public inbox for gdb-patches@sourceware.org
 help / color / mirror / Atom feed
From: Alexander Smundak <asmundak@google.com>
To: Tom Tromey <tromey@redhat.com>
Cc: Eli Zaretskii <eliz@gnu.org>, gdb-patches <gdb-patches@sourceware.org>
Subject: Re: [PATCH] Add Frame.read_register to Python API
Date: Wed, 11 Jun 2014 23:52:00 -0000	[thread overview]
Message-ID: <CAHQ51u71A-sBCW_em68nHsFTXmHZCsYudKYtZm8MfC=rTGGAOg@mail.gmail.com> (raw)
In-Reply-To: <87ioo7uuqm.fsf@fleche.redhat.com>

[-- Attachment #1: Type: text/plain, Size: 2814 bytes --]

> Alexander>      def __init__(self, fobj):
> Alexander>          super(InlinedFrameDecorator, self).__init__(fobj)
> Alexander> +        self.fobj = fobj
>
> Alexander>      def function(self):
> Alexander> -        frame = fobj.inferior_frame()
> Alexander> +        frame = self.fobj.inferior_frame()
> Alexander>          name = str(frame.name())
>
> I think this is a nice fix but it seems unrelated to the patch at hand.
>
> Alexander>  @defun Frame.find_sal ()
> Alexander> -Return the frame's symtab and line object.
> Alexander> +Return the frame's @code{gdb.Symtab_and_line} object.
>
> Likewise.

Should I mail these two as a single patch or as two separate patches?

> Alexander> +      FRAPY_REQUIRE_VALID (self, frame);
> Alexander> +      if (!PyArg_ParseTuple (args, "i", &regnum))
> Alexander> +    {
> Alexander> +      const char *regnum_str;
> Alexander> +      PyErr_Clear();  /* Clear PyArg_ParseTuple failure above.  */
> Alexander> +      if (PyArg_ParseTuple (args, "s", &regnum_str))
> Alexander> +        {
> Alexander> +          regnum = user_reg_map_name_to_regnum (get_frame_arch (frame),
> Alexander> +                                                regnum_str,
> Alexander> +                                                strlen (regnum_str));
> Alexander> +        }
> Alexander> +    }
>
> I tend to think this would be clearer if the arguments were only parsed
> once and then explicit type checks were applied to the resulting object.

Did that, and then started doubting whether it is really necessary to read
a register by its (very arch-specific) number. The new version supports
reading the register by the name. Another change is that it now throws
an exception if the name is wrong.

> Alexander> +# On x86-64, PC is register 16.
> Alexander> +gdb_test "python print ('result = %s' % ((f0.architecture().name() != 'i386:x86-64') or f0.read_register('pc') == f0.read_register(16)))" \
> Alexander> +  "True" \
> Alexander> +  "test Frame.read_register(regnum)"
>
> A test that is arch-specific needs to be conditionalized somehow.
IMHO it's borderline arch-specific -- it is runnable on any platform,
although it will not be testing much on any but x86-64. There hasn't
been any arch-specific tests for Python so far, so I am not sure what to do.

Here's the new version (style violations have been addressed, too):

The ability to read registers is needed to use Frame Filter API to
display the frames created by JIT compilers.

gdb/Changelog
2014-06-11  Sasha Smundak  <asmundak@google.com>

* python/py-frame.c (frapy_read_register): New function.

2014-06-11  Sasha Smundak  <asmundak@google.com>

* python.texi (Frames in Python): Add read_register description.

2014-06-11  Sasha Smundak  <asmundak@google.com>

* gdb.python/py-frame.exp: Test Frame.read_register.

[-- Attachment #2: frame-read-register-patch-3.txt --]
[-- Type: text/plain, Size: 4078 bytes --]

diff --git a/gdb/NEWS b/gdb/NEWS
index 1397e8b..3db458b 100644
--- a/gdb/NEWS
+++ b/gdb/NEWS
@@ -3,6 +3,9 @@
 
 *** Changes since GDB 7.7
 
+* Python Scripting
+  You can now access frame registers from Python scripts.
+
 * New command line options
 
 -D data-directory
diff --git a/gdb/doc/python.texi b/gdb/doc/python.texi
index 4688783..3cb6bf8 100644
--- a/gdb/doc/python.texi
+++ b/gdb/doc/python.texi
@@ -3589,6 +3589,13 @@ Return the frame's symtab and line object.
 @xref{Symbol Tables In Python}.
 @end defun
 
+@defun Frame.read_register (register)
+Return the value of @var{register} in this frame.  The @var{register}
+argument must be a string (e.g., @code{'sp'} or @code{'rax'}).
+Returns a @code{Gdb.Value} object.  Throws an exception if @var{register}
+does not exist.
+@end defun
+
 @defun Frame.read_var (variable @r{[}, block@r{]})
 Return the value of @var{variable} in this frame.  If the optional
 argument @var{block} is provided, search for the variable from that
diff --git a/gdb/python/py-frame.c b/gdb/python/py-frame.c
index 77077d3..73bffb1 100644
--- a/gdb/python/py-frame.c
+++ b/gdb/python/py-frame.c
@@ -28,6 +28,7 @@
 #include "python-internal.h"
 #include "symfile.h"
 #include "objfiles.h"
+#include "user-regs.h"
 
 typedef struct {
   PyObject_HEAD
@@ -235,6 +236,39 @@ frapy_pc (PyObject *self, PyObject *args)
   return gdb_py_long_from_ulongest (pc);
 }
 
+/* Implementation of gdb.Frame.read_register (self, register) -> gdb.Value.
+   Returns the value of a register in this frame.  */
+
+static PyObject *
+frapy_read_register (PyObject *self, PyObject *args)
+{
+  struct frame_info *frame;
+  volatile struct gdb_exception except;
+  int regnum = -1;
+  struct value *val = NULL;
+  const char *regnum_str;
+
+  if (!PyArg_ParseTuple (args, "s", &regnum_str))
+    return NULL;
+
+  TRY_CATCH (except, RETURN_MASK_ALL)
+    {
+      FRAPY_REQUIRE_VALID (self, frame);
+
+      regnum = user_reg_map_name_to_regnum (get_frame_arch (frame),
+                                            regnum_str,
+                                            strlen (regnum_str));
+      if (regnum >= 0)
+        val = value_of_register (regnum, frame);
+
+      if (val == NULL)
+        PyErr_SetString (PyExc_ValueError, _("Unknown register."));
+    }
+  GDB_PY_HANDLE_EXCEPTION (except);
+
+  return val == NULL ? NULL : value_to_value_object (val);
+}
+
 /* Implementation of gdb.Frame.block (self) -> gdb.Block.
    Returns the frame's code block.  */
 
@@ -674,6 +708,9 @@ Return the reason why it's not possible to find frames older than this." },
   { "pc", frapy_pc, METH_NOARGS,
     "pc () -> Long.\n\
 Return the frame's resume address." },
+  { "read_register", frapy_read_register, METH_VARARGS,
+    "read_register (register) -> gdb.Value\n\
+Return the value of the register in the frame." },
   { "block", frapy_block, METH_NOARGS,
     "block () -> gdb.Block.\n\
 Return the frame's code block." },
diff --git a/gdb/testsuite/gdb.python/py-frame.exp b/gdb/testsuite/gdb.python/py-frame.exp
index 3517824..9eeebba 100644
--- a/gdb/testsuite/gdb.python/py-frame.exp
+++ b/gdb/testsuite/gdb.python/py-frame.exp
@@ -94,3 +94,16 @@ gdb_test "python print ('result = %s' % f0.read_var ('variable_which_surely_does
 gdb_test "python print ('result = %s' % f0.read_var ('a'))" " = 1" "test Frame.read_var - success"
 
 gdb_test "python print ('result = %s' % (gdb.selected_frame () == f1))" " = True" "test gdb.selected_frame"
+
+# Can read SP register.
+gdb_test "python print ('result = %s' % f0.read_register('sp'))" \
+  " = 0x\[0-9a-fA-F\]+" \
+  "test Frame.read_register(fp)"
+# PC value obtained via read_register is as expected.
+gdb_test "python print ('result = %s' % (f0.read_register('pc').cast(gdb.lookup_type('unsigned long')) == f0.pc()))" \
+  "True" \
+  "test Frame.read_register(pc)"
+# On x86-64, PC is in $rip register.
+gdb_test "python print ('result = %s' % ((f0.architecture().name() != 'i386:x86-64') or f0.read_register('pc') == f0.read_register('rip')))" \
+  "True" \
+  "test Frame.read_register(regnum)"

  reply	other threads:[~2014-06-11 23:52 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-06-09 19:15 Alexander Smundak
2014-06-09 19:26 ` Eli Zaretskii
2014-06-09 21:16   ` Alexander Smundak
2014-06-11 19:11     ` Tom Tromey
2014-06-11 23:52       ` Alexander Smundak [this message]
2014-06-18 17:18         ` Alexander Smundak
2014-06-18 17:33           ` Eli Zaretskii
2014-06-23 22:46           ` Alexander Smundak
2014-06-30 16:22             ` Alexander Smundak
2014-07-07 20:59               ` Alexander Smundak
2014-07-14 16:24                 ` Alexander Smundak
2014-07-24 17:45                   ` Doug Evans
2014-07-31 18:53                     ` Doug Evans
2014-07-31 20:05                       ` Tom Tromey
2014-08-21 18:44         ` Doug Evans
2014-08-26 20:31           ` Alexander Smundak
2014-08-29 13:39             ` Doug Evans
2014-08-29 23:32               ` Sasha Smundak
2014-08-29 23:36                 ` Doug Evans
2014-09-03 23:46                   ` Doug Evans

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='CAHQ51u71A-sBCW_em68nHsFTXmHZCsYudKYtZm8MfC=rTGGAOg@mail.gmail.com' \
    --to=asmundak@google.com \
    --cc=eliz@gnu.org \
    --cc=gdb-patches@sourceware.org \
    --cc=tromey@redhat.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).