> > I have attached a patch based on Derrick's suggestion to use the > AFS_CACHE_VNODE_PATH code.
Excellent. > o There is code in afs/SOLARIS for a Xvfs file system. > Rather then overloading the osi_UfsOpen, it might be a > better choice to create a osi_ZfsOpen much like the > osi_XvfsOpen. Let me know if that would be better, > and I can rework the patch. It might be possible > to use VOP_GETATTR to get an inode as even with ZFS > they to generate inode numbers. Given that afsd's behavior changes, this is all or none. Not much reason not to change things in the UfsOpen case. > o The Solaris vn_open will use the user's CRED, and > it was found that permission denied was possible, > so pn_get_buf, lookuppnvp and VOP_OPEN where used > instead, using the cred of afs_osi_cred. That seems wise. > o The ois_DisableAtime routine was trying to update > the inode. I skipped that part of the code. Is this > OK or not? Is there another way to update the flag? That's not ideal but it will work until we find another way. > o The original patch would set the ainode to a signed > value, but on Solairs with 64 bit, the ainode was > defined to be unsigned long. This caused problems > with if (ainode < 0) and switch (ainode). > So the patch goes back to using afs_int32 for ainode. Perhaps we should use values just shy of 0xffffffff. _______________________________________________ OpenAFS-info mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-info
