[Bug 260375] NFS server truncates directory cookies to 32-bits
- Reply: bugzilla-noreply_a_freebsd.org: "[Bug 260375] NFS server truncates directory cookies to 32-bits"
- Reply: bugzilla-noreply_a_freebsd.org: "[Bug 260375] NFS server truncates directory cookies to 32-bits"
- Reply: bugzilla-noreply_a_freebsd.org: "[Bug 260375] NFS server truncates directory cookies to 32-bits"
- Reply: bugzilla-noreply_a_freebsd.org: "[Bug 260375] NFS server truncates directory cookies to 32-bits"
- Reply: bugzilla-noreply_a_freebsd.org: "[Bug 260375] NFS server truncates directory cookies to 32-bits"
- Reply: bugzilla-noreply_a_freebsd.org: "[Bug 260375] NFS server truncates directory cookies to 32-bits"
- Reply: bugzilla-noreply_a_freebsd.org: "[Bug 260375] NFS server truncates directory cookies to 32-bits"
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Mon, 13 Dec 2021 03:37:48 UTC
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260375 Bug ID: 260375 Summary: NFS server truncates directory cookies to 32-bits Product: Base System Version: 13.0-RELEASE Hardware: Any OS: Any Status: New Severity: Affects Only Me Priority: --- Component: kern Assignee: bugs@FreeBSD.org Reporter: asomers@FreeBSD.org VOP_READDIR can return a list of cookies that store seek offsets for the directory entries returned in uio. Each cookie is a 64-bit value, matching the 64-bit d_off field in struct dirent. However, The NFS server truncates these cookies to 32-bits. It happens in both nfsrvd_readdir and nfsrvd_readdirplus. The relevant lines are these: *tl++ = 0; *tl = txdr_unsigned(*cookiep); That works just fine for UFS, where dirent offsets correspond to byte offsets within a densely packed array. And it works just fine for ZFS, where dirent offsets seem to come from a 29-bit hash function. But it fails for the FUSE file system I'm developing, where the dirent offset uses all 64 bits. According to RFC-1813, the NFSv3 standard, the cookie3 data type is a uint64 variable. I can't figure out any good reason why the NFS server should truncate it. It's probably just a legacy thing. I'll try to come up with a patch. -- You are receiving this mail because: You are the assignee for the bug.