Neil Brown wrote: > On Monday February 26, SteveD@redhat.com wrote: >> commit 806b02f63fe6e01946d84a1f14c74e977be76530 >> Author: Amit Gud >> Date: Sat Feb 24 14:31:13 2007 -0500 >> >> Added the following non-nfs mount options as suggested >> but upstream code reviews: >> >> [no]auto, [no]users, [no]user, [no]owner, [no]group >> _netdev, comment > > This patch does a lot more than that brief description suggests. > > - The 'users' / 'user' changes are enough to be a patch in itself, > particularly as it is a security issue and needs careful review. > mount and umount do similar tests here and it would be good to > include them in the library (nfs/fstab.c). I believe this was just a port of nfs/fstab.c which was need, but I forget why... Amit would you happen to remember? > > - 'const' is removed from nfs_mntent_s, and copy_mntent is introduced > .. are those related? Yes... these have to do with the remount support, which was part of getting all the non-nfs options to work.. The 'const' was remove to stop the following warning: mount.c: In function 'copy_mntent': mount.c:168: warning: passing argument 1 of '__builtin___strcpy_chk' discards qualifiers from pointer target type > > - You copy instead of link the 4 different mount.* programs, which > might be a good thing, but deserves a comment in the changelog. This was because of set-uid issues... I believe Amit tried to use links but there was an issue with it that cause us to do the copies... > > - Some printf change to fprintf, which is good, but seems to be > independent of the rest. This was general clean up... > > and possible some other things that I didn't notice. As this is a > setuid program, I really want to be able to review changes in > isolation please. Ok... I broke the original patch up into 3 separate ones. Pleas see attachments... steved.