018 11:06:21 +0200	[thread overview]
Message-ID: <64684c59-492c-3310-a5d2-14b467602acc@suse.cz> (raw)
In-Reply-To: <20180823141219.GA4766@kernel.org>

[-- Attachment #1: Type: text/plain, Size: 1887 bytes --]

On 08/23/2018 04:12 PM, Arnaldo Carvalho de Melo wrote:
> Em Thu, Aug 23, 2018 at 02:29:34PM +0200, Martin Liška escreveu:
>> The patch changes interpretation of:
>> callq  *0x8(%rbx)
>>
>> from:
>>   0.26 │     → callq  *8
>> to:
>>   0.26 │     → callq  *0x8(%rbx)
>>
>> in this can an address is followed by a register, thus
>> one can't parse only address.
> 
> Please mention one or two functions where such sequence appears, so that
> others can reproduce your before/after more quickly,

Sure, there's self-contained example on can compile (-O2) and test.
It's following call in test function:

test:
.LFB1:
        .cfi_startproc
        movq    %rdi, %rax
        subq    $8, %rsp
        .cfi_def_cfa_offset 16
        movq    %rsi, %rdi
        movq    %rdx, %rsi
        call    *8(%rax) <---- here
        cmpl    $1, %eax
        adcl    $-1, %eax
        addq    $8, %rsp
        .cfi_def_cfa_offset 8
        ret
        .cfi_endproc

Martin

> 
> - Arnaldo
>  
>> Signed-off-by: Martin Liška <mliska@suse.cz>
>> ---
>>  tools/perf/util/annotate.c | 10 ++++++++--
>>  1 file changed, 8 insertions(+), 2 deletions(-)
>>
>>
> 
>> diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c
>> index e4268b948e0e..e32ead4744bd 100644
>> --- a/tools/perf/util/annotate.c
>> +++ b/tools/perf/util/annotate.c
>> @@ -246,8 +246,14 @@ static int call__parse(struct arch *arch, struct ins_operands *ops, struct map_s
>>  
>>  indirect_call:
>>  	tok = strchr(endptr, '*');
>> -	if (tok != NULL)
>> -		ops->target.addr = strtoull(tok + 1, NULL, 16);
>> +	if (tok != NULL) {
>> +		endptr++;
>> +
>> +		/* Indirect call can use a non-rip register and offset: callq  *0x8(%rbx).
>> +		 * Do not parse such instruction.  */
>> +		if (strstr(endptr, "(%r") == NULL)
>> +			ops->target.addr = strtoull(endptr, NULL, 16);
>> +	}
>>  	goto find_target;
>>  }
>>  
>>
> 

[-- Attachment #2: perf-callq.c --]
[-- Type: text/x-csrc, Size: 1271 bytes --]

typedef int (*htab_eq) (const void *, const void *);

static int eq (const void *a, const void *b)
{
  return a - b;
}

struct htab
{
  /* Current size (in entries) of the hash table */
  int size;

  /* Pointer to comparison function.  */
  htab_eq eq_f;

  /* Table itself.  */
  void **entries;

  /* Current number of elements including also deleted elements */
  int n_elements;

  /* Current number of deleted elements in the table */
  int n_deleted;

  /* The following member is used for debugging. Its value is number
     of all calls of `htab_find_slot' for the hash table. */
  unsigned int searches;

  /* The following member is used for debugging.  Its value is number
     of collisions fixed for time of work with the hash table. */
  unsigned int collisions;

  unsigned int max_size;

  /* This is non-zero if we are allowed to return NULL for function calls
     that allocate memory.  */
  int return_allocation_failure;
};

int
__attribute__ ((noinline))
test(struct htab *t, int *a, int *b)
{
  int r = t->eq_f (a, b);
  if (r)
    return r - 1;

  return 0;
}

struct htab mytable;
int r;

int main(int argc, char **argv)
{
  mytable.eq_f = &eq;
  for (unsigned i = 0; i < 100000000; i++)
    r += test (&mytable, &argc, &argc);
  
  return 0;
}

  reply	other threads:[~2018-08-27  9:06 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-08-23 12:29 [PATCH] Properly interpret indirect call in perf annotate Martin Liška
2018-08-23 14:12 ` Arnaldo Carvalho de Melo
2018-08-27  9:06   ` Martin Liška [this message]
2018-08-28 14:10     ` Arnaldo Carvalho de Melo
2018-08-28 14:18       ` Arnaldo Carvalho de Melo
2018-08-28 17:55         ` Martin Liška
2018-08-23 23:00 ` Kim Phillips
2018-08-27 10:37 ` Namhyung Kim
2018-08-27 14:28   ` Martin Liška
2018-08-28 14:10 ` Arnaldo Carvalho de Melo

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=64684c59-492c-3310-a5d2-14b467602acc@suse.cz \
    --to=mliska@suse.cz \
    --cc=acme@kernel.org \
    --cc=jolsa@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-perf-users@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).