nfsd: clear NFSEXP_PNFS flags if not supports pnfs in kernel
Subsystems (2)
PATCH 1/1 nfsd: clear NFSEXP_PNFS flags if not supports pnfs in kernel
nfsd should not show "pnfs" flag in proc file if the filesystem does not support pnfs.
# cat /proc/fs/nfsd/exports # Version 1.1 # Path Client(Flags) # IPs /testfs *(rw,insecure,no_root_squash,sync,wdelay,no_subtree_check,pnfs,fsid=0,uuid=37b07525:dc5e4203:99cc71b0:3df9dfa5,sec=1)
Signed-off-by Kinglong Mee <[email protected]>
--- fs/nfsd/nfs4layouts.c | 2 ++ 1 file changed, 2 insertions(+)
fs/nfsd/nfs4layouts.c +2 -0
@@ -125,6 +125,8 @@ void nfsd4_setup_layout_type(struct svc_export *exp) sb->s_export_op->map_blocks && sb->s_export_op->commit_blocks) exp->ex_layout_type = LAYOUT_BLOCK_VOLUME; + else + exp->ex_flags &= ~NFSEXP_PNFS; } static void
Signature · 1 line
2.3.4
Quoted 2 lines · Kinglong Mee
nfsd should not show "pnfs" flag in proc file if the filesystem does not support pnfs.
Hmm, do we want to silently clear or reject it? Either way the !CONFIG_NFSD_PNFS stub needs to do the same.
Quoted 6 lines · Christoph Hellwig
Quoted 2 lines · Kinglong Mee
nfsd should not show "pnfs" flag in proc file if the filesystem does not support pnfs.
Hmm, do we want to silently clear or reject it? Either way the !CONFIG_NFSD_PNFS stub needs to do the same.
There's a note:
/* quietly ignore unknown words and anything * following. Newer user-space can try to set * new values, then see what the result was. */
Not sure if that's the best approach, but I guess we should stick with it.
--b.