Skip to contentSkip to conversation

synergy: Fix build with gcc 6

PATCH New Peter Korsgaard 2 participants 3 messages Feb 25, 2017 – Feb 26, 2017

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 synergy: Fix build with gcc 6

Fixes:
http://autobuild.buildroot.net/results/05b/05bbc22dd6cd5564462226e612ab1e95778fe935/
http://autobuild.buildroot.net/results/edf/edfe58749d1b7a1bb2e0184a6824a74b9d38ddb9/
http://autobuild.buildroot.net/results/139/1395eca13ca537bde928ddd68a5bc6e130e82ba3/
http://autobuild.buildroot.net/results/94a/94ad6e8bbbeb926ea834d9d6e3ba87d0398acb86/

Gcc 6.x defaults to C++14, and the iostream operator bool behaviour changed in C++11. In previous versions, a somewhat odd operator void* was used to return the status of the stream as a pointer. Since C++11 a more sensible operator bool is used to return the stream staus.

For details, see:

http://en.cppreference.com/w/cpp/io/basic_ios/operator_bool

The code in CConfigReadContext assumes the pre-C++11 behaviour and provides its own operator void overload to return the status of the embedded iostream. With C++11, iostream no longer provides this overload, breaking the build:

CConfig.cpp: In member function 'CConfigReadContext::operator void*() const':
CConfig.cpp:1851:9: error: cannot convert 'std::istream {aka std::basic_istream<char>}' to 'void*' in return
return m_stream;

To fix it, backport part of upstream commit 3d963bfbe7897d0a33ad (possible fix for mavericks) which changes the code to simply provide a getStream() method which returns a reference to the embedded stream and the calling code is changed to use operator bool on the returned stream, making the code compatible with both old and new compilers.

This upstream commit is part of the 1.6.0 release, so can be dropped when the version is bumped.

---
 package/synergy/0003-Fix-build-with-gcc-6.patch | 86 +++++++++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 package/synergy/0003-Fix-build-with-gcc-6.patch
package/synergy/0003-Fix-build-with-gcc-6.patch new file +86 -0
new file mode 100644
index 000000000..843078fe6
@@ -0,0 +1,86 @@
+From b79d478c21ed6af20554ead97da72de845dc3206 Mon Sep 17 00:00:00 2001
+From: Peter Korsgaard <[email protected]>
+Date: Sat, 25 Feb 2017 21:57:19 +0100
+Subject: [PATCH] Fix build with gcc 6
+
+Gcc 6.x defaults to C++14, and the iostream operator bool behaviour changed
+in C++11.  In previous versions, a somewhat odd operator void* was used to
+return the status of the stream as a pointer.  Since C++11 a more sensible
+operator bool is used to return the stream staus.
+
+For details, see:
+
+http://en.cppreference.com/w/cpp/io/basic_ios/operator_bool
+
+The code in CConfigReadContext assumes the pre-C++11 behaviour and provides
+its own operator void overload to return the status of the embedded
+iostream.  With C++11, iostream no longer provides this overload, breaking
+the build:
+
+CConfig.cpp: In member function 'CConfigReadContext::operator void*() const':
+CConfig.cpp:1851:9: error: cannot convert 'std::istream {aka std::basic_istream<char>}' to 'void*' in return
+  return m_stream;
+
+To fix it, backport part of upstream commit 3d963bfbe7897d0a33ad (possible
+fix for mavericks) which changes the code to simply provide a getStream()
+method which returns a reference to the embedded stream and the calling code
+is changed to use operator bool on the returned stream, making the code
+compatible with both old and new compilers.
+
+This upstream commit is part of the 1.6.0 release, so can be dropped when
+the version is bumped.
+
+Signed-off-by: Peter Korsgaard <[email protected]>
+---
+ lib/server/CConfig.cpp | 7 +------
+ lib/server/CConfig.h   | 2 +-
+ 2 files changed, 2 insertions(+), 7 deletions(-)
+
+diff --git a/lib/server/CConfig.cpp b/lib/server/CConfig.cpp
+index a502fe78..d67dde20 100644
+--- a/lib/server/CConfig.cpp
++++ b/lib/server/CConfig.cpp
+@@ -607,7 +607,7 @@ void
+ CConfig::read(CConfigReadContext& context)
+ {
+ 	CConfig tmp;
+-	while (context) {
++	while (context.getStream()) {
+ 		tmp.readSection(context);
+ 	}
+ 	*this = tmp;
+@@ -1846,11 +1846,6 @@ CConfigReadContext::getLineNumber() const
+ 	return m_line;
+ }
+ 
+-CConfigReadContext::operator void*() const
+-{
+-	return m_stream;
+-}
+-
+ bool
+ CConfigReadContext::operator!() const
+ {
+diff --git a/lib/server/CConfig.h b/lib/server/CConfig.h
+index c0d2faa8..0ee453cb 100644
+--- a/lib/server/CConfig.h
++++ b/lib/server/CConfig.h
+@@ -480,7 +480,6 @@ public:
+ 	bool			readLine(CString&);
+ 	UInt32			getLineNumber() const;
+ 
+-	operator void*() const;
+ 	bool			operator!() const;
+ 
+ 	OptionValue		parseBoolean(const CString&) const;
+@@ -502,6 +501,7 @@ public:
+ 	IPlatformScreen::CButtonInfo*
+ 					parseMouse(const CString& mouse) const;
+ 	KeyModifierMask	parseModifier(const CString& modifiers) const;
++	std::istream&	getStream() const { return m_stream; };
+ 
+ private:
+ 	// not implemented
+-- 
+2.11.0
+
Signature · 1 line
2.11.0
#1 lore raw

PATCH 1/1 synergy: Fix build with gcc 6

Hello,

Quoted 30 lines · Peter Korsgaard

Fixes:
http://autobuild.buildroot.net/results/05b/05bbc22dd6cd5564462226e612ab1e95778fe935/
http://autobuild.buildroot.net/results/edf/edfe58749d1b7a1bb2e0184a6824a74b9d38ddb9/
http://autobuild.buildroot.net/results/139/1395eca13ca537bde928ddd68a5bc6e130e82ba3/
http://autobuild.buildroot.net/results/94a/94ad6e8bbbeb926ea834d9d6e3ba87d0398acb86/

Gcc 6.x defaults to C++14, and the iostream operator bool behaviour changed in C++11. In previous versions, a somewhat odd operator void* was used to return the status of the stream as a pointer. Since C++11 a more sensible operator bool is used to return the stream staus.

For details, see:

http://en.cppreference.com/w/cpp/io/basic_ios/operator_bool

The code in CConfigReadContext assumes the pre-C++11 behaviour and provides its own operator void overload to return the status of the embedded iostream. With C++11, iostream no longer provides this overload, breaking the build:

CConfig.cpp: In member function 'CConfigReadContext::operator void*() const':
CConfig.cpp:1851:9: error: cannot convert 'std::istream {aka std::basic_istream<char>}' to 'void*' in return
return m_stream;

To fix it, backport part of upstream commit 3d963bfbe7897d0a33ad (possible fix for mavericks) which changes the code to simply provide a getStream() method which returns a reference to the embedded stream and the calling code is changed to use operator bool on the returned stream, making the code compatible with both old and new compilers.

Quoted 8 lines

This upstream commit is part of the 1.6.0 release, so can be dropped when the version is bumped.

---
 package/synergy/0003-Fix-build-with-gcc-6.patch | 86 +++++++++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 package/synergy/0003-Fix-build-with-gcc-6.patch

Applied to master, thanks. However, did you notice that the hash we have in Buildroot for synergy-1.3.1.tar.gz does not patch the tarball available from Sourceforge, and therefore we always fallback to downloading from sources.b.n ?

See:

Location: https://sourceforge.net/publish/ [following]
--2017-02-26 14:15:18-- https://sourceforge.net/publish/
Connecting to sourceforge.net (sourceforge.net)|216.34.181.60|:443... connected. HTTP request sent, awaiting response... 200 OK
Length: 43482 (42K) [text/html]
Saving to: ?/home/thomas/projets/buildroot/output/build/.synergy-1.3.1.tar.gz.LebFB4/output?

/home/thomas/projets/buildroot/output/build/.synergy 100%[===================================================================================================================>] 42,46K 172KB/s in 0,2s

2017-02-26 14:15:19 (172 KB/s) - ?/home/thomas/projets/buildroot/output/build/.synergy-1.3.1.tar.gz.LebFB4/output? saved [43482/43482]

ERROR: synergy-1.3.1.tar.gz has wrong sha256 hash:
ERROR: expected: 4a15e26456498dda5f3aa7d688a4892e0c50394a2f01685136fbc954bacc9691
ERROR: got : 062a0b0841486719a474df96c42381910dbc75447ce41e4a2059bef36829b259
ERROR: Incomplete download, or man-in-the-middle (MITM) attack
--2017-02-26 14:15:19-- http://sources.buildroot.net/synergy-1.3.1.tar.gz
Resolving sources.buildroot.net (sources.buildroot.net)... 176.9.16.109 Connecting to sources.buildroot.net (sources.buildroot.net)|176.9.16.109|:80... connected. HTTP request sent, awaiting response... 200 OK
Length: 793172 (775K) [application/x-gzip]
Saving to: ?/home/thomas/projets/buildroot/output/build/.synergy-1.3.1.tar.gz.sA8PnB/output?

/home/thomas/projets/buildroot/output/build/.synergy 100%[===================================================================================================================>] 774,58K 1,60MB/s in 0,5s

2017-02-26 14:15:20 (1,60 MB/s) - ?/home/thomas/projets/buildroot/output/build/.synergy-1.3.1.tar.gz.sA8PnB/output? saved [793172/793172]

synergy-1.3.1.tar.gz: OK (sha256: 4a15e26456498dda5f3aa7d688a4892e0c50394a2f01685136fbc954bacc9691)

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com
#2 lore raw

PATCH 1/1 synergy: Fix build with gcc 6

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Hi,

Quoted 4 lines

Applied to master, thanks. However, did you notice that the hash we have in Buildroot for synergy-1.3.1.tar.gz does not patch the tarball available from Sourceforge, and therefore we always fallback to downloading from sources.b.n ?

Hmm, no - I didn't. I must have had the tarball locally from earlier.

Quoted 4 lines

See:

Location: https://sourceforge.net/publish/ [following] --2017-02-26 14:15:18-- https://sourceforge.net/publish/

That /publish thing is afaik a new change on sourceforge.

Quoted 9 lines

Connecting to sourceforge.net (sourceforge.net)|216.34.181.60|:443... connected. HTTP request sent, awaiting response... 200 OK
Length: 43482 (42K) [text/html]
Saving to: ?/home/thomas/projets/buildroot/output/build/.synergy-1.3.1.tar.gz.LebFB4/output?

/home/thomas/projets/buildroot/output/build/.synergy 100%[===================================================================================================================>] 42,46K 172KB/s in 0,2s

Quoted 7 lines
2017-02-26 14:15:19 (172 KB/s) -

?/home/thomas/projets/buildroot/output/build/.synergy-1.3.1.tar.gz.LebFB4/output? saved [43482/43482]

ERROR: synergy-1.3.1.tar.gz has wrong sha256 hash:
ERROR: expected: 4a15e26456498dda5f3aa7d688a4892e0c50394a2f01685136fbc954bacc9691
ERROR: got : 062a0b0841486719a474df96c42381910dbc75447ce41e4a2059bef36829b259

It isn't just upstream who have uploaded a new tarball. Apparently there are some recent changes to sourceforge that causes it to return a HTML page instead:

file synergy-1.3.1.tar.gz
synergy-1.3.1.tar.gz: HTML document, UTF-8 Unicode text, with very long lines

I checked with another package from sourceforge (c-icap), and that one is OK.

It seems like synergy is no longer hosted on sourceforge. Accessing the project page just redirects to symless.com.

It looks like the tarball on github has identical content (but the tarball doesn't have the same sha256sum), so perhaps we should move to that:

https://github.com/symless/synergy/releases/tag/1.3.1

Signature · 1 line
Bye, Peter Korsgaard