On Wed, Jan 12, 2022 at 5:44 PM Ilya Maximets <[email protected]> wrote:
> Thanks for the report.  And thanks, David, for checking,
> I can not verify the change right now, I just wanted to rant for a
> bit about system-dpdk testsuite being highly unstable.  It depends
> on the system configuration too much.  Today I touched my iommu
> configuration and now all the tests are failing due to tons of
> virtual address hint warnings:
>
> +2022-01-12T16:25:40.190Z|00023|dpdk|WARN|EAL: WARNING! Base virtual address 
> hint (0x100005000 != 0x7f7855c90000) not respected!
> +2022-01-12T16:25:40.190Z|00024|dpdk|WARN|EAL:    This may cause issues with 
> mapping memory into secondary processes
> +2022-01-12T16:25:40.191Z|00025|dpdk|WARN|EAL: WARNING! Base virtual address 
> hint (0x10000b000 != 0x7f7855e0c000) not respected!
> +2022-01-12T16:25:40.191Z|00026|dpdk|WARN|EAL:    This may cause issues with 
> mapping memory into secondary processes
> +2022-01-12T16:25:40.191Z|00027|dpdk|WARN|EAL: WARNING! Base virtual address 
> hint (0x24000c000 != 0x7f7040000000) not respected!
> +2022-01-12T16:25:40.191Z|00028|dpdk|WARN|EAL:    This may cause issues with 
> mapping memory into secondary processes
>
> Maybe we should just disable all DPDK warnings?

About disabling all warnings, I am unsure.
As a DPDK contributor, this test helps me to catch unwanted DPDK logs
from the core libraries...

No strong opinion.


> BTW, DPDK should probably not emit a warning about secondary process
> for applications with disabled multiprocessing.

For this log here, that's something to fix/enhance.
There was a patch from Anatoly that made those logs debug level
(unless a base-virtaddr option was passed). I can revive the thread.
That would hide this issue, but the memory allocator still expects mp
to be available, regardless of --in-memory or rte_mp_disable calls.


-- 
David Marchand

_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to