public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "tanksherman27 at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/105645] Template specializations are not hidden with fvisibility=hidden
Date: Fri, 23 Feb 2024 13:40:57 +0000	[thread overview]
Message-ID: <bug-105645-4-IPvDuaN7OR@http.gcc.gnu.org/bugzilla/> (raw)
In-Reply-To: <bug-105645-4@http.gcc.gnu.org/bugzilla/>

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105645

Julian Waters <tanksherman27 at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tanksherman27 at gmail dot com

--- Comment #3 from Julian Waters <tanksherman27 at gmail dot com> ---
I can confirm this on gcc 11 and higher:

#include <cstdlib>

template<typename T>
void method(T* src, T* dst, size_t length);

template<>
void method<short>(short* src, short* dst, size_t length) {

}

template<>
void method<int>(int* src, int* dst, size_t length) {

}

Compiled using command line:
g++ -O3 -fvisibility=hidden -std=c++14 -pedantic -Wpedantic -Wall -Werror
templates.cpp

Linked using:
g++ -shared -o libtemplates.so templates.o

When nm is run over it using nm -gDC libtemplates.so, the following symbols are
seen:
                 w _ITM_deregisterTMCloneTable
                 w _ITM_registerTMCloneTable
0000000000001110 T void method<int>(int*, int*, unsigned long)
0000000000001100 T void method<short>(short*, short*, unsigned long)
                 w __cxa_finalize
                 w __gmon_start__

T void method<int>(int*, int*, unsigned long) and T void method<short>(short*,
short*, unsigned long) should not be in the symbol table, but they are even
with -fvisibility=hidden.

Strangely, marking both methods with [[gnu::visibility("hidden")]] works and
results in both not being exported.

This was recently observed in the JDK, inside the Java HotSpot VM (see
https://github.com/openjdk/jdk/pull/17955/files#r1498933843)

Is this enough of a reproducer to get this confirmed?

  parent reply	other threads:[~2024-02-23 13:41 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-05-18 14:02 [Bug c++/105645] New: " kndevl at outlook dot com
2022-05-18 14:25 ` [Bug c++/105645] " pinskia at gcc dot gnu.org
2022-05-19  6:10 ` kndevl at outlook dot com
2024-02-23 13:40 ` tanksherman27 at gmail dot com [this message]
2024-05-26  7:43 ` pinskia at gcc dot gnu.org
2024-05-26  7:48 ` tanksherman27 at gmail dot com

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=bug-105645-4-IPvDuaN7OR@http.gcc.gnu.org/bugzilla/ \
    --to=gcc-bugzilla@gcc.gnu.org \
    --cc=gcc-bugs@gcc.gnu.org \
    /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).