public inbox for gdb-patches@sourceware.org
 help / color / mirror / Atom feed
From: Alan Hayward <Alan.Hayward@arm.com>
To: Simon Marchi <simon.marchi@ericsson.com>
Cc: "gdb-patches@sourceware.org" <gdb-patches@sourceware.org>,
	nd <nd@arm.com>
Subject: Re: [PATCH v2 04/10] Add regcache raw_compare method
Date: Fri, 08 Jun 2018 15:16:00 -0000	[thread overview]
Message-ID: <B2E1E96D-A235-4CE2-9CF9-217FBE23C50A@arm.com> (raw)
In-Reply-To: <ada727b3-a7d5-bc04-afd8-4b32f8ec67ac@ericsson.com>



> On 7 Jun 2018, at 21:55, Simon Marchi <simon.marchi@ericsson.com> wrote:
> 
> On 2018-06-06 11:16 AM, Alan Hayward wrote:
>> raw_compare returns the same as a memcmp (0 for success,
>> the diff otherwise). Kept with tristate return as this
>> feels potentally more useful than a simple true/false return.
>> (Happy to change if not).
> 
> I would err on the bool side, but I don't really mind, the important is
> that it's documented properly.  We still use many ints as bool in GDB, so
> someone seeing an int return value could easily think it's 0 -> not equal,
> 1 -> equal.


Ok, updated.

> 
> Speaking of doc, I would suggest (as in the previous patch) to centralize
> the doc in the class/struct at the root, so we don't duplicate it.

Done.

> 
>> 2018-06-06  Alan Hayward  <alan.hayward@arm.com>
>> 
>> gdb/
>> 	* common/common-regcache.h (raw_compare): New function.
>> 	* regcache.c (regcache::raw_compare): Likewise.
>> 	* regcache.h (regcache::raw_compare): New declaration.
>> 
>> gdbserver/
>> 	* regcache.c (regcache::raw_compare): New function.
>> 	* regcache.h (regcache::raw_compare): New declaration.
>> ---
>> gdb/common/common-regcache.h |  1 +
>> gdb/gdbserver/regcache.c     | 10 ++++++++++
>> gdb/gdbserver/regcache.h     |  5 +++++
>> gdb/regcache.c               | 15 +++++++++++++++
>> gdb/regcache.h               | 11 +++++++++++
>> 5 files changed, 42 insertions(+)
>> 
>> diff --git a/gdb/common/common-regcache.h b/gdb/common/common-regcache.h
>> index 487da0a7fb..e91439bec5 100644
>> --- a/gdb/common/common-regcache.h
>> +++ b/gdb/common/common-regcache.h
>> @@ -67,6 +67,7 @@ struct reg_buffer_common
>>   virtual ~reg_buffer_common () = default;
>>   virtual void raw_supply (int regnum, const void *buf) = 0;
>>   virtual void raw_collect (int regnum, void *buf) const = 0;
>> +  virtual int raw_compare (int regnum, const void *buf, int offset) const = 0;
>>   virtual register_status get_register_status (int regnum) const = 0;
>> };
>> 
>> diff --git a/gdb/gdbserver/regcache.c b/gdb/gdbserver/regcache.c
>> index 857721ee3d..9648428349 100644
>> --- a/gdb/gdbserver/regcache.c
>> +++ b/gdb/gdbserver/regcache.c
>> @@ -501,3 +501,13 @@ regcache::get_register_status (int regnum) const
>>   return REG_VALID;
>> #endif
>> }
>> +
>> +/* See gdbserver/regcache.h.  */
>> +
>> +int
>> +regcache::raw_compare (int regnum, const void *buf, int offset) const
>> +{
>> +  gdb_assert (register_size (tdesc, regnum) > offset);
> 
> Should we check ">= offset"?  I think it could be useful for some edge cases
> where we could avoid an "if (offset < size)" in the caller, if we know offset
> could be equal to size.  memcmp would return 0 (equal), which is fine.
> 

Given that memcmp can cope with this, then ok.
(I need to make sure I update the patch that uses this function too!)

> Maybe assign "register_size (tdesc, regnum)" to a variable (e.g. reg_size)
> and use it in both places?

Done.

> 
>> +  return memcmp (buf, register_data (this, regnum, 1) + offset,
>> +		 register_size (tdesc, regnum) - offset);
>> +}
>> diff --git a/gdb/gdbserver/regcache.h b/gdb/gdbserver/regcache.h
>> index 1842f1d9cf..b26f39a8ad 100644
>> --- a/gdb/gdbserver/regcache.h
>> +++ b/gdb/gdbserver/regcache.h
>> @@ -50,6 +50,11 @@ struct regcache : public reg_buffer_common
>> 
>>   void raw_collect (int regnum, void *buf) const override;
>> 
>> +  /* Compare the contents of the register stored in the regcache (ignoring the
>> +     first OFFSET bytes) to the contents of BUF (without any offset). Returns
>> +     the same result as memcmp.  */
>> +  int raw_compare (int regnum, const void *buf, int offset) const override;
>> +
>>   enum register_status get_register_status (int regnum) const override;
>> };
>> 
>> diff --git a/gdb/regcache.c b/gdb/regcache.c
>> index a914b548ca..383e355e9f 100644
>> --- a/gdb/regcache.c
>> +++ b/gdb/regcache.c
>> @@ -1082,6 +1082,21 @@ regcache::collect_regset (const struct regset *regset,
>>   transfer_regset (regset, NULL, regnum, NULL, buf, size);
>> }
>> 
>> +/* See regcache.h.  */
>> +
>> +int
>> +regcache::raw_compare (int regnum, const void *buf, int offset) const
>> +{
>> +  const char *regbuf;
>> +  size_t size;
>> +
>> +  gdb_assert (buf != NULL);
>> +  assert_regnum (regnum);
> 
> Should we assert that offset is < or <= size here too?
> 

Not sure why I didn’t add this.


Patch update with changes above. Are you ok with this version?


diff --git a/gdb/common/common-regcache.h b/gdb/common/common-regcache.h
index 29d9a81182ad1a5797b080e136b682fe59ecef37..fe3ece7ac52d60d07d718ad617f30be2f7133b5f 100644
--- a/gdb/common/common-regcache.h
+++ b/gdb/common/common-regcache.h
@@ -75,6 +75,11 @@ struct reg_buffer_common

   /* Collect register REGNUM from REGCACHE and store its contents in BUF.  */
   virtual void raw_collect (int regnum, void *buf) const = 0;
+
+  /* Compare the contents of the register stored in the regcache (ignoring the
+     first OFFSET bytes) to the contents of BUF (without any offset).  Returns
+     True if the same.  */
+  virtual bool raw_compare (int regnum, const void *buf, int offset) const = 0;
 };

 #endif /* COMMON_REGCACHE_H */
diff --git a/gdb/gdbserver/regcache.h b/gdb/gdbserver/regcache.h
index 352c1df3f9eeacd622a3d5d668ff4ea98aea9c6f..b4c4c20ebd368f56f14af758301ff7d55fd16dae 100644
--- a/gdb/gdbserver/regcache.h
+++ b/gdb/gdbserver/regcache.h
@@ -54,6 +54,9 @@ struct regcache : public reg_buffer_common

   /* See common/common-regcache.h.  */
   void raw_collect (int regnum, void *buf) const override;
+
+  /* See common/common-regcache.h.  */
+  bool raw_compare (int regnum, const void *buf, int offset) const override;
 };

 struct regcache *init_register_cache (struct regcache *regcache,
diff --git a/gdb/gdbserver/regcache.c b/gdb/gdbserver/regcache.c
index 83837525a18c1dc96b6e9d24397064e247ced459..735ce7bccfc7a058f8604620928f3894d4655887 100644
--- a/gdb/gdbserver/regcache.c
+++ b/gdb/gdbserver/regcache.c
@@ -502,3 +502,17 @@ regcache::get_register_status (int regnum) const
   return REG_VALID;
 #endif
 }
+
+/* See common/common-regcache.h.  */
+
+bool
+regcache::raw_compare (int regnum, const void *buf, int offset) const
+{
+  gdb_assert (buf != NULL);
+
+  const unsigned char *regbuf = register_data (this, regnum, 1);
+  int size = register_size (tdesc, regnum);
+  gdb_assert (size >= offset);
+
+  return (memcmp (buf, regbuf + offset, size - offset) == 0);
+}
diff --git a/gdb/regcache.h b/gdb/regcache.h
index 4e5659b25bc530a4ced32520825136534be2e642..511864858d24a7fa53047fdc247846a551b121af 100644
--- a/gdb/regcache.h
+++ b/gdb/regcache.h
@@ -191,6 +191,10 @@ public:
     xfree (m_registers);
     xfree (m_register_status);
   }
+
+  /* See common/common-regcache.h.  */
+  bool raw_compare (int regnum, const void *buf, int offset) const override;
+
 protected:
   /* Assert on the range of REGNUM.  */
   void assert_regnum (int regnum) const;
diff --git a/gdb/regcache.c b/gdb/regcache.c
index 032fef0d34ac635c96dd6c39426f5c6d04f28095..fd0fd7318c8cca0083c897d2bd6ba1361dd627c0 100644
--- a/gdb/regcache.c
+++ b/gdb/regcache.c
@@ -1078,6 +1078,20 @@ regcache::collect_regset (const struct regset *regset,
   transfer_regset (regset, NULL, regnum, NULL, buf, size);
 }

+/* See common/common-regcache.h.  */
+
+bool
+reg_buffer::raw_compare (int regnum, const void *buf, int offset) const
+{
+  gdb_assert (buf != NULL);
+  assert_regnum (regnum);
+
+  const char *regbuf = (const char *) register_buffer (regnum);
+  size_t size = m_descr->sizeof_register[regnum];
+  gdb_assert (size >= offset);
+
+  return (memcmp (buf, regbuf + offset, size - offset) == 0);
+}

 /* Special handling for register PC.  */



  reply	other threads:[~2018-06-08 15:16 UTC|newest]

Thread overview: 56+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-06-06 15:16 [PATCH v2 00/10] gdb/gdbserver support for aarch64 SVE Alan Hayward
2018-06-06 15:17 ` [PATCH v2 09/10] Ptrace support for AArch64 SVE gdbsever Alan Hayward
2018-06-11  2:43   ` Simon Marchi
2018-06-11  2:44   ` Simon Marchi
2018-06-06 15:17 ` [PATCH v2 02/10] Add Aarch64 SVE Linux headers Alan Hayward
2018-06-08 14:13   ` Alan Hayward
2018-06-08 14:37     ` Simon Marchi
2018-06-08 15:23       ` Simon Marchi
2018-06-12 14:37         ` Alan Hayward
2018-06-12 14:43           ` Pedro Alves
2018-06-12 15:06             ` Simon Marchi
2018-06-12 15:11               ` Pedro Alves
2018-06-12 15:21                 ` Simon Marchi
2018-06-12 15:09             ` Alan Hayward
2018-06-12 14:51           ` Simon Marchi
2018-06-12 16:34             ` Sergio Durigan Junior
2018-06-12 17:51               ` Alan Hayward
2018-06-12 20:29                 ` Sergio Durigan Junior
2018-06-15  9:45                   ` Ramana Radhakrishnan
2018-06-15 17:14                     ` Alan Hayward
2018-09-20 21:16                       ` Status of the AArch* builders (was: Re: [PATCH v2 02/10] Add Aarch64 SVE Linux headers) Sergio Durigan Junior
2018-09-24 14:16                         ` Alan Hayward
2018-09-24 14:42                           ` Status of the AArch* builders Sergio Durigan Junior
2018-10-11  9:23                             ` Alan Hayward
2018-10-12 19:06                               ` Sergio Durigan Junior
2018-10-15 10:16                                 ` Alan Hayward
2018-10-15 12:42                                   ` Sergio Durigan Junior
2018-10-15 14:02                                     ` Alan Hayward
2018-10-15 15:32                                       ` Sergio Durigan Junior
2018-10-17 18:46                                         ` Sergio Durigan Junior
2018-10-24  9:56                                           ` Alan Hayward
2018-10-25 16:26                                             ` Sergio Durigan Junior
2018-06-08 15:27       ` [PATCH v2 02/10] Add Aarch64 SVE Linux headers Alan Hayward
2018-06-06 15:17 ` [PATCH v2 01/10] Aarch64 SVE pseudo register support Alan Hayward
2018-06-06 22:17   ` Simon Marchi
2018-06-07  9:34     ` Alan Hayward
2018-06-06 15:17 ` [PATCH v2 07/10] Increase gdbsever PBUFSIZ Alan Hayward
2018-06-11  0:46   ` Simon Marchi
2018-06-06 15:17 ` [PATCH v2 08/10] Enable Aarch64 SVE for gdbserver Alan Hayward
2018-06-11  0:49   ` Simon Marchi
2018-06-06 15:17 ` [PATCH v2 10/10] Remove reg2 section from Aarch64 SVE cores Alan Hayward
2018-06-11  2:47   ` Simon Marchi
2018-06-11 16:37     ` Alan Hayward
2018-06-06 15:17 ` [PATCH v2 05/10] Ptrace support for Aarch64 SVE Alan Hayward
2018-06-10 22:52   ` Simon Marchi
2018-06-06 15:17 ` [PATCH v2 06/10] Add Aarch64 SVE dwarf regnums Alan Hayward
2018-06-11  0:43   ` Simon Marchi
2018-06-06 15:17 ` [PATCH v2 03/10] Add reg_buffer_common Alan Hayward
2018-06-07 20:19   ` Simon Marchi
2018-06-07 20:42     ` Simon Marchi
2018-06-08 14:14     ` Alan Hayward
2018-06-10 22:21       ` Simon Marchi
2018-06-06 15:17 ` [PATCH v2 04/10] Add regcache raw_compare method Alan Hayward
2018-06-07 20:56   ` Simon Marchi
2018-06-08 15:16     ` Alan Hayward [this message]
2018-06-10 22:26       ` Simon Marchi

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=B2E1E96D-A235-4CE2-9CF9-217FBE23C50A@arm.com \
    --to=alan.hayward@arm.com \
    --cc=gdb-patches@sourceware.org \
    --cc=nd@arm.com \
    --cc=simon.marchi@ericsson.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;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).