netdev - Re: [PATCH net-next] net: stmmac: allow ethtool action on PCI devices if device is down

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <ZCqfwxr2I7xt6zon@calimero.vinschen.de>
Date:   Mon, 3 Apr 2023 11:43:31 +0200
From:   Corinna Vinschen <vinschen@...hat.com>
To:     Heiner Kallweit <hkallweit1@...il.com>
Cc:     Giuseppe Cavallaro <peppe.cavallaro@...com>,
        Alexandre Torgue <alexandre.torgue@...s.st.com>,
        Jose Abreu <joabreu@...opsys.com>, netdev@...r.kernel.org
Subject: Re: [PATCH net-next] net: stmmac: allow ethtool action on PCI
 devices if device is down

On Mar 31 12:48, Heiner Kallweit wrote:
> On 31.03.2023 11:23, Corinna Vinschen wrote:
> > So far stmmac is only able to handle ethtool commands if the device
> > is UP.  However, PCI devices usually just have to be in the active
> > state for ethtool commands.
> > 
> What do mean with "active state" here? D0? Or, as you say "connected
> to PCI" a few lines later, do you refer to hot-plugging?
> 
> PCI being in D0 often isn't sufficient, especially if interface is down.
> Then required resources like clocks may be disabled by the driver.
> 
> A driver may runtime-suspend a device for multiple reasons, e.g.
> interface down or link down. Then the device may be put to D3hot
> to save power.
> If we say that it's worth to wake a device for an ethtool command,
> then I wonder whether this is something that should be done in net
> core. E.g. calling pm_runtime_get_sync() in __dev_ethtool, similar to
> what we do in __dev_open() to deal with run-time suspended and
> therefore potentially detached devices.

Actually, I'm not sure how to reply to your question.  I replicated the
behaviour of the igb and igc drivers, because that looked like the right
thing to do for a PCIe device.  It seems a bit awkward that the UP/DOWN
state allows or denies device settings.  


Corinna



> 
> > Rename stmmac_check_if_running to stmmac_ethtool_begin and add a
> > stmmac_ethtool_complete action.  Check if the device is connected
> > to PCI and if so, just make sure the device is active.  Reset it
> > to idle state as complete action.
> > 
> > Tested on Intel Elkhart Lake system.
> > 
> > Signed-off-by: Corinna Vinschen <vinschen@...hat.com>
> > ---
> >  .../ethernet/stmicro/stmmac/stmmac_ethtool.c  | 20 +++++++++++++++++--
> >  1 file changed, 18 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
> > index 35c8dd92d369..5a57970dc06a 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
> > @@ -14,6 +14,7 @@
> >  #include <linux/mii.h>
> >  #include <linux/phylink.h>
> >  #include <linux/net_tstamp.h>
> > +#include <linux/pm_runtime.h>
> >  #include <asm/io.h>
> >  
> >  #include "stmmac.h"
> > @@ -429,13 +430,27 @@ static void stmmac_ethtool_setmsglevel(struct net_device *dev, u32 level)
> >  
> >  }
> >  
> > -static int stmmac_check_if_running(struct net_device *dev)
> > +static int stmmac_ethtool_begin(struct net_device *dev)
> >  {
> > +	struct stmmac_priv *priv = netdev_priv(dev);
> > +
> > +	if (priv->plat->pdev) {
> > +		pm_runtime_get_sync(&priv->plat->pdev->dev);
> > +		return 0;
> > +	}
> >  	if (!netif_running(dev))
> >  		return -EBUSY;
> >  	return 0;
> >  }
> >  
> > +static void stmmac_ethtool_complete(struct net_device *dev)
> > +{
> > +	struct stmmac_priv *priv = netdev_priv(dev);
> > +
> > +	if (priv->plat->pdev)
> > +		pm_runtime_put(&priv->plat->pdev->dev);
> > +}
> > +
> >  static int stmmac_ethtool_get_regs_len(struct net_device *dev)
> >  {
> >  	struct stmmac_priv *priv = netdev_priv(dev);
> > @@ -1152,7 +1167,8 @@ static int stmmac_set_tunable(struct net_device *dev,
> >  static const struct ethtool_ops stmmac_ethtool_ops = {
> >  	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
> >  				     ETHTOOL_COALESCE_MAX_FRAMES,
> > -	.begin = stmmac_check_if_running,
> > +	.begin = stmmac_ethtool_begin,
> > +	.complete = stmmac_ethtool_complete,
> >  	.get_drvinfo = stmmac_ethtool_getdrvinfo,
> >  	.get_msglevel = stmmac_ethtool_getmsglevel,
> >  	.set_msglevel = stmmac_ethtool_setmsglevel,