On Fri, 01 Oct 2021, Jason Gunthorpe <[email protected]> wrote: > On Fri, Oct 01, 2021 at 02:04:04PM +0300, Jani Nikula wrote: >> On Fri, 27 Aug 2021, Jason Gunthorpe <[email protected]> wrote: >> > On Fri, Aug 27, 2021 at 03:12:36PM +0000, kernel test robot wrote: >> >> tree: https://github.com/awilliam/linux-vfio.git next >> >> head: ea870730d83fc13a5fa2bd0e175176d7ac8a400a >> >> commit: 343b7258687ecfbb363bfda8833a7cf641aac524 [33/38] PCI: Add >> >> 'override_only' field to struct pci_device_id >> >> config: i386-randconfig-a004-20210827 (attached as .config) >> >> compiler: clang version 14.0.0 (https://github.com/llvm/llvm-project >> >> 1076082a0d97bd5c16a25ee7cf3dbb6ee4b5a9fe) >> >> 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 >> >> # >> >> https://github.com/awilliam/linux-vfio/commit/343b7258687ecfbb363bfda8833a7cf641aac524 >> >> git remote add vfio https://github.com/awilliam/linux-vfio.git >> >> git fetch --no-tags vfio next >> >> git checkout 343b7258687ecfbb363bfda8833a7cf641aac524 >> >> # save the attached .config to linux build tree >> >> COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross >> >> ARCH=i386 >> >> >> >> If you fix the issue, kindly add following tag as appropriate >> >> Reported-by: kernel test robot <[email protected]> >> > >> > Ugh, this is due to this code: >> > >> > #define INTEL_VGA_DEVICE(id, info) { \ >> > 0x8086, id, \ >> > ~0, ~0, \ >> > 0x030000, 0xff0000, \ >> > (unsigned long) info } >> > >> > #define INTEL_QUANTA_VGA_DEVICE(info) { \ >> > 0x8086, 0x16a, \ >> > 0x152d, 0x8990, \ >> > 0x030000, 0xff0000, \ >> > (unsigned long) info } >> > >> > >> > Which really should be using the normal pattern for defining these >> > structs: >> > >> > #define PCI_DEVICE_CLASS(dev_class,dev_class_mask) \ >> > .class = (dev_class), .class_mask = (dev_class_mask), \ >> > .vendor = PCI_ANY_ID, .device = PCI_ANY_ID, \ >> > .subvendor = PCI_ANY_ID, .subdevice = PCI_ANY_ID >> > >> > The warning is also not a real issue, just clang being overzealous. >> >> Stumbled upon this old report, sorry for the delayed response. >> >> The reason it's not using designated initializers is that the same file >> gets synced to some userspace projects (at least libdrm and >> igt-gpu-tools) which use the macros to initialize slightly different >> structs. For example, igt uses struct pci_id_match from libpciaccess-dev >> (/usr/include/pciaccess.h) and can't easily adapt to different member >> names. > > Do it like this: > > > #ifdef __KERNEL__ > #define INTEL_VGA_DEVICE(..) > #endif > > > And userspace does > > #define INTEL_VGA_DEVICE(..) > #include <foo.h>
Sure. >> Anyway, we've got >> >> subdir-ccflags-y += $(call cc-disable-warning, missing-field-initializers) >> subdir-ccflags-y += $(call cc-disable-warning, initializer-overrides) >> >> in drivers/gpu/drm/i915/Makefile, so I wonder why they're not respected. > > Disabling kernel warnings because some userspace wants to copy a > kernel header is horrific, don't do that. We've disabled some warnings because those lines are preceded by subdir-ccflags-y := -Wall -Wextra enabling more warnings than the kernel build generally does. BR, Jani. -- Jani Nikula, Intel Open Source Graphics Center
