Subject: Re: README: lchmod(2), lchown(2) and lutimes(2) is added, and now readlink(2) checks read permission of link.
To: None <current-users@NetBSD.ORG>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: current-users
Date: 10/09/1997 14:29:50
>> Hmm? What does ufs have to do with it? When I did this, I put it
>> in kern/vfs_syscalls.c, and whether the underlying filesystem is ufs
>> or not has nothing to do with it.
> Does filesystem other than local unix filesystem actually need it?
Well, putting it in the ufs layer has implications for disklesses: it
means that the semantics of symlink permissions are those of the server
OS, not those of the client OS.
> # btw, if vfs layer is modified, vfs_lookup.c also.
Well, yes; this is actually a stronger reason to do it in the vfs
layer. The check you put in ufs always uses the read bit, since the
same VOP_READLINK operation is used for readlink(2) and during path
walks. However, this doesn't make much sense; it makes more sense to
me to have the read bit control readlink(2) and the execute bit control
path walk operations - which is the semantics I implemented, and yes,
the latter is in vfs_lookup.c.
der Mouse
mouse@rodents.montreal.qc.ca
7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B