4:13 a.m.
On Tue, Jul 7, 2020 at 6:29 PM Philip Li <philip.li(a)intel.com> wrote:
On Tue, Jul 07, 2020 at 10:01:12AM -0700, Nick Desaulniers wrote:
> Looks like c4e5c229b610, should we ask stable to backport to 5.4? I
> don't think we had clang builds working for 5.4 for s390, so maybe we
> should just Philip to disable Clang builds of this stable branch for
> s390? (There may not be too many backports to support, but I don't
Got it, we will disable s390 for the linux-stable-rc.
Yeah, it's hard for s390 and probably riscv support on stable as we
only got those working on mainline recently.
BTW: does the bot need switch to use clang's as instead of
gnu's?
No; in the process of whipping LLVM's binutils substitutes into shape,
we're doing pretty well, except for clang's integrated assembler (IA).
Ironically, right now we can only build risc-v with clang's IA.
I would recommend `make LLVM=1` for most cases, though `make CC=clang
LD=ld.lld ...` may be all that we can get away with for testing stable
tree's branches.
Will you, Rong, and the rest of the 0day bot team be attending Linux
Plumbers Conference this year? I would like to put together a session
on "CI and Clang" with the kernelci folks and tuxbuild folks to go
over our current compatibility table, which is a little complex ATM.
If you're attending the conference, would you be interested in
attending such a session?
--
Thanks,
~Nick Desaulniers