From: Joel Brobecker <brobecker@adacore.com>
To: Simon Marchi <simon.marchi@ericsson.com>
Cc: gdb-patches@sourceware.org
Subject: Re: [PATCH] Define GNULIB_NAMESPACE in unittests/string_view-selftests.c
Date: Fri, 04 May 2018 16:59:00 -0000 [thread overview]
Message-ID: <20180504165944.vflxxmzv66f74fsk@adacore.com> (raw)
In-Reply-To: <20180504165552.f4y7zcxzmgvwngel@adacore.com>
[-- Attachment #1: Type: text/plain, Size: 1477 bytes --]
> > gdb/ChangeLog:
> >
> > * unittests/string_view-selftests.c: Define GNULIB_NAMESPACE.
>
> I noticed the exact same problem, and came up with the exact same
> solution. Sorry for not reporting right away, I have been swamped
> :-( and also wanted to spend some more time investigating the bigger
> picture; it seems to me like we may be having a huge time bomb
> in our hands?!?
>
> In other words, either we don't define GNULIB_NAMESPACE, but then
> we run into unwanted renames of class method names; or we defined
> GNULIB_NAMESPACE, but that means we must be prefixing all the symbols
> we call that are covered by the imported part of GNULIB with that
> namespace.
>
> What worries me is that I don't see what's preventing us from hitting
> that issue outside of the unittests code? We know we can adjust our
> own classes, but this problem occured with a system class, so we had
> no choice but to use GNULIB_NAMESPACE. I worry that the transition
> from no GNULIB_NAMESPACE to using GNULIB_NAMESPACE in a given unit
> will leave some C system calls that should normally be covered by
> gnulib silently now reverting to the system (buggy) version.
Actually, for the record, here is the patch I came up with.
Similar in nature, but additional comments and also only defining
the GNULIB_NAMESPACE macro on Windows hosts -- as a strategy to
minimize impact.
I think your approach is best, as we'll at least get the same
conditions regardless of the host.
--
Joel
[-- Attachment #2: 0001-avoid-undefined-reference-to-basic_ofstream-rpl_clos.patch --]
[-- Type: text/x-diff, Size: 2620 bytes --]
From 711720d2b969d679880383cafe48fb09a1f21262 Mon Sep 17 00:00:00 2001
From: Joel Brobecker <brobecker@adacore.com>
Date: Tue, 24 Apr 2018 13:07:19 -0400
Subject: [PATCH] avoid undefined reference to basic_ofstream::rpl_close durin
GDB link
On Windows hosts, GDB currently fails to link with the following error:
| CXXLD gdb.exe
|unittests/string_view-selftests.o: In function `ZN9selftests11string_view11inserters_26test05Ej':
|/[...]/gdb/unittests/basic_string_view/inserters/char/2.cc:60: undefined reference to `std::basic_ofstream<char, std::char_traits<char> >::rpl_close()'
|collect2.exe: error: ld returned 1 exit status
This is due to gnulib's unistd.h defining "close" as "rpl_close",
which interferes with our use of ofstream::close.
This patch works around the issue by defining GNULIB_NAMESPACE
inside the unit test itself, when building on Windows hosts.
gdb/ChangeLog:
* unittests/string_view-selftests.c [_WIN32]: Define GNULIB_NAMESPACE.
TN: R424-035
---
gdb/unittests/string_view-selftests.c | 27 +++++++++++++++++++++++++++
1 file changed, 27 insertions(+)
diff --git a/gdb/unittests/string_view-selftests.c b/gdb/unittests/string_view-selftests.c
index 182a5df9e4d..dd8bf8b8c52 100644
--- a/gdb/unittests/string_view-selftests.c
+++ b/gdb/unittests/string_view-selftests.c
@@ -21,6 +21,33 @@
the "real" version. */
#if __cplusplus < 201703L
+#if defined (_WIN32)
+/* On Windows hosts, gnulib's unistd.h uses a macro to redefine "close"
+ into "rpl_close", because of:
+
+ # if !(defined __cplusplus && defined GNULIB_NAMESPACE)
+ # undef close
+ # define close rpl_close
+ # endif
+
+ This interferes with one of the tests below that calls class
+ std::ofstream's "close" method, which gets renamed into a call
+ to non-existant "rpl_close" method instead, thus leading to
+ an undefined symbol error during the link:
+
+ unittests/basic_string_view/inserters/char/2.cc:60:
+ undefined reference toi
+ `std::basic_ofstream<char, std::char_traits<char> >::rpl_close()'
+
+ The gnulib documentation discusses this exact issue and says
+ we can define GNULIB_NAMESPACE to put all gnulib symbols into
+ the given namespace, but this is at the cost of gnulib no longer
+ replacing possibly faulty system functions. We have no real choice
+ here, and this is about unittest, so define GNULIB_NAMESPACE,
+ solely for this unit (the gnulib manual confirms this is OK). */
+#define GNULIB_NAMESPACE gnulib
+#endif
+
#include "defs.h"
#include "selftest.h"
#include "common/gdb_string_view.h"
--
2.11.0
next prev parent reply other threads:[~2018-05-04 16:59 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-05-03 21:25 Simon Marchi
2018-05-04 16:55 ` Joel Brobecker
2018-05-04 16:59 ` Joel Brobecker [this message]
2018-05-04 17:11 ` Pedro Alves
2018-05-04 17:24 ` Joel Brobecker
2018-05-08 20:47 ` Simon Marchi
2018-05-08 21:22 ` Joel Brobecker
2018-05-09 13:50 ` Pedro Alves
2018-05-09 14:00 ` Joel Brobecker
2018-05-09 15:19 ` Simon Marchi
2018-05-04 17:14 ` Simon Marchi
2018-05-04 17:20 ` Joel Brobecker
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=20180504165944.vflxxmzv66f74fsk@adacore.com \
--to=brobecker@adacore.com \
--cc=gdb-patches@sourceware.org \
--cc=simon.marchi@ericsson.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