On 11/27/08 11:48, Christian Lohmaier wrote:
My understanding is that the gcc_linux_powerpc one is enough to fix
both problems, and the "don't strip executables" is not necessary (but
I could be wrong on that)

Yes, forget about my "don't strip executables" stuff again, I would say.

Sorry, misread that - you mention linux ppc :-)

I think linux x64 is not 100% correct either,  since I often get
regcomp failures in testtools (running build often enough helps)
But I guess this is related to gcj/gij
regcomp -register [...]
"error (RuntimeException): [jni_uno bridge error] UNO calling Java
method initialize: non-UNO exception occurred:
java.lang.NullPointerException"
e.g.
http://tinderbox.go-oo.org/gunzip.cgi?tree=DEV300_m35&brief-log=1227726602.22207#err117

Might or might not be a similar problem. However, my understanding always was that on x86 instruction memory synchronization comes for free, and my assumption always was that this is the same for x86-64 (and indeed, bridges::cpp_uno::shared::VtableFactory::flushCode at bridges/source/cpp_uno/gcc3_linux_x86-64/cpp2uno.cxx is a no-op).

-Stephan

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to