Hello guys,
I was looking into our failures and kdenetwork3 fails for quite time with OOM.
Possibly it is caused by gcc4.8. But I have not enough C++ foo to figure out how to fix this. We need it rolling so we can still keep kde3 stuff in 13.1 so anyone feels like sorting it out?
Cheers
Tom
On Mon, 30 Sep 2013, Tom?? Chv?tal wrote:
Hello guys,
I was looking into our failures and kdenetwork3 fails for quite time with OOM.
Possibly it is caused by gcc4.8. But I have not enough C++ foo to figure out how to fix this. We need it rolling so we can still keep kde3 stuff in 13.1 so anyone feels like sorting it out?
Does adding -fno-var-tracking fix the build?
Richard.
Dne Po 30. září 2013 14:41:54, Richard Biener napsal(a):
On Mon, 30 Sep 2013, Tom?? Chv?tal wrote:
Hello guys,
I was looking into our failures and kdenetwork3 fails for quite time with OOM.
Possibly it is caused by gcc4.8. But I have not enough C++ foo to figure out how to fix this. We need it rolling so we can still keep kde3 stuff in 13.1 so anyone feels like sorting it out?
Does adding -fno-var-tracking fix the build?
Richard.
Still the same [1].
Tom
[1] https://build.opensuse.org/package/live_build_log/home:scarabeus_iv/kdenetwo...
On Mon, 30 Sep 2013, Tom?? Chv?tal wrote:
Dne Po 30. z??? 2013 14:41:54, Richard Biener napsal(a):
On Mon, 30 Sep 2013, Tom?? Chv?tal wrote:
Hello guys,
I was looking into our failures and kdenetwork3 fails for quite time with OOM.
Possibly it is caused by gcc4.8. But I have not enough C++ foo to figure out how to fix this. We need it rolling so we can still keep kde3 stuff in 13.1 so anyone feels like sorting it out?
Does adding -fno-var-tracking fix the build?
Richard.
Still the same [1].
Tom
[1] https://build.opensuse.org/package/live_build_log/home:scarabeus_iv/kdenetwo...
Seems to be preprocessing of kcm_kcmsambaconf_la.all_cpp.cpp that fails, tracked down to preprocessing of kcminterface.cpp. It works for me with adding -w, testcase:
pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) );
so it's the control sequence that confuses GCC.
Richard.
On Mon, 30 Sep 2013, Richard Biener wrote:
On Mon, 30 Sep 2013, Tom?? Chv?tal wrote:
Dne Po 30. z??? 2013 14:41:54, Richard Biener napsal(a):
On Mon, 30 Sep 2013, Tom?? Chv?tal wrote:
Hello guys,
I was looking into our failures and kdenetwork3 fails for quite time with OOM.
Possibly it is caused by gcc4.8. But I have not enough C++ foo to figure out how to fix this. We need it rolling so we can still keep kde3 stuff in 13.1 so anyone feels like sorting it out?
Does adding -fno-var-tracking fix the build?
Richard.
Still the same [1].
Tom
[1] https://build.opensuse.org/package/live_build_log/home:scarabeus_iv/kdenetwo...
Seems to be preprocessing of kcm_kcmsambaconf_la.all_cpp.cpp that fails, tracked down to preprocessing of kcminterface.cpp. It works for me with adding -w, testcase:
pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) );so it's the control sequence that confuses GCC.
Which are actually literal zero-bytes.
Richard.
On Monday 30 September 2013 18:18:53 Richard Biener wrote:
so it's the control sequence that confuses GCC.
Which are actually literal zero-bytes.
I have added the option -fno-diagnostics-show-caret and now it builds well.
It is really a GCC bug: this option should not affect the build result.
On Mon, 30 Sep 2013, Richard Biener wrote:
On Mon, 30 Sep 2013, Tom?? Chv?tal wrote:
Dne Po 30. z??? 2013 14:41:54, Richard Biener napsal(a):
On Mon, 30 Sep 2013, Tom?? Chv?tal wrote:
Hello guys,
I was looking into our failures and kdenetwork3 fails for quite time with OOM.
Possibly it is caused by gcc4.8. But I have not enough C++ foo to figure out how to fix this. We need it rolling so we can still keep kde3 stuff in 13.1 so anyone feels like sorting it out?
Does adding -fno-var-tracking fix the build?
Richard.
Still the same [1].
Tom
[1] https://build.opensuse.org/package/live_build_log/home:scarabeus_iv/kdenetwo...
Seems to be preprocessing of kcm_kcmsambaconf_la.all_cpp.cpp that fails, tracked down to preprocessing of kcminterface.cpp. It works for me with adding -w, testcase:
pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) ); pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+^@" ) ) );so it's the control sequence that confuses GCC.
That is, change the above to
pamPasswordChangeChk->setAccel( QKeySequence( tr2i18n( "Alt+\0" ) ) );
and all other occurances that trigger a warning like
test.cpp:5:1: warning: null character(s) preserved in literal [enabled by default]
Richard.
On Monday 30 September 2013 15:24:23 Tomáš Chvátal wrote:
I was looking into our failures and kdenetwork3 fails for quite time with OOM.
Possibly it is caused by gcc4.8. But I have not enough C++ foo to figure out how to fix this. We need it rolling so we can still keep kde3 stuff in 13.1 so anyone feels like sorting it out?
I made a bug report about this issue a while ago:
https://bugzilla.novell.com/show_bug.cgi?id=834155
On Monday 30 September 2013 15:24:23 Tomáš Chvátal wrote:
I was looking into our failures and kdenetwork3 fails for quite time with OOM.
Possibly it is caused by gcc4.8. But I have not enough C++ foo to figure out how to fix this. We need it rolling so we can still keep kde3 stuff in 13.1 so anyone feels like sorting it out?
If nothing else helps, I think we can use an older GCC compliner version when building for 32-bit architecture.
I still hope the origin of the problem can be tracked.
-
Ilya Chernykh -
Richard Biener -
Tomáš Chvátal