Re: [PATCH v2 0/2] fix race in rte_thread_create failure path

[Tyler Retzlaff <roretzla@linux.microsoft.com>]

On Tue, Mar 14, 2023 at 12:47:54PM +0100, David Marchand wrote:
> On Tue, Mar 14, 2023 at 12:31 AM Tyler Retzlaff
> <roretzla@linux.microsoft.com> wrote:
> >
> > v2:
> >   * new approach over v1 of the patch to avoid using pthread np API that
> >     is not available on Alpine Linux.
> >   * to conform to rte_thread_create parameter const qualification include
> >     an additional patch to const qualify rte_thread_set_affinity cpusetp
> >     parameter.
> >
> > Tyler Retzlaff (2):
> >   eal: make cpusetp to rte thread set affinity const
> >   eal: fix failure path race setting new thread affinity
> >
> >  lib/eal/common/eal_common_thread.c |  6 ++---
> >  lib/eal/include/rte_thread.h       |  2 +-
> >  lib/eal/unix/rte_thread.c          | 52 ++++++++++++++++++++++++++++++--------
> >  3 files changed, 46 insertions(+), 14 deletions(-)
> 
> ASan flagged some use after free.
> See logs https://github.com/ovsrobot/dpdk/suites/11537702259/artifacts/597032673
> 
> 24/90 DPDK:fast-tests / lcores_autotest       FAIL     1.72 s (exit status 1)

thanks, i'm waiting on a system to test plan to investigate today if it
becomes available.

ty