Re: [PATCH] arm reversible : <phase_2_complete>

[oza Pawandeep <oza dot pawandeep at gmail dot com>]

Hi Yao,

Thanks for your comments. please find my comments below.

1)
Yao:  x86 assembly will not compiled and run for arm target, so they don't
> make fails here.  Beside these fails, I find there are some GDB internal
> errors in test result

Oza: what I meant here was: x86 assembly might have failed to compile
even and gdb test results might be including that.
anyway, arm assembly has to be written separately, which I guess
Chandra K. might be doing.

2)
Yao:
Looks like most of indentaiton of your 2nd line of comment is incorrect.
 I'll point some of them out, but I may miss some.

I suggest that you can find some good editor which can highlight
trailing spaces and too-long line easily.

Oza: this is one thing I have been struggling with, I have been moving
this patch back and forth from windows to linux and using
source insight, vi, xemacs
an d finally when I paste it to google mail the space gets distorted.
I need to find out what I can do.
sorry about repeating space mistakes.

3) Yes, only 16-bit thumb is supported, but you should report a warning or
error when encounter a 32-bit thumb insn, and skip it.  In
arm_process_record, you assume that all thumb insns are 16-bit, which is
not correct.  Please reference "thumb_process_displaced_insn".

Oza: I will include this logic.

4) Yao:
I don't understand why coproc insn is not handled for process record
here.  is it in phase_3?

Oza: initially I had planned to put both coprocessor and
arm_extension_space insn in phase 3.
but I am going to update this patch with arm__extension_space insn atleast.

as far as coprocessor insns are concerned, I could not find any API
which could be used to read coprocessor register values.
I had sent email regarding the same long back: but I did not get
response, so I moved it to phase 3.

If you could point me out the way to read coprocessor register values:
I will try to implement on phase 2.

Thanks Yao,
Regards,
Oza.





On Mon, Oct 17, 2011 at 7:25 AM, Yao Qi <yao@codesourcery.com> wrote:
> On 10/16/2011 12:34 AM, oza Pawandeep wrote:
>> Hi Yao,
>>
>> first of all thank you for your comments, will be sending the patch
>> soon with your comments incorporated as much as possible.
>> thank you again for sending test results;
>>
>> I suppose failed test case might be including
>>
>> 1) system call support
>> 2) signal support
>> 3) any other linux ABI support
>> 4) there are some programs on x86 assembly which needs to be written
>> for ARM and put separately.
>
> x86 assembly will not compiled and run for arm target, so they don't
> make fails here. ÂBeside these fails, I find there are some GDB internal
> errors in test result:
>
> stepi^M
> ../../gdb/gdb/breakpoint.c:12523: internal-error:
> insert_single_step_breakpoint: Assertion `single_step_breakpoints[1] ==
> NULL' failed.
>
> stepi^M
> ../../gdb/gdb/infrun.c:1804: internal-error: resume: Assertion
> `!(singlestep_breakpoints_inserted_p && step)' failed.
>
> --
> Yao (éå)
>