Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
From: Antoine Tremblay <antoine.tremblay@ericsson.com>
To: Yao Qi <qiyaoltc@gmail.com>
Cc: <gdb-patches@sourceware.org>
Subject: Re: [PATCH v6 5/7] Support software single step on ARM in GDBServer.
Date: Mon, 07 Dec 2015 15:58:00 -0000	[thread overview]
Message-ID: <5665AC96.1090808@ericsson.com> (raw)
In-Reply-To: <8637ves2xp.fsf@gmail.com>



On 12/07/2015 10:17 AM, Yao Qi wrote:
> Antoine Tremblay <antoine.tremblay@ericsson.com> writes:
>
>> gdb/ChangeLog:
>> 	* Makefile.in: Add arm-get-next-pcs.c/o, arm-linux.c/o.
>> 	to all arm targets.
>
> Please describe how Makefile.in is changed, as I suggested in previous patch.
>
OK.

>> 	* aarch32-linux-nat.c: Include arch/arm-get-next-pcs.h.
>> 	* arch/arm-get-next-pcs.c: New file.
>> 	* arch/arm-get-next-pcs.h: New file.
>> 	* arch/arm-linux.h: New file.
>> 	* arch/arm-linux.c: New file.
>> 	* arm-linux-nat.c: Include arch/arm.h, arch/arm-get-next-pcs.h
>> 	arch/arm-linux.h.
>> 	* arm-linux-tdep.c: Include arch/arm.h,arch/arm-get-next-pcs.h
>> 	arch/arm-linux.h.
>> 	(arm_linux_get_next_pcs_ops): New struct.
>> 	(ARM_SIGCONTEXT_R0, ARM_UCONTEXT_SIGCONTEXT,
>> 	ARM_OLD_RT_SIGFRAME_SIGINFO, ARM_OLD_RT_SIGFRAME_UCONTEXT,
>> 	ARM_NEW_RT_SIGFRAME_UCONTEXT, ARM_NEW_SIGFRAME_MAGIC): Move stack
>> 	layout defines to arch/arm-linux.h.
>> 	(arm_linux_sigreturn_next_pc_offset): Move to arch/arm-linux.c.
>> 	(arm_linux_software_single_step): Adjust for
>> 	arm_get_next_pcs implementation.
>> 	* arm-tdep.c: Include arch/arm-get-next-pcs.h.
>> 	(arm_get_next_pcs_ops): New struct.
>> 	(submask): Move macro to arm-get-next-pcs.h.
>> 	(bit): Likewise.
>> 	(bits): Likewise.
>> 	(sbits): Likewise.
>> 	(BranchDest): Likewise.
>
>> 	(thumb_instruction_changes_pc): Move to arm-get-next-pcs.c
>> 	(thumb2_instruction_changes_pc): Likewise.
>> 	(arm_instruction_changes_pc): Likewise.
>
> I feel they should be moved to arch/arm.c, because they are not about
> getting the next pcs.
>

OK. I went at first with the idea that if they were only used in 
get_next_pc to leave them with it as the criteria since I was not sure 
where to draw the line. But these do feel more generic.

>> 	(shifted_reg_val): Likewise.
>> 	(thumb_advance_itstate): Likewise.
>
> and this one.
>
OK.

>> 	(thumb_get_next_pc_raw): Likewise.
>> 	(arm_get_next_pc_raw): Likewise.
>> 	(arm_get_next_pc): Likewise.
>> 	(thumb_deal_with_atomic_sequence_raw): Likewise.
>> 	(arm_deal_with_atomic_sequence_raw): Likewise.
>> 	(arm_deal_with_atomic_sequence): Likewise.
>> 	(arm_get_next_pcs_read_memory_unsigned_integer): New function.
>> 	(arm_get_next_pcs_addr_bits_remove): Likewise.
>> 	(arm_get_next_pcs_syscall_next_pc): Likewise.
>> 	(arm_get_next_pcs_is_thumb): Likewise.
>> 	(arm_software_single_step): Adjust for arm_get_next_pcs
>> 	implementation.
>> 	* arm-tdep.h: (arm_get_next_pc): Remove declaration.
>> 	(arm_get_next_pcs_read_memory_unsigned_integer):
>> 	New declaration.
>> 	(arm_get_next_pcs_addr_bits_remove): Likewise.
>> 	(arm_get_next_pcs_syscall_next_pc): Likewise.
>> 	(arm_software_single_step): Likewise.
>> 	(arm_deal_with_atomic_sequence: Remove declaration.
>> 	(arm_get_next_pcs_is_thumb): New delcaration.
>> 	* arm-wince-tdep.c: Include arch/arm-get-next-pcs.h.
>> 	* armbsd-tdep.c: Likewise.
>> 	* armnbsd-tdep.c: Likewise.
>> 	* armobsd-tdep.c: Likewise.
>
> Why do we need to include arm-get-next-pcs.h in these *-tdep.c files?
>

We don't now, some defines were needed before, I forgot to clean that up 
sorry.

>> 	* common/gdb_vecs.h: Add CORE_ADDR vector definition.
>> 	* configure.tgt: Add arm-get-next-pcs.o, arm-linux.o to all
>> 	relevent arm targets.
>> 	* symtab.h: Move CORE_ADDR vector definition to gdb_vecs.h.
>>
>> gdb/gdbserver/ChangeLog:
>> 	* Makefile.in : Add arm-get-next-pcs.c/o, arm-linux.c/o.
>
> Please describe what are changed.
>

OK.

>> 	* configure.srv: Add arm-get-next-pcs.o, arm-linux.o
>> 	to needed arm targets.
>> 	* linux-arm-low.c: Include arch/arm-linux.h,
>> 	aarch/arm-get-next-pcs.h.
>> 	(get_next_pcs_ops): New struct.
>> 	(get_next_pcs_addr_bits_remove): New function.
>> 	(get_next_pcs_is_thumb): New function.
>> 	(get_next_pcs_read_memory_unsigned_integer): Likewise.
>> 	(arm_sigreturn_next_pc): Likewise.
>> 	(get_next_pcs_syscall_next_pc): Likewise.
>> 	(arm_gdbserver_get_next_pcs): Likewise.
>> 	(struct linux_target_ops) <arm_gdbserver_get_next_pcs>:
>> 	Initialize.
>> 	* linux-low.h: Move CORE_ADDR vector definition to gdb_vecs.h.
>> 	* server.h: Include gdb_vecs.h.
>> ---
>>   gdb/Makefile.in               |   16 +-
>>   gdb/aarch32-linux-nat.c       |    1 +
>>   gdb/arch/arm-get-next-pcs.c   | 1216 +++++++++++++++++++++++++++++++++
>>   gdb/arch/arm-get-next-pcs.h   |   95 +++
>>   gdb/arch/arm-linux.c          |   59 ++
>>   gdb/arch/arm-linux.h          |   73 ++
>>   gdb/arm-linux-nat.c           |    2 +
>>   gdb/arm-linux-tdep.c          |  136 ++--
>>   gdb/arm-tdep.c                | 1495 +++++------------------------------------
>>   gdb/arm-tdep.h                |   18 +-
>>   gdb/arm-wince-tdep.c          |    1 +
>>   gdb/armbsd-tdep.c             |    1 +
>>   gdb/armnbsd-tdep.c            |    1 +
>>   gdb/armobsd-tdep.c            |    1 +
>>   gdb/common/gdb_vecs.h         |    2 +
>>   gdb/configure.tgt             |   20 +-
>>   gdb/gdbserver/Makefile.in     |    9 +-
>>   gdb/gdbserver/configure.srv   |    2 +
>>   gdb/gdbserver/linux-arm-low.c |  144 +++-
>>   gdb/gdbserver/linux-low.h     |    2 -
>>   gdb/gdbserver/server.h        |    1 +
>>   gdb/symtab.h                  |    2 -
>>   22 files changed, 1871 insertions(+), 1426 deletions(-)
>>   create mode 100644 gdb/arch/arm-get-next-pcs.c
>>   create mode 100644 gdb/arch/arm-get-next-pcs.h
>>   create mode 100644 gdb/arch/arm-linux.c
>>   create mode 100644 gdb/arch/arm-linux.h
>>
>
>
>
>> +
>> +/* Checks for an atomic sequence of instructions beginning with a LDREX{,B,H,D}
>> +   instruction and ending with a STREX{,B,H,D} instruction.  If such a sequence
>> +   is found, attempt to step through it.  The end of the sequence address is
>> +   added to the next_pcs list.  */
>> +
>> +static int
>> +thumb_deal_with_atomic_sequence_raw (struct arm_get_next_pcs *self,
>> +				     CORE_ADDR pc,
>> +				     VEC (CORE_ADDR) **next_pcs)
>
> We can return VEC (CORE_ADDR) * here, if it is NULL, the sequence is not
> found, otherwise, the end of the sequence addresses are returned in the vector.
>

OK.

>> +
>> +/* Checks for an atomic sequence of instructions beginning with a LDREX{,B,H,D}
>> +   instruction and ending with a STREX{,B,H,D} instruction.  If such a sequence
>> +   is found, attempt to step through it.  The end of the sequence address is
>> +   added to the next_pcs list.  */
>> +
>> +static int
>> +arm_deal_with_atomic_sequence_raw (struct arm_get_next_pcs *self,
>> +				   CORE_ADDR pc,
>> +				   VEC (CORE_ADDR) **next_pcs)
>
> Likewise.
>
>> +
>> +/* See arm-get-next-pcs.h.  */
>> +
>> +VEC (CORE_ADDR) *
>> +arm_get_next_pcs (struct arm_get_next_pcs *self, CORE_ADDR pc)
>> +{
>> +  VEC (CORE_ADDR) *next_pcs = NULL;
>> +
>> +  if (self->ops->is_thumb (self))
>> +    {
>> +      if (!thumb_deal_with_atomic_sequence_raw (self, pc, &next_pcs))
>> +	return thumb_get_next_pcs_raw (self, pc, &next_pcs);
>
> Why do we need to pass next_pcs to thumb_get_next_pcs_raw?  Isn't
> simpler to write it like this,
>
>          next_pcs = thumb_deal_with_atomic_sequence_raw (self, pc);
>          if (next_pcs == NULL)
>            next_pcs = thumb_get_next_pcs_raw (self, pc);
>

Indeed, I was focused on not losing the VEC allocation/resize and forgot 
  thumb_deal_with_atomic_sequence_raw and thumb_get_next_pcs_raw  will 
not combine their vectors.

Thanks.

>> +    }
>> +  else
>> +    {
>> +      if (!arm_deal_with_atomic_sequence_raw (self, pc, &next_pcs))
>> +	return arm_get_next_pcs_raw (self, pc, &next_pcs);
>> +    }
>> +
>> +  return next_pcs;
>> +}
>
>> diff --git a/gdb/arch/arm-get-next-pcs.h b/gdb/arch/arm-get-next-pcs.h
>> new file mode 100644
>> index 0000000..39921e0
>> --- /dev/null
>> +++ b/gdb/arch/arm-get-next-pcs.h
>> @@ -0,0 +1,95 @@
>> +/* Common code for ARM software single stepping support.
>> +
>> +   Copyright (C) 1988-2015 Free Software Foundation, Inc.
>> +
>> +   This file is part of GDB.
>> +
>> +   This program is free software; you can redistribute it and/or modify
>> +   it under the terms of the GNU General Public License as published by
>> +   the Free Software Foundation; either version 3 of the License, or
>> +   (at your option) any later version.
>> +
>> +   This program is distributed in the hope that it will be useful,
>> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
>> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> +   GNU General Public License for more details.
>> +
>> +   You should have received a copy of the GNU General Public License
>> +   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
>> +
>> +#ifndef ARM_GET_NEXT_PCS_H
>> +#define ARM_GET_NEXT_PCS_H 1
>> +
>> +/* Support routines for instruction parsing.  */
>> +#define submask(x) ((1L << ((x) + 1)) - 1)
>> +#define bits(obj,st,fn) (((obj) >> (st)) & submask ((fn) - (st)))
>> +#define bit(obj,st) (((obj) >> (st)) & 1)
>> +#define sbits(obj,st,fn) \
>> +  ((long) (bits(obj,st,fn) | ((long) bit(obj,fn) * ~ submask (fn - st))))
>> +#define BranchDest(addr,instr) \
>> +  ((CORE_ADDR) (((unsigned long) (addr)) + 8 + (sbits (instr, 0, 23) << 2)))
>> +
>
> These macros should be moved to arch/arm.h.
>

OK.


      reply	other threads:[~2015-12-07 15:58 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-12-04 18:46 [PATCH v6 0/6] Support software single step and conditional breakpoints " Antoine Tremblay
2015-12-04 18:46 ` [PATCH v6 1/7] Replace breakpoint_reinsert_addr by get_next_pcs operation " Antoine Tremblay
2015-12-04 18:46 ` [PATCH v6 4/7] Share regcache function regcache_raw_read_unsigned Antoine Tremblay
2015-12-07 14:40   ` Yao Qi
2015-12-07 14:49     ` Antoine Tremblay
2015-12-04 18:46 ` [PATCH v6 2/7] Share some ARM target dependent code from GDB with GDBServer Antoine Tremblay
2015-12-04 18:46 ` [PATCH v6 3/7] Refactor arm_software_single_step to use regcache Antoine Tremblay
2015-12-07 14:32   ` Yao Qi
2015-12-07 14:48     ` Antoine Tremblay
2015-12-07 15:21       ` Yao Qi
2015-12-07 17:37         ` Antoine Tremblay
2015-12-07 14:51   ` Yao Qi
2015-12-07 14:51     ` Antoine Tremblay
2015-12-04 18:47 ` [PATCH v6 6/7] Enable software single stepping for while-stepping actions in GDBServer Antoine Tremblay
2015-12-04 18:47 ` [PATCH v6 7/7] Enable conditional breakpoints for targets that support software single step " Antoine Tremblay
2015-12-04 18:47 ` [PATCH v6 5/7] Support software single step on ARM " Antoine Tremblay
2015-12-07 15:18   ` Yao Qi
2015-12-07 15:58     ` Antoine Tremblay [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=5665AC96.1090808@ericsson.com \
    --to=antoine.tremblay@ericsson.com \
    --cc=gdb-patches@sourceware.org \
    --cc=qiyaoltc@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox