Hi, On 08/10/17 14:20, Dr. Tobias Quathamer wrote: > the latest upload of dh-make-golang FTBFS on mips and mipsel. This is > due to the "mips-linux-gnu-gccgo-7: waitid: bad address" kernel bug, see > https://bugs.debian.org/867358. The bug seems to have been fixed, but > the fix has not yet reached the buildd-network. > > gb dh-make-golang_0.0~git20171007.f9dfad6-1 . mips mipsel
Given back. > Also, I'd like to get a feedback if you (as wb-team) agree to give back > those packages, maybe even a few times in a row. An alternative approach > to circumvent this bug is to try several builds on mips directly in the > d/rules file, like this: > > # Build/test in single thread and allow retries on mips and mipsel > # to try to get over the "mips-linux-gnu-gccgo-7: waitid: bad address" > # error. See https://bugs.debian.org/867358 > ifeq ($(findstring mips,$(DEB_BUILD_ARCH)),mips) > override_dh_auto_build: > dh_auto_build -O--buildsystem=golang -O--no-parallel || \ > dh_auto_build -O--buildsystem=golang -O--no-parallel || \ > dh_auto_build -O--buildsystem=golang -O--no-parallel || \ > dh_auto_build -O--buildsystem=golang -O--no-parallel || \ > dh_auto_build -O--buildsystem=golang -O--no-parallel || \ > dh_auto_build -O--buildsystem=golang -O--no-parallel || \ > dh_auto_build -O--buildsystem=golang -O--no-parallel > > override_dh_auto_test: > dh_auto_test -O--buildsystem=golang -O--no-parallel || \ > dh_auto_test -O--buildsystem=golang -O--no-parallel || \ > dh_auto_test -O--buildsystem=golang -O--no-parallel || \ > dh_auto_test -O--buildsystem=golang -O--no-parallel || \ > dh_auto_test -O--buildsystem=golang -O--no-parallel || \ > dh_auto_test -O--buildsystem=golang -O--no-parallel || \ > dh_auto_test -O--buildsystem=golang -O--no-parallel > endif > > However, I'd rather like to avoid this "hack", because it's no longer > needed when the autobuilders run the fixed kernel. [Not speaking for the wb-team here] I'm not a fan of adding hacks like this to packages to workaround bugs elsewhere. I am happy to give back the go packages affected by this bug multiple times if necessary, although I might need prodding occasionally. The kernel bug which causes this is now fixed in unstable, however unfortunately it missed the recent stretch point release. It's fixed in the linux git packaging repo so should be available at least by the next update. Thanks, James
signature.asc
Description: OpenPGP digital signature
