On Mon, 23 Apr 2001 23:41:32 BST, dawat_private (David Wagner) said: > You seem to have some misconceptions about how kernel code works. > For starters, errno is a purely user-space construction. In the kernel, > a syscall returns, e.g., -EPERM to indicate a permission error; then > user-level libc wrappers change this into errno=EPERM, retval=-1. > If you want to add a global kernel variable that holds the most recent > error, and you want to add a getlatesterror() syscall (or /proc entry, > or ioctl, or whatever the best interface is), feel free to do so in > your module. I don't want this in my module. And, I'm not convinced > this is even a good idea. For instance, this is not thread-safe. And as I've said several times, modules that don't want to do it will just provide a null-stub and not do it... > For these reasons, I don't think this proposal is mature enough to > merit inclusion in the general kernel patch (as opposed to your favorite > policy module). And since it doesn't need to be in the general kernel > patch -- since it can be handled by policy modules -- maybe we should > leave it out of the general kernel patch. Agreed? I'll meet you half-way on this one - I admit doing most of my coding over on the userland side of the fence, and I don't really care if it's actually implemented in the kernel patch or not. So... As long as we define an API, and have at least *some* hand-waving of "a module could do this-or-that to provide it", I'll shut up about the exact implementation. My requirement is "the LSM interface needs the moral equivalent of the PAM pam_strerror() routine", and that I know that if I get an EPERM error in errno, that calling lsm_perror() will: a) be guaranteed to be there (even if only as a stub) b) Return either a more detailed description than EPERM or 'no further info'. /Valdis _______________________________________________ linux-security-module mailing list linux-security-moduleat_private http://mail.wirex.com/mailman/listinfo/linux-security-module
This archive was generated by hypermail 2b30 : Mon Apr 23 2001 - 19:35:56 PDT