Re: Getting rid of freezer for suspend [was Re: [fuse-devel] [PATCH] fuse: make fuse daemon frozen along with kernel threads]

["Rafael J. Wysocki" <rjw@sisk.pl>]

On Thursday, February 14, 2013 11:41:16 AM Miklos Szeredi wrote:
> On Wed, Feb 13, 2013 at 10:21 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> > On Wednesday, February 13, 2013 06:34:16 PM Miklos Szeredi wrote:
> 
> >>
> >> So I think the PF_FREEZE_DAEMON idea (the patch from Li Fei that
> >> started this thread) may still be our best bet at handling this
> >> situation.  The idea being that pure "originator" processes (ones that
> >> take no part in serving filesystem syscalls) can be frozen up-front.
> >> Then the "fuse daemon" (or "server") processes are hopefully in a
> >> quiescent state and can be frozen without difficulty.
> >>
> >> Unfortunately it needs help from userspace: the kernel can't easily
> >> guess which processes are part of a "fuse daemon" and which aren't.
> >> Fortunately we have a standard library (libfuse) that can tell it to
> >> the kernel for the vast majority of cases.
> >
> > So basically the idea would be to introduce something like PF_FREEZE_LATE
> > for user space processes that need to be frozen after all of the other
> > (non-PF_FREEZE_LATE) user space processes have been frozen and hack fuse
> > to use that flag?
> 
> Yes.
> 
> It is essentially the same mechanism that is used to delay the
> freezing of kernel threads after userspace tasks have been frozen.
> Except it's a lot more difficult to determine which userspace tasks
> need to be suspended late and which aren't.

Well, I suppose that information is available to user space.

Do we need an interface for a process to mark itself as PF_FREEZE_LATE or
do we need an interface for one process to mark another process as
PF_FREEZE_LATE, or both?

Rafael


-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.