Le 29/04/2014 21:36, Pierre Labastie a écrit : > Le 29/04/2014 18:57, Bruce Dubbs a écrit : >> Fernando de Oliveira wrote: >>> Each time, I remember a recent post, I think is from Armin. >>> >>> Problem solved, but reporting,so perhaps a better solution can be given >>> (as Bruce recently did). >>> >>> It is always something to do with sse, gcc, xmmintrin.h and now also >>> emmintrin.h, if I remove xmmintrin.h. So, removed from configure and >>> everything built and first tests ran correctly. First, simple patch I >>> built to fix, after that, error log. >>> >>> If there are no objections, I will submit this patch later and fix the >>> book (for gcc-4.9.0 and i686, unless someone can confirm it is also in >>> x86_64). >>> >>> Should I keep posting these problems here even if I solve them? >> >> Posting solutions to problems in general terms doesn't hurt. It provides >> some >> context for others searching for the same problem. I would just mention the >> patch, but not always include it. It is available (or will be) in the >> patches >> repo. The log is probably not necessary. >> >> On the other hand, posting the patch earlier today led to a simpler solution, >> so the situation may vary. >> >> -- Bruce > Another solution to the kind of problems reported by Fernado could be to set > -msse2 for i686. > > For example, the following sed allows lame to build on a 32 bit (virtual) > machine: > sed -i -e "s/fast-math/& -msse2/" configure > > On x86_64, -msse2 is the default, while it is not for i?86. > > I think sse2 is pretty common those days, so the sed above should do no harm. > It is better IMHO to keep the speed of sse2, rather than removing totally the > XMM intrinsics. > > I'll test gst-plugins soon. >
The following sed allows to build gst-plugins-base-1.2.4: sed -i -e "s/-O2/& -msse2/" configure Pierre -- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
