From: Alexander Lukichev <alexander.lukichev@gmail.com>
To: buildroot@busybox.net
Subject: [Buildroot] [autobuild.buildroot.net] Build results for 2015-09-27
Date: Fri, 02 Oct 2015 12:33:13 +0300 [thread overview]
Message-ID: <560E4F59.2050506@gmail.com> (raw)
In-Reply-To: <20150928063014.85E79102A48@stock.ovh.net>
Hi,
On 09/28/2015 09:30 AM, Thomas Petazzoni wrote:
> arm | qt-4.8.7 | NOK | http://autobuild.buildroot.net/results/11a4c19638ca211bd2a424fd92602166235c9433/
This, and some other builds [1], fail because of the BR2_STATIC_LIBS configuration. The pre-built toolchain in this case does not have dlfcn.h. The failing source file src/corelib/plugin/qlibrary_unix.cpp has a QT_NO_DYNAMIC_LIBRARY guard constant for this case, which was added to qt by the patch [2]. It makes the module behave correctly when plugins cannot be loaded dynamically, which is the case for BR2_STATIC_LIBS in Buildroot:
#if (defined(Q_OS_VXWORKS) && !defined(VXWORKS_RTP)) || defined (Q_OS_NACL)
#define QT_NO_DYNAMIC_LIBRARY
#endif
QT_BEGIN_NAMESPACE
#if !defined(QT_HPUX_LD) && !defined(QT_NO_DYNAMIC_LIBRARY)
QT_BEGIN_INCLUDE_NAMESPACE
#include <dlfcn.h>
QT_END_INCLUDE_NAMESPACE
#endif
static QString qdlerror()
{
#if defined(QT_NO_DYNAMIC_LIBRARY)
const char *err = "This platform does not support dynamic libraries.";
#elif !defined(QT_HPUX_LD)
const char *err = dlerror();
#else
const char *err = strerror(errno);
#endif
return err ? QLatin1Char('(') + QString::fromLocal8Bit(err) + QLatin1Char(')'): QString();
}
...and so on.
The QT_NO_DYNAMIC_LIBRARY constant is defined in the same file for Q_OS_VXWORKS or Q_OS_NACL (from src/corelib/global/qglobal.h). There are also two other constants related to inclusion of dlfcn.h: QT_HPUX_LD (from a platform mkspec's qplatformdefs.h, e.g. for hpux-acc-o64) and VXWORKS_RTP (defined elsewhere than qt package).
One way to fix the failing builds [1] would be to make this QT_NO_DYNAMIC_LIBRARY defined if BR2_STATIC_LIBS is chosen in the Buildroot configuration, and to do this preferably in an upstream-friendly way but I don't see how this could be done. One option could be to define an mkspec for Buildroot, which seems too heavy. Another is to make a "constant defined elsewhere" available in the qt's compilation environment (like VXWORKS_RTP above). How this could be done? Does adding a Qt-wide BUILDROOT_STATIC (or some such) constant to QT_CXXFLAGS in qt.mk in seem a good idea? Would such a patch (that examines this Buildroot-specific constant in src/corelib/plugin/qlibrary_unix.cpp) look good in Qt's upstream?
References:
[1] http://autobuild.buildroot.net/?status=NOK&reason=qt-4.8.7&arch=arm
[2] https://github.com/mirror/qt/commit/468142db15fe99c70505f507f6c2941706172853
--
Best regards,
Alexander Lukichev
prev parent reply other threads:[~2015-10-02 9:33 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-09-28 6:30 [Buildroot] [autobuild.buildroot.net] Build results for 2015-09-27 Thomas Petazzoni
2015-10-02 9:33 ` Alexander Lukichev [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=560E4F59.2050506@gmail.com \
--to=alexander.lukichev@gmail.com \
--cc=buildroot@busybox.net \
/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