http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c2 --- Comment #2 from Ronald Maas --- (In reply to Matthias Brugger from comment #1)

Did you tried with 4.5.0-3-default?

Yes. Tried today with latest snapshot http://download.opensuse.org/ports/aarch64/tumbleweed/iso/openSUSE-Tumblewee...

I just tried this kernel on xgene1 and wasn't able to reproduce the behaviour. Do you see any significant information in the logs?

... url mount: trying http://download.opensuse.org/ports/aa64/tubleweed/repo/oss?device=eth0 (path = ports/aa64/tubleweed/repo/oss/) (ARM ethernet 0) url_setup_device: http://download.opensuse.org/ports/aa64/tubleweed/repo/oss?device=eth0 interface setup: eth0 ifcfg_write: device = eth0, ifcfg = (null) creating ifcfg-eth0 BOOTPROTO='dhcp' STARTMODE='auto' network setup failed ... repository: not found no openSUSE Tumbleweed repository found Automatic setup not possible. Only eth0 (1 Gbps ethernet adapter) is used. The other 1 Gbps ethernet and 2 x 10 Gbps ethernet adapters are not connected. I tried swapping eth0 and eth1 but that made no difference. Note someone else experienced the exact same issue: https://lists.centos.org/pipermail/arm-dev/2016-April/001969.html -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 Matthias Brugger changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |alau@suse.com -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 Dirk Mueller changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |ronaldjmaas@gmail.com Flags| |needinfo?(ronaldjmaas@gmail | |.com) -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c4 --- Comment #4 from Dirk Mueller --- Also I wonder, does the interface look like it has a proper MAC address assigned when it is being loaded as a module? can you also provide the dmesg output (if any) when loading the module and or the boot.log ? -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c6 --- Comment #6 from Ronald Maas --- (In reply to Ronald Maas from comment #5)

(In reply to Dirk Mueller from comment #3)

...

There is a typo here (/ports/aarch64/tumbleweed/..)

Yes I typed it manually by looking at the screen

...

...

(ARM ethernet 0) url_setup_device: http://download.opensuse.org/ports/aa64/tubleweed/repo/oss?device=eth0 interface setup: eth0 ifcfg_write: device = eth0, ifcfg = (null) creating ifcfg-eth0 BOOTPROTO='dhcp' STARTMODE='auto' network setup failed

is this generated ? are you in the installed system or in the installer still?

I started the NET installer. But without network connection it was not able to progress. I found this errors on one of the other terminals (Ctrl+Alt+F3 or Ctrl+Alt+F4)

Yes it has the proper MAC address(In reply to Ronald Maas from comment #5)

(In reply to Dirk Mueller from comment #3)

...

There is a typo here (/ports/aarch64/tumbleweed/..)

Yes I typed it manually by looking at the screen

...

...

(ARM ethernet 0) url_setup_device: http://download.opensuse.org/ports/aa64/tubleweed/repo/oss?device=eth0 interface setup: eth0 ifcfg_write: device = eth0, ifcfg = (null) creating ifcfg-eth0 BOOTPROTO='dhcp' STARTMODE='auto' network setup failed

is this generated ? are you in the installed system or in the installer still?

I started the NET installer. But without network connection it was not able to progress. I found this errors on one of the other terminals (Ctrl+Alt+F3 or Ctrl+Alt+F4)

(In reply to Dirk Mueller from comment #4)

Also I wonder, does the interface look like it has a proper MAC address assigned when it is being loaded as a module?

I reinstalled using the latest Tumbleweed installer DVD (dated 17-Apr-2016). And except for not being able to recognize the network, I was able to successfully install and boot OpenSuse. eth0 was configured to use DHCP. After reboot, no IP address was obtained from DHCP. Running ifup eth0 did not return an error but no IP address was obtained either. ifstatus eth0 -o debug returned the correct MAC address.

can you also provide the dmesg output (if any) when loading the module and or the boot.log ?

Not able to find boot.log. But here some other files that may help: dmesg http://pastebin.com/8FET235k lsmod http://pastebin.com/gcKfDaDp ifstatus http://pastebin.com/CCfQgKdm ifcfg-eth0 http://pastebin.com/LbsvriW7 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 Dirk Mueller changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |loic.devulder@laposte.net -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c8 --- Comment #8 from Matthias Brugger --- (In reply to Loic Devulder from comment #7)

Hello all,

I have the same problem with 2 Gigabyte R120-P30 servers (MP30-AR0 motherboard).

After flashing the same version of Tianocore uEFI, I was able to install the openSUSE Tumbleweed but I don't have network connectivity :-(

I have recompiled the kernel on one server to directly include the xgene-enet driver and I can confirm that the network stack is working well after this change.

I let the other server as is for the moment.

I can't give access to this server :-( but if you need to do some test or see the difference on a working or non-working server you may ask me.

Can you try to unload and then load the driver module "xgene-enet" and see if this changes anything? -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c10 --- Comment #10 from Alexander Graf --- I'll leave it to Matthias to comment on the network bug, but for the shim-install one, a fix is on its way: https://github.com/yast/yast-bootloader/pull/330 Thanks for reporting all of these issues! -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c12 --- Comment #12 from Loic Devulder --- (In reply to Matthias Brugger from comment #8)

(In reply to Loic Devulder from comment #7)

...

Hello all,

I have the same problem with 2 Gigabyte R120-P30 servers (MP30-AR0 motherboard).

After flashing the same version of Tianocore uEFI, I was able to install the openSUSE Tumbleweed but I don't have network connectivity :-(

I have recompiled the kernel on one server to directly include the xgene-enet driver and I can confirm that the network stack is working well after this change.

I let the other server as is for the moment.

I can't give access to this server :-( but if you need to do some test or see the difference on a working or non-working server you may ask me.

Can you try to unload and then load the driver module "xgene-enet" and see if this changes anything?

I have tried like Ronald and I also got 2 segfaults. I've attached the logfile to this bug. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c14 --- Comment #14 from Loic Devulder --- Created attachment 676107 --> http://bugzilla.suse.com/attachment.cgi?id=676107&action=edit rmmod xgene_enet segfaults -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c16 --- Comment #16 from Ronald Maas --- This is information I got from Phong Vo (AppliedMicro): ------------ Yes, it's a known issue and we are working on it. It's not specific to MP30-AR0 board. xgene_enet driver would need to be compiled as built-in for now. -Phong ------------ I do not have a date when the fix will be available on mainline. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c18 --- Comment #18 from Ronald Maas --- (In reply to Matthias Brugger from comment #17)

This is a bug in the clk tree implementation.

As a workaround you can pass "clk_ignore_unused" as kernel parameter via grub.

After adding "clk_ignore_unused" as kernel parameter, all reported network issues are resolved. After installation and reboot, OPenSuse Tumbleweed was able to obtain a valid IP address from DHCP. Both zypper dup and incoming ssh connections work fine. ISO http://download.opensuse.org/ports/aarch64/tumbleweed/iso/openSUSE-Tumblewee... was used for the installation. Thanks -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c20 --- Comment #20 from Ronald Maas --- (In reply to Matthias Brugger from comment #19)

(In reply to Matthias Brugger from comment #17)

...

This is a bug in the clk tree implementation.

After digging deeper into this, I don't think that it is a clk tree implementation bug, but a bug in the device tree node passed from the FW to the kernel.

Can you please check the status of the xgene-menet node: cat /proc/device-tree/soc/ethernet@17020000/status

I can see this node is disabled on the Gigabyte machine. This leads to the fact, that the kernel does not enable the clocks connected with this device (ethclk, menetclk).

Can you please double check this on your machine, as I suppose that the FW is slightly different.

There are 5 ethernet adapters on the board: ls -la /proc/device-tree/soc/ | grep ether drwxr-xr-x 3 root root 0 May 16 19:49 ethernet@17020000 ---> BMC management drwxr-xr-x 2 root root 0 May 16 19:50 ethernet@1f210000 ---> eth0 1Gbps drwxr-xr-x 2 root root 0 May 16 19:50 ethernet@1f210030 ---> eth1 1Gbps drwxr-xr-x 2 root root 0 May 16 19:50 ethernet@1f610000 ---> eth2 10Gbps drwxr-xr-x 2 root root 0 May 16 19:50 ethernet@1f620000 ---> eth3 10Gbps cat /proc/device-tree/soc/ethernet@17020000/status disabled cat /proc/device-tree/soc/ethernet@1f210000/status ok cat /proc/device-tree/soc/ethernet@1f210030/status ok cat /proc/device-tree/soc/ethernet@1f610000/status ok cat /proc/device-tree/soc/ethernet@1f620000/status ok So only the ethernet adapter used to connect to the BMC web console is disabled. The other ethernet adapters eth0 up to eth3 are properly enabled by UEFI. Same result with and without "clk_ignore_unused" kernel parameter. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c22 --- Comment #22 from Matthias Brugger --- (In reply to Loic Devulder from comment #21)

The device tree bug sounds good by why all are working well when the driver is compiled in the kernel?

My theory is, that the clocks get enabled correctly by the FW. All clocks that got not enabled by any driver will be explicitly disabled by the common clock framework. When the driver is build in the configuration of the network device is done before the clocks get disabled. In the case the driver is loaded as a module, the initialization of the network device is done after the common clock framework disabled the unused clocks. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c24 --- Comment #24 from Ronald Maas --- (In reply to Matthias Brugger from comment #23)

Can you please provide "cat /sys/kernel/debug/clk/clk_summary"

clock enable_cnt prepare_cnt rate accuracy phase ---------------------------------------------------------------------------------------- refclk 1 1 100000000 0 0 socpll 1 1 2000000000 0 0 socplldiv2 10 10 1000000000 0 0 pcie4clk 1 1 1000000000 0 0 pcie3clk 1 1 1000000000 0 0 pcie2clk 1 1 1000000000 0 0 pcie1clk 1 1 1000000000 0 0 pcie0clk 1 1 1000000000 0 0 rngpkaclk 1 1 1000000000 0 0 sata01clk 0 0 1000000000 0 0 xge1clk 1 1 1000000000 0 0 xge0clk 1 1 1000000000 0 0 sge1clk 1 1 1000000000 0 0 sge0clk 1 1 1000000000 0 0 ethclk 0 0 125000000 0 0 menetclk 0 0 125000000 0 0 qmlclk 0 0 1000000000 0 0 pcppll 0 0 2400000000 0 0 No change when booted with and without "clk_ignore_unused" kernel parameter.

How many network devices do you see?

YAST Hardware Information reports 4 network adapters: eth0, eth1, eth2, eth3

When booting with clk_ignore_unused, did you try to bring up the network manually, e.g. "wicked ifup eth0"?

Network starts automatically after reboot - getting IP address from DHCP server. I do not need to run any commands. Furthermore I did not make any changes to the existing network configuration. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c26 --- Comment #26 from Ronald Maas --- I installed the Leap 42.2 kernel using command: rpm -i --oldpackage kernel-default/kernel-default-4.4.11-2.1.aarch64.rpm But after selecting this kernel from Grub boot menu and removing clk_ignore_unused kernel parameter, it does not boot successfully. Instead it prints 1000s of lines like: [ 333.610762] pcieport 0002:00:00.0: PCIe Bus Error: severity=Uncorrected (Non-Fatal), type=Transaction Layer, id=0000(Requester ID) [ 333.610764] pcieport 0002:00:00.0: device [10e8:e004] error status/mask=00100000/00000000 [ 333.610765] pcieport 0002:00:00.0: [20] Unsupported Request (First) [ 333.610767] pcieport 0002:00:00.0: TLP Header: 60001008 000000ff 00000090 80000000 [ 333.610778] pcieport 0002:00:00.0: AER: Device recovery failed [ 333.810740] pcieport 0002:00:00.0: AER: Multiple Uncorrected (Non-Fatal) error received: id=0000 Note I used the Tumbleweed distribution with all the latest updates. I probably do something wrong. Any advise? -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c28 --- Comment #28 from Ronald Maas --- (In reply to Matthias Brugger from comment #27)

(In reply to Ronald Maas from comment #26)

...

I installed the Leap 42.2 kernel using command:

rpm -i --oldpackage kernel-default/kernel-default-4.4.11-2.1.aarch64.rpm

But after selecting this kernel from Grub boot menu and removing clk_ignore_unused kernel parameter, it does not boot successfully. Instead it prints 1000s of lines like:

[ 333.610762] pcieport 0002:00:00.0: PCIe Bus Error: severity=Uncorrected (Non-Fatal), type=Transaction Layer, id=0000(Requester ID) [ 333.610764] pcieport 0002:00:00.0: device [10e8:e004] error status/mask=00100000/00000000 [ 333.610765] pcieport 0002:00:00.0: [20] Unsupported Request (First) [ 333.610767] pcieport 0002:00:00.0: TLP Header: 60001008 000000ff 00000090 80000000 [ 333.610778] pcieport 0002:00:00.0: AER: Device recovery failed [ 333.810740] pcieport 0002:00:00.0: AER: Multiple Uncorrected (Non-Fatal) error received: id=0000

Note I used the Tumbleweed distribution with all the latest updates.

That shouldn't be a problem.

...

I probably do something wrong. Any advise?

Which firmware version do you use? Are you still using the Tianocore version you mentioned in your first comment?

yes -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c30 --- Comment #30 from Ronald Maas --- I can also confirm the networking works fine with latest Tumbleweed kernel 4.6.0-1-default using the AMI APTIO D1e UEFI firmware without kernel parameter "clk_ignore_unused". I also tested with Leap 42.2 kernel 4.4.11-2-default with the same firmware, but networking only works if "clk_ignore_unused" is provided. Without this, the network adapter fails to obtain a valid IP address. dmesg output for 4.4.11-2-default: http://pastebin.com/MD8R2EhZ Thanks for all the help resolving the issue. If you need any assistence for testing Leap 42.2 on Gigabyte MP30-AR0 board, just let me know. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.suse.com/show_bug.cgi?id=973756 http://bugzilla.suse.com/show_bug.cgi?id=973756#c32 --- Comment #32 from Matthias Brugger --- (In reply to Ronald Maas from comment #30)

Thanks for all the help resolving the issue. If you need any assistence for testing Leap 42.2 on Gigabyte MP30-AR0 board, just let me know.

Can you try to unload/load the module? If this helps, would you mind to open a new bug for Leap 42.2 with your findings? -- You are receiving this mail because: You are on the CC list for the bug.