mbox series

[v3,0/4] Use correct memory ordering in eal functions

Message ID 20211025045237.19878-1-honnappa.nagarahalli@arm.com (mailing list archive)
Headers
Series Use correct memory ordering in eal functions |

Message

Honnappa Nagarahalli Oct. 25, 2021, 4:52 a.m. UTC
  v3:
a) Added Fixes, Cc:stable#dpdk.org in 1/6
b) Merged 3/6 & 4/6 and moved after the first commit in the series
c) Merged 2/6 & 5/6 as they need to be in a single commit
d) Removed use of volatile in 6/6 (Konstantin)

rte_eal_remote_launch and rte_eal_wait_lcore need to provide
correct memory ordering to address the data communication from
main core to worker core.

There are 2 use cases:
1) All the store operations (meant for worker) by main core
should be visible to worker core before the worker core runs the
assigned function

2) All the store operations by the worker core should be visible
to the main core after rte_eal_wait_lcore returns.

For the data that needs to be communicated to the worker after
the rte_eal_remote_launch returns, load-acquire and store-release
semantics should be used.

For the main to worker communication, the pointer to function
to execute is used as the guard variable. Hence, resetting of
the function pointer is important.

For the worker to main communication, the existing code uses the
lcore state as the guard variable. However, it looks like
the FINISHED state is not really required. Hence the FINISHED state
is removed before using the state as the guard variable.

Honnappa Nagarahalli (4):
  eal: reset lcore function pointer and argument
  eal: lcore state FINISHED is not required
  eal: ensure correct memory ordering
  test/ring: use relaxed barriers for ring stress test

 app/test/test_ring_stress_impl.h              | 18 +++----
 drivers/event/dpaa2/dpaa2_eventdev_selftest.c |  2 +-
 drivers/event/octeontx/ssovf_evdev_selftest.c |  2 +-
 drivers/event/sw/sw_evdev_selftest.c          |  4 +-
 examples/l2fwd-keepalive/main.c               |  3 +-
 lib/eal/common/eal_common_launch.c            | 13 ++---
 lib/eal/freebsd/eal_thread.c                  | 45 +++++++++++++----
 lib/eal/include/rte_launch.h                  | 21 ++++----
 lib/eal/include/rte_service.h                 |  4 +-
 lib/eal/linux/eal_thread.c                    | 48 +++++++++++++------
 lib/eal/windows/eal_thread.c                  | 48 +++++++++++++------
 11 files changed, 132 insertions(+), 76 deletions(-)
  

Comments

David Marchand Oct. 25, 2021, 4:23 p.m. UTC | #1
On Mon, Oct 25, 2021 at 6:53 AM Honnappa Nagarahalli
<honnappa.nagarahalli@arm.com> wrote:
>
> v3:
> a) Added Fixes, Cc:stable#dpdk.org in 1/6
> b) Merged 3/6 & 4/6 and moved after the first commit in the series
> c) Merged 2/6 & 5/6 as they need to be in a single commit
> d) Removed use of volatile in 6/6 (Konstantin)
>
> rte_eal_remote_launch and rte_eal_wait_lcore need to provide
> correct memory ordering to address the data communication from
> main core to worker core.
>
> There are 2 use cases:
> 1) All the store operations (meant for worker) by main core
> should be visible to worker core before the worker core runs the
> assigned function
>
> 2) All the store operations by the worker core should be visible
> to the main core after rte_eal_wait_lcore returns.
>
> For the data that needs to be communicated to the worker after
> the rte_eal_remote_launch returns, load-acquire and store-release
> semantics should be used.
>
> For the main to worker communication, the pointer to function
> to execute is used as the guard variable. Hence, resetting of
> the function pointer is important.
>
> For the worker to main communication, the existing code uses the
> lcore state as the guard variable. However, it looks like
> the FINISHED state is not really required. Hence the FINISHED state
> is removed before using the state as the guard variable.
>
> Honnappa Nagarahalli (4):
>   eal: reset lcore function pointer and argument
>   eal: lcore state FINISHED is not required
>   eal: ensure correct memory ordering
>   test/ring: use relaxed barriers for ring stress test
>
>  app/test/test_ring_stress_impl.h              | 18 +++----
>  drivers/event/dpaa2/dpaa2_eventdev_selftest.c |  2 +-
>  drivers/event/octeontx/ssovf_evdev_selftest.c |  2 +-
>  drivers/event/sw/sw_evdev_selftest.c          |  4 +-
>  examples/l2fwd-keepalive/main.c               |  3 +-
>  lib/eal/common/eal_common_launch.c            | 13 ++---
>  lib/eal/freebsd/eal_thread.c                  | 45 +++++++++++++----
>  lib/eal/include/rte_launch.h                  | 21 ++++----
>  lib/eal/include/rte_service.h                 |  4 +-
>  lib/eal/linux/eal_thread.c                    | 48 +++++++++++++------
>  lib/eal/windows/eal_thread.c                  | 48 +++++++++++++------
>  11 files changed, 132 insertions(+), 76 deletions(-)

Tweaked commit titles, removed deprecation notice and updated release
notes in patch 2.
Series applied, thanks.