On Wed, Feb 19, 2020 at 1:45 AM Leo Yan <leo....@linaro.org> wrote: > > Hi Khem, > > On Tue, Feb 18, 2020 at 03:27:26PM +0800, Leo Yan wrote: > > Hi Khem, > > > > On Mon, Feb 17, 2020 at 10:30:46PM -0800, Khem Raj wrote: > > > > [...] > > > > > > +DEPENDS += "bison-native \ > > > > + ninja-native \ > > > > + elfutils-native \ > > > > + flex-native \ > > > > + ${MLPREFIX}elfutils \ > > > > + ${MLPREFIX}binutils \ > > > > + ${MLPREFIX}flex \ > > > > + luajit \ > > > > + clang \ > > > > + " > > > > + > > > > > > this will add meta-clang dependency directly into meta-oe, which is not > > > desired > > > is it possible to make it optional or perhaps marked when meta-clang > > > is not in BBLAYERS > > > > Thanks a lot for reviewing. clang is mandotory for BCC to compile > > eBPF programs, so I think we cannot build clang as optional for BCC. > > > > For marking the recipe when meta-clang is not in BBLAYERS, I searched > > with google but seems 'LAYERDEPENDS' and 'BBMASK' both are not suitable > > for this case. Could you give more detailed info for how to do this? > > Just want to ask a following question. > > Since BCC and bpftrace both are heavily relying on LLVM/Clang for their > tracing, do you think it's a good idea to merge BCC and bpftrace into > meta-clang layer, e.g. place into /meta-clang/recipes-devtools/? > If so, we can easily resolve the dependency issue and avoid to introduce > mess between meta-oe and meta-clang. > > For the tool ply, it's simple and doesn't depend on Clang. So should > keep it into meta-oe/recipes-devtools/ folder? Also welcome any > suggestion for this.
Yeah I think thats a fine suggestion. meta-clang is good for it. > > Thanks, > Leo Yan -- _______________________________________________ Openembedded-devel mailing list Openembedded-devel@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-devel