[dpdk-dev] [Bug 807] mbuf couldn't be allocated in socket1 in dynmen mode, but in legency mode, mbuf could be allocated correctly in remote socket1.

[dpdk-dev] [Bug 807] mbuf couldn't be allocated in socket1 in dynmen mode, but in legency mode, mbuf could be allocated correctly in remote socket1.

[bugzilla]

https://bugs.dpdk.org/show_bug.cgi?id=807

            Bug ID: 807
           Summary: mbuf couldn't be allocated in socket1 in dynmen mode,
                    but in legency mode, mbuf could be allocated correctly
                    in remote socket1.
           Product: DPDK
           Version: unspecified
          Hardware: All
                OS: All
            Status: UNCONFIRMED
          Severity: major
          Priority: Normal
         Component: core
          Assignee: dev@dpdk.org
          Reporter: yzhu@amperecomputing.com
  Target Milestone: ---

my setup is OVS+DPDK. 

when I configed memory in dyn mode , only
"other_config:dpdk-socket-mem=8192,8192"
I also configured OVS PMD core in socket1, and 2VM also in socket1, I expect
all traffic will be in socket1, and won't touch any resource in socket0, But
after I printed the dpdk's mbuf address(pkts) in virtio_dev_rx_split's
copy_mbuf_to_desc, I found  mbufs  are in socket0.

then I configure memory in legency mode. added
"other_config:dpdk-extra=--legacy-mem ", I checked rte_eal_hugepage_init,
eal_legacy_hugepage_init is called, and also print mbuf address in 
virtio_dev_rx_split's copy_mbuf_to_desc.  All mbufs are in socket1, it's what
expected .

So the dyn mode couldn't correctly allocated mbuf into socket1  should be a bug

-- 
You are receiving this mail because:
You are the assignee for the bug.