Re: [opencontrail-dev] Need help in setting up vRouter in DPDK-enabled mode

2017-04-09 Thread Deepak Tiwari 
Hi Raja,

As I updated last week, I am no longer getting this error (after I switched 
from trunk to 3.2 and back to trunk) now for some unknown reason. Now when I 
start the contrail-dpdk binary, I get following error—

contrail@contrail-23:~/dev/contrail-installer$ /usr/bin/contrail-vrouter-dpdk 
--no-daemon --socket-mem 1024,1024
ERROR: this system does not support “SSE4_2”.
Please check that RTE_MACHINE is set correctly.

After debugging the root-cause of above issue, I found that in file 
“third_party/dpdk/mk/machine/native/rte.vars.mk”, following check (marked in 
red) is missing. I came to this conclusion because—


1.  On my setup, compiler doesn’t report SSE4_2 support

a.  checked with command “gcc -march=native -dM -E - < /dev/null | grep 
SSE4_2”



2.  File “/var/run/dmesg.boot” is not present

Based on above points and following code, “MACHINE_CFLAGS” will get set to 
“-march=corei7”

# on FreeBSD systems, sometimes the correct cputype is not picked up.
# To get everything to compile, we need SSE4.2 support, so check if that is
# reported by compiler. If not, check if the CPU actually supports it, and if
# so, set the compilation target to be a corei7, minimum target with SSE4.2
SSE42_SUPPORT=$(shell $(CC) -march=native -dM -E - < /dev/null | grep SSE4_2)
ifeq ($(SSE42_SUPPORT),)
  CPU_SSE42_SUPPORT=$(shell if [ -f /var/run/dmesg.boot ] ; then grep SSE4\.2 
/var/run/dmesg.boot ; fi)
  ifneq ($(CPU_SSE42_SUPPORT),)
MACHINE_CFLAGS= -march=corei7
  endif
endif

Ref: http://dpdk.org/ml/archives/dev/2014-July/004053.html

So I added above additional code as highlighted in RED. Recompiled DPDK library 
as well as contrail-vrouter-dpdk binary and tried to run it. However still same 
error is received

Br, Deepak

From: Rajagopalan Sivaramakrishnan [mailto:r...@juniper.net]
Sent: 08 April 2017 05:29
To: Deepak Tiwari ; dev@lists.opencontrail.org
Subject: Re: [opencontrail-dev] Need help in setting up vRouter in DPDK-enabled 
mode

Hi Deepak,
It might be worth hardcoding core_mask_count just to see if the issue is 
with using __builtin_popcountll
in the gcc version you are using. Do you still see the crash without the call 
to __builtin_popcountll?

Raja

"DISCLAIMER: This message is proprietary to Aricent and is intended solely for 
the use of the individual to whom it is addressed. It may contain privileged or 
confidential information and should not be circulated or used for any purpose 
other than for what it is intended. If you have received this message in error, 
please notify the originator immediately. If you are not the intended 
recipient, you are notified that you are strictly prohibited from using, 
copying, altering, or disclosing the contents of this message. Aricent accepts 
no responsibility for loss or damage arising from the use of the information 
transmitted by this email including damage from virus."
___
Dev mailing list
Dev@lists.opencontrail.org
http://lists.opencontrail.org/mailman/listinfo/dev_lists.opencontrail.org

Re: [opencontrail-dev] An error in "Configuring SR-IOV section" in "Contrail Feature Guide" ? (/etc/nova/nova.conf settings in the compute node)

2017-04-09 Thread Kevin Wilson 
Yes, it works without this
Kevin

On Tue, Mar 14, 2017 at 5:05 PM, Jakub Pavlik  wrote:
> Hi Kevin,
>
> I am not sure, but based on this
> https://www.juniper.net/techpubs/en_US/contrail3.0/topics/concept/sriov-with-vrouter-vnc.html#enabling-aspm-in-bios
> it seems needed.
>
> Does SRIOV works?
>
> Jakub
>
> On Fri, Mar 10, 2017 at 6:18 PM, Kevin Wilson  wrote:
>>
>> Thanks, Jakub !
>>
>> Are you sure sire ASPM is needed ?
>>
>> For me,
>> lspci -nvvvs 04:00.0| grep ASPM | grep Enabled
>> returns nothing.
>> And:
>> lspci -nvvvs 04:00.0 | grep ASPM | grep -i disabled
>> returns:
>> LnkCtl: ASPM Disabled; RCB 64 bytes Disabled- CommClk+
>>
>> 04:00.0  is the BDF I use.
>>
>> And I am using 04:00.0 for SRIOV with Qemu.
>>
>> I don't think I have "Enable ASPM in BIOS".
>>
>> Are you sure ASPM is needed ? or maybe this is for very specific
>> configurations ?
>>
>> Also dmesg shows that ASPM is disabled:
>>
>> $dmesg | grep  ASPM
>> [0.114703] ACPI FADT declares the system doesn't support PCIe
>> ASPM, so disable it
>> [0.135999] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM
>> ClockPM Segments MSI]
>> [0.136237] acpi PNP0A08:00: FADT indicates ASPM is unsupported,
>> using BIOS configuration
>> [0.222695] acpi PNP0A08:01: _OSC: OS supports [ExtendedConfig ASPM
>> ClockPM Segments MSI]
>> [0.223101] acpi PNP0A08:01: FADT indicates ASPM is unsupported,
>> using BIOS configuration
>> [0.226173] acpi PNP0A08:02: _OSC: OS supports [ExtendedConfig ASPM
>> ClockPM Segments MSI]
>> [0.226573] acpi PNP0A08:02: FADT indicates ASPM is unsupported,
>> using BIOS configuration
>>
>>
>> Regards,
>> Kevin
>>
>> On Sun, Feb 19, 2017 at 2:10 PM, Jakub Pavlik 
>> wrote:
>> > Yes,
>> >
>> > do not forget to check following parameters on compute nodes:
>> >
>> > Enable VT-d in BIOS
>> >
>> > check:
>> > # dmesg | grep -e DMAR -e IOMMU
>> >
>> >
>> > Enable ASPM in BIOS
>> >
>> > check:
>> > # lspci -vvv | grep ASPM | grep Enabled
>> >
>> >
>> >
>> > nova.conf on controller
>> >
>> > scheduler_default_filters = RetryFilter, AvailabilityZoneFilter,
>> > RamFilter,
>> > DiskFilter,
>> >
>> >ComputeFilter,
>> > ComputeCapabilitiesFilter,
>> > ImagePropertiesFilter,
>> >
>> >ServerGroupAntiAffinityFilter,
>> > ServerGroupAffinityFilter, PciPassthroughFilter
>> >
>> > scheduler_available_filters = nova.scheduler.filters.all_filters
>> >
>> > scheduler_available_filters =
>> > nova.scheduler.filters.pci_passthrough_filter.PciPassthroughFilter
>> >
>> >
>> >
>> > nova.conf on compute node
>> > pci_passthrough_whitelist = { "devname": "eth1", "physical_network":
>> > "physnet1"}
>> >
>> > On Mon, Feb 13, 2017 at 11:24 AM, Kevin Wilson 
>> > wrote:
>> >>
>> >> Hello, OpenContrail experts,
>> >>
>> >> I found the following in the "Contrail Feature Guide",
>> >>
>> >>
>> >> https://www.juniper.net/techpubs/en_US/contrail3.0/information-products/pathway-pages/contrail-feature-guide-pwp.pdf
>> >>
>> >> It says:
>> >>
>> >> ...
>> >> Overview: Configuring SR-IOV
>> >> ...
>> >> Configuring SR-IOV Features Without env.sriov in testbed.py
>> >> ...
>> >> In the Nova compute config file, configure a Nova Scheduler filter
>> >> based on the new
>> >> PCI configuration, as in the following example:
>> >> /etc/nova/nova.conf
>> >> [default]
>> >> scheduler_default_filters = PciPassthroughFilter
>> >> scheduler_available_filters = nova.scheduler.filters.all_filters
>> >> scheduler_available_filters =
>> >> nova.scheduler.filters.pci_passthrough_filter.PciPassthroughFilter
>> >> 6. Restart Nova Scheduler.
>> >> service nova-scheduler restart
>> >>
>> >> It seems to me that there is an error here: it says to make the
>> >> changes in the Nova compute config file and
>> >> then restart nova-scheduler. However, there is no nova-scheduler
>> >> service in the Nova compute node
>> >> at all:
>> >> service nova-scheduler status
>> >> nova-scheduler: unrecognized service
>> >>
>> >> So it seems to me that this change, of adding
>> >> scheduler_default_filters/scheduler_default_filters/... settings,
>> >> should be done in /etc/nova/nova.conf of the *control* node and *not*
>> >> /etc/nova/nova.conf of the compute node.
>> >>
>> >> Does this sounds reasonable ?
>> >>
>> >> Kevin
>> >>
>> >> ___
>> >> Dev mailing list
>> >> Dev@lists.opencontrail.org
>> >>
>> >> http://lists.opencontrail.org/mailman/listinfo/dev_lists.opencontrail.org
>> >
>> >
>> >
>> >
>> > --
>> > Jakub Pavlik
>> > +420 602 177 027
>> > jpav...@mirantis.com
>
>
>
>
> --
> Jakub Pavlik
> +420 602 177 027
> jpav...@mirantis.com

___
Dev mailing list
Dev@lists.opencontrail.org
http://lists.opencontrail.org/mailman/listinfo/dev_lists.opencontrail.org