September 2007 - openSUSE Programming - openSUSE Mailing Lists

[opensuse-programming] pthread cleanup handler not called on pthread_cancel and sometimes pthread_exit
by K.R. Foley 17 Sep '07

17 Sep '07

Hi, Since going to opensuse 10.2 we have found a problem with pthread cleanup handlers not being called on pthread_cancel and under certain conditions on pthread_exit from within C++ programs. The issue is that cleanup handlers put onto a threads cleanup stack via pthread_cleanup_push() should be called when a thread is canceled or exits. In some cases that is not happening. This only appears to be an issue with C++ programs that have exceptions enabled (default). It doesn't happen with C programs or if exceptions are disabled by using -fno-exceptions. This is not a viable solution. The issue is well documented as a glibc bug here: http://sourceware.org/bugzilla/show_bug.cgi?id=4123, so I won't repeat all of the details here. Unfortunately, this is a critical bug and it has been sitting open since March 2 without any visible activity other than my recent activity. Does anyone have suggestions about how to possibly work around this issue or to help move things along? Thanks, -- kr --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-programming+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-programming+help(a)opensuse.org

2 3

[opensuse-programming] RPM spec macro for distribution
by Morten Bjørnsvik 12 Sep '07

12 Sep '07

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---------------------------------------------------------------------To unsubscribe, e-mail: opensuse-programming+unsubscribe(a)opensuse.orgFor additional commands, e-mail: opensuse-programming+help(a)opensuse.org

3 2

[opensuse-programming] how to stop instrumentation of inlined fucntions [gcc -finstrumentation-functions]
by Alok Ranjan Singh 12 Sep '07

12 Sep '07

Hi Folks, While using '-finstrumentation-functions' of gcc, the inlined functions are also instrumented automatically. I understand that it has been done purposely to get trace data for inlined functions as well and to not lose data about them. But what I want is different as I am generating a call graph in which I don't want the inlined functions to figure if they have been inlined. Is there any flag available to turn off the feature of instrumenting the inlined functions? (Note that adding 'no_instrument_function' won't server the purpose as it will need modification of each and every such inlinable function in code.) Or can anybody think of any workaround for preventing their instrumentation? Thanks, Alok --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-programming+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-programming+help(a)opensuse.org

1 0