On Tuesday, July 07, 2015 11:03:20 AM Alan Stern wrote:

On Tue, 7 Jul 2015, Oliver Neukum wrote:

quoted

quoted

he (or she) pulls the storage device out of the system, moves it to another
system, makes changes (say removes the file written to by the process above,
so the blocks previously occupied by that file are now used for some metadata)
and moves the storage back to the suspended system.  The system is resumed
and the writing process continues writing possibly to the wrong blocks and
corrupts the filesystem.

That is a tough nut. But that's not a reason to make it worse.
I'd say there's no reason not to use a secondary interface to
suspend without syncing or to extend or introduce such an interface
if the API is deficient.

Indeed, the problem Rafael outlined always exists whether or not the
kernel does a sync.  Even if no I/O is in progress when the system goes
to sleep, if the user moves a portable storage device with a mounted
filesystem to another computer and updates it before waking the system
up, corruption is highly likely.

In principle this could be solved by adding suspend/resume callbacks to
filesystems.  For example, the resume callback could verify that the
superblock had not been changed since the suspend occurred.  Or there 
could be some other simple way of determining that the filesystem had 
not been remounted and changed.

Either way, this is irrelevant to the question of whether the kernel 
should issue a sync when suspending.

well, that depends on what the purpose of the sync is supposed to be.

If it is there to prevent users from corrupting their filesystems as a result
of a mistake, it is insufficient.  If it's there for other reasons, I'm wondering
what those reasons are (on systems that suspend and resume reliably, because the
original reason to put it in there was to reduce the damage from suspend/resume
crashes).

Thanks,
Rafael