On Sun, Jul 26, 2020 at 06:03:58PM +0200, Christoph Hellwig wrote: > Always use <linux/compat.h> so that we can move more declarations to > common code. In two of the three cases the asm include was in addition > to an existing one for <linux/compat.h> anyway. > > Signed-off-by: Christoph Hellwig <hch@xxxxxx> > --- > arch/arm64/include/asm/stat.h | 2 +- > arch/arm64/kernel/process.c | 1 - > arch/arm64/kernel/ptrace.c | 1 - > 3 files changed, 1 insertion(+), 3 deletions(-) > > diff --git a/arch/arm64/include/asm/stat.h b/arch/arm64/include/asm/stat.h > index 3b4a62f5aeb0c3..1b5ac1ef5d04cc 100644 > --- a/arch/arm64/include/asm/stat.h > +++ b/arch/arm64/include/asm/stat.h > @@ -10,7 +10,7 @@ > #ifdef CONFIG_COMPAT > > #include <linux/time.h> > -#include <asm/compat.h> > +#include <linux/compat.h> This breaks arm64 defconfig: $ make -skj"$(nproc)" ARCH=arm64 CROSS_COMPILE=aarch64-linux- distclean defconfig init/main.o In file included from ./include/linux/compat.h:17, from ./arch/arm64/include/asm/stat.h:13, from ./include/linux/stat.h:6, from ./include/linux/sysfs.h:22, from ./include/linux/kobject.h:20, from ./include/linux/of.h:17, from ./include/linux/irqdomain.h:35, from ./include/linux/acpi.h:13, from ./include/acpi/apei.h:9, from ./include/acpi/ghes.h:5, from ./include/linux/arm_sdei.h:8, from arch/arm64/kernel/asm-offsets.c:10: ./include/linux/fs.h: In function 'vfs_whiteout': ./include/linux/fs.h:1736:32: error: 'S_IFCHR' undeclared (first use in this function) 1736 | return vfs_mknod(dir, dentry, S_IFCHR | WHITEOUT_MODE, WHITEOUT_DEV); | ^~~~~~~ ./include/linux/fs.h:1736:32: note: each undeclared identifier is reported only once for each function it appears in ./include/linux/fs.h: At top level: ./include/linux/fs.h:1886:46: warning: 'struct kstat' declared inside parameter list will not be visible outside of this definition or declaration 1886 | int (*getattr) (const struct path *, struct kstat *, u32, unsigned int); | ^~~~~ ./include/linux/fs.h: In function '__mandatory_lock': ./include/linux/fs.h:2372:25: error: 'S_ISGID' undeclared (first use in this function); did you mean 'SIGIO'? 2372 | return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; | ^~~~~~~ | SIGIO ./include/linux/fs.h:2372:35: error: 'S_IXGRP' undeclared (first use in this function) 2372 | return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; | ^~~~~~~ ... $ git bisect log # bad: [7b287a5c6ac518c415a258f2aa7b1ebb25c263d2] Add linux-next specific files for 20200730 # good: [d3590ebf6f91350192737dd1d1b219c05277f067] Merge tag 'audit-pr-20200729' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/audit git bisect start '7b287a5c6ac518c415a258f2aa7b1ebb25c263d2' 'd3590ebf6f91350192737dd1d1b219c05277f067' # bad: [1f1ed12be70e9eb4e05ac206c6ad6a5a31f5b921] Merge remote-tracking branch 'crypto/master' git bisect bad 1f1ed12be70e9eb4e05ac206c6ad6a5a31f5b921 # bad: [07fad673c2f1a02440c879c34f8182b12786a735] Merge remote-tracking branch 'hid/for-next' git bisect bad 07fad673c2f1a02440c879c34f8182b12786a735 # good: [7a77c92312546a74d3507484b256ae17bfb2cfe2] Merge remote-tracking branch 'm68knommu/for-next' git bisect good 7a77c92312546a74d3507484b256ae17bfb2cfe2 # good: [40dd62f180e38317e744d7f82c98af31a24fd2c9] Merge remote-tracking branch 'f2fs/dev' git bisect good 40dd62f180e38317e744d7f82c98af31a24fd2c9 # bad: [52138dfdd2192bcfc7d3bc2e79475966ee4b20c4] Merge remote-tracking branch 'printk/for-next' git bisect bad 52138dfdd2192bcfc7d3bc2e79475966ee4b20c4 # good: [a37c3e37fa3fa1381e03d918d708f82927ddd160] Merge remote-tracking branch 'xfs/for-next' git bisect good a37c3e37fa3fa1381e03d918d708f82927ddd160 # good: [4e523547e2bf755d40cb10e85795c2f9620ff3fb] nvme-pci: add a blank line after declarations git bisect good 4e523547e2bf755d40cb10e85795c2f9620ff3fb # bad: [5066741180729f7bad9401de34efda3766c3274a] Merge branches 'fixes' and 'work.quota-compat' into for-next git bisect bad 5066741180729f7bad9401de34efda3766c3274a # good: [4ff8a356daafaafbf90141ee7a3b8fdc18e560a8] ia64: switch to ->regset_get() git bisect good 4ff8a356daafaafbf90141ee7a3b8fdc18e560a8 # good: [ce327e1c54119179066d6f3573a28001febc9265] regset: kill user_regset_copyout{,_zero}() git bisect good ce327e1c54119179066d6f3573a28001febc9265 # good: [1697a322e28ba96d35953c5d824540d172546d36] [elf-fdpic] switch coredump to regsets git bisect good 1697a322e28ba96d35953c5d824540d172546d36 # good: [259bf01c1bd1f049958496a089c4f334fe0c8a48] Merge branches 'work.misc', 'work.regset' and 'work.fdpic' into for-next git bisect good 259bf01c1bd1f049958496a089c4f334fe0c8a48 # bad: [0a3a4497a1de8e68e809a693b549c7ec2f195301] compat: lift compat_s64 and compat_u64 to <linux/compat.h> git bisect bad 0a3a4497a1de8e68e809a693b549c7ec2f195301 # bad: [b902bfb3f0e9d07ec9f48256e57e5c5de6108f8c] arm64: stop using <asm/compat.h> directly git bisect bad b902bfb3f0e9d07ec9f48256e57e5c5de6108f8c # first bad commit: [b902bfb3f0e9d07ec9f48256e57e5c5de6108f8c] arm64: stop using <asm/compat.h> directly I assume the stat header order should be messed around with but I am not sure what exactly that would entail to make sure that nothing else breaks, hence just the report. Cheers, Nathan