[PATCH v4 0/2] Fix testpmd interrupt regression

[Thomas Monjalon]

16/03/2023 18:01, Ferruh Yigit:
> On 3/16/2023 3:31 PM, Stephen Hemminger wrote:
> > On Thu, 16 Mar 2023 12:20:41 +0000
> > Ferruh Yigit <ferruh.yigit at amd.com> wrote:
> > 
> >> On 3/15/2023 5:31 PM, Stephen Hemminger wrote:
> >>> Resolve issues from using control-C in testpmd.
> >>> Fixes regression from recent change to use cmdline_poll().
> >>>
> >>> v4 - drop sig_atomic_t. Not required requires changes on
> >>>      some platforms.
> >>>
> >>> Stephen Hemminger (2):
> >>>   testpmd: go back to using cmdline_interact
> >>>   testpmd: enable interrupt in interactive mode
> >>>
> >>>  app/test-pmd/cmdline.c           | 27 ++++++++++++++-------------
> >>>  app/test-pmd/testpmd.c           | 11 +++++++++++
> >>>  lib/cmdline/cmdline.h            | 10 ++++++++++
> >>>  lib/cmdline/cmdline_os_unix.c    |  8 +++++++-
> >>>  lib/cmdline/cmdline_os_windows.c | 18 ++++++++++++++++--
> >>>  lib/cmdline/cmdline_private.h    |  2 +-
> >>>  lib/cmdline/version.map          |  3 +++
> >>>  7 files changed, 62 insertions(+), 17 deletions(-)
> >>>   
> >>
> >> This solution is adding new cmdline API for -rc3 and there are some
> >> testpmd changes,
> >> can it be possible to have a simple workaround specific to window for
> >> this release and get this set for next release?
> > 
> > 
> > Not really. cmdline_poll() is broken in several ways.
> > Don't want to fix it or use it.
> 
> What about to revert the original fix [^1] in this release and get a new
> version of it at early next release?
> 
> @Thomas, what do you think? I think better to decide before -rc3.

We should not add risky changes at this stage.
Also I would prefer to see a review from Olivier, the cmdline maintainer.

The reason of this breakage was a cleanup.
I think it is more reasonnable to revert.