Re: [PATCH 2/4] Implements aarch64 process record and reverse debugging support

[Yao Qi <qiyaoltc@gmail.com>]

Yao Qi <qiyaoltc@gmail.com> writes:

>> gcc -g -O2 -mminimal-toc  -I. -I../../gdb -I../../gdb/common -I../../gdb/config -DLOCALEDIR="\"/usr/local/share/locale\"" -DHAVE_CONFIG_H -I../../gdb/../include/opcode -I../../gdb/../opcodes/.. -I../../gdb/../readline/.. -I../../gdb/../zlib -I../bfd -I../../gdb/../bfd -I../../gdb/../include -I../libdecnumber -I../../gdb/../libdecnumber  -I../../gdb/gnulib/import -Ibuild-gnulib/import   -DTUI=1  -I/usr/include/python2.7 -I/usr/include/python2.7 -Wall -Wpointer-arith -Wno-unused -Wunused-value -Wunused-function -Wno-switch -Wno-char-subscripts -Wempty-body -Wpointer-sign -Wmissing-prototypes -Wdeclaration-after-statement -Wmissing-parameter-type -Wold-style-declaration -Wold-style-definition -Wformat-nonliteral -Werror -c -o aarch64-tdep.o -MT aarch64-tdep.o -MMD -MP -MF .deps/aarch64-tdep.Tpo ../../gdb/aarch64-tdep.c
>> ../../gdb/aarch64-tdep.c: In function ‘aarch64_process_record’:
>> ../../gdb/aarch64-tdep.c:2823:23: error: ‘record_buf[0]’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
>>                  memcpy(&REGS[0], &RECORD_BUF[0], sizeof(uint32_t)*LENGTH); \
>>                        ^
>> ../../gdb/aarch64-tdep.c:3529:12: note: ‘record_buf[0]’ was declared here
>>    uint32_t record_buf[2];
>>             ^
>>
>> The simple patch below intends to fix this. Ok?
>
> This build error reveals some logic error inside function
> aarch64_record_data_proc_simd_fp on decoding instructions.  I'll take a look.

Here is the patch below, it fixes the build failure on my machine.  Does
it work for you too?

-- 
Yao (齐尧)

From: Yao Qi <yao.qi@linaro.org>
Date: Wed, 27 May 2015 11:08:40 +0100
Subject: [PATCH] [aarch64] Handle unknown instruction encoding

We see the following build error:

 gdb/aarch64-tdep.c: In function ‘aarch64_process_record’:
 gdb/aarch64-tdep.c:2823:17: error: ‘record_buf[0]’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
                 memcpy(&REGS[0], &RECORD_BUF[0], sizeof(uint32_t)*LENGTH); \
                 ^
 gdb/aarch64-tdep.c:3529:12: note: ‘record_buf[0]’ was declared here
   uint32_t record_buf[2];

since record_buf[0] isn't initialized in some paths in which
instruction encodings doesn't exist in current ARMv8 reference manual.
This patch is to let the function return AARCH64_RECORD_UNKNOWN for
these paths.

gdb:

2015-05-27  Yao Qi  <yao.qi@linaro.org>

	* aarch64-tdep.c (aarch64_record_data_proc_simd_fp): Return
	AARCH64_RECORD_UNKNOWN for unknown instruction encoding.

diff --git a/gdb/aarch64-tdep.c b/gdb/aarch64-tdep.c
index 8597497..d07b26c 100644
--- a/gdb/aarch64-tdep.c
+++ b/gdb/aarch64-tdep.c
@@ -3592,7 +3592,7 @@ aarch64_record_data_proc_simd_fp (insn_decode_record *aarch64_insn_r)
 	      record_buf[0] = AARCH64_CPSR_REGNUM;
 	    }
 	  /* Floating point - integer conversions instructions.  */
-	  if (insn_bits12_15 == 0x00)
+	  else if (insn_bits12_15 == 0x00)
 	    {
 	      /* Convert float to integer instruction.  */
 	      if (!(opcode >> 1) || ((opcode >> 1) == 0x02 && !rmode))
@@ -3621,8 +3621,14 @@ aarch64_record_data_proc_simd_fp (insn_decode_record *aarch64_insn_r)
 		  else
 		    record_buf[0] = reg_rd + AARCH64_V0_REGNUM;
 		}
+	      else
+		return AARCH64_RECORD_UNKNOWN;
             }
+	  else
+	    return AARCH64_RECORD_UNKNOWN;
         }
+      else
+	return AARCH64_RECORD_UNKNOWN;
     }
   else if ((insn_bits28_31 & 0x09) == 0x00 && insn_bits24_27 == 0x0e)
     {