From: Jason Merrill <jason@redhat.com>
To: Jakub Jelinek <jakub@redhat.com>
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [PATCH] c++, v2: Fix ICE with __builtin_bit_cast [PR98469]
Date: Thu, 7 Jan 2021 16:11:16 -0500 [thread overview]
Message-ID: <9271ff46-f54a-8c6d-62fc-c7aec2e7af63@redhat.com> (raw)
In-Reply-To: <20210105152624.GR725145@tucnak>
[-- Attachment #1: Type: text/plain, Size: 1647 bytes --]
On 1/5/21 10:26 AM, Jakub Jelinek wrote:
> On Mon, Jan 04, 2021 at 04:01:25PM -0500, Jason Merrill via Gcc-patches wrote:
>> On 1/4/21 3:48 PM, Jakub Jelinek wrote:
>>> On Mon, Jan 04, 2021 at 03:44:46PM -0500, Jason Merrill wrote:
>>>> This change is OK, but part of the problem is that we're trying to do
>>>> overload resolution for an S copy/move constructor, which we shouldn't be
>>>> because bit_cast is a prvalue, so in C++17 and up we should use it to
>>>> directly initialize the target without any implied constructor call.
>>>>
>>>> It seems we're mishandling this because the code in
>>>> build_special_member_call specifically looks for TARGET_EXPR or CONSTRUCTOR,
>>>> and BIT_CAST_EXPR is neither of those.
>>>>
>>>> Wrapping a BIT_CAST_EXPR of aggregate type in a TARGET_EXPR would address
>>>> this, and any other places that expect a class prvalue to come in the form
>>>> of a TARGET_EXPR.
>>>
>>> I can try that tomorrow. Won't that cause copying through extra temporary
>>> in some cases though, or is that guaranteed to be optimized?
>>
>> It won't cause any extra copying when it's used to initialize another object
>> (like the return value of std::bit_cast). Class prvalues are always
>> expressed with a TARGET_EXPR in the front end; the TARGET_EXPR melts away
>> when used as an initializer, it only creates a temporary when it's used in
>> another way.
>
> Ok, this version wraps it into a TARGET_EXPR then, it alone fixes the bug,
> but I've kept the constexpr.c change too.
This patch corrects this and one other place to not be as dependent on
TARGET_EXPR, but I think I'm going to save it for stage 1.
Jason
[-- Attachment #2: glvp.diff --]
[-- Type: text/x-patch, Size: 1952 bytes --]
commit 0d732b8c7fb3f8378dc1c894358bb5d766e6be5d
Author: Jason Merrill <jason@redhat.com>
Date: Mon Jan 4 16:11:08 2021 -0500
c++: Tweak prvalue test [PR98469]
Discussing the 98469 patch and class prvalues with Jakub also inspired me to
change the place that was mishandling BIT_CAST_EXPR and one other to use the
lvalue_kind machinery to decide whether something is a prvalue, instead of
looking specifically for a TARGET_EXPR.
gcc/cp/ChangeLog:
* call.c (build_special_member_call): Use !glvalue_p rather
than specific tree codes to test for prvalue.
(conv_is_prvalue): Likewise.
(implicit_conversion): Check CLASS_TYPE_P first.
diff --git a/gcc/cp/call.c b/gcc/cp/call.c
index 218157088ef..e2d2b23e449 100644
--- a/gcc/cp/call.c
+++ b/gcc/cp/call.c
@@ -2118,8 +2118,8 @@ implicit_conversion (tree to, tree from, tree expr, bool c_cast_p,
flags, complain);
if (!conv || conv->bad_p)
return conv;
- if (conv_is_prvalue (conv)
- && CLASS_TYPE_P (conv->type)
+ if (CLASS_TYPE_P (conv->type)
+ && conv_is_prvalue (conv)
&& CLASSTYPE_PURE_VIRTUALS (conv->type))
conv->bad_p = true;
return conv;
@@ -8500,8 +8500,7 @@ conv_is_prvalue (conversion *c)
return true;
if (c->kind == ck_user && !TYPE_REF_P (c->type))
return true;
- if (c->kind == ck_identity && c->u.expr
- && TREE_CODE (c->u.expr) == TARGET_EXPR)
+ if (c->kind == ck_identity && c->u.expr && !glvalue_p (c->u.expr))
return true;
return false;
@@ -9950,8 +9949,7 @@ build_special_member_call (tree instance, tree name, vec<tree, va_gc> **args,
&& CONSTRUCTOR_NELTS (arg) == 1)
arg = CONSTRUCTOR_ELT (arg, 0)->value;
- if ((TREE_CODE (arg) == TARGET_EXPR
- || TREE_CODE (arg) == CONSTRUCTOR)
+ if (!glvalue_p (arg)
&& (same_type_ignoring_top_level_qualifiers_p
(class_type, TREE_TYPE (arg))))
{
prev parent reply other threads:[~2021-01-07 21:11 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-12-30 9:13 [PATCH] c++: " Jakub Jelinek
2021-01-04 20:44 ` Jason Merrill
2021-01-04 20:48 ` Jakub Jelinek
2021-01-04 21:01 ` Jason Merrill
2021-01-05 15:26 ` [PATCH] c++, v2: " Jakub Jelinek
2021-01-05 16:00 ` Jason Merrill
2021-01-07 21:11 ` Jason Merrill [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=9271ff46-f54a-8c6d-62fc-c7aec2e7af63@redhat.com \
--to=jason@redhat.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=jakub@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).