Linus Torvalds writes: (Summary) wrote:
This really seems idiotic.
This really seems idiotic.
Let me rephrase this commit message so that you can see why I think it's wrong: Let me rephrase this commit message so that you can see why I think it's wrong: "NFSd uses a single hash chain for all dentries, which can cause horrible lock contention, in ways that the normal hashing does not. In pretty much all cases, we could just hash the damn dentry, much all cases, we could just hash the damn dentry, The only reason for actually having s_anon seems to be that we want some per-superblock list of these unconnected dentries for shrink_dcache_for_umount().
shrink_dcache_for_umount().
Everything else would actually be *much* happier with just having the dentry on the regular hash table.
[...]
in soft-lockup warnings.This really seems idiotic.
This really seems idiotic.
Let me rephrase this commit message so that you can see why I think it's wrong: Let me rephrase this commit message so that you can see why I think it's wrong: "NFSd uses a single hash chain for all dentries, which can cause horrible lock contention, in ways that the normal hashing does not. In pretty much all cases, we could just hash the damn dentry, much all cases, we could just hash the damn dentry, The only reason for actually having s_anon seems to be that we want some per-superblock list of these unconnected dentries for shrink_dcache_for_umount().
shrink_dcache_for_umount().
Everything else would actually be *much* happier with just having the dentry on the regular hash table.