On 2015/1/4 13:05, Masami Hiramatsu wrote:
> Hi Wang,
> 
> (2014/12/29 13:07), Wang Nan wrote:
>> This patch utilizes previous introduced checker to check register usage
>> for probed ARM instruction and saves it in a mask. Futher patch will
>> use such information to avoid simuation or emulation.
>>
>> Signed-off-by: Wang Nan <[email protected]>
>> ---
>>  arch/arm/include/asm/probes.h          |  12 ++++
>>  arch/arm/probes/decode.c               |   7 ++
>>  arch/arm/probes/kprobes/actions-arm.c  |   2 +-
>>  arch/arm/probes/kprobes/checkers-arm.c | 124 
>> +++++++++++++++++++++++++++++++++
>>  arch/arm/probes/kprobes/checkers.h     |   1 +
>>  5 files changed, 145 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/arm/include/asm/probes.h b/arch/arm/include/asm/probes.h
>> index f0a1ee8..ee04067 100644
>> --- a/arch/arm/include/asm/probes.h
>> +++ b/arch/arm/include/asm/probes.h
>> @@ -41,6 +41,18 @@ struct arch_probes_insn {
>>      probes_insn_singlestep_t        *insn_singlestep;
>>      probes_insn_fn_t                *insn_fn;
>>      int stack_space;
>> +
>> +    /* Use 2 bits for a register. One more bit for extension */
> 
> Would you have any concrete idea for the extend bits? If not, we don't need
> it at this point. I think we don't need to care about future binary 
> compatibility :)
> (moreover, if you need another bitflag, you can add another flag)
> 

2 bits can describe the read/write direction of a register. With such 
information,
futher code is possible to utilize unused register to do some optimization. 
However,
as you pointed, it is not a very concrete idea.


>> +#define REG_NO_USE  (0)
>> +#define REG_USE             (1)
>> +#define REG_MASK    (3)
>> +#define __register_usage_flag(n, f) ((f) << ((n) * 2))
>> +#define __register_usage_mask(n)    (REG_MASK << ((n) * 2))
>> +#define __clean_register_flag(m, n) ((m) & (~(__register_usage_mask(n))))
>> +#define __set_register_flag(m, n, f)        (__clean_register_flag(m, n) | 
>> __register_usage_flag(n, f))
>> +#define set_register_nouse(m, n)    do {(m) = __set_register_flag(m, n, 
>> REG_NO_USE);} while(0)
>> +#define set_register_use(m, n)              do {(m) = 
>> __set_register_flag(m, n, REG_USE);} while(0)
>> +    int register_usage_mask;
> 
> Is this a mask or flag? It seems a bit flag, if so, it should be 
> "register_usage_flag".
> 
> Thank you,
> 

OK, I'll rename it.

Thanks to your comment.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Reply via email to