lkml.org 
[lkml]   [2020]   [Mar]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: [PATCH net] net: phy: Avoid multiple suspends
Hi Heiner,

On Wed, Mar 11, 2020 at 10:22 PM Heiner Kallweit <hkallweit1@gmail.com> wrote:
> On 11.03.2020 10:17, Geert Uytterhoeven wrote:
> > On Tue, Mar 10, 2020 at 5:47 PM Florian Fainelli <f.fainelli@gmail.com> wrote:
> >> On 3/10/20 7:16 AM, Geert Uytterhoeven wrote:
> >>> On Mon, Feb 24, 2020 at 5:59 AM David Miller <davem@davemloft.net> wrote:
> >>>> From: Florian Fainelli <f.fainelli@gmail.com>
> >>>> Date: Thu, 20 Feb 2020 15:34:53 -0800
> >>>>
> >>>>> It is currently possible for a PHY device to be suspended as part of a
> >>>>> network device driver's suspend call while it is still being attached to
> >>>>> that net_device, either via phy_suspend() or implicitly via phy_stop().
> >>>>>
> >>>>> Later on, when the MDIO bus controller get suspended, we would attempt
> >>>>> to suspend again the PHY because it is still attached to a network
> >>>>> device.
> >>>>>
> >>>>> This is both a waste of time and creates an opportunity for improper
> >>>>> clock/power management bugs to creep in.
> >>>>>
> >>>>> Fixes: 803dd9c77ac3 ("net: phy: avoid suspending twice a PHY")
> >>>>> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
> >>>>
> >>>> Applied, and queued up for -stable, thanks Florian.
> >>>
> >>> This patch causes a regression on r8a73a4/ape6evm and sh73a0/kzm9g.
> >>> After resume from s2ram, Ethernet no longer works:
> >>>
> >>> PM: suspend exit
> >>> nfs: server aaa.bbb.ccc.ddd not responding, still trying
> >>> ...
> >>>
> >>> Reverting commit 503ba7c6961034ff ("net: phy: Avoid multiple suspends")
> >>> fixes the issue.
> >>>
> >>> On both boards, an SMSC LAN9220 is connected to a power-managed local
> >>> bus.
> >>>
> >>> I added some debug code to check when the clock driving the local bus
> >>> is stopped and started, but I see no difference before/after. Hence I
> >>> suspect the Ethernet chip is no longer reinitialized after resume.
> >>
> >> Can you provide a complete log?
> >
> > With some debug info:
> >
> > SDHI0 Vcc: disabling
> > PM: suspend entry (deep)
> > Filesystems sync: 0.002 seconds
> > Freezing user space processes ... (elapsed 0.001 seconds) done.
> > OOM killer disabled.
> > Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done.
> > PM: ==== a3sp/ee120000.sd: stop
> > PM: ==== a3sp/ee100000.sd: stop
> > smsc911x 8000000.ethernet: smsc911x_suspend:2577
> > smsc911x 8000000.ethernet: smsc911x_suspend:2579 running
> > smsc911x 8000000.ethernet: smsc911x_suspend:2584
> > PM: ==== a3sp/ee200000.mmc: stop
> > PM: ==== c4/fec10000.bus: stop
> > PM: ==== a3sp/e6c40000.serial: stop
> > PM: ==== c5/e61f0000.thermal: stop
> > PM: ==== c4/e61c0200.interrupt-controller: stop
> > PM: == a3sp: power off
> > rmobile_pd_power_down: a3sp
> > Disabling non-boot CPUs ...
> > PM: ==== c4/e61c0200.interrupt-controller: start
> > PM: ==== c5/e61f0000.thermal: start
> > PM: ==== a3sp/e6c40000.serial: start
> > PM: ==== c4/fec10000.bus: start
> > PM: ==== a3sp/ee200000.mmc: start
> > smsc911x 8000000.ethernet: smsc911x_resume:2606
> > smsc911x 8000000.ethernet: smsc911x_resume:2625 running
> > PM: ==== a3sp/ee100000.sd: start
> > OOM killer enabled.
> > Restarting tasks ... done.
> > PM: ==== a3sp/ee120000.sd: start
> > PM: suspend exit
> > nfs: server aaa.bbb.ccc.ddd not responding, still trying
> > ...
> >
> > But no difference between the good and the bad case, except for the nfs
> > failures.
> >
> >> Do you use the Generic PHY driver or a
> >> specialized one?
> >
> > CONFIG_FIXED_PHY=y
> > CONFIG_SMSC_PHY=y
> >
> > Just the smsc,lan9115 node, cfr. arch/arm/boot/dts/r8a73a4-ape6evm.dts
> >
> >> Do you have a way to dump the registers at the time of
> >> failure and see if BMCR.PDOWN is still set somehow?
> >
> > Added a hook into "nfs: server not responding", which prints:
> >
> > MII_BMCR = 0x1900
> >
> > i.e. BMCR_PDOWN = 0x0800 is still set.
> >
> >> Does the following help:
> >>
> >> diff --git a/drivers/net/ethernet/smsc/smsc911x.c
> >> b/drivers/net/ethernet/smsc/smsc911x.c
> >> index 49a6a9167af4..df17190c76c0 100644
> >> --- a/drivers/net/ethernet/smsc/smsc911x.c
> >> +++ b/drivers/net/ethernet/smsc/smsc911x.c
> >> @@ -2618,6 +2618,7 @@ static int smsc911x_resume(struct device *dev)
> >> if (netif_running(ndev)) {
> >> netif_device_attach(ndev);
> >> netif_start_queue(ndev);
> >> + phy_resume(dev->phydev);
> >> }
> >>
> >
> > Yes i does, after s/dev->/ndev->/.
> > Thanks!
>
> This seems to be a workaround. And the same issue we may have with

I agree.

> other drivers too. Could you please alternatively test the following?
> It tackles the issue that mdio_bus_phy_may_suspend() is used in
> suspend AND resume, and both calls may return different values.
>
> With this patch we call mdio_bus_phy_may_suspend() only when
> suspending, and let the phy_device store whether it was suspended
> by MDIO bus PM.

Thanks, your patch fixes the issue, too.

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

\
 
 \ /
  Last update: 2020-03-12 09:28    [W:0.053 / U:0.436 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site