* [PATCH] Fix recent thinko in operand_equal_p
@ 2022-11-04 9:27 Eric Botcazou
2022-11-04 9:40 ` Richard Biener
0 siblings, 1 reply; 2+ messages in thread
From: Eric Botcazou @ 2022-11-04 9:27 UTC (permalink / raw)
To: gcc-patches
[-- Attachment #1: Type: text/plain, Size: 1367 bytes --]
Hi,
there is a thinko in the recent improvement by Jan:
2020-11-19 Jan Hubicka <jh@suse.cz>
* fold-const.c (operand_compare::operand_equal_p): Fix thinko in
COMPONENT_REF handling and guard types_same_for_odr by
virtual_method_call_p.
(operand_compare::hash_operand): Likewise.
where the code just looks at operand 2 of COMPONENT_REF, if it is present, to
compare addresses. That's wrong because operand 2 contains the number of
DECL_OFFSET_ALIGN-bit-sized words so, when DECL_OFFSET_ALIGN > 8, not all the
bytes are included and some of them are in DECL_FIELD_BIT_OFFSET instead, see
get_inner_reference for the model computation.
In other words, you would need to compare operand 2 and DECL_OFFSET_ALIGN and
DECL_FIELD_BIT_OFFSET in this situation, but I'm not sure this is worth the
hassle in practice so the attached fix just removes this alternate handling.
Tested on x86-64/Linux, OK for mainline, 12 and 11 branches?
2022-11-04 Eric Botcazou <ebotcazou@adacore.com>
* fold-const.cc (operand_compare::operand_equal_p) <COMPONENT_REF>:
Do not take into account operand 2.
(operand_compare::hash_operand) <COMPONENT_REF>: Likewise.
2022-11-04 Eric Botcazou <ebotcazou@adacore.com>
* gnat.dg/opt99.adb: New test.
* gnat.dg/opt99_pkg1.ads, gnat.dg/opt99_pkg1.adb: New helper.
* gnat.dg/opt99_pkg2.ads: Likewise.
--
Eric Botcazou
[-- Attachment #2: p.diff --]
[-- Type: text/x-patch, Size: 1304 bytes --]
diff --git a/gcc/fold-const.cc b/gcc/fold-const.cc
index 9f7beae14e5..e7a9d5215a5 100644
--- a/gcc/fold-const.cc
+++ b/gcc/fold-const.cc
@@ -3351,9 +3351,6 @@ operand_compare::operand_equal_p (const_tree arg0, const_tree arg1,
if (compare_address
&& (flags & OEP_ADDRESS_OF_SAME_FIELD) == 0)
{
- if (TREE_OPERAND (arg0, 2)
- || TREE_OPERAND (arg1, 2))
- return OP_SAME_WITH_NULL (2);
tree field0 = TREE_OPERAND (arg0, 1);
tree field1 = TREE_OPERAND (arg1, 1);
@@ -3864,17 +3861,10 @@ operand_compare::hash_operand (const_tree t, inchash::hash &hstate,
if (sflags & OEP_ADDRESS_OF)
{
hash_operand (TREE_OPERAND (t, 0), hstate, flags);
- if (TREE_OPERAND (t, 2))
- hash_operand (TREE_OPERAND (t, 2), hstate,
- flags & ~OEP_ADDRESS_OF);
- else
- {
- tree field = TREE_OPERAND (t, 1);
- hash_operand (DECL_FIELD_OFFSET (field),
- hstate, flags & ~OEP_ADDRESS_OF);
- hash_operand (DECL_FIELD_BIT_OFFSET (field),
- hstate, flags & ~OEP_ADDRESS_OF);
- }
+ hash_operand (DECL_FIELD_OFFSET (TREE_OPERAND (t, 1)),
+ hstate, flags & ~OEP_ADDRESS_OF);
+ hash_operand (DECL_FIELD_BIT_OFFSET (TREE_OPERAND (t, 1)),
+ hstate, flags & ~OEP_ADDRESS_OF);
return;
}
break;
[-- Attachment #3: opt99.adb --]
[-- Type: text/x-adasrc, Size: 248 bytes --]
-- { dg-do run }
-- { dg-options "-O" }
with Opt99_Pkg1; use Opt99_Pkg1;
procedure Opt99 is
C : constant My_Character := (D => True, C => ' ');
D : Derived;
begin
Set (D, C, C);
if not D.C2.D then
raise Program_Error;
end if;
end;
[-- Attachment #4: opt99_pkg1.adb --]
[-- Type: text/x-adasrc, Size: 169 bytes --]
package body Opt99_Pkg1 is
procedure Set (D: in out Derived; C1, C2 : My_Character) is
begin
D.I := 0;
D.C1 := C1;
D.C2 := C2;
end;
end Opt99_Pkg1;
[-- Attachment #5: opt99_pkg2.ads --]
[-- Type: text/x-adasrc, Size: 222 bytes --]
package Opt99_Pkg2 is
function Get_Max return Positive is (4);
C : constant Positive := Get_Max;
type Arr is array (1 .. C) of Integer;
type Root is tagged record
Data : Arr;
end record;
end Opt99_Pkg2;
[-- Attachment #6: opt99_pkg1.ads --]
[-- Type: text/x-adasrc, Size: 384 bytes --]
with Opt99_Pkg2;
package Opt99_Pkg1 is
type My_Character (D : Boolean := False) is record
case D is
when False => null;
when True => C : Character;
end case;
end record;
type Derived is new Opt99_Pkg2.Root with record
I : Integer;
C1, C2 : My_Character;
end record;
procedure Set (D: in out Derived; C1, C2 : My_Character);
end Opt99_Pkg1;
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] Fix recent thinko in operand_equal_p
2022-11-04 9:27 [PATCH] Fix recent thinko in operand_equal_p Eric Botcazou
@ 2022-11-04 9:40 ` Richard Biener
0 siblings, 0 replies; 2+ messages in thread
From: Richard Biener @ 2022-11-04 9:40 UTC (permalink / raw)
To: Eric Botcazou; +Cc: gcc-patches
> Am 04.11.2022 um 10:29 schrieb Eric Botcazou via Gcc-patches <gcc-patches@gcc.gnu.org>:
>
> Hi,
>
> there is a thinko in the recent improvement by Jan:
>
> 2020-11-19 Jan Hubicka <jh@suse.cz>
>
> * fold-const.c (operand_compare::operand_equal_p): Fix thinko in
> COMPONENT_REF handling and guard types_same_for_odr by
> virtual_method_call_p.
> (operand_compare::hash_operand): Likewise.
>
> where the code just looks at operand 2 of COMPONENT_REF, if it is present, to
> compare addresses. That's wrong because operand 2 contains the number of
> DECL_OFFSET_ALIGN-bit-sized words so, when DECL_OFFSET_ALIGN > 8, not all the
> bytes are included and some of them are in DECL_FIELD_BIT_OFFSET instead, see
> get_inner_reference for the model computation.
>
> In other words, you would need to compare operand 2 and DECL_OFFSET_ALIGN and
> DECL_FIELD_BIT_OFFSET in this situation, but I'm not sure this is worth the
> hassle in practice so the attached fix just removes this alternate handling.
> Tested on x86-64/Linux, OK for mainline, 12 and 11 branches?
Ok.
Thanks,
Richard
>
> 2022-11-04 Eric Botcazou <ebotcazou@adacore.com>
>
> * fold-const.cc (operand_compare::operand_equal_p) <COMPONENT_REF>:
> Do not take into account operand 2.
> (operand_compare::hash_operand) <COMPONENT_REF>: Likewise.
>
>
> 2022-11-04 Eric Botcazou <ebotcazou@adacore.com>
>
> * gnat.dg/opt99.adb: New test.
> * gnat.dg/opt99_pkg1.ads, gnat.dg/opt99_pkg1.adb: New helper.
> * gnat.dg/opt99_pkg2.ads: Likewise.
>
> --
> Eric Botcazou
> <p.diff>
> <opt99.adb>
> <opt99_pkg1.adb>
> <opt99_pkg2.ads>
> <opt99_pkg1.ads>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-11-04 9:40 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-11-04 9:27 [PATCH] Fix recent thinko in operand_equal_p Eric Botcazou
2022-11-04 9:40 ` Richard Biener
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).