On 6/9/19 7:57 PM, David Miller wrote: > From: Matteo Croce <mcroce@xxxxxxxxxx> > Date: Sat, 8 Jun 2019 14:50:19 +0200 > >> MPLS routing code relies on sysctl to work, so let it select PROC_SYSCTL. >> >> Reported-by: Randy Dunlap <rdunlap@xxxxxxxxxxxxx> >> Suggested-by: David Ahern <dsahern@xxxxxxxxx> >> Signed-off-by: Matteo Croce <mcroce@xxxxxxxxxx> > > Applied, thanks. > This patch causes build errors when # CONFIG_PROC_FS is not set because PROC_SYSCTL depends on PROC_FS. The build errors are not in fs/proc/ but in other places in the kernel that never expect to see PROC_FS not set but PROC_SYSCTL=y. I see the following 2 build errors: ../kernel/sysctl_binary.c: In function 'binary_sysctl': ../kernel/sysctl_binary.c:1305:37: error: 'struct pid_namespace' has no member named 'proc_mnt'; did you mean 'proc_work'? mnt = task_active_pid_ns(current)->proc_mnt; ^~~~~~~~ ../fs/xfs/xfs_sysctl.c:80:19: error: 'xfs_panic_mask_proc_handler' undeclared here (not in a function); did you mean 'xfs_panic_mask'? .proc_handler = xfs_panic_mask_proc_handler, ^~~~~~~~~~~~~~~~~~~~~~~~~~~ The patch's line: + select PROC_SYSCTL should not be done unless PROC_FS is enabled, e.g.: select PROC_SYSCTL if PROC_FS but that still doesn't help the mpls driver operate as it should. The patch should have been depends on PROC_SYSCTL As it stands now (in linux-next), this patch should be reverted IMO. -- ~Randy