exports(5) mentions a kernel option KERBNFS (it really exists) but this options is not even mentioned in a comment in LINT. I think it would be a good idea to add a line "options KERBNFS" to LINT with a small comment to what it does as that way it gets a bit more attention and maybe more people make use of it.
Hi while poking around among the PRs I found this PR. The option is called NFSKERB _but_, I found in src/sys/nfs/nfs_socket.c from rev 1.8 that this option is unimplemented and should therefore not be visible in LINT. ==== src/sys/nfs/nfs_socket.c === 1297: #ifdef NFSKERB 1298: XXX 1299: #endif Using this option would make the kernel not compile. Doug, you intorduced the code in question. Do you this this ever will be implemented? If not I think it should be removed along with the reference in exports(5). Thanks Johan K
On Wed, 9 Aug 2000, Johan Karlsson wrote: > > Hi while poking around among the PRs I found this PR. > > The option is called NFSKERB _but_, > > I found in src/sys/nfs/nfs_socket.c from rev 1.8 that > this option is unimplemented and should therefore not be > visible in LINT. > > ==== src/sys/nfs/nfs_socket.c === > 1297: #ifdef NFSKERB > 1298: XXX > 1299: #endif > > Using this option would make the kernel not compile. > > > > Doug, you intorduced the code in question. > Do you this this ever will be implemented? > > If not I think it should be removed along with the reference > in exports(5). That is a marker to point out code which was removed to allow the code in question to be exported. It shouldn't be removed (unless all of the NFSKERB code is being removed) since it needs to be replaced with an appropriate call to an encryption routine. -- Doug Rabson Mail: dfr@nlsystems.com Nonlinear Systems Ltd. Phone: +44 20 8348 3944
At Thu, 10 Aug 2000 09:24:57 BST, Doug Rabson wrote: > On Wed, 9 Aug 2000, Johan Karlsson wrote: > > ==== src/sys/nfs/nfs_socket.c === > > 1297: #ifdef NFSKERB > > 1298: XXX > > 1299: #endif > > > > Using this option would make the kernel not compile. > > > > If not I think it should be removed along with the reference > > in exports(5). > > That is a marker to point out code which was removed to allow the code in > question to be exported. It shouldn't be removed (unless all of the > NFSKERB code is being removed) since it needs to be replaced with an > appropriate call to an encryption routine. > Ok, thanks for the explanation. What do you think about removing the reference in exports(5) or at least adding something like "Using this option will make the kernel not compile, if one do not also take care of adding calls to the appropriate kerberos encryption routins in the NFS source." /Johan K
On Thu, 10 Aug 2000, Johan Karlsson wrote: > At Thu, 10 Aug 2000 09:24:57 BST, Doug Rabson wrote: > > On Wed, 9 Aug 2000, Johan Karlsson wrote: > > > ==== src/sys/nfs/nfs_socket.c === > > > 1297: #ifdef NFSKERB > > > 1298: XXX > > > 1299: #endif > > > > > > Using this option would make the kernel not compile. > > > > > > If not I think it should be removed along with the reference > > > in exports(5). > > > > That is a marker to point out code which was removed to allow the code in > > question to be exported. It shouldn't be removed (unless all of the > > NFSKERB code is being removed) since it needs to be replaced with an > > appropriate call to an encryption routine. > > > > Ok, thanks for the explanation. > > What do you think about removing the reference in exports(5) or at least > adding something like > > "Using this option will make the kernel not compile, if one do not also > take care of adding calls to the appropriate kerberos encryption routins > in the NFS source." That sounds about right. -- Doug Rabson Mail: dfr@nlsystems.com Nonlinear Systems Ltd. Phone: +44 20 8348 3944
On Fri, 11 Aug 2000 11:12:05 +0200, Johan Karlsson wrote: > +Using this option will make the kernel not compile, if one do not also > +take care of adding calls to the appropriate Kerberos encryption routines > +in the NFS source. Rather this, which addresses the grammar and the line breaking style: The use of this option will prevent the kernel from compiling unless calls to the appropriate Kerberos encryption routines are provided in the NFS source. I wonder whether it'd be a good idea to provide paths to the affected source files? Ciao, Sheldon.
At Fri, 11 Aug 2000 12:15:17 +0200, Sheldon Hearn wrote: > I wonder whether it'd be a good idea to provide paths to the affected > source files? The files using the option are: sys/nfs/nfs_nqlease.c sys/nfs/nfs_socket.c sys/nfs/nfs_syscalls.c sys/nfs/rpcv2.h sbin/mount_nfs/mount_nfs.c sbin/nfsd/nfsd.c where implementation is missing in: sys/nfs/nfs_socket.c sys/nfs/nfs_syscalls.c sys/nfs/rpcv2.h However, since sbin/munt_nfs and sbin/nfsd uses the option they will also need to be recompiled. /Johan K
On Fri, 11 Aug 2000 04:40:03 MST, Johan Karlsson wrote: > However, since sbin/munt_nfs and sbin/nfsd uses the option > they will also need to be recompiled. Hmmm, probably just to use the generic text, then. Ciao, Sheldon.
Responsible Changed From-To: freebsd-bugs->sheldonh I'll take this one.
State Changed From-To: open->closed Explanation added in both HEAD and RELENG_4.