* [dpdk-dev] why all the other threads except lcore-slave pinned to master lcore?
@ 2016-10-24 11:10 ychen
2016-10-24 18:25 ` Kevin Traynor
0 siblings, 1 reply; 2+ messages in thread
From: ychen @ 2016-10-24 11:10 UTC (permalink / raw)
To: dev
Hi, I am a freshman learning DPDK, when I followed the document INSTALL.DPDK.md to launch openvswitch with dpdk inited, I found that all the threads are pinned to master lcore except lcore-slave and vfio-sync, but I can not find any code to set the affinity for these threads.
Here is my question:
1. why vfio-sync is pinned to a core which is not included in the slave lcore nor master lcore?
2. why all other threads pinned to master lcore? is anything I am setting wrong?
Here is some logs:
2016-10-24T10:42:03Z|00001|vlog|INFO|opened log file /var/log/openvswitch/ovs-vswitchd.log
2016-10-24T10:42:03Z|00002|ovs_numa|INFO|Discovered 24 CPU cores on NUMA node 0
2016-10-24T10:42:03Z|00003|ovs_numa|INFO|Discovered 24 CPU cores on NUMA node 1
2016-10-24T10:42:03Z|00004|ovs_numa|INFO|Discovered 2 NUMA nodes and 48 CPU cores
2016-10-24T10:42:03Z|00005|reconnect|INFO|unix:/var/run/openvswitch/db.sock: connecting...
2016-10-24T10:42:03Z|00006|reconnect|INFO|unix:/var/run/openvswitch/db.sock: connected
2016-10-24T10:42:03Z|00007|dpdk|INFO|DPDK Enabled, initializing
2016-10-24T10:42:03Z|00008|dpdk|INFO|No vhost-sock-dir provided - defaulting to /var/run/openvswitch
2016-10-24T10:42:03Z|00009|dpdk|INFO|EAL ARGS: ovs-vswitchd -c 0xf --socket-mem 1024,1024
EAL: Detected 48 lcore(s)
EAL: Probing VFIO support...
EAL: VFIO support initialized
PMD: bnxt_rte_pmd_init() called for (null)
EAL: PCI device 0000:01:00.0 on NUMA socket 0
EAL: probe driver: 8086:10fb rte_ixgbe_pmd
EAL: PCI device 0000:01:00.1 on NUMA socket 0
EAL: probe driver: 8086:10fb rte_ixgbe_pmd
EAL: PCI device 0000:06:00.0 on NUMA socket 0
EAL: probe driver: 8086:1521 rte_igb_pmd
EAL: PCI device 0000:06:00.1 on NUMA socket 0
EAL: probe driver: 8086:1521 rte_igb_pmd
2016-10-24T10:42:06Z|00010|dpdk|INFO|DPDK pdump packet capture enable
and the output of the cpu_layout:
cores = [0, 1, 2, 3, 4, 5, 8, 9, 10, 11, 12, 13]
sockets = [0, 1]
Socket 0 Socket 1
-------- --------
Core 0 [0, 24] [1, 25]
Core 1 [2, 26] [3, 27]
Core 2 [4, 28] [5, 29]
Core 3 [6, 30] [7, 31]
Core 4 [8, 32] [9, 33]
Core 5 [10, 34] [11, 35]
Core 8 [12, 36] [13, 37]
Core 9 [14, 38] [15, 39]
Core 10 [16, 40] [17, 41]
Core 11 [18, 42] [19, 43]
Core 12 [20, 44] [21, 45]
Core 13 [22, 46] [23, 47]
All the threads and their pinned core for vswitchd:
28262 28262 0 ovs-vswitchd
28263 28262 39 vfio-sync
28297 28262 0 eal-intr-thread
28298 28262 1 lcore-slave-1
28299 28262 2 lcore-slave-2
28300 28262 3 lcore-slave-3
28301 28262 0 dpdk_watchdog2
28302 28262 0 vhost_thread1
28303 28262 0 pdump-thread
28304 28262 0 ct_clean3
28305 28262 0 urcu4
28744 28262 0 handler101
28745 28262 0 handler100
28746 28262 0 handler99
28747 28262 0 handler98
28748 28262 0 handler95
28749 28262 0 handler77
28750 28262 0 handler79
28751 28262 0 handler80
28752 28262 0 handler81
28753 28262 0 handler73
28756 28262 0 handler92
28757 28262 0 handler82
28758 28262 0 handler96
28759 28262 0 handler71
28760 28262 0 handler61
28761 28262 0 handler62
28762 28262 0 handler83
28763 28262 0 handler63
28764 28262 0 handler84
28765 28262 0 handler93
28766 28262 0 handler64
28767 28262 0 handler85
28768 28262 0 handler74
28769 28262 0 handler65
28770 28262 0 handler66
28771 28262 0 handler78
28772 28262 0 handler86
28773 28262 0 handler87
28774 28262 0 handler97
28775 28262 0 handler88
28776 28262 0 handler56
28777 28262 0 handler76
28778 28262 0 handler67
28779 28262 0 handler60
28780 28262 0 handler68
28781 28262 0 revalidator75
28782 28262 0 revalidator57
28783 28262 0 revalidator89
28784 28262 0 revalidator69
28785 28262 0 revalidator54
28786 28262 0 revalidator90
28787 28262 0 revalidator55
28788 28262 0 revalidator58
28789 28262 0 revalidator59
28790 28262 0 revalidator70
28791 28262 0 revalidator94
28792 28262 0 revalidator91
28793 28262 0 revalidator72
28827 28262 4 pmd103
28829 28262 6 pmd102
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [dpdk-dev] why all the other threads except lcore-slave pinned to master lcore?
2016-10-24 11:10 [dpdk-dev] why all the other threads except lcore-slave pinned to master lcore? ychen
@ 2016-10-24 18:25 ` Kevin Traynor
0 siblings, 0 replies; 2+ messages in thread
From: Kevin Traynor @ 2016-10-24 18:25 UTC (permalink / raw)
To: ychen, dev
On 10/24/2016 12:10 PM, ychen wrote:
> Hi, I am a freshman learning DPDK, when I followed the document INSTALL.DPDK.md to launch openvswitch with dpdk inited, I found that all the threads are pinned to master lcore except lcore-slave and vfio-sync, but I can not find any code to set the affinity for these threads.
> Here is my question:
> 1. why vfio-sync is pinned to a core which is not included in the slave lcore nor master lcore?
> 2. why all other threads pinned to master lcore? is anything I am setting wrong?
Hi - these are probably more appropriate for the ovs-dev list. I
answered the post you put over there.
>
>
> Here is some logs:
> 2016-10-24T10:42:03Z|00001|vlog|INFO|opened log file /var/log/openvswitch/ovs-vswitchd.log
> 2016-10-24T10:42:03Z|00002|ovs_numa|INFO|Discovered 24 CPU cores on NUMA node 0
> 2016-10-24T10:42:03Z|00003|ovs_numa|INFO|Discovered 24 CPU cores on NUMA node 1
> 2016-10-24T10:42:03Z|00004|ovs_numa|INFO|Discovered 2 NUMA nodes and 48 CPU cores
> 2016-10-24T10:42:03Z|00005|reconnect|INFO|unix:/var/run/openvswitch/db.sock: connecting...
> 2016-10-24T10:42:03Z|00006|reconnect|INFO|unix:/var/run/openvswitch/db.sock: connected
> 2016-10-24T10:42:03Z|00007|dpdk|INFO|DPDK Enabled, initializing
> 2016-10-24T10:42:03Z|00008|dpdk|INFO|No vhost-sock-dir provided - defaulting to /var/run/openvswitch
> 2016-10-24T10:42:03Z|00009|dpdk|INFO|EAL ARGS: ovs-vswitchd -c 0xf --socket-mem 1024,1024
> EAL: Detected 48 lcore(s)
> EAL: Probing VFIO support...
> EAL: VFIO support initialized
> PMD: bnxt_rte_pmd_init() called for (null)
> EAL: PCI device 0000:01:00.0 on NUMA socket 0
> EAL: probe driver: 8086:10fb rte_ixgbe_pmd
> EAL: PCI device 0000:01:00.1 on NUMA socket 0
> EAL: probe driver: 8086:10fb rte_ixgbe_pmd
> EAL: PCI device 0000:06:00.0 on NUMA socket 0
> EAL: probe driver: 8086:1521 rte_igb_pmd
> EAL: PCI device 0000:06:00.1 on NUMA socket 0
> EAL: probe driver: 8086:1521 rte_igb_pmd
> 2016-10-24T10:42:06Z|00010|dpdk|INFO|DPDK pdump packet capture enable
>
>
> and the output of the cpu_layout:
> cores = [0, 1, 2, 3, 4, 5, 8, 9, 10, 11, 12, 13]
> sockets = [0, 1]
>
>
> Socket 0 Socket 1
> -------- --------
> Core 0 [0, 24] [1, 25]
> Core 1 [2, 26] [3, 27]
> Core 2 [4, 28] [5, 29]
> Core 3 [6, 30] [7, 31]
> Core 4 [8, 32] [9, 33]
> Core 5 [10, 34] [11, 35]
> Core 8 [12, 36] [13, 37]
> Core 9 [14, 38] [15, 39]
> Core 10 [16, 40] [17, 41]
> Core 11 [18, 42] [19, 43]
> Core 12 [20, 44] [21, 45]
> Core 13 [22, 46] [23, 47]
> All the threads and their pinned core for vswitchd:
> 28262 28262 0 ovs-vswitchd
> 28263 28262 39 vfio-sync
> 28297 28262 0 eal-intr-thread
> 28298 28262 1 lcore-slave-1
> 28299 28262 2 lcore-slave-2
> 28300 28262 3 lcore-slave-3
> 28301 28262 0 dpdk_watchdog2
> 28302 28262 0 vhost_thread1
> 28303 28262 0 pdump-thread
> 28304 28262 0 ct_clean3
> 28305 28262 0 urcu4
> 28744 28262 0 handler101
> 28745 28262 0 handler100
> 28746 28262 0 handler99
> 28747 28262 0 handler98
> 28748 28262 0 handler95
> 28749 28262 0 handler77
> 28750 28262 0 handler79
> 28751 28262 0 handler80
> 28752 28262 0 handler81
> 28753 28262 0 handler73
> 28756 28262 0 handler92
> 28757 28262 0 handler82
> 28758 28262 0 handler96
> 28759 28262 0 handler71
> 28760 28262 0 handler61
> 28761 28262 0 handler62
> 28762 28262 0 handler83
> 28763 28262 0 handler63
> 28764 28262 0 handler84
> 28765 28262 0 handler93
> 28766 28262 0 handler64
> 28767 28262 0 handler85
> 28768 28262 0 handler74
> 28769 28262 0 handler65
> 28770 28262 0 handler66
> 28771 28262 0 handler78
> 28772 28262 0 handler86
> 28773 28262 0 handler87
> 28774 28262 0 handler97
> 28775 28262 0 handler88
> 28776 28262 0 handler56
> 28777 28262 0 handler76
> 28778 28262 0 handler67
> 28779 28262 0 handler60
> 28780 28262 0 handler68
> 28781 28262 0 revalidator75
> 28782 28262 0 revalidator57
> 28783 28262 0 revalidator89
> 28784 28262 0 revalidator69
> 28785 28262 0 revalidator54
> 28786 28262 0 revalidator90
> 28787 28262 0 revalidator55
> 28788 28262 0 revalidator58
> 28789 28262 0 revalidator59
> 28790 28262 0 revalidator70
> 28791 28262 0 revalidator94
> 28792 28262 0 revalidator91
> 28793 28262 0 revalidator72
> 28827 28262 4 pmd103
> 28829 28262 6 pmd102
>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2016-10-24 18:25 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-10-24 11:10 [dpdk-dev] why all the other threads except lcore-slave pinned to master lcore? ychen
2016-10-24 18:25 ` Kevin Traynor
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).