Thanks, so how can I make sure my platform support NUMA or kernel supports NUMA. I ran the command below:
[changczh@t480-OL7 ~]$ numactl --show policy: default preferred node: current physcpubind: 0 1 2 3 cpubind: 0 nodebind: 0 membind: 0 [changczh@t480-OL7 ~]$ numactl --hardware available: 1 nodes (0) node 0 cpus: 0 1 2 3 node 0 size: 2047 MB node 0 free: 146 MB node distances: node 0 0: 10 [changczh@t480-OL7 ~]$ grep -i numa /var/log/dmesg [ 0.000000] No NUMA configuration found Changchun (Alex) From: Dheeraj Dang [mailto:[email protected]] Sent: Tuesday, March 05, 2019 1:36 AM To: Changchun Zhang <[email protected]> Cc: [email protected] Subject: Re: [dpdk-users] Invalid NUMA socket when running helloworld on VM linux Yes, it makes sense to ignore this log when the platform does not support NUMA or kernel is compiled without NUMA support. Regards Dheeraj Dang On Tue, Mar 5, 2019 at 1:02 AM Changchun Zhang <HYPERLINK "mailto:[email protected]"[email protected]> wrote: Thank you Dheeraj, but looks like the system is not allowing me to change it. I saw in preivous thread that it is suggested to ignore this log at all. HYPERLINK "https://urldefense.proofpoint.com/v2/url?u=http-3A__mails.dpdk.org_archives_users_2017-2DDecember_002742.html&d=DwMFaQ&c=RoP1YumCXCgaWHvlZYR8PZh8Bv7qIrMUB65eapI_JnE&r=VCO7jqQFACS5pfBnqRHRln4bZ3htFERYZ3PkX0ytpns&m=19j8M4hWMogD8-1YEFedt2R5_Q8wpPScygHeGb4FB0E&s=XRhPXPv7gXLn2bt7T1no0hdADSN9VavpseWUhAc66Mg&e="http://mails.dpdk.org/archives/users/2017-December/002742.html Does it make sense? Changchun (Alex) From: Dheeraj Dang [mailto:HYPERLINK "mailto:[email protected]"[email protected]] Sent: Sunday, March 03, 2019 4:58 AM To: Changchun Zhang <HYPERLINK "mailto:[email protected]"[email protected]> Cc: HYPERLINK "mailto:[email protected]"[email protected] Subject: Re: [dpdk-users] Invalid NUMA socket when running helloworld on VM linux Hi Changchun, You can set numa_node using this command echo 0 > /sys/devices/pci0000\:00/<pci_id>/numa_node (if numa node is 0) echo 1 > /sys/devices/pci0000\:00/<pci_id>/numa_node (if numa node is 1) In logs, you are seeing its invalid value i.e -1. Regards Dheeraj Dang On Sun, Mar 3, 2019 at 4:22 AM Changchun Zhang <HYPERLINK "mailto:[email protected]"[email protected]> wrote: Hi Gurus, I launched a virtual box based linux on my windows, and my PC is thinkpad T480. After configured the huge pages as number of 2MBs and the cpu number as 4 on the vm linux, I build DPDK and run the example of hellowworld but get the below output, in which I saw the "invalid NUMA socket". What does it mean and does it matter, as I see the hello from core # is also printed. [~ build]$ sudo ./helloworld EAL: Detected 4 lcore(s) EAL: Detected 1 NUMA nodes EAL: Multi-process socket /var/run/dpdk/rte/mp_socket EAL: Probing VFIO support... EAL: PCI device 0000:00:03.0 on NUMA socket -1 EAL: Invalid NUMA socket, default to 0 EAL: probe driver: 8086:100e net_e1000_em hello from core 1 hello from core 2 hello from core 3 hello from core 0
