Hi Rafael,
On Wed, Aug 24, 2022 at 10:11 PM Rafael J. Wysocki <raf...@kernel.org> wrote:
>
> On Wed, Aug 24, 2022 at 6:41 AM Juhyung Park <qkrwngud...@gmail.com> wrote:
> >
> > Commit 2fd77fff4b44 ("PM / suspend: make sync() on suspend-to-RAM build-time
> > optional") added an option to skip sync() on suspend entry to avoid heavy
> > overhead on platforms with frequent suspends.
> >
> > Years later, commit 261e224d6a5c ("pm/sleep: Add PM_USERSPACE_AUTOSLEEP
> > Kconfig") added a dedicated config for indicating that the kernel is
> > subject to
> > frequent suspends.
> >
> > While SUSPEND_SKIP_SYNC is also available as a knob that the userspace can
> > configure, it makes sense to enable this by default if
> > PM_USERSPACE_AUTOSLEEP
> > is selected already.
> >
> > Signed-off-by: Juhyung Park <qkrwngud...@gmail.com>
> > ---
> > kernel/power/Kconfig | 1 +
> > 1 file changed, 1 insertion(+)
> >
> > diff --git a/kernel/power/Kconfig b/kernel/power/Kconfig
> > index 60a1d3051cc7..5725df6c573b 100644
> > --- a/kernel/power/Kconfig
> > +++ b/kernel/power/Kconfig
> > @@ -23,6 +23,7 @@ config SUSPEND_SKIP_SYNC
> > bool "Skip kernel's sys_sync() on suspend to RAM/standby"
> > depends on SUSPEND
> > depends on EXPERT
> > + default PM_USERSPACE_AUTOSLEEP
>
> Why is this better than selecting SUSPEND_SKIP_SYNC from
> PM_USERSPACE_AUTOSLEEP?
That won't allow developers to opt-out from SUSPEND_SKIP_SYNC when
they still want PM_USERSPACE_AUTOSLEEP. (Can't think of a valid reason
for this though, as PM_USERSPACE_AUTOSLEEP is only used by Android and
probably Chromium, afaik.)
I don't think SUSPEND_SKIP_SYNC is critical enough to enforce when
PM_USERSPACE_AUTOSLEEP is enabled, but I don't have a strong opinion
on this either.
(We could do `imply SUSPEND_SKIP_SYNC` from PM_USERSPACE_AUTOSLEEP,
but that doesn't look good semantically imho.)
If you want, I can send a v2 with 'PM_USERSPACE_AUTOSLEEP select
SUSPEND_SKIP_SYNC'.
Thanks.
>
> > help
> > Skip the kernel sys_sync() before freezing user processes.
> > Some systems prefer not to pay this cost on every invocation
> > --
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
