Re: [EXT] Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
On Thu, May 14, 2020 at 06:01:52PM +0200, Andrew Lunn wrote: > On Thu, May 14, 2020 at 03:47:16PM +, Christian Herber wrote: > > Hi Andrew, > > > > > On Wed, May 13, 2020 at 03:39:00PM +0200, Andrew Lunn wrote: > > >> On Thu, May 14, 2020 at 02:09:59PM +0200, Oleksij Rempel wrote: > > >> ETHTOOL_A_CABLE_RESULT_CODE_ACTIVE_PARTNER - the link partner is active. > > >> > > >> The TJA1102 is able to detect it if partner link is master. > > >> > > > master is not a cable diagnostics issue. This is a configuration > > > issue. > > > > > Master is very relevant for cable diagnostics, as a cable > > measurement should not be done with an active link partner on the > > other end (i.e. a PHY in master mode trying to train the link). > > > So if the measurement detects an active link partner disturbing the > > measurement, it is important to report this to the user. > > So with 'normal' PHYs, we use autoneg to make the link go quiet. But > you don't have autoneg. > > If there is no way to force the link quiet, then > ETHTOOL_A_CABLE_RESULT_CODE_ACTIVE_PARTNER makes sense. But we need to > keep the meaning generic. I don't want it to mean a T1 PHY with an > active master peer. It should be usable for any reason the link cannot > be made to go quiet. It looks for me, like AT803X_CDT_STATUS_STAT_FAIL has the same reason as ACTIVE PERTNER (Master) on TJA11xx. What kind of meaning, naming would be generic? -- Pengutronix e.K. | | Steuerwalder Str. 21 | http://www.pengutronix.de/ | 31137 Hildesheim, Germany | Phone: +49-5121-206917-0| Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917- | signature.asc Description: PGP signature
Re: [EXT] Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
On Thu, May 14, 2020 at 03:47:16PM +, Christian Herber wrote: > Hi Andrew, > > > On Wed, May 13, 2020 at 03:39:00PM +0200, Andrew Lunn wrote: > >> On Thu, May 14, 2020 at 02:09:59PM +0200, Oleksij Rempel wrote: > >> ETHTOOL_A_CABLE_RESULT_CODE_ACTIVE_PARTNER - the link partner is active. > >> > >> The TJA1102 is able to detect it if partner link is master. > >> > > master is not a cable diagnostics issue. This is a configuration > > issue. > > Master is very relevant for cable diagnostics, as a cable > measurement should not be done with an active link partner on the > other end (i.e. a PHY in master mode trying to train the link). > So if the measurement detects an active link partner disturbing the > measurement, it is important to report this to the user. So with 'normal' PHYs, we use autoneg to make the link go quiet. But you don't have autoneg. If there is no way to force the link quiet, then ETHTOOL_A_CABLE_RESULT_CODE_ACTIVE_PARTNER makes sense. But we need to keep the meaning generic. I don't want it to mean a T1 PHY with an active master peer. It should be usable for any reason the link cannot be made to go quiet. Andrew
RE: [EXT] Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
Hi Andrew, > On Wed, May 13, 2020 at 03:39:00PM +0200, Andrew Lunn wrote: >> On Thu, May 14, 2020 at 02:09:59PM +0200, Oleksij Rempel wrote: >> ETHTOOL_A_CABLE_RESULT_CODE_ACTIVE_PARTNER - the link partner is active. >> >> The TJA1102 is able to detect it if partner link is master. >> > master is not a cable diagnostics issue. This is a configuration > issue. Master is very relevant for cable diagnostics, as a cable measurement should not be done with an active link partner on the other end (i.e. a PHY in master mode trying to train the link). So if the measurement detects an active link partner disturbing the measurement, it is important to report this to the user. Christian
Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
On Thu, May 14, 2020 at 02:09:59PM +0200, Oleksij Rempel wrote: > On Wed, May 13, 2020 at 08:01:40PM +0200, Andrew Lunn wrote: > > > What would be the best place to do a test before the link is getting up? > > > Can it be done in the phy core, or it should be done in the PHY driver? > > > > > > So far, no action except of logging these errors is needed. > > > > You could do it in the config_aneg callback. > > In this case I get two test cycles if the test was requested from user > space: from .cable_test_get_status and from .config_aneg Oh yes. Forgot about the restore after the test completes. When do you want to run the test? When the interface is administratively configured up, or when the link actually goes up? You could do it in read_status(), when the link changes status. > > A kernel log entry is not very easy to use. You might want to see how > > easy it is to send a cable test result to userspace. Anything which is > > interested in this information can then listen for it. All the needed > > code is there, you will just need to rearrange it a bit. > > Indeed. I discovered" ethtool --monitor" for me. And the code is some > thing like this: > ethnl_cable_test_alloc(phydev); > phydev->drv->cable_test_start(phydev); > usleep_range(100, 200); > phydev->drv->cable_test_get_status(phydev, &finished); > if (finished) > ethnl_cable_test_finished(phydev); Yes, something like that. > Beside, what do you think about new result codes: > ETHTOOL_A_CABLE_RESULT_CODE_POLARITY - if cable polarity is wrong (- > connected to +) Polarity should be a whole new nested pair property, not a status extension. I think most PHYs are happy to work with the polarity swapped. So we don't want it to sounds like an error. So status should be OK and then the polarity property should then indicate it i swapped. And you can only detected swapped polarity when the link is up. So in most cases, you would not even include the properties, since cable tests is normally performed on a down'ed link. > ETHTOOL_A_CABLE_RESULT_CODE_ACTIVE_PARTNER - the link partner is active. > The TJA1102 is able to detect it if partner link is master. master is not a cable diagnostics issue. This is a configuration issue. Andrew
Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
On Wed, May 13, 2020 at 08:01:40PM +0200, Andrew Lunn wrote: > > What would be the best place to do a test before the link is getting up? > > Can it be done in the phy core, or it should be done in the PHY driver? > > > > So far, no action except of logging these errors is needed. > > You could do it in the config_aneg callback. In this case I get two test cycles if the test was requested from user space: from .cable_test_get_status and from .config_aneg > A kernel log entry is not very easy to use. You might want to see how > easy it is to send a cable test result to userspace. Anything which is > interested in this information can then listen for it. All the needed > code is there, you will just need to rearrange it a bit. Indeed. I discovered" ethtool --monitor" for me. And the code is some thing like this: ethnl_cable_test_alloc(phydev); phydev->drv->cable_test_start(phydev); usleep_range(100, 200); phydev->drv->cable_test_get_status(phydev, &finished); if (finished) ethnl_cable_test_finished(phydev); Beside, what do you think about new result codes: ETHTOOL_A_CABLE_RESULT_CODE_POLARITY - if cable polarity is wrong (- connected to +) ETHTOOL_A_CABLE_RESULT_CODE_ACTIVE_PARTNER - the link partner is active. The TJA1102 is able to detect it if partner link is master. Regards, Oleksij -- Pengutronix e.K. | | Steuerwalder Str. 21 | http://www.pengutronix.de/ | 31137 Hildesheim, Germany | Phone: +49-5121-206917-0| Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917- | signature.asc Description: PGP signature
Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
From: Oleksij Rempel Date: Wed, 13 May 2020 14:34:40 +0200 > Add initial cable testing support. > This PHY needs only 100usec for this test and it is recommended to run it > before the link is up. For now, provide at least ethtool support, so it > can be tested by more developers. > > This patch was tested with TJA1102 PHY with following results: > - No cable, is detected as open > - 1m cable, with no connected other end and detected as open > - a 40m cable (out of spec, max lenght should be 15m) is detected as OK. > > Current patch do not provide polarity test support. This test would > indicate not proper wire connection, where "+" wire of main phy is > connected to the "-" wire of the link partner. > > Signed-off-by: Oleksij Rempel Applied, thanks.
Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
On 5/13/2020 5:34 AM, Oleksij Rempel wrote: > Add initial cable testing support. > This PHY needs only 100usec for this test and it is recommended to run it > before the link is up. For now, provide at least ethtool support, so it > can be tested by more developers. > > This patch was tested with TJA1102 PHY with following results: > - No cable, is detected as open > - 1m cable, with no connected other end and detected as open > - a 40m cable (out of spec, max lenght should be 15m) is detected as OK. > > Current patch do not provide polarity test support. This test would > indicate not proper wire connection, where "+" wire of main phy is > connected to the "-" wire of the link partner. > > Signed-off-by: Oleksij Rempel Reviewed-by: Florian Fainelli -- Florian
Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
> > Do these registers all conform to the standard? Can we pull this code > > out into a library which all standards conformant PHY drivers can use? > > According to opensig, this functionality should be present on all new T1 PHYs. > But the register/bit layout is no specified as standard. At least I was not > able > to find it. I assume, current layout is TJA11xx specific. O.K. then: Reviewed-by: Andrew Lunn Andrew
Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
> What would be the best place to do a test before the link is getting up? > Can it be done in the phy core, or it should be done in the PHY driver? > > So far, no action except of logging these errors is needed. You could do it in the config_aneg callback. A kernel log entry is not very easy to use. You might want to see how easy it is to send a cable test result to userspace. Anything which is interested in this information can then listen for it. All the needed code is there, you will just need to rearrange it a bit. Andrew
Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
On Wed, May 13, 2020 at 03:39:25PM +0200, Andrew Lunn wrote: > On Wed, May 13, 2020 at 02:34:40PM +0200, Oleksij Rempel wrote: > > Add initial cable testing support. > > This PHY needs only 100usec for this test and it is recommended to run it > > before the link is up. For now, provide at least ethtool support, so it > > can be tested by more developers. > > > > This patch was tested with TJA1102 PHY with following results: > > - No cable, is detected as open > > - 1m cable, with no connected other end and detected as open > > - a 40m cable (out of spec, max lenght should be 15m) is detected as OK. > > > > Current patch do not provide polarity test support. This test would > > indicate not proper wire connection, where "+" wire of main phy is > > connected to the "-" wire of the link partner. > > > > Signed-off-by: Oleksij Rempel > > --- > > drivers/net/phy/nxp-tja11xx.c | 106 +- > > 1 file changed, 105 insertions(+), 1 deletion(-) > > > > diff --git a/drivers/net/phy/nxp-tja11xx.c b/drivers/net/phy/nxp-tja11xx.c > > index ca5f9d4dc57ed..8b743d25002b9 100644 > > --- a/drivers/net/phy/nxp-tja11xx.c > > +++ b/drivers/net/phy/nxp-tja11xx.c > > @@ -5,6 +5,7 @@ > > */ > > #include > > #include > > +#include > > #include > > #include > > #include > > @@ -26,6 +27,7 @@ > > #define MII_ECTRL_POWER_MODE_NO_CHANGE (0x0 << 11) > > #define MII_ECTRL_POWER_MODE_NORMAL(0x3 << 11) > > #define MII_ECTRL_POWER_MODE_STANDBY (0xc << 11) > > +#define MII_ECTRL_CABLE_TEST BIT(5) > > #define MII_ECTRL_CONFIG_ENBIT(2) > > #define MII_ECTRL_WAKE_REQUEST BIT(0) > > > > @@ -55,6 +57,11 @@ > > #define MII_GENSTAT24 > > #define MII_GENSTAT_PLL_LOCKED BIT(14) > > > > +#define MII_EXTSTAT25 > > +#define MII_EXTSTAT_SHORT_DETECT BIT(8) > > +#define MII_EXTSTAT_OPEN_DETECTBIT(7) > > +#define MII_EXTSTAT_POLARITY_DETECTBIT(6) > > + > > Do these registers all conform to the standard? Can we pull this code > out into a library which all standards conformant PHY drivers can use? According to opensig, this functionality should be present on all new T1 PHYs. But the register/bit layout is no specified as standard. At least I was not able to find it. I assume, current layout is TJA11xx specific. > The code itself looks O.K. What would be the best place to do a test before the link is getting up? Can it be done in the phy core, or it should be done in the PHY driver? So far, no action except of logging these errors is needed. -- Pengutronix e.K. | | Steuerwalder Str. 21 | http://www.pengutronix.de/ | 31137 Hildesheim, Germany | Phone: +49-5121-206917-0| Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917- |
Re: [PATCH net-next v1] net: phy: tja11xx: add cable-test support
On Wed, May 13, 2020 at 02:34:40PM +0200, Oleksij Rempel wrote: > Add initial cable testing support. > This PHY needs only 100usec for this test and it is recommended to run it > before the link is up. For now, provide at least ethtool support, so it > can be tested by more developers. > > This patch was tested with TJA1102 PHY with following results: > - No cable, is detected as open > - 1m cable, with no connected other end and detected as open > - a 40m cable (out of spec, max lenght should be 15m) is detected as OK. > > Current patch do not provide polarity test support. This test would > indicate not proper wire connection, where "+" wire of main phy is > connected to the "-" wire of the link partner. > > Signed-off-by: Oleksij Rempel > --- > drivers/net/phy/nxp-tja11xx.c | 106 +- > 1 file changed, 105 insertions(+), 1 deletion(-) > > diff --git a/drivers/net/phy/nxp-tja11xx.c b/drivers/net/phy/nxp-tja11xx.c > index ca5f9d4dc57ed..8b743d25002b9 100644 > --- a/drivers/net/phy/nxp-tja11xx.c > +++ b/drivers/net/phy/nxp-tja11xx.c > @@ -5,6 +5,7 @@ > */ > #include > #include > +#include > #include > #include > #include > @@ -26,6 +27,7 @@ > #define MII_ECTRL_POWER_MODE_NO_CHANGE (0x0 << 11) > #define MII_ECTRL_POWER_MODE_NORMAL (0x3 << 11) > #define MII_ECTRL_POWER_MODE_STANDBY (0xc << 11) > +#define MII_ECTRL_CABLE_TEST BIT(5) > #define MII_ECTRL_CONFIG_EN BIT(2) > #define MII_ECTRL_WAKE_REQUEST BIT(0) > > @@ -55,6 +57,11 @@ > #define MII_GENSTAT 24 > #define MII_GENSTAT_PLL_LOCKED BIT(14) > > +#define MII_EXTSTAT 25 > +#define MII_EXTSTAT_SHORT_DETECT BIT(8) > +#define MII_EXTSTAT_OPEN_DETECT BIT(7) > +#define MII_EXTSTAT_POLARITY_DETECT BIT(6) > + Do these registers all conform to the standard? Can we pull this code out into a library which all standards conformant PHY drivers can use? The code itself looks O.K. Andrew