Re: [dpdk-dev] Marking symbols as experimental in the headers only

[Neil Horman <nhorman@tuxdriver.com>]

On Wed, Dec 19, 2018 at 10:12:06AM +0100, David Marchand wrote:
> On Tue, Dec 18, 2018 at 1:27 PM David Marchand <david.marchand@redhat.com>
> wrote:
> 
> >
> > On Tue, Dec 18, 2018 at 1:25 PM Neil Horman <nhorman@tuxdriver.com> wrote:
> >
> >> On Tue, Dec 18, 2018 at 11:41:34AM +0100, David Marchand wrote:
> >>
> >> > The real issue in the end is that the __rte_experimental in headers is
> >> the
> >> > most important thing and can be missed during reviews.
> >> > But I found no easy way to detect this.
> >> >
> >> > Do you have any idea ?
> >> >
> >> The most direct way is to add a regular expression search to the script
> >> that
> >> checks the object files.  That would be some tricky grep/awk magic, but it
> >> should be possible
> >>
> >
> So, wrote something that I hooked in rte.lib.mk.
> On a fresh master branch, I can see:
> 
> == Build lib/librte_bbdev
> ERROR: rte_bbdev_dequeue_dec_ops is not marked as experimental in this
> library headers
> ERROR: rte_bbdev_dequeue_enc_ops is not marked as experimental in this
> library headers
> ERROR: rte_bbdev_devices is not marked as experimental in this library
> headers
> ERROR: rte_bbdev_enqueue_dec_ops is not marked as experimental in this
> library headers
> ERROR: rte_bbdev_enqueue_enc_ops is not marked as experimental in this
> library headers
> ERROR: rte_bbdev_op_pool_create is not marked as experimental in this
> library headers
> ERROR: rte_bbdev_op_type_str is not marked as experimental in this library
> headers
> 
> == Build lib/librte_cryptodev
> ERROR: rte_crypto_asym_op_strings is not marked as experimental in this
> library headers
> ERROR: rte_crypto_asym_xform_strings is not marked as experimental in this
> library headers
> 
> == Build lib/librte_vhost
> ERROR: rte_vhost_va_from_guest_pa is not marked as experimental in this
> library headers
> 
> Those warnings seem valid, need to double check (bbdev is already known).
> 
> But my script still needs some work to make it lighter...
> A fresh build went from:
> real    3m25.823s
> user    2m42.026s
> sys    1m2.730s
> to:
> real    3m42.442s
> user    2m56.733s
> sys    1m5.565s
> 
> I think I'd rather adapt it to hook in checkpatches.sh.
> Maintainers can then ignore it when the check is broken (my regexp skills
> are lacking :)).
> 
The fact that you got it to work this well, this quickly, says they're not that
bad to me :).

If you want to post a draft of your patch here, perhaps we can help find ways to
speed it up.

Well done.
Neil

> 
> -- 
> David Marchand