Re: [Qemu-devel] [PATCH v4 09/11] target-arm: dfilter support for in_asm, op, opt_op

[Richard Henderson <rth@twiddle.net>]

On 08/04/2015 10:26 AM, Alex Bennée wrote:
> 
> Richard Henderson <rth@twiddle.net> writes:
> 
>> On 08/03/2015 02:14 AM, Alex Bennée wrote:
>>> Each individual architecture needs to use the qemu_log_in_addr_range()
>>> feature for enabling in_asm and marking blocks for op/opt_op output.
>>>
>>> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
>>> ---
>>>  target-arm/translate-a64.c | 6 ++++--
>>>  target-arm/translate.c     | 6 ++++--
>>>  2 files changed, 8 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/target-arm/translate-a64.c b/target-arm/translate-a64.c
>>> index 689f2be..0b0f4ae 100644
>>> --- a/target-arm/translate-a64.c
>>> +++ b/target-arm/translate-a64.c
>>> @@ -11026,7 +11026,8 @@ void gen_intermediate_code_internal_a64(ARMCPU *cpu,
>>>              gen_io_start();
>>>          }
>>>  
>>> -        if (unlikely(qemu_loglevel_mask(CPU_LOG_TB_OP | CPU_LOG_TB_OP_OPT))) {
>>> +        if (unlikely(qemu_loglevel_mask(CPU_LOG_TB_OP | CPU_LOG_TB_OP_OPT) &&
>>> +                     qemu_log_in_addr_range(dc->pc))) {
>>>              tcg_gen_debug_insn_start(dc->pc);
>>>          }
>>
>> If there's more than one or two ranges, it's probably quicker to
>> generate the debug opcode regardless of the range.  Remember, this
>> check is happening once per insn, not once per tb.
> 
> Maybe I should hoist the check up to the start of a block? This would
> mean we would dump all instructions in a block even if they went past
> the end-point but the reverse case is probably just confusing.
> 
> We'll still not dump anything that starts outside the range.

Why hoist when the loglevel_mask check is so quick?
Processing of these debug opcodes is equally quick.

It's really only the dumping of the opcodes elsewhere
that needs to check the addr_range.


r~