* [PATCH] Unary plus
@ 2005-02-09 13:37 Nathan Sidwell
2005-02-27 4:22 ` Daniel Jacobowitz
0 siblings, 1 reply; 4+ messages in thread
From: Nathan Sidwell @ 2005-02-09 13:37 UTC (permalink / raw)
To: gdb-patches
[-- Attachment #1: Type: text/plain, Size: 448 bytes --]
Hi,
gdb doesn't parse or evaluate the unary plus operator (but does have one).
gdb.trace/collection.exp contains such an expression -- 'a[+b]'. This patch
adds the necessary parsing and evaluation machinery.
built & tested on i686-px-linux-gnu and an unreleased architecture. ok?
nathan
--
Nathan Sidwell :: http://www.codesourcery.com :: CodeSourcery LLC
nathan@codesourcery.com :: http://www.planetfall.pwp.blueyonder.co.uk
[-- Attachment #2: unop_plus.patch --]
[-- Type: text/plain, Size: 4363 bytes --]
2005-02-09 Nathan Sidwell <nathan@codesourcery.com>
* ax-gdb.c (gen_expr): Add UNOP_PLUS case.
* c-exp.y (exp): Add unary plus.
* eval.c (evaluate_subexp_standard): Add UNOP_PLUS case.
* valarith.c (value_pos): New.
* value.h (value_pos): Declare.
Index: ax-gdb.c
===================================================================
RCS file: /cvs/src/src/gdb/ax-gdb.c,v
retrieving revision 1.27
diff -c -3 -p -r1.27 ax-gdb.c
*** ax-gdb.c 29 Jan 2005 17:53:25 -0000 1.27
--- ax-gdb.c 9 Feb 2005 09:55:10 -0000
*************** gen_expr (union exp_element **pc, struct
*** 1642,1647 ****
--- 1642,1654 ----
}
break;
+ case UNOP_PLUS:
+ (*pc)++;
+ /* + FOO is equivalent to 0 + FOO, which can be optimized. */
+ gen_expr (pc, ax, value);
+ gen_usual_unary (ax, value);
+ break;
+
case UNOP_NEG:
(*pc)++;
/* -FOO is equivalent to 0 - FOO. */
Index: c-exp.y
===================================================================
RCS file: /cvs/src/src/gdb/c-exp.y,v
retrieving revision 1.28
diff -c -3 -p -r1.28 c-exp.y
*** c-exp.y 7 Aug 2004 19:45:45 -0000 1.28
--- c-exp.y 9 Feb 2005 09:55:11 -0000
*************** exp : '-' exp %prec UNARY
*** 256,261 ****
--- 256,265 ----
{ write_exp_elt_opcode (UNOP_NEG); }
;
+ exp : '+' exp %prec UNARY
+ { write_exp_elt_opcode (UNOP_PLUS); }
+ ;
+
exp : '!' exp %prec UNARY
{ write_exp_elt_opcode (UNOP_LOGICAL_NOT); }
;
Index: eval.c
===================================================================
RCS file: /cvs/src/src/gdb/eval.c,v
retrieving revision 1.49
diff -c -3 -p -r1.49 eval.c
*** eval.c 7 Feb 2005 00:09:53 -0000 1.49
--- eval.c 9 Feb 2005 09:55:16 -0000
*************** evaluate_subexp_standard (struct type *e
*** 1854,1859 ****
--- 1854,1868 ----
evaluate_subexp (NULL_TYPE, exp, pos, noside);
return evaluate_subexp (NULL_TYPE, exp, pos, noside);
+ case UNOP_PLUS:
+ arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
+ if (noside == EVAL_SKIP)
+ goto nosideret;
+ if (unop_user_defined_p (op, arg1))
+ return value_x_unop (arg1, op, noside);
+ else
+ return value_pos (arg1);
+
case UNOP_NEG:
arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
if (noside == EVAL_SKIP)
Index: valarith.c
===================================================================
RCS file: /cvs/src/src/gdb/valarith.c,v
retrieving revision 1.35
diff -c -3 -p -r1.35 valarith.c
*** valarith.c 7 Feb 2005 15:04:43 -0000 1.35
--- valarith.c 9 Feb 2005 09:55:25 -0000
*************** value_less (struct value *arg1, struct v
*** 1313,1319 ****
}
}
\f
! /* The unary operators - and ~. Both free the argument ARG1. */
struct value *
value_neg (struct value *arg1)
--- 1313,1349 ----
}
}
\f
! /* The unary operators +, - and ~. They free the argument ARG1
! (unless it is returned). */
!
! struct value *
! value_pos (struct value *arg1)
! {
! struct type *type;
!
! arg1 = coerce_ref (arg1);
!
! type = check_typedef (value_type (arg1));
!
! if (TYPE_CODE (type) == TYPE_CODE_FLT)
! return arg1;
! else if (is_integral_type (type))
! {
! /* Perform integral promotion for ANSI C/C++. FIXME: What about
! FORTRAN and (the deleted) chill ? */
! if (TYPE_LENGTH (type) >= TYPE_LENGTH (builtin_type_int))
! return arg1;
!
! type = builtin_type_int;
!
! return value_from_longest (type, value_as_long (arg1));
! }
! else
! {
! error ("Argument to positive operation not a number.");
! return 0; /* For lint -- never reached */
! }
! }
struct value *
value_neg (struct value *arg1)
Index: value.h
===================================================================
RCS file: /cvs/src/src/gdb/value.h,v
retrieving revision 1.72
diff -c -3 -p -r1.72 value.h
*** value.h 7 Feb 2005 15:04:43 -0000 1.72
--- value.h 9 Feb 2005 09:55:26 -0000
*************** extern struct value *value_addr (struct
*** 332,337 ****
--- 332,339 ----
extern struct value *value_assign (struct value *toval,
struct value *fromval);
+ extern struct value *value_pos (struct value *arg1);
+
extern struct value *value_neg (struct value *arg1);
extern struct value *value_complement (struct value *arg1);
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] Unary plus 2005-02-09 13:37 [PATCH] Unary plus Nathan Sidwell @ 2005-02-27 4:22 ` Daniel Jacobowitz 2005-03-08 10:51 ` Nathan Sidwell 0 siblings, 1 reply; 4+ messages in thread From: Daniel Jacobowitz @ 2005-02-27 4:22 UTC (permalink / raw) To: Nathan Sidwell; +Cc: gdb-patches On Wed, Feb 09, 2005 at 10:16:17AM +0000, Nathan Sidwell wrote: > Hi, > gdb doesn't parse or evaluate the unary plus operator (but does have one). > gdb.trace/collection.exp contains such an expression -- 'a[+b]'. This patch > adds the necessary parsing and evaluation machinery. > > built & tested on i686-px-linux-gnu and an unreleased architecture. ok? Two comments: - Could you add a non-tracepoint testcase for this? I think one good place would be gdb.cp/userdef.exp. > ! /* The unary operators +, - and ~. They free the argument ARG1 > ! (unless it is returned). */ > ! > ! struct value * > ! value_pos (struct value *arg1) > ! { > ! struct type *type; > ! > ! arg1 = coerce_ref (arg1); > ! > ! type = check_typedef (value_type (arg1)); > ! > ! if (TYPE_CODE (type) == TYPE_CODE_FLT) > ! return arg1; - IIUC, that's not quite right - we need an rvalue here. -- Daniel Jacobowitz CodeSourcery, LLC ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] Unary plus 2005-02-27 4:22 ` Daniel Jacobowitz @ 2005-03-08 10:51 ` Nathan Sidwell 2005-03-08 13:40 ` Daniel Jacobowitz 0 siblings, 1 reply; 4+ messages in thread From: Nathan Sidwell @ 2005-03-08 10:51 UTC (permalink / raw) To: Daniel Jacobowitz; +Cc: gdb-patches [-- Attachment #1: Type: text/plain, Size: 595 bytes --] Daniel Jacobowitz wrote: > Two comments: > > - Could you add a non-tracepoint testcase for this? I think one good > place would be gdb.cp/userdef.exp. done. As this overloads operator+, I introduced a new class for the breakpoint test. >>! if (TYPE_CODE (type) == TYPE_CODE_FLT) >>! return arg1; > > > - IIUC, that's not quite right - we need an rvalue here. Like this? built and tested on i686-pc-linux-gnu, ok? nathan -- Nathan Sidwell :: http://www.codesourcery.com :: CodeSourcery LLC nathan@codesourcery.com :: http://www.planetfall.pwp.blueyonder.co.uk [-- Attachment #2: unop_plus-2.patch --] [-- Type: text/plain, Size: 7600 bytes --] 2005-03-08 Nathan Sidwell <nathan@codesourcery.com> * ax-gdb.c (gen_expr): Add UNOP_PLUS case. * c-exp.y (exp): Add unary plus. * eval.c (evaluate_subexp_standard): Add UNOP_PLUS case. * valarith.c (value_x_unop): Add UNOP_PLUS case. (value_pos): New. * value.h (value_pos): Declare. * gdb.cp/userdef.cc (A1::operator+): New unary plus. (A2): New class. (main): Test operator+. * gdb.cp/userdef.exp: Test unary plus. Use A2::operator+ for breakpoint test. Index: ax-gdb.c =================================================================== RCS file: /cvs/src/src/gdb/ax-gdb.c,v retrieving revision 1.28 diff -c -3 -p -r1.28 ax-gdb.c *** ax-gdb.c 7 Feb 2005 23:51:02 -0000 1.28 --- ax-gdb.c 8 Mar 2005 10:40:15 -0000 *************** gen_expr (union exp_element **pc, struct *** 1642,1647 **** --- 1642,1654 ---- } break; + case UNOP_PLUS: + (*pc)++; + /* + FOO is equivalent to 0 + FOO, which can be optimized. */ + gen_expr (pc, ax, value); + gen_usual_unary (ax, value); + break; + case UNOP_NEG: (*pc)++; /* -FOO is equivalent to 0 - FOO. */ Index: c-exp.y =================================================================== RCS file: /cvs/src/src/gdb/c-exp.y,v retrieving revision 1.28 diff -c -3 -p -r1.28 c-exp.y *** c-exp.y 7 Aug 2004 19:45:45 -0000 1.28 --- c-exp.y 8 Mar 2005 10:40:20 -0000 *************** exp : '-' exp %prec UNARY *** 256,261 **** --- 256,265 ---- { write_exp_elt_opcode (UNOP_NEG); } ; + exp : '+' exp %prec UNARY + { write_exp_elt_opcode (UNOP_PLUS); } + ; + exp : '!' exp %prec UNARY { write_exp_elt_opcode (UNOP_LOGICAL_NOT); } ; Index: eval.c =================================================================== RCS file: /cvs/src/src/gdb/eval.c,v retrieving revision 1.53 diff -c -3 -p -r1.53 eval.c *** eval.c 11 Feb 2005 04:05:46 -0000 1.53 --- eval.c 8 Mar 2005 10:40:25 -0000 *************** evaluate_subexp_standard (struct type *e *** 1855,1860 **** --- 1855,1869 ---- evaluate_subexp (NULL_TYPE, exp, pos, noside); return evaluate_subexp (NULL_TYPE, exp, pos, noside); + case UNOP_PLUS: + arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside); + if (noside == EVAL_SKIP) + goto nosideret; + if (unop_user_defined_p (op, arg1)) + return value_x_unop (arg1, op, noside); + else + return value_pos (arg1); + case UNOP_NEG: arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside); if (noside == EVAL_SKIP) Index: valarith.c =================================================================== RCS file: /cvs/src/src/gdb/valarith.c,v retrieving revision 1.38 diff -c -3 -p -r1.38 valarith.c *** valarith.c 11 Feb 2005 04:06:08 -0000 1.38 --- valarith.c 8 Mar 2005 10:40:29 -0000 *************** value_x_unop (struct value *arg1, enum e *** 560,565 **** --- 560,568 ---- case UNOP_NEG: strcpy (ptr, "-"); break; + case UNOP_PLUS: + strcpy (ptr, "+"); + break; case UNOP_IND: strcpy (ptr, "*"); break; *************** value_less (struct value *arg1, struct v *** 1313,1319 **** } } \f ! /* The unary operators - and ~. Both free the argument ARG1. */ struct value * value_neg (struct value *arg1) --- 1316,1350 ---- } } \f ! /* The unary operators +, - and ~. They free the argument ARG1 ! (unless it is returned). */ ! ! struct value * ! value_pos (struct value *arg1) ! { ! struct type *type; ! ! arg1 = coerce_ref (arg1); ! ! type = check_typedef (value_type (arg1)); ! ! if (TYPE_CODE (type) == TYPE_CODE_FLT) ! return value_from_double (type, value_as_double (arg1)); ! else if (is_integral_type (type)) ! { ! /* Perform integral promotion for ANSI C/C++. FIXME: What about ! FORTRAN and (the deleted) chill ? */ ! if (TYPE_LENGTH (type) < TYPE_LENGTH (builtin_type_int)) ! type = builtin_type_int; ! ! return value_from_longest (type, value_as_long (arg1)); ! } ! else ! { ! error ("Argument to positive operation not a number."); ! return 0; /* For lint -- never reached */ ! } ! } struct value * value_neg (struct value *arg1) Index: value.h =================================================================== RCS file: /cvs/src/src/gdb/value.h,v retrieving revision 1.84 diff -c -3 -p -r1.84 value.h *** value.h 28 Feb 2005 17:00:49 -0000 1.84 --- value.h 8 Mar 2005 10:40:30 -0000 *************** extern struct value *value_addr (struct *** 326,331 **** --- 326,333 ---- extern struct value *value_assign (struct value *toval, struct value *fromval); + extern struct value *value_pos (struct value *arg1); + extern struct value *value_neg (struct value *arg1); extern struct value *value_complement (struct value *arg1); Index: testsuite/gdb.cp/userdef.cc =================================================================== RCS file: /cvs/src/src/gdb/testsuite/gdb.cp/userdef.cc,v retrieving revision 1.2 diff -c -3 -p -r1.2 userdef.cc *** testsuite/gdb.cp/userdef.cc 12 Jun 2004 16:45:49 -0000 1.2 --- testsuite/gdb.cp/userdef.cc 8 Mar 2005 10:40:35 -0000 *************** A1 operator/(const A1&); *** 63,68 **** --- 63,69 ---- A1 operator=(const A1&); A1 operator~(); + A1 operator+(); A1 operator-(); int operator!(); A1 operator++(); *************** A1 A1::operator-(void) *** 225,230 **** --- 226,240 ---- return (neg); } + A1 A1::operator+(void) + { + A1 pos(0,0); + pos.x = +x; + pos.y = +y; + + return (pos); + } + A1 A1::operator~(void) { A1 acompl(0,0); *************** ostream& operator<<(ostream& outs, A1 on *** 286,291 **** --- 296,312 ---- return (outs << endl << "x = " << one.x << endl << "y = " << one.y << endl << "-------" << endl); } + class A2 { + public: + A2 operator+(); + }; + + A2 A2::operator+() + { + return A2 (); + } + + int main (void) { A1 one(2,3); *************** int main (void) *** 342,347 **** --- 363,370 ---- val = (!one); cout << "! " << val << endl << "-----"<<endl; + three = (+one); + cout << "+ " << three; three = (-one); cout << "- " << three; three = (~one); Index: testsuite/gdb.cp/userdef.exp =================================================================== RCS file: /cvs/src/src/gdb/testsuite/gdb.cp/userdef.exp,v retrieving revision 1.2 diff -c -3 -p -r1.2 userdef.exp *** testsuite/gdb.cp/userdef.exp 11 Feb 2004 14:01:25 -0000 1.2 --- testsuite/gdb.cp/userdef.exp 8 Mar 2005 10:40:35 -0000 *************** gdb_test "print one >> 31" "\\\$\[0-9\]* *** 121,126 **** --- 121,128 ---- gdb_test "print !one" "\\\$\[0-9\]* = 0\[\r\n\]" # Assumes 2's complement. So does everything... + gdb_test "print +one" "\\\$\[0-9\]* = {x = 2, y = 3}" + gdb_test "print ~one" "\\\$\[0-9\]* = {x = -3, y = -4}" gdb_test "print -one" "\\\$\[0-9\]* = {x = -2, y = -3}" *************** gdb_test "print one += 7" "\\\$\[0-9\]* *** 138,145 **** gdb_test "print two = one" "\\\$\[0-9\]* = {x = 9, y = 10}" # Check that GDB tolerates whitespace in operator names. ! gdb_test "break A1::'operator+'" ".*Breakpoint $decimal at.*" ! gdb_test "break A1::'operator +'" ".*Breakpoint $decimal at.*" gdb_exit return 0 --- 140,147 ---- gdb_test "print two = one" "\\\$\[0-9\]* = {x = 9, y = 10}" # Check that GDB tolerates whitespace in operator names. ! gdb_test "break A2::'operator+'" ".*Breakpoint $decimal at.*" ! gdb_test "break A2::'operator +'" ".*Breakpoint $decimal at.*" gdb_exit return 0 ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] Unary plus 2005-03-08 10:51 ` Nathan Sidwell @ 2005-03-08 13:40 ` Daniel Jacobowitz 0 siblings, 0 replies; 4+ messages in thread From: Daniel Jacobowitz @ 2005-03-08 13:40 UTC (permalink / raw) To: Nathan Sidwell; +Cc: gdb-patches On Tue, Mar 08, 2005 at 10:50:53AM +0000, Nathan Sidwell wrote: > Daniel Jacobowitz wrote: > >Two comments: > > > >- Could you add a non-tracepoint testcase for this? I think one good > >place would be gdb.cp/userdef.exp. > done. As this overloads operator+, I introduced a new class for > the breakpoint test. > > > >>! if (TYPE_CODE (type) == TYPE_CODE_FLT) > >>! return arg1; > > > > > >- IIUC, that's not quite right - we need an rvalue here. > > Like this? > > built and tested on i686-pc-linux-gnu, ok? OK. This comment isn't quite right after the rvalue fix: > ! /* The unary operators - and ~. Both free the argument ARG1. */ > > struct value * > value_neg (struct value *arg1) > --- 1316,1350 ---- > } > } > \f > ! /* The unary operators +, - and ~. They free the argument ARG1 > ! (unless it is returned). */ -- Daniel Jacobowitz CodeSourcery, LLC ^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2005-03-08 13:40 UTC | newest] Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2005-02-09 13:37 [PATCH] Unary plus Nathan Sidwell 2005-02-27 4:22 ` Daniel Jacobowitz 2005-03-08 10:51 ` Nathan Sidwell 2005-03-08 13:40 ` Daniel Jacobowitz
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox