From: "Arsen Arsenović" <arsen@aarsen.me>
To: Jason Merrill <jason@redhat.com>
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [PATCH] contracts: Stop relying on mangling for naming .pre/.post clones
Date: Thu, 15 Dec 2022 19:00:56 +0100 [thread overview]
Message-ID: <86bko4ttho.fsf@aarsen.me> (raw)
In-Reply-To: <5cc46cd6-5f08-e73a-0db2-ea165a6e7734@redhat.com>
[-- Attachment #1.1: Type: text/plain, Size: 717 bytes --]
Hi Jason,
Jason Merrill <jason@redhat.com> writes:
> On 12/10/22 08:13, Arsen Arsenović wrote:
>> If the mangler is relied on, functions with extern "C" on them emit multiple
>> definitions of the same name.
>
> But doing it here interferes with lazy mangling. How about appending the
> suffix into write_mangled_name instead of write_encoding? The demangler
> already expects "clone" suffixes at the end of the mangled name.
Ah, sorry. I'm not well versed in the mangler code, so I didn't realize
(frankly, I was initially surprised when I saw that DECL_ASSEMBLER_NAME
was set that early, but went with it). That makes sense.
How about this? Tested on x86_64-pc-linux-gnu via check-g++.
[-- Attachment #1.2: c++: Mangle contracts in write_mangled_name unconditionally --]
[-- Type: text/plain, Size: 2644 bytes --]
From 2a2d98e94bdd7a8d7f862b2accda849927e4509e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Arsen=20Arsenovi=C4=87?= <arsen@aarsen.me>
Date: Thu, 15 Dec 2022 18:56:59 +0100
Subject: [PATCH v2] c++: Mangle contracts in write_mangled_name
unconditionally
This fixes contract-checked extern "C" functions.
gcc/cp/ChangeLog:
* mangle.cc (write_encoding): Move contract pre/post function mangling
from here...
(write_mangled_name): ... to here, and make it happen always.
gcc/testsuite/ChangeLog:
* g++.dg/contracts/contracts-externC.C: New test.
---
gcc/cp/mangle.cc | 14 +++++++-------
.../g++.dg/contracts/contracts-externC.C | 19 +++++++++++++++++++
2 files changed, 26 insertions(+), 7 deletions(-)
create mode 100644 gcc/testsuite/g++.dg/contracts/contracts-externC.C
diff --git a/gcc/cp/mangle.cc b/gcc/cp/mangle.cc
index e363ef35b9f..074cf27ec7a 100644
--- a/gcc/cp/mangle.cc
+++ b/gcc/cp/mangle.cc
@@ -798,6 +798,13 @@ write_mangled_name (const tree decl, bool top_level)
write_string ("_Z");
write_encoding (decl);
}
+
+ /* If this is the pre/post function for a guarded function, append
+ .pre/post, like something from create_virtual_clone. */
+ if (DECL_IS_PRE_FN_P (decl))
+ write_string (".pre");
+ else if (DECL_IS_POST_FN_P (decl))
+ write_string (".post");
}
/* Returns true if the return type of DECL is part of its signature, and
@@ -856,13 +863,6 @@ write_encoding (const tree decl)
mangle_return_type_p (decl),
d);
- /* If this is the pre/post function for a guarded function, append
- .pre/post, like something from create_virtual_clone. */
- if (DECL_IS_PRE_FN_P (decl))
- write_string (".pre");
- else if (DECL_IS_POST_FN_P (decl))
- write_string (".post");
-
/* If this is a coroutine helper, then append an appropriate string to
identify which. */
if (tree ramp = DECL_RAMP_FN (decl))
diff --git a/gcc/testsuite/g++.dg/contracts/contracts-externC.C b/gcc/testsuite/g++.dg/contracts/contracts-externC.C
new file mode 100644
index 00000000000..873056b742b
--- /dev/null
+++ b/gcc/testsuite/g++.dg/contracts/contracts-externC.C
@@ -0,0 +1,19 @@
+// simple check to ensure we don't emit a function with the same name twice,
+// when wrapping functions in pre- and postconditions.
+// { dg-do link }
+// { dg-options "-std=c++2a -fcontracts -fcontract-continuation-mode=on" }
+
+volatile int x = 10;
+
+extern "C" void
+f ()
+ [[ pre: x < 10 ]]
+{
+}
+
+int
+main ()
+ [[ post: x > 10 ]]
+{
+ f();
+}
--
2.39.0
[-- Attachment #1.3: Type: text/plain, Size: 554 bytes --]
I did run c++filt (afaik, it uses the libiberty demangler) on this
revision, and I got:
.type f(int) [clone .pre], @function
f(int) [clone .pre]:
out of ``void f(int x) [[ pre: x > 10 ]] {}'', which seems to match your
description.
If I understand this right, write_xxx corresponds to xxx in the Itanium
ABI mangling BNF, in which case, I believe I have the correct spot here.
In that case, a similar change should happen for coroutines; I think
Iain was working on that.
Thanks, have a great day.
--
Arsen Arsenović
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 381 bytes --]
next prev parent reply other threads:[~2022-12-15 18:25 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-12-10 13:13 Arsen Arsenović
2022-12-10 13:14 ` Arsen Arsenović
2022-12-10 19:51 ` Bernhard Reutner-Fischer
2022-12-11 15:56 ` Arsen Arsenović
2022-12-15 15:40 ` Jason Merrill
2022-12-15 18:00 ` Arsen Arsenović [this message]
2022-12-15 20:29 ` Iain Sandoe
2022-12-17 11:30 ` Arsen Arsenović
2022-12-15 21:48 ` Jason Merrill
2022-12-17 11:30 ` Arsen Arsenović
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=86bko4ttho.fsf@aarsen.me \
--to=arsen@aarsen.me \
--cc=gcc-patches@gcc.gnu.org \
--cc=jason@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).