Re: [dpdk-dev] [PATCH v2 3/4] examples/fips_validation: remove illegal usage of APIs

[Akhil Goyal <gakhil@marvell.com>]

Hi Fan,
> > > > Some of the cryptodev APIs are not allowed to be used by application
> > > > directly. Hence removing the usage of 1. queue_pair_release: it is not
> > > > required, as configure
> > > >    of queue pair release the previous queue pairs and the
> > > >    dev is not directly exposed to application, hence cannot
> > > >    use its ops from app.
> > > > 2. rte_cryptodev_stop: it can be used directly without
> > > >    checking if the device is started or not.
> > > > 3. rte_cryptodev_pmd_destroy: application should use
> > > >    rte_cryptodev_close instead.
> > > >
> > > > Signed-off-by: Akhil Goyal <gakhil@marvell.com>
> > >
> > > Look's like it should be backported to stable releases with a Fixes
> reference.
> > > What do you think?
> > >
> > > Besides,
> > > Acked-by: Matan Azrad <matan@nvidia.com>
> > >
> > Yes, Agreed.
> > I wanted to get opinion from the Maintainer of fips app first if the changes
> > are
> > Correct or not.
> 
> Hi Akhil,
> 
> FIPS requirement is upon failure of running self-test the device memory
> shall be destroyed completely and not visible by the application at all -
> rte_cryptodev_close does not provide this functionality.
> 
> In this case we may need new API rte_cryptodev_destroy() to replace
> rte_cryptodev_pmd_destroy().
> 
_create and _destroy are called from device probe and remove.
If the requirement is to use pmd_destroy, how is the driver created again?
I do not see the use of rte_cryptodev_pmd_create() function.

The use of pmd APIs is not allowed from the application and it looks something
Is not correct here for the use of destroy API.
Please check and fix as required.

Regards,
Akhil