On (31/12/09 16:56), Guo Hongruan wrote: > I am afraid that the failure is not so easy. > > Today, I recompile the following builds: > MACHINE=qemux86 DISTRO=micro-uclibc > MACHINE=qemuarm DISTRO=micro-uclibc > MACHINE=qemumips DISTRO=micro-uclibc > > only qemuarm works. But the problem is that the uclibc configuration > in qemuarm building has not defined UCLIBC_HAS_FENV, and the > configuration options of gcc-cross-intermedia lacks > --disable-decimal-float. So according the previous analysis, the > qemuarm building should failed and reported the same reason. > > Why did qemuarm build pass? I can not figure it out.
decimal float is enabled by default for powerpc, x86 and x86_64 in gcc thats why you see that it works ok on arm because its disabled by default for arm. So yes for uclibc on these architectures you need to disable it because uclibc does not have decimal floating library implementation. Thats why in my patch in previous email I disabled it for all uclibc based builds of gcc. It should be disabled for mips as the code in gcc seems like. If you could post your gcc build logs somewhere I can have a look. -Khem _______________________________________________ Openembedded-devel mailing list Openembedded-devel@lists.openembedded.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel