perf annotate: Fix branch instruction with multiple operands
authorKim Phillips <kim.phillips@arm.com>
Thu, 1 Jun 2017 14:29:59 +0000 (09:29 -0500)
committerArnaldo Carvalho de Melo <acme@redhat.com>
Thu, 1 Jun 2017 17:48:36 +0000 (14:48 -0300)
commitb13bbeee5ee606cfb57ddcf47e66802f9aa7273e
treec8bdb79efed4b81c223eb40d7dda092c13f07280
parent54265664c15a68905d8d67d19205e9a767636434
perf annotate: Fix branch instruction with multiple operands

'perf annotate' is dropping the cr* fields from branch instructions.

Fix it by adding support to display branch instructions having
multiple operands.

Power Arch objdump of int_sqrt:

 20.36 | c0000000004d2694:   subf   r10,r10,r3
       | c0000000004d2698: v bgt    cr6,c0000000004d26a0 <int_sqrt+0x40>
  1.82 | c0000000004d269c:   mr     r3,r10
 29.18 | c0000000004d26a0:   mr     r10,r8
       | c0000000004d26a4: v bgt    cr7,c0000000004d26ac <int_sqrt+0x4c>
       | c0000000004d26a8:   mr     r10,r7

Power Arch Before Patch:

 20.36 |       subf   r10,r10,r3
       |     v bgt    40
  1.82 |       mr     r3,r10
 29.18 | 40:   mr     r10,r8
       |     v bgt    4c
       |       mr     r10,r7

Power Arch After patch:

 20.36 |       subf   r10,r10,r3
       |     v bgt    cr6,40
  1.82 |       mr     r3,r10
 29.18 | 40:   mr     r10,r8
       |     v bgt    cr7,4c
       |       mr     r10,r7

Also support AArch64 conditional branch instructions, which can
have up to three operands:

Aarch64 Non-simplified (raw objdump) view:

       │ffff0000083cd11c: ↑ cbz    w0, ffff0000083cd100 <security_fil▒
...
  4.44 │ffff000083cd134: ↓ tbnz   w0, #26, ffff0000083cd190 <securit▒
...
  1.37 │ffff000083cd144: ↓ tbnz   w22, #5, ffff0000083cd1a4 <securit▒
       │ffff000083cd148:   mov    w19, #0x20000                   //▒
  1.02 │ffff000083cd14c: ↓ tbz    w22, #2, ffff0000083cd1ac <securit▒
...
  0.68 │ffff000└──3cd16c: ↑ cbnz   w0, ffff0000083cd120 <security_fil▒

Aarch64 Simplified, before this patch:

       │    ↑ cbz    40
...
  4.44 │   │↓ tbnz   w0, #26, ffff0000083cd190 <security_file_permiss▒
...
  1.37 │   │↓ tbnz   w22, #5, ffff0000083cd1a4 <security_file_permiss▒
       │   │  mov    w19, #0x20000                   // #131072
  1.02 │   │↓ tbz    w22, #2, ffff0000083cd1ac <security_file_permiss▒
...
  0.68 │   └──cbnz   60

the cbz operand is missing, and the tbz doesn't get simplified processing
at all because the parsing function failed to match an address.

Aarch64 Simplified, After this patch applied:

       │    ↑ cbz    w0, 40
...
  4.44 │   │↓ tbnz   w0, #26, d0
...
  1.37 │   │↓ tbnz   w22, #5, e4
       │   │  mov    w19, #0x20000                   // #131072
  1.02 │   │↓ tbz    w22, #2, ec
...
  0.68 │   └──cbnz   w0, 60

Originally-by: Ravi Bangoria <ravi.bangoria@linux.vnet.ibm.com>
Tested-by: Ravi Bangoria <ravi.bangoria@linux.vnet.ibm.com>
Reported-by: Anton Blanchard <anton@samba.org>
Reported-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Kim Phillips <kim.phillips@arm.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Taeung Song <treeze.taeung@gmail.com>
Link: http://lkml.kernel.org/r/20170601092959.f60d98912e8a1b66fd1e4c0e@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
tools/perf/util/annotate.c