[1/2] net/ixgbe: fix x550 code to handle unidentified PHY
diff mbox series

Message ID 20181101140413.18665-1-bluca@debian.org
State Superseded, archived
Delegated to: Qi Zhang
Headers show
Series
  • [1/2] net/ixgbe: fix x550 code to handle unidentified PHY
Related show

Checks

Context Check Description
ci/checkpatch success coding style OK
ci/Intel-compilation success Compilation OK

Commit Message

Luca Boccassi Nov. 1, 2018, 2:04 p.m. UTC
ixgbe_identify_phy_x550em() was missing the code to handle unidentified
PHY that has been there in 82599 so it was not able to complete
initialization of ixgbe sequence if no sfp plugged in.
Port it over to return an appropriate type and complete init sequence
properly.

Fixes: d2e72774e58c ("ixgbe/base: support X550")
Cc: stable@dpdk.org

Signed-off-by: Luca Boccassi <bluca@debian.org>
---
v2: refresh to remove merge conflict with master

 drivers/net/ixgbe/base/ixgbe_x550.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)

Comments

Luca Boccassi Nov. 1, 2018, 2:41 p.m. UTC | #1
On Thu, 2018-11-01 at 14:04 +0000, Luca Boccassi wrote:
> ixgbe_identify_phy_x550em() was missing the code to handle
> unidentified
> PHY that has been there in 82599 so it was not able to complete
> initialization of ixgbe sequence if no sfp plugged in.
> Port it over to return an appropriate type and complete init sequence
> properly.
> 
> Fixes: d2e72774e58c ("ixgbe/base: support X550")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Luca Boccassi <bluca@debian.org>
> ---
> v2: refresh to remove merge conflict with master
> 
>  drivers/net/ixgbe/base/ixgbe_x550.c | 19 ++++++++++++++++---
>  1 file changed, 16 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/ixgbe/base/ixgbe_x550.c
> b/drivers/net/ixgbe/base/ixgbe_x550.c
> index f7b98af52..83b394861 100644

Sorry, finger slipped and missed the -v2 from the command so it's not
in the subject - it's the same, just rebased to fix a conflict in the
first patch.
Zhang, Qi Z Nov. 2, 2018, 2:11 p.m. UTC | #2
> -----Original Message-----
> From: Luca Boccassi [mailto:bluca@debian.org]
> Sent: Thursday, November 1, 2018 9:04 AM
> To: dev@dpdk.org
> Cc: Lu, Wenzhuo <wenzhuo.lu@intel.com>; Ananyev, Konstantin
> <konstantin.ananyev@intel.com>; Zhang, Qi Z <qi.z.zhang@intel.com>; Luca
> Boccassi <bluca@debian.org>; stable@dpdk.org
> Subject: [PATCH 1/2] net/ixgbe: fix x550 code to handle unidentified PHY
> 
> ixgbe_identify_phy_x550em() was missing the code to handle unidentified
> PHY that has been there in 82599 so it was not able to complete initialization
> of ixgbe sequence if no sfp plugged in.
> Port it over to return an appropriate type and complete init sequence
> properly.
> 
> Fixes: d2e72774e58c ("ixgbe/base: support X550")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Luca Boccassi <bluca@debian.org>
> ---
> v2: refresh to remove merge conflict with master
> 
>  drivers/net/ixgbe/base/ixgbe_x550.c | 19 ++++++++++++++++---
>  1 file changed, 16 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/ixgbe/base/ixgbe_x550.c
> b/drivers/net/ixgbe/base/ixgbe_x550.c
> index f7b98af52..83b394861 100644
> --- a/drivers/net/ixgbe/base/ixgbe_x550.c
> +++ b/drivers/net/ixgbe/base/ixgbe_x550.c
> @@ -315,13 +315,21 @@ STATIC void ixgbe_setup_mux_ctl(struct ixgbe_hw
> *hw)
>   */
>  STATIC s32 ixgbe_identify_phy_x550em(struct ixgbe_hw *hw)  {
> +	s32 status;
> +
>  	hw->mac.ops.set_lan_id(hw);
> 
>  	ixgbe_read_mng_if_sel_x550em(hw);
> 
>  	switch (hw->device_id) {
>  	case IXGBE_DEV_ID_X550EM_A_SFP:
> -		return ixgbe_identify_sfp_module_X550em(hw);
> +		status = ixgbe_identify_sfp_module_X550em(hw);
> +		/* Set PHY type none if no PHY detected */
> +		if (hw->phy.type == ixgbe_phy_unknown) {
> +			hw->phy.type = ixgbe_phy_none;
> +			return IXGBE_SUCCESS;
> +		}

Why this can't be handled at caller, why we replace phy_unknown by phy_none only for x550?

> +		return status;
>  	case IXGBE_DEV_ID_X550EM_X_SFP:
>  		/* set up for CS4227 usage */
>  		ixgbe_setup_mux_ctl(hw);
> @@ -329,8 +337,13 @@ STATIC s32 ixgbe_identify_phy_x550em(struct
> ixgbe_hw *hw)
>  		/* Fallthrough */
> 
>  	case IXGBE_DEV_ID_X550EM_A_SFP_N:
> -		return ixgbe_identify_sfp_module_X550em(hw);
> -		break;
> +		status = ixgbe_identify_sfp_module_X550em(hw);
> +		/* Set PHY type none if no PHY detected */
> +		if (hw->phy.type == ixgbe_phy_unknown) {
> +			hw->phy.type = ixgbe_phy_none;
> +			return IXGBE_SUCCESS;
> +		}
> +		return status;
>  	case IXGBE_DEV_ID_X550EM_X_KX4:
>  		hw->phy.type = ixgbe_phy_x550em_kx4;
>  		break;
> --
> 2.19.1
Luca Boccassi Nov. 2, 2018, 3:21 p.m. UTC | #3
On Fri, 2018-11-02 at 14:11 +0000, Zhang, Qi Z wrote:
> > -----Original Message-----
> > From: Luca Boccassi [mailto:bluca@debian.org]
> > Sent: Thursday, November 1, 2018 9:04 AM
> > To: dev@dpdk.org
> > Cc: Lu, Wenzhuo <wenzhuo.lu@intel.com>; Ananyev, Konstantin
> > <konstantin.ananyev@intel.com>; Zhang, Qi Z <qi.z.zhang@intel.com>;
> > Luca
> > Boccassi <bluca@debian.org>; stable@dpdk.org
> > Subject: [PATCH 1/2] net/ixgbe: fix x550 code to handle
> > unidentified PHY
> > 
> > ixgbe_identify_phy_x550em() was missing the code to handle
> > unidentified
> > PHY that has been there in 82599 so it was not able to complete
> > initialization
> > of ixgbe sequence if no sfp plugged in.
> > Port it over to return an appropriate type and complete init
> > sequence
> > properly.
> > 
> > Fixes: d2e72774e58c ("ixgbe/base: support X550")
> > Cc: stable@dpdk.org
> > 
> > Signed-off-by: Luca Boccassi <bluca@debian.org>
> > ---
> > v2: refresh to remove merge conflict with master
> > 
> >  drivers/net/ixgbe/base/ixgbe_x550.c | 19 ++++++++++++++++---
> >  1 file changed, 16 insertions(+), 3 deletions(-)
> > 
> > diff --git a/drivers/net/ixgbe/base/ixgbe_x550.c
> > b/drivers/net/ixgbe/base/ixgbe_x550.c
> > index f7b98af52..83b394861 100644
> > --- a/drivers/net/ixgbe/base/ixgbe_x550.c
> > +++ b/drivers/net/ixgbe/base/ixgbe_x550.c
> > @@ -315,13 +315,21 @@ STATIC void ixgbe_setup_mux_ctl(struct
> > ixgbe_hw
> > *hw)
> >   */
> >  STATIC s32 ixgbe_identify_phy_x550em(struct ixgbe_hw *hw)  {
> > +	s32 status;
> > +
> >  	hw->mac.ops.set_lan_id(hw);
> > 
> >  	ixgbe_read_mng_if_sel_x550em(hw);
> > 
> >  	switch (hw->device_id) {
> >  	case IXGBE_DEV_ID_X550EM_A_SFP:
> > -		return ixgbe_identify_sfp_module_X550em(hw);
> > +		status = ixgbe_identify_sfp_module_X550em(hw);
> > +		/* Set PHY type none if no PHY detected */
> > +		if (hw->phy.type == ixgbe_phy_unknown) {
> > +			hw->phy.type = ixgbe_phy_none;
> > +			return IXGBE_SUCCESS;
> > +		}
> 
> Why this can't be handled at caller, why we replace phy_unknown by
> phy_none only for x550?

Hi, thanks for the review.

I've moved the code into the caller in v3.

It's done for x550 simply because that's the hardware that we have and
that this was tested with, I didn't want to take extra risks. It's also
the hardware that our customer reported the issue with.
Zhang, Qi Z Nov. 2, 2018, 4:49 p.m. UTC | #4
> -----Original Message-----
> From: Luca Boccassi [mailto:bluca@debian.org]
> Sent: Friday, November 2, 2018 10:21 AM
> To: Zhang, Qi Z <qi.z.zhang@intel.com>; dev@dpdk.org
> Cc: Lu, Wenzhuo <wenzhuo.lu@intel.com>; Ananyev, Konstantin
> <konstantin.ananyev@intel.com>; stable@dpdk.org
> Subject: Re: [PATCH 1/2] net/ixgbe: fix x550 code to handle unidentified PHY
> 
> On Fri, 2018-11-02 at 14:11 +0000, Zhang, Qi Z wrote:
> > > -----Original Message-----
> > > From: Luca Boccassi [mailto:bluca@debian.org]
> > > Sent: Thursday, November 1, 2018 9:04 AM
> > > To: dev@dpdk.org
> > > Cc: Lu, Wenzhuo <wenzhuo.lu@intel.com>; Ananyev, Konstantin
> > > <konstantin.ananyev@intel.com>; Zhang, Qi Z <qi.z.zhang@intel.com>;
> > > Luca Boccassi <bluca@debian.org>; stable@dpdk.org
> > > Subject: [PATCH 1/2] net/ixgbe: fix x550 code to handle unidentified
> > > PHY
> > >
> > > ixgbe_identify_phy_x550em() was missing the code to handle
> > > unidentified PHY that has been there in 82599 so it was not able to
> > > complete initialization of ixgbe sequence if no sfp plugged in.
> > > Port it over to return an appropriate type and complete init
> > > sequence properly.
> > >
> > > Fixes: d2e72774e58c ("ixgbe/base: support X550")
> > > Cc: stable@dpdk.org
> > >
> > > Signed-off-by: Luca Boccassi <bluca@debian.org>
> > > ---
> > > v2: refresh to remove merge conflict with master
> > >
> > >  drivers/net/ixgbe/base/ixgbe_x550.c | 19 ++++++++++++++++---
> > >  1 file changed, 16 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/drivers/net/ixgbe/base/ixgbe_x550.c
> > > b/drivers/net/ixgbe/base/ixgbe_x550.c
> > > index f7b98af52..83b394861 100644
> > > --- a/drivers/net/ixgbe/base/ixgbe_x550.c
> > > +++ b/drivers/net/ixgbe/base/ixgbe_x550.c
> > > @@ -315,13 +315,21 @@ STATIC void ixgbe_setup_mux_ctl(struct
> > > ixgbe_hw
> > > *hw)
> > >   */
> > >  STATIC s32 ixgbe_identify_phy_x550em(struct ixgbe_hw *hw)  {
> > > +	s32 status;
> > > +
> > >  	hw->mac.ops.set_lan_id(hw);
> > >
> > >  	ixgbe_read_mng_if_sel_x550em(hw);
> > >
> > >  	switch (hw->device_id) {
> > >  	case IXGBE_DEV_ID_X550EM_A_SFP:
> > > -		return ixgbe_identify_sfp_module_X550em(hw);
> > > +		status = ixgbe_identify_sfp_module_X550em(hw);
> > > +		/* Set PHY type none if no PHY detected */
> > > +		if (hw->phy.type == ixgbe_phy_unknown) {
> > > +			hw->phy.type = ixgbe_phy_none;
> > > +			return IXGBE_SUCCESS;
> > > +		}
> >
> > Why this can't be handled at caller, why we replace phy_unknown by
> > phy_none only for x550?
> 
> Hi, thanks for the review.
> 
> I've moved the code into the caller in v3.

Sorry, this is not what I suggested. 
I'm not sure for X550EM_A_SFP, it is the case that the device does not have PHY so we should correct it (by replace it with no_phy) 
or it is the case that we can't identify PHY so we just replace it by no_phy to bypass the check for workaround?
If the second case, is it possible not to do replacement and handle it at upper layer (caller or caller's caller ... ) and keep the information more accurate?

> 
> It's done for x550 simply because that's the hardware that we have and that
> this was tested with, I didn't want to take extra risks. It's also the hardware
> that our customer reported the issue with.


> 
> --
> Kind regards,
> Luca Boccassi
Luca Boccassi Nov. 2, 2018, 5:08 p.m. UTC | #5
On Fri, 2018-11-02 at 16:49 +0000, Zhang, Qi Z wrote:
> > -----Original Message-----
> > From: Luca Boccassi [mailto:bluca@debian.org]
> > Sent: Friday, November 2, 2018 10:21 AM
> > To: Zhang, Qi Z <qi.z.zhang@intel.com>; dev@dpdk.org
> > Cc: Lu, Wenzhuo <wenzhuo.lu@intel.com>; Ananyev, Konstantin
> > <konstantin.ananyev@intel.com>; stable@dpdk.org
> > Subject: Re: [PATCH 1/2] net/ixgbe: fix x550 code to handle
> > unidentified PHY
> > 
> > On Fri, 2018-11-02 at 14:11 +0000, Zhang, Qi Z wrote:
> > > > -----Original Message-----
> > > > From: Luca Boccassi [mailto:bluca@debian.org]
> > > > Sent: Thursday, November 1, 2018 9:04 AM
> > > > To: dev@dpdk.org
> > > > Cc: Lu, Wenzhuo <wenzhuo.lu@intel.com>; Ananyev, Konstantin
> > > > <konstantin.ananyev@intel.com>; Zhang, Qi Z <qi.z.zhang@intel.c
> > > > om>;
> > > > Luca Boccassi <bluca@debian.org>; stable@dpdk.org
> > > > Subject: [PATCH 1/2] net/ixgbe: fix x550 code to handle
> > > > unidentified
> > > > PHY
> > > > 
> > > > ixgbe_identify_phy_x550em() was missing the code to handle
> > > > unidentified PHY that has been there in 82599 so it was not
> > > > able to
> > > > complete initialization of ixgbe sequence if no sfp plugged in.
> > > > Port it over to return an appropriate type and complete init
> > > > sequence properly.
> > > > 
> > > > Fixes: d2e72774e58c ("ixgbe/base: support X550")
> > > > Cc: stable@dpdk.org
> > > > 
> > > > Signed-off-by: Luca Boccassi <bluca@debian.org>
> > > > ---
> > > > v2: refresh to remove merge conflict with master
> > > > 
> > > >  drivers/net/ixgbe/base/ixgbe_x550.c | 19 ++++++++++++++++---
> > > >  1 file changed, 16 insertions(+), 3 deletions(-)
> > > > 
> > > > diff --git a/drivers/net/ixgbe/base/ixgbe_x550.c
> > > > b/drivers/net/ixgbe/base/ixgbe_x550.c
> > > > index f7b98af52..83b394861 100644
> > > > --- a/drivers/net/ixgbe/base/ixgbe_x550.c
> > > > +++ b/drivers/net/ixgbe/base/ixgbe_x550.c
> > > > @@ -315,13 +315,21 @@ STATIC void ixgbe_setup_mux_ctl(struct
> > > > ixgbe_hw
> > > > *hw)
> > > >   */
> > > >  STATIC s32 ixgbe_identify_phy_x550em(struct ixgbe_hw *hw)  {
> > > > +	s32 status;
> > > > +
> > > >  	hw->mac.ops.set_lan_id(hw);
> > > > 
> > > >  	ixgbe_read_mng_if_sel_x550em(hw);
> > > > 
> > > >  	switch (hw->device_id) {
> > > >  	case IXGBE_DEV_ID_X550EM_A_SFP:
> > > > -		return ixgbe_identify_sfp_module_X550em(hw);
> > > > +		status = ixgbe_identify_sfp_module_X550em(hw);
> > > > +		/* Set PHY type none if no PHY detected */
> > > > +		if (hw->phy.type == ixgbe_phy_unknown) {
> > > > +			hw->phy.type = ixgbe_phy_none;
> > > > +			return IXGBE_SUCCESS;
> > > > +		}
> > > 
> > > Why this can't be handled at caller, why we replace phy_unknown
> > > by
> > > phy_none only for x550?
> > 
> > Hi, thanks for the review.
> > 
> > I've moved the code into the caller in v3.
> 
> Sorry, this is not what I suggested. 
> I'm not sure for X550EM_A_SFP, it is the case that the device does
> not have PHY so we should correct it (by replace it with no_phy) 
> or it is the case that we can't identify PHY so we just replace it by
> no_phy to bypass the check for workaround?
> If the second case, is it possible not to do replacement and handle
> it at upper layer (caller or caller's caller ... ) and keep the
> information more accurate?

It is the second case - I've amended the comment in v3, the issue is
that the driver will fail to initialise if there is no SFP present at
boot (not very hotplug-friendly!).
The issue with calling it one layer above is that this function is
saved in a table, and there are at least 3-4 places where it is called
from. So it would have to be repeated many times, and also I'm not 100%
sure of all the places where to do this, as the function table is
updated several times across the base driver.

So, when you say caller or caller's caller, do you have identified
precisely where? I'm happy to change it if you can point me exactly to
the right places.

Thanks!

> > 
> > It's done for x550 simply because that's the hardware that we have
> > and that
> > this was tested with, I didn't want to take extra risks. It's also
> > the hardware
> > that our customer reported the issue with.
> 
> 
> > 
> > --
> > Kind regards,
> > Luca Boccassi

Patch
diff mbox series

diff --git a/drivers/net/ixgbe/base/ixgbe_x550.c b/drivers/net/ixgbe/base/ixgbe_x550.c
index f7b98af52..83b394861 100644
--- a/drivers/net/ixgbe/base/ixgbe_x550.c
+++ b/drivers/net/ixgbe/base/ixgbe_x550.c
@@ -315,13 +315,21 @@  STATIC void ixgbe_setup_mux_ctl(struct ixgbe_hw *hw)
  */
 STATIC s32 ixgbe_identify_phy_x550em(struct ixgbe_hw *hw)
 {
+	s32 status;
+
 	hw->mac.ops.set_lan_id(hw);
 
 	ixgbe_read_mng_if_sel_x550em(hw);
 
 	switch (hw->device_id) {
 	case IXGBE_DEV_ID_X550EM_A_SFP:
-		return ixgbe_identify_sfp_module_X550em(hw);
+		status = ixgbe_identify_sfp_module_X550em(hw);
+		/* Set PHY type none if no PHY detected */
+		if (hw->phy.type == ixgbe_phy_unknown) {
+			hw->phy.type = ixgbe_phy_none;
+			return IXGBE_SUCCESS;
+		}
+		return status;
 	case IXGBE_DEV_ID_X550EM_X_SFP:
 		/* set up for CS4227 usage */
 		ixgbe_setup_mux_ctl(hw);
@@ -329,8 +337,13 @@  STATIC s32 ixgbe_identify_phy_x550em(struct ixgbe_hw *hw)
 		/* Fallthrough */
 
 	case IXGBE_DEV_ID_X550EM_A_SFP_N:
-		return ixgbe_identify_sfp_module_X550em(hw);
-		break;
+		status = ixgbe_identify_sfp_module_X550em(hw);
+		/* Set PHY type none if no PHY detected */
+		if (hw->phy.type == ixgbe_phy_unknown) {
+			hw->phy.type = ixgbe_phy_none;
+			return IXGBE_SUCCESS;
+		}
+		return status;
 	case IXGBE_DEV_ID_X550EM_X_KX4:
 		hw->phy.type = ixgbe_phy_x550em_kx4;
 		break;