[PATCH] Record and replay: ARM thumb 32bit instructions support

[Yao Qi]

On 10/07/2013 05:16 AM, Omair Javaid wrote:
> * Without thumb32 Patch
> Native GDB (with support process record enabled) # of expected passes 32
> Native GDB (with support process record and reverse debug enabled) #
> of expected passes 518
> Remote GDB (with support process record enabled) # of expected passes 506
> Remote GDB (with support process record and reverse debug enabled) #
> of expected passes 506
>
> * With thumb32 Patch
> Native GDB (with support process record enabled) # of expected passes
> 105 (73 new passed tests after thumb32 patch)
> Native GDB (with support process record and reverse debug enabled) #
> of expected passes 1133 (627 new passed tests after thumb32 patch)
> Remote GDB (with support process record enabled) # of expected passes
> 973 (467 new passed tests after thumb32 patch)
> Remote GDB (with support process record and reverse debug enabled) #
> of expected passes 978 (472 new passed tests after thumb32 patch)
>

The result looks good.

> gdb/testsuite/ChangeLog:
>
> 2013-10-06  Omair Javaid<Omair.Javaid@linaro.org>
>
>          * arm-tdep.c: Adds support for thumb32 instructions in
>            arm process record.

Your changelog entry is too simple.  We still need to list the added new 
function names here, even it is quite mechanical.

We still need a NEWS entry.

> +      else
> +        {
> +          offset_imm = bits (thumb2_insn_r->arm_insn, 0, 7);
> +
> +          if (bit (thumb2_insn_r->arm_insn, 24))
> +            {
> +              if (bit (thumb2_insn_r->arm_insn, 23))
> +                {
> +                  offset_addr = u_regval[0] + (offset_imm * 4);
> +                }

braces are not needed if the statement is single-line.  Here and
elsewhere.

Why don't you add "arm*-*-linux*" triplet into proc 
supports_process_record and supports_reverse in testsuite/lib/gdb.exp?
-- 
Yao (齐尧)