On Mon, Jul 6, 2020 at 7:33 PM Nathan Chancellor <natechancellor(a)gmail.com&gt; wrote: > > On Mon, Jul 06, 2020 at 10:01:07AM -0700, 'Nick Desaulniers' via Clang Built Linux wrote: > > On Mon, Jul 6, 2020 at 12:20 AM kernel test robot <lkp(a)intel.com&gt; wrote: > > > > > > tree: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git queue/5.4 > > > head: bb26a37d5a4bf25b8377c83fa407a219b3822ed1 > > > commit: 5cee9e67596f7d694e94d7def7cd9d65a7599f66 [109/220] i2c: core: check returned size of emulated smbus block read > > > config: s390-randconfig-r024-20200706 (attached as .config) > > > compiler: clang version 11.0.0 (https://github.com/llvm/llvm-project a378c0449507e00e96534ff9ce9034e185425182) > > > reproduce (this is a W=1 build): > > > wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross > > > chmod +x ~/bin/make.cross > > > # install s390 cross compiling tool for clang build > > > # apt-get install binutils-s390x-linux-gnu > > > git checkout 5cee9e67596f7d694e94d7def7cd9d65a7599f66 > > > # save the attached .config to linux build tree > > > COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross ARCH=s390 > > > > > > If you fix the issue, kindly add following tag as appropriate > > > Reported-by: kernel test robot <lkp(a)intel.com&gt; > > > > > > Note: the linux-stable-rc/queue/5.4 HEAD bb26a37d5a4bf25b8377c83fa407a219b3822ed1 builds fine. > > > It only hurts bisectibility. > > Is this note talking about the config in general? Because that config > does not build at all... I see this over and over, even on mainline > (note that this report is initially against 5.4, which does not build > with versions of clang that support asm goto): > > ./arch/s390/include/asm/bitops.h:69:4: error: invalid operand in inline asm: 'oi $0,${1:b}' > "oi %0,%b1\n" > ^ > ./arch/s390/include/asm/bitops.h:90:4: error: invalid operand in inline asm: 'ni $0,${1:b}' > "ni %0,%b1\n" > ^ > > This might be solved with commit c4e5c229b610 ("s390/jump_label: use "i" > constraint for clang"). 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

On Tue, Jul 7, 2020 at 6:29 PM Philip Li <philip.li(a)intel.com&gt; 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?