On 03/05/2012 08:04 PM, Myklebust, Trond wrote: >>> Is there any reason why we actually care about checking the crap >>> in /etc/mtab on umount? >>> >> Yeah... its called backwards compatibility with older distros... >> Believe, if I could bury mtab I would... in a New York minute! >> I just don't see it happening... > > Yes, but despite all your work, you are just replacing one broken model > by another. > > > At least with the current code, they can _see_ that the model is broken > and have an immediate incentive to move to the > mtab-is-a-symlink-to-/proc/mounts based model. The latter is in any case > the only one that is valid in the mount-namespace based world in which > we've been living for the past 5 years or so... > I do not disagree with what you are saying... The fact the code/model is broken is unarguably true... Relying on info in the mtab verses /proc/mounts is completely brain dead... But those bits are on the street and need to be fixed... So the question is how to fix them... Sending out a patch that symbolically links mtab to /proc/mounts would solve this entire problem! But I just can't make major change like that in established worlds... Who know what people use mtab for... I certainly don't. Changing mtab to read-only is recipe for disaster... IMHO... Now if we don't want to take a patch like this in upstream that's a different story... This patch is basically meaningless when it comes to new Linux distros... I'm ok with that.. But I do think its wise for us to at least try to be somewhat backwards compatible.... steved. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
