On Fri, Jul 13, 2018 at 11:10 AM, Tiwei Bie <tiwei....@intel.com> wrote: > On Fri, Jul 13, 2018 at 10:09:42AM +0200, David Marchand wrote: > [...] >> Well, there was an issue vith virtio at one time (interrupt handler >> did not have the right iopl for virtio callback). >> http://git.dpdk.org/dpdk/commit/lib/librte_eal/linuxapp/eal/eal.c?id=fd6949c55c9a48e81c625138679159829d51ac51 >> >> Now... time has passed since then. >> It is worth checking the issue is still here, and if it is the case, >> revisit this. > > Now rte_virtio_pmd_init() which calls rte_eal_iopl_init() > is called as a constructor [1], so the issue isn't there > any more.
If the virtio pmd is loaded dynamically, eal_plugins_init() is still called before rte_eal_intr_init(). Yes, for this case, it looks to be fine with Anatoly change. -- David Marchand
