On Tue, Nov 10, 2020 at 4:42 AM Jean-Philippe Brucker <jean-philippe@xxxxxxxxxx> wrote: > > On Mon, Nov 09, 2020 at 10:11:52AM -0800, Andrii Nakryiko wrote: > > On Mon, Nov 9, 2020 at 3:11 AM Jean-Philippe Brucker > > <jean-philippe@xxxxxxxxxx> wrote: > > > > > > A few fixes for cross and out-of-tree build of bpftool and runqslower. > > > These changes allow to build for different target architectures, using > > > the same source tree. > > > > > > I sent [v1] ages ago but haven't found time to resend. No change except > > > rebasing on the latest bpf-next/master. > > > > > > [v1] https://lore.kernel.org/bpf/20200827153629.3820891-1-jean-philippe@xxxxxxxxxx/ > > > > > > > While you are looking at bpftool builds... Seems like it regressed > > recently and doesn't honor -jX setting. Either way the build is > > sequential (and rather slow). Do you mind checking if your changes > > could fix the regression (I haven't had a chance to bisect the > > offending change causing regression). > > I bisected it to ba2fd563b740 ("tools/bpftool: Support passing > BPFTOOL_VERSION to make"), in v5.9. As BPFTOOL_VERSION became a recursivly > expanded variable, the shell function is evaluated on every expansion of > CFLAGS. I'll add the fix to v3: > > -BPFTOOL_VERSION ?= $(shell make -rR --no-print-directory -sC ../../.. kernelversion) > +ifeq ($(BPFTOOL_VERSION),) > +BPFTOOL_VERSION := $(shell make -rR --no-print-directory -sC ../../.. kernelversion) > +endif > Great, thanks! > Thanks, > Jean
