On 4/8/25 4:45 PM, Eric Biggers wrote: > On Wed, Apr 02, 2025 at 09:29:41AM -0700, Eric Biggers wrote: >> On Wed, Apr 02, 2025 at 09:51:05AM -0400, Chuck Lever wrote: >>> On 4/1/25 6:02 PM, Eric Biggers wrote: >>>> From: Eric Biggers <ebiggers@xxxxxxxxxx> >>>> >>>> nfs.ko, nfsd.ko, and lockd.ko all use crc32_le(), which is available >>>> only when CONFIG_CRC32 is enabled. But the only NFS kconfig option that >>>> selected CONFIG_CRC32 was CONFIG_NFS_DEBUG, which is client-specific and >>>> did not actually guard the use of crc32_le() even on the client. >>>> >>>> The code worked around this bug by only actually calling crc32_le() when >>>> CONFIG_CRC32 is built-in, instead hard-coding '0' in other cases. This >>>> avoided randconfig build errors, and in real kernels the fallback code >>>> was unlikely to be reached since CONFIG_CRC32 is 'default y'. But, this >>>> really needs to just be done properly, especially now that I'm planning >>>> to update CONFIG_CRC32 to not be 'default y'. >>> >>> It's interesting that no-one has noticed this before. dprintk is not the >>> only consumer of the FH hash function: NFS/NFSD trace points also use >>> it. >>> >>> Eric, assuming you would like to carry this patch forward instead of us >>> taking it through one of the NFS client or server trees: >>> >>> Acked-by: Chuck Lever <chuck.lever@xxxxxxxxxx> >>> >>> for the hunks related to nfsd and lockd. >> >> Please go ahead and take it through one of the NFS trees. Thanks! >> > > I ended up sending in the removal of 'default y' from CONFIG_CRC32 for 6.15. So > I recommend that you send this NFS patch in for 6.15 as well, as it's now > slightly more likely that people can end up with CONFIG_CRC32 disabled (though > many other parts of the kernel select it anyway, so it still tends to be > enabled). I've moved the patch to nfsd-next, which has exposure to community automated testing robots. If no problems are reported, I will move the patch to nfsd-fixes and push it to v6.15-rc in a couple of weeks. -- Chuck Lever
