Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

classic Classic list List threaded Threaded
10 messages Options
Reply | Threaded
Open this post in threaded view
|

Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Axel Rau
Hi All,

while installing 5.5-RELEASE on a ATOM C2000 based Axiomtek NA361,
I get
em2 at pci0 dev 20 function 0 „Intel I354 SGMII“ rev 0x03: msiem2: Hardware Initialization Failedem2: Unable to initialize the hardware
on all 4 nics.

Installing a recent snapshot from 5.5-CURRENT does not show this problem.

Looking at HEAD
--- src/sys/dev/pci/if_em.c 2014/02/22 04:41:31 1.277
+++ src/sys/dev/pci/if_em.c 2014/08/26 11:01:21 1.288
shows lot og activity.

Any patch for 5.5 welcome.

Thanks, Axel
---
PGP-Key:29E99DD6  ☀ +49 151 2300 9283  ☀ computing @ chaos claudius

Reply | Threaded
Open this post in threaded view
|

Re: Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Gregor Best
Hi Axel,

since you seem to be deploying a new setup, I'd simply install a
snapshot. The release of 5.6 is soon(-ish), so I doubt there will
be lots of functional changes until then, and it'd be wise to upgrade
anyway once 5.6 is out.

--
        Gregor Best

Reply | Threaded
Open this post in threaded view
|

Re: Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Axel Rau
Am 28.08.2014 um 12:36 schrieb Gregor Best <[hidden email]>:

> since you seem to be deploying a new setup, I'd simply install a
> snapshot. The release of 5.6 is soon(-ish), so I doubt there will
> be lots of functional changes until then, and it'd be wise to upgrade
> anyway once 5.6 is out.
Ports/packages are not yet ready for 5.6 and I wanted to avoid the porting
effort, which was significant in the past.

I’m just trying a patch against CURRENT. . .

Axel
-—
PGP-Key:29E99DD6  ☀ +49 151 2300 9283  ☀ computing @ chaos claudius

Reply | Threaded
Open this post in threaded view
|

Re: Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Jonathan Gray-11
In reply to this post by Axel Rau
On Thu, Aug 28, 2014 at 11:50:32AM +0200, Axel Rau wrote:

> Hi All,
>
> while installing 5.5-RELEASE on a ATOM C2000 based Axiomtek NA361,
> I get
> em2 at pci0 dev 20 function 0 „Intel I354 SGMII“ rev 0x03: msiem2: Hardware Initialization Failedem2: Unable to initialize the hardware
> on all 4 nics.
>
> Installing a recent snapshot from 5.5-CURRENT does not show this problem.
>
> Looking at HEAD
> --- src/sys/dev/pci/if_em.c 2014/02/22 04:41:31 1.277
> +++ src/sys/dev/pci/if_em.c 2014/08/26 11:01:21 1.288
> shows lot og activity.
>
> Any patch for 5.5 welcome.

Start with the following patch, perhaps there needs to be
some additional i347 specific handling.

Index: if_em.h
===================================================================
RCS file: /cvs/src/sys/dev/pci/if_em.h,v
retrieving revision 1.53
diff -u -p -r1.53 if_em.h
--- if_em.h 22 Jul 2014 13:12:11 -0000 1.53
+++ if_em.h 28 Aug 2014 11:40:48 -0000
@@ -248,7 +248,7 @@ typedef int boolean_t;
 #define SPEED_MODE_BIT (1<<21) /* On PCI-E MACs only */
 
 /* Defines for printing debug information */
-#define DEBUG_INIT 0
+#define DEBUG_INIT 1
 #define DEBUG_IOCTL 0
 #define DEBUG_HW 0
 
Index: if_em_osdep.h
===================================================================
RCS file: /cvs/src/sys/dev/pci/if_em_osdep.h,v
retrieving revision 1.12
diff -u -p -r1.12 if_em_osdep.h
--- if_em_osdep.h 5 Oct 2011 02:52:10 -0000 1.12
+++ if_em_osdep.h 28 Aug 2014 11:41:34 -0000
@@ -44,7 +44,8 @@ POSSIBILITY OF SUCH DAMAGE.
 
 #define MSGOUT(S, A, B) printf(S "\n", A, B)
 #define DEBUGFUNC(F) DEBUGOUT(F);
-#ifdef DBG
+//#ifdef DBG
+#if 1
  #define DEBUGOUT(S) printf(S "\n")
  #define DEBUGOUT1(S,A) printf(S "\n",A)
  #define DEBUGOUT2(S,A,B) printf(S "\n",A,B)

Reply | Threaded
Open this post in threaded view
|

Re: Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Axel Rau
Am 28.08.2014 um 13:51 schrieb Jonathan Gray <[hidden email]>:

> Start with the following patch, perhaps there needs to be
> some additional i347 specific handling.
I’m seeing now:
- - -
em_set_phy_type
Invalid PHY ID 0x1410DC0
 
Error, did not detect valid phy.
 
em2: Hardware Initialization Failedem2: Unable to initialize the hardware
- - -
Complete debug output available on request.

Axel
---
PGP-Key:29E99DD6  ☀ +49 151 2300 9283  ☀ computing @ chaos claudius

Reply | Threaded
Open this post in threaded view
|

Re: Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Jonathan Gray-11
On Thu, Aug 28, 2014 at 09:21:15PM +0200, Axel Rau wrote:

> Am 28.08.2014 um 13:51 schrieb Jonathan Gray <[hidden email]>:
>
> > Start with the following patch, perhaps there needs to be
> > some additional i347 specific handling.
> I’m seeing now:
> - - -
> em_set_phy_type
> Invalid PHY ID 0x1410DC0
>  
> Error, did not detect valid phy.
>  
> em2: Hardware Initialization Failedem2: Unable to initialize the hardware
> - - -
> Complete debug output available on request.

Initial support for the i347 phy was added back in March but that wasn't
part of 5.5.  I suspect you want something along the lines of the
following patch:

Index: if_em_hw.c
===================================================================
RCS file: /cvs/src/sys/dev/pci/if_em_hw.c,v
retrieving revision 1.77
diff -u -p -r1.77 if_em_hw.c
--- if_em_hw.c 22 Feb 2014 04:41:31 -0000 1.77
+++ if_em_hw.c 29 Aug 2014 06:02:46 -0000
@@ -225,8 +225,10 @@ em_set_phy_type(struct em_hw *hw)
  case M88E1000_I_PHY_ID:
  case M88E1011_I_PHY_ID:
  case M88E1111_I_PHY_ID:
+ case M88E1112_E_PHY_ID:
  case M88E1543_E_PHY_ID:
  case I210_I_PHY_ID:
+ case I347AT4_E_PHY_ID:
  hw->phy_type = em_phy_m88;
  break;
  case IGP01E1000_I_PHY_ID:
@@ -5199,7 +5201,9 @@ em_match_gig_phy(struct em_hw *hw)
  case em_i350:
  if (hw->phy_id == I82580_I_PHY_ID ||
     hw->phy_id == I210_I_PHY_ID ||
+    hw->phy_id == I347AT4_E_PHY_ID ||
     hw->phy_id == I350_I_PHY_ID ||
+    hw->phy_id == M88E1112_E_PHY_ID ||
     hw->phy_id == M88E1543_E_PHY_ID) {
  uint32_t mdic;
 
Index: if_em_hw.h
===================================================================
RCS file: /cvs/src/sys/dev/pci/if_em_hw.h,v
retrieving revision 1.58
diff -u -p -r1.58 if_em_hw.h
--- if_em_hw.h 22 Feb 2014 04:41:32 -0000 1.58
+++ if_em_hw.h 29 Aug 2014 06:02:46 -0000
@@ -3381,6 +3381,8 @@ struct em_host_command_info {
 #define M88E1000_14_PHY_ID   M88E1000_E_PHY_ID
 #define M88E1011_I_REV_4     0x04
 #define M88E1111_I_PHY_ID    0x01410CC0
+#define M88E1112_E_PHY_ID    0x01410C90
+#define I347AT4_E_PHY_ID     0x01410DC0
 #define L1LXT971A_PHY_ID     0x001378E0
 #define GG82563_E_PHY_ID     0x01410CA0
 #define BME1000_E_PHY_ID     0x01410CB0

Reply | Threaded
Open this post in threaded view
|

[RESOLVED] Re: Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Axel Rau
Am 29.08.2014 um 08:11 schrieb Jonathan Gray <[hidden email]>:

> Initial support for the i347 phy was added back in March but that wasn't
> part of 5.5.  I suspect you want something along the lines of the
> following patch:
Yes, this patch worked (does at least initialization of em2-em5,
more testing to follow).

Thanks, Axel
---
PGP-Key:29E99DD6  ☀ +49 151 2300 9283  ☀ computing @ chaos claudius

Reply | Threaded
Open this post in threaded view
|

Re: Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Axel Rau
Am 30.08.2014 um 13:46 schrieb Axel Rau <[hidden email]>:

> Am 29.08.2014 um 08:11 schrieb Jonathan Gray <[hidden email]>:
>
>> Initial support for the i347 phy was added back in March but that wasn't
>> part of 5.5.  I suspect you want something along the lines of the
>> following patch:
> Yes, this patch worked (does at least initialization of em2-em5,
> more testing to follow).
Next problem shows up with sppp over vlan from MAC 00:60:e0:5a:75:45:
- - - -
13:49:38.170666 00:60:e0:5a:75:39 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q (0x8100)
                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADI
                [Service-Name] [Host-Uniq 0x95F818D3]
13:49:38.313082 00:30:88:1f:18:9a > 00:60:e0:5a:75:39, ethertype 802.1Q (0x8100)
                , length 87: vlan 7, p 6, ethertype PPPoE D, PPPoE PADO
                [Host-Uniq 0x95F818D3] [AC-Name "FFMR71-se800-B2224180702381"]
                [AC-Cookie "..lvg}..Bv.>.;8."] [Service-Name]
13:49:38.313093 00:60:e0:5a:75:39 > 00:30:75:39:00:30, ethertype 802.1Q (0x8100)
                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADR
                [Service-Name] [AC-Cookie "..lvg}..Bv.>.;8."]
                [Host-Uniq 0x95F818D3]
13:49:43.310779 00:60:e0:5a:75:39 > 00:30:75:39:00:30, ethertype 802.1Q (0x8100)
                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADR
                [Service-Name] [AC-Cookie "..lvg}..Bv.>.;8."]
                [Host-Uniq 0x95F818D3]
13:49:53.311256 00:60:e0:5a:75:39 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q (0x8100)
                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADI
                [Service-Name] [Host-Uniq 0x95F818D3]
13:49:53.339482 00:30:88:1f:18:9a > 00:60:e0:5a:75:39, ethertype 802.1Q (0x8100)
                , length 87: vlan 7, p 6, ethertype PPPoE D, PPPoE PADO
                [Host-Uniq 0x95F818D3] [AC-Name "FFMR71-se800-B2224180702381"]
                [AC-Cookie "..lvg}..Bv.>.;8."] [Service-Name]
13:49:53.339492 00:60:e0:5a:75:39 > 00:30:75:39:00:30, ethertype 802.1Q (0x8100)
                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADR
                [Service-Name] [AC-Cookie "..lvg}..Bv.>.;8."]
                [Host-Uniq 0x95F818D3]
13:50:01.214264 00:60:e0:5a:75:39 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q (0x8100)
                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADI
                [Service-Name] [Host-Uniq 0x95F818D3]
- - -
The i347 device (em5) has a hardware-MAC of 00:60:e0:5a:75:45, but shows up
above as 00:60:e0:5a:75:39.
The answer to the pppoe server with MAC address 00:30:88:1f:18:9a is being
sent to MAC 00:30:75:39:00:30 instead.

Do I need more patches (perhaps VLAN related) for the i347 ?

Any help welcome,
Axel
---
PGP-Key:29E99DD6  ☀ +49 151 2300 9283  ☀ computing @ chaos claudius

Reply | Threaded
Open this post in threaded view
|

Re: Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Axel Rau
Am 02.09.2014 um 15:45 schrieb Axel Rau <[hidden email]>:

> The i347 device (em5) has a hardware-MAC of 00:60:e0:5a:75:45, but shows up
> above as 00:60:e0:5a:75:39.
> The answer to the pppoe server with MAC address 00:30:88:1f:18:9a is being
> sent to MAC 00:30:75:39:00:30 instead.
Nobody any idea?

Axel
---
PGP-Key:29E99DD6  ☀ +49 151 2300 9283  ☀ computing @ chaos claudius

Reply | Threaded
Open this post in threaded view
|

Re: Intel i354 Quad GbE network adapter failed on 5.5-RELEASE

Axel Rau
In reply to this post by Axel Rau
I tested this on other hardware: It has nothing to do with i354.
It’s a bug in the vlan driver which has already been reported here
        http://marc.info/?l=openbsd-misc&m=139903544321689&w=2

Axel

Am 02.09.2014 um 15:45 schrieb Axel Rau <[hidden email]>:

> Am 30.08.2014 um 13:46 schrieb Axel Rau <[hidden email]>:
>
>> Am 29.08.2014 um 08:11 schrieb Jonathan Gray <[hidden email]>:
>>
>>> Initial support for the i347 phy was added back in March but that wasn't
>>> part of 5.5.  I suspect you want something along the lines of the
>>> following patch:
>> Yes, this patch worked (does at least initialization of em2-em5,
>> more testing to follow).
> Next problem shows up with sppp over vlan from MAC 00:60:e0:5a:75:45:
> - - - -
> 13:49:38.170666 00:60:e0:5a:75:39 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q (0x8100)
>                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADI
>                [Service-Name] [Host-Uniq 0x95F818D3]
> 13:49:38.313082 00:30:88:1f:18:9a > 00:60:e0:5a:75:39, ethertype 802.1Q (0x8100)
>                , length 87: vlan 7, p 6, ethertype PPPoE D, PPPoE PADO
>                [Host-Uniq 0x95F818D3] [AC-Name "FFMR71-se800-B2224180702381"]
>                [AC-Cookie "..lvg}..Bv.>.;8."] [Service-Name]
> 13:49:38.313093 00:60:e0:5a:75:39 > 00:30:75:39:00:30, ethertype 802.1Q (0x8100)
>                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADR
>                [Service-Name] [AC-Cookie "..lvg}..Bv.>.;8."]
>                [Host-Uniq 0x95F818D3]
> 13:49:43.310779 00:60:e0:5a:75:39 > 00:30:75:39:00:30, ethertype 802.1Q (0x8100)
>                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADR
>                [Service-Name] [AC-Cookie "..lvg}..Bv.>.;8."]
>                [Host-Uniq 0x95F818D3]
> 13:49:53.311256 00:60:e0:5a:75:39 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q (0x8100)
>                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADI
>                [Service-Name] [Host-Uniq 0x95F818D3]
> 13:49:53.339482 00:30:88:1f:18:9a > 00:60:e0:5a:75:39, ethertype 802.1Q (0x8100)
>                , length 87: vlan 7, p 6, ethertype PPPoE D, PPPoE PADO
>                [Host-Uniq 0x95F818D3] [AC-Name "FFMR71-se800-B2224180702381"]
>                [AC-Cookie "..lvg}..Bv.>.;8."] [Service-Name]
> 13:49:53.339492 00:60:e0:5a:75:39 > 00:30:75:39:00:30, ethertype 802.1Q (0x8100)
>                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADR
>                [Service-Name] [AC-Cookie "..lvg}..Bv.>.;8."]
>                [Host-Uniq 0x95F818D3]
> 13:50:01.214264 00:60:e0:5a:75:39 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q (0x8100)
>                , length 60: vlan 7, p 3, ethertype PPPoE D, PPPoE PADI
>                [Service-Name] [Host-Uniq 0x95F818D3]
> - - -
> The i347 device (em5) has a hardware-MAC of 00:60:e0:5a:75:45, but shows up
> above as 00:60:e0:5a:75:39.
> The answer to the pppoe server with MAC address 00:30:88:1f:18:9a is being
> sent to MAC 00:30:75:39:00:30 instead.
>
> Do I need more patches (perhaps VLAN related) for the i347 ?
>
> Any help welcome,
> Axel
> ---
> PGP-Key:29E99DD6  ☀ +49 151 2300 9283  ☀ computing @ chaos claudius
>

---
PGP-Key:29E99DD6  ☀ +49 151 2300 9283  ☀ computing @ chaos claudius