On Mon, Mar 22, 2021 at 11:23 PM Dmitry Kozlyuk <dmitry.kozl...@gmail.com> wrote: > > 2021-03-22 21:55 (UTC+0000), Jie Zhou: > > Thank you Dmitry! > > > > Regarding " I can make eal_mem_free() behave as expected at issue 2. Or > > should we simple disable this code when where's no multiprocess support > > (how do we test for it properly, BTW)?". The issue hit when I ran testpmd > > (with i40e pmd) and at "quit" (which calling into rte_eal_cleanup and > > reached this rte_mem_unmap). I can test your patch with my local testpmd > > (which I will add back the rte_eal_cleanup to exercise the code path), if > > that is what you are asking about test. > > Thanks, but it really was a question to Anatoly about how to test if > multiprocess is supported. I can send a patch that kind of works, if you > wish, but I wonder if there's a simpler solution. > > Please don't drop Cc so that everyone can see the progress. >
When in --in-memory mode, would it be simpler to avoid the cleaning up code entirely? It should (I would not say solve but) avoid your issue on Windows which only supports --in-memory. If this does not help... Anatoly, the floor is yours. -- David Marchand