[dpdk-users] i40e with DPDK exits abruptly in rte_eal_init()

Xu, Qian Q qian.q.xu at intel.com
Tue Jan 26 02:09:22 CET 2016


Mishra
Could you tell me your exact command when running primary and secondary commands? Do you change the code when running dpdk app? Thx.


Thanks
Qian

From: Saurabh Mishra [mailto:saurabh.globe at gmail.com]
Sent: Tuesday, January 26, 2016 1:20 AM
To: Xu, Qian Q
Cc: Zhang, Helin; Thomas Monjalon; users at dpdk.org; Mcnamara, John
Subject: Re: [dpdk-users] i40e with DPDK exits abruptly in rte_eal_init()

Hi Qian --

>Mishra, could you help answer below questions?
> What’s your i40e’s vmware driver on host?

We are doing PCI-passthrough of i40e.

vmnic7  0000:07:00.1 i40e        Up   10000Mbps Full   3c:fd:fe:04:11:c2 1500   Intel Corporation Ethernet Controller X710 for 10GbE SFP+
[root at localhost:~] vmkload_mod -s i40e |grep Version
 Version: Version 1.3.38, Build: 1331820, Interface: 9.2 Built on: Aug  5 2015
[root at localhost:~] vmkload_mod -s i40e
vmkload_mod module information
 input file: /usr/lib/vmware/vmkmod/i40e
 Version: Version 1.3.38, Build: 1331820, Interface: 9.2 Built on: Aug  5 2015
 License: GPL
 Required name-spaces:
  com.vmware.driverAPI#9.2.2.0
  com.vmware.vmkapi#v2_2_0_0
 Parameters:
  skb_mpool_max: int
    Maximum attainable private socket buffer memory pool size for the driver.
  skb_mpool_initial: int
    Driver's minimum private socket buffer memory pool size.
  heap_max: int
    Maximum attainable heap size for the driver.
  heap_initial: int
    Initial heap size allocated for the driver.
  debug: int
    Debug level (0=none,...,16=all)
  RSS: array of int
    Number of Receive-Side Scaling Descriptor Queues: 0 = disable/default, 1-4 = enable (number of cpus)
  VMDQ: array of int
    Number of Virtual Machine Device Queues: 0/1 = disable, 2-16 enable (default =I40E_ESX_DEFAULT_VMDQ)
  max_vfs: array of int
    Number of Virtual Functions: 0 = disable (default), 1-128 = enable this many VFs
[root at localhost:~]
>What’s the firmware of FVL? Is your FVL 4X10G or 2x10G?
Hmm..how do I check that? It has two ports.

>Could you tell us your detailed test step? I only launch one dpdk app, no secondary process, but your case have 2 processes. When pass through the PCI device, I just pass through 2 ports to the VM and start the VM in my case.

So I was passing one port to VM. I do launch primary and a secondary process.

This is the error we got. It seems we were not able to map to the VA. So I moved the rte_eal_init() and mbuf alloc to a separate process and now it seems to work with one primary and six secondary process.

EAL:   PCI memory mapped at 0x7ffff7faa000
EAL: Cannot mmap device resource file /sys/bus/pci/devices/0000:1b:00.0/resource3 to address: 0x7ffff7fac000
EAL: Error - exiting with code: 1
  Cause: Requested device 0000:1b:00.0 cannot be used


Thanks,
/Saurabh


On Sun, Jan 24, 2016 at 5:17 PM, Xu, Qian Q <qian.q.xu at intel.com<mailto:qian.q.xu at intel.com>> wrote:
We have tried the I40E driver with FVL4 on ESXi in r2.2 testing and PCI pass through can work but SRIOV will have issues. I didn’t see Mishra’s error . Now we are stuck at setting up the ESXi environment, and need
Vmware’s support to set up the env again.
Mishra, could you help answer below questions?

1.       What’s your i40e’s vmware driver on host?

2.       What’s the firmware of FVL? Is your FVL 4X10G or 2x10G?

3.       Could you tell us your detailed test step? I only launch one dpdk app, no secondary process, but your case have 2 processes. When pass through the PCI device, I just pass through 2 ports to the VM and start the VM in my case.

Helin, rombar=0 seems for KVM only, on VMWARE, not see the issue.


Thanks
Qian

From: Zhang, Helin
Sent: Sunday, January 24, 2016 11:17 AM
To: Saurabh Mishra; Thomas Monjalon; Xu, Qian Q
Cc: users at dpdk.org<mailto:users at dpdk.org>; Mcnamara, John
Subject: RE: [dpdk-users] i40e with DPDK exits abruptly in rte_eal_init()

Hi Saurabh

Were you talking about assigning NIC PF to the VMWARE guest?
I remember that there is an issue of assigning PF to KVM guest, then rombar=0 is needed for grub.
Thank you for the issue report!

Can Qian tell the details of that?

Regards,
Helin

From: Saurabh Mishra [mailto:saurabh.globe at gmail.com]
Sent: Saturday, January 23, 2016 1:30 AM
To: Thomas Monjalon <thomas.monjalon at 6wind.com<mailto:thomas.monjalon at 6wind.com>>
Cc: users at dpdk.org<mailto:users at dpdk.org>; Zhang, Helin <helin.zhang at intel.com<mailto:helin.zhang at intel.com>>; Mcnamara, John <john.mcnamara at intel.com<mailto:john.mcnamara at intel.com>>
Subject: Re: [dpdk-users] i40e with DPDK exits abruptly in rte_eal_init()

Hi,

Thanks Thomas.. So this is the error we see on i40e PCI-passthrough of whole NIC:


Secondary Process:

EAL: Detected lcore 0 as core 0 on socket 0
EAL: Detected lcore 1 as core 0 on socket 0
EAL: Support maximum 128 logical core(s) by configuration.
EAL: Detected 2 lcore(s)
EAL: Setting up physically contiguous memory...
EAL: Analysing 1024 files
EAL: Mapped segment 0 of size 0x22800000
EAL: Mapped segment 1 of size 0x200000
EAL: Mapped segment 2 of size 0x200000
EAL: Mapped segment 3 of size 0x57600000
EAL: Mapped segment 4 of size 0x400000
EAL: Mapped segment 5 of size 0x400000
EAL: Mapped segment 6 of size 0x400000
EAL: Mapped segment 7 of size 0x200000
EAL: Mapped segment 8 of size 0x2200000
EAL: Mapped segment 9 of size 0x200000
EAL: Mapped segment 10 of size 0x800000
EAL: Mapped segment 11 of size 0x600000
EAL: Mapped segment 12 of size 0x800000
EAL: Mapped segment 13 of size 0xa00000
EAL: Mapped segment 14 of size 0x400000
EAL: Mapped segment 15 of size 0x200000
EAL: Mapped segment 16 of size 0x200000
EAL: Mapped segment 17 of size 0x200000
EAL: Mapped segment 18 of size 0x200000
EAL: memzone_reserve_aligned_thread_unsafe(): memzone <RG_MP_log_history> already exists
RING: Cannot reserve memory
EAL: TSC frequency is ~1799997 KHz
EAL: Master lcore 1 is ready (tid=f7fe78c0;cpuset=[1])
EAL: PCI device 0000:03:00.0 on NUMA socket 0
EAL:   probe driver: 15ad:7b0 rte_vmxnet3_pmd
EAL:   Not managed by a supported kernel driver, skipped
EAL: PCI device 0000:1b:00.0 on NUMA socket 0
EAL:   probe driver: 8086:1572 rte_i40e_pmd
EAL:   PCI memory mapped at 0x7fff6f1f3000
EAL:   PCI memory mapped at 0x7ffff7faa000
EAL: Cannot mmap device resource file /sys/bus/pci/devices/0000:1b:00.0/resource3 to address: 0x7ffff7fac000
EAL: Error - exiting with code: 1
  Cause: Requested device 0000:1b:00.0 cannot be used


# ./dpdk-2.2.0/tools/dpdk_nic_bind.py --status

Network devices using DPDK-compatible driver
============================================
0000:1b:00.0 'Device 1572' drv=uio_pci_generic unused=i40e

Network devices using kernel driver
===================================
0000:03:00.0 'VMXNET3 Ethernet Controller' if=eth0 drv=vmxnet3 unused=uio_pci_generic *Active*

Other network devices
=====================
<none>

# grep Huge /proc/meminfo
AnonHugePages:    118784 kB
HugePages_Total:    1024
HugePages_Free:        0
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
#


Primary Process:

EAL: Detected lcore 0 as core 0 on socket 0
EAL: Detected lcore 1 as core 0 on socket 0
EAL: Support maximum 128 logical core(s) by configuration.
EAL: Detected 2 lcore(s)
EAL: Setting up physically contiguous memory...
EAL: cannot open /proc/self/numa_maps, consider that all memory is in socket_id 0
EAL: Ask a virtual area of 0x22800000 bytes
EAL: Virtual area found at 0x7fffd0a00000 (size = 0x22800000)
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7fffd0600000 (size = 0x200000)
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7fffd0200000 (size = 0x200000)
EAL: Ask a virtual area of 0x57600000 bytes
EAL: Virtual area found at 0x7fff78a00000 (size = 0x57600000)
EAL: Ask a virtual area of 0x400000 bytes
EAL: Virtual area found at 0x7fff78400000 (size = 0x400000)
EAL: Ask a virtual area of 0x400000 bytes
EAL: Virtual area found at 0x7fff77e00000 (size = 0x400000)
EAL: Ask a virtual area of 0x400000 bytes
EAL: Virtual area found at 0x7fff77800000 (size = 0x400000)
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7fff77400000 (size = 0x200000)
EAL: Ask a virtual area of 0x2200000 bytes
EAL: Virtual area found at 0x7fff75000000 (size = 0x2200000)
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7fff74c00000 (size = 0x200000)
EAL: Ask a virtual area of 0x800000 bytes
EAL: Virtual area found at 0x7fff74200000 (size = 0x800000)
EAL: Ask a virtual area of 0x600000 bytes
EAL: Virtual area found at 0x7fff73a00000 (size = 0x600000)
EAL: Ask a virtual area of 0x800000 bytes
EAL: Virtual area found at 0x7fff73000000 (size = 0x800000)
EAL: Ask a virtual area of 0xa00000 bytes
EAL: Virtual area found at 0x7fff72400000 (size = 0xa00000)
EAL: Ask a virtual area of 0x400000 bytes
EAL: Virtual area found at 0x7fff71e00000 (size = 0x400000)
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7fff71a00000 (size = 0x200000)
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7fff71600000 (size = 0x200000)
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7fff71200000 (size = 0x200000)
EAL: Ask a virtual area of 0x200000 bytes
EAL: Virtual area found at 0x7fff70e00000 (size = 0x200000)
EAL: Requesting 1024 pages of size 2MB from socket 0
EAL: TSC frequency is ~1799997 KHz
EAL: Master lcore 1 is ready (tid=f7fe78a0;cpuset=[1])
EAL: PCI device 0000:03:00.0 on NUMA socket 0
EAL:   probe driver: 15ad:7b0 rte_vmxnet3_pmd
EAL:   Not managed by a supported kernel driver, skipped
EAL: PCI device 0000:1b:00.0 on NUMA socket 0
EAL:   probe driver: 8086:1572 rte_i40e_pmd
EAL:   PCI memory mapped at 0x7fff6f1f3000
EAL:   PCI memory mapped at 0x7ffff7fac000
#

On Fri, Jan 22, 2016 at 2:12 AM, Thomas Monjalon <thomas.monjalon at 6wind.com<mailto:thomas.monjalon at 6wind.com>> wrote:
Hi,

Thanks for asking.
There are a couple of requests hidden in this message.
The maintainer of i40e is Helin (CC'ed).

2016-01-21 14:54, Saurabh Mishra:
> Hi,
>
> We have noticed that i40e if we do PCI-pass through of whole NIC to VM on
> ESXi 6.0, the DPDK exits abruptly in rte_eal_init()?

This looks to be a bug report :)

> We are passing following parameters:
>
>     char *eal_argv[] = {"fakeelf",
>
>                         "-c2",
>
>                         "-n4",
>
>                         "--proc-type=primary",};
>
>
> int ret = rte_eal_init(4, eal_argv);
>
> The code works with Intel '82599ES 10-Gigabit SFI/SFP+ ' adapter in
> PCI-passthrough or SR-IOV mode however i40e it does not work.

You probably have the feeling that i40e does not work as other PMDs,
maybe wondering what are the special extended PCI configs.

> [root at localhost:~] esxcfg-nics -l
>
> [.]
>
> vmnic6  0000:07:00.0 i40e        Up   10000Mbps Full   3c:fd:fe:04:11:c0
> 1500   Intel Corporation Ethernet Controller X710 for 10GbE SFP+
>
> vmnic7  0000:07:00.1 i40e        Up   10000Mbps Full   3c:fd:fe:04:11:c2
> 1500   Intel Corporation Ethernet Controller X710 for 10GbE SFP+
>
>
> We have turned on following config in DPDK:
>
> CONFIG_RTE_PCI_CONFIG=y
>
> CONFIG_RTE_PCI_EXTENDED_TAG="on"
> CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC=y
>
>
> Is there any special handling in DPDK for i40e adapter in terms of config?

So you had no help when reading the code comments neither in the doc.
Indeed the only doc about i40e is the SR-IOV VF page:
        http://dpdk.org/doc/guides/nics/intel_vf.html

Please Helin, check the issue and the lack of documentation.
Thanks




More information about the users mailing list