Re: [dpdk-dev] [PATCH v3 4/4] compressdev: add huffman encoding flags

["De Lara Guarch, Pablo" <pablo.de.lara.guarch@intel.com>]

> -----Original Message-----
> From: Verma, Shally [mailto:Shally.Verma@cavium.com]
> Sent: Thursday, July 5, 2018 9:14 AM
> To: De Lara Guarch, Pablo <pablo.de.lara.guarch@intel.com>; Gupta, Ashish
> <Ashish.Gupta@cavium.com>; Trahe, Fiona <fiona.trahe@intel.com>; Daly, Lee
> <lee.daly@intel.com>
> Cc: dev@dpdk.org
> Subject: RE: [PATCH v3 4/4] compressdev: add huffman encoding flags
> 
> 
> 
> >-----Original Message-----
> >From: Pablo de Lara [mailto:pablo.de.lara.guarch@intel.com]
> >Sent: 04 July 2018 19:41
> >To: Verma, Shally <Shally.Verma@cavium.com>; Gupta, Ashish
> ><Ashish.Gupta@cavium.com>; fiona.trahe@intel.com; lee.daly@intel.com
> >Cc: dev@dpdk.org; Pablo de Lara <pablo.de.lara.guarch@intel.com>
> >Subject: [PATCH v3 4/4] compressdev: add huffman encoding flags
> >
> >External Email
> >
> >Added Huffman fixed and dynamic encoding feature flags, so an
> >application can query if a device supports these two types, when
> >performing DEFLATE compression.
> >
> >Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
> >Acked-by: Fiona Trahe <fiona.trahe@intel.com>
> >---
> >
> >Changes in v3:
> >
> >- No change
> >
> >Changes in v2:
> >
> >- Fixed typo
> >
> > drivers/compress/isal/isal_compress_pmd_ops.c |  4 +++-
> > lib/librte_compressdev/rte_comp.c             |  4 ++++
> > lib/librte_compressdev/rte_comp.h             |  4 ++++
> > test/test/test_compressdev.c                  | 16 ++++++++++++++++
> > 4 files changed, 27 insertions(+), 1 deletion(-)
> >
> 
> //snip
> 
> >diff --git a/lib/librte_compressdev/rte_comp.c
> >b/lib/librte_compressdev/rte_comp.c
> >index f5bd3a6c0..5ed1d0daa 100644
> >--- a/lib/librte_compressdev/rte_comp.c
> >+++ b/lib/librte_compressdev/rte_comp.c
> >@@ -36,6 +36,10 @@ rte_comp_get_feature_name(uint64_t flag)
> >                return "SHA2_SHA256_HASH";
> >        case RTE_COMP_FF_SHAREABLE_PRIV_XFORM:
> >                return "SHAREABLE_PRIV_XFORM";
> >+       case RTE_COMP_FF_HUFFMAN_FIXED:
> >+               return "HUFFMAN_FIXED";
> >+       case RTE_COMP_FF_HUFFMAN_DYNAMIC:
> >+               return "HUFFMAN_DYNAMIC";
> >        default:
> >                return NULL;
> >        }
> >diff --git a/lib/librte_compressdev/rte_comp.h
> >b/lib/librte_compressdev/rte_comp.h
> >index 6660cee82..c9245cce1 100644
> >--- a/lib/librte_compressdev/rte_comp.h
> >+++ b/lib/librte_compressdev/rte_comp.h
> >@@ -62,6 +62,10 @@ extern "C" {
> >  * to create as many priv_xforms as it expects to have stateless
> >  * operations in-flight.
> >  */
> >+#define RTE_COMP_FF_HUFFMAN_FIXED              (1ULL << 13)
> >+/**< Fixed huffman encoding is supported */
> >+#define RTE_COMP_FF_HUFFMAN_DYNAMIC            (1ULL << 14)
> >+/**< Dynamic huffman encoding is supported */
> >
> 
> [Shally] As such okay to have this feature. But while looking at this, got a
> question:
> 
> rte_compressdev_info_get() returns feature flags of type
> RTE_COMPDEV_FF_xxx and, rte_compressdev_capability_get() returns PMD
> capability for specific algo using feature flags of type RTE_COMP_FF_xxx. So,
> 
> 1. should rte_compressdev_capability_get() and "struct
> rte_compressdev_capabilities" be changed to
> rte_compressdev_comp_capability_get() or
> rte_compressdev_algo_capability_get()?

I would prefer to leave it as it is, as it matches the function name in cryptodev, which is doing something similar.
However, I don't have an strong opinion on this. Maybe Fiona can break the tie? :)

> 
> 2. where does RTE_COMPDEV_FF_HW_ACCELERATED be set? in dev_info-
> >feature flag or capability->feature_flag?
> What if PMD support hw acceleration of one algo but have sw support of
> another. (say, deflate HW accelerated and LZS sw?)

It is set in dev_info->feature_flag.
As far as I know, this is not expected. HW_ACCELERATED should apply to the whole device.
The device will be either hardware or software. If it is hardware, but it uses software-based
Implementation for an algorithm, still the device is hardware.

> 
> Thanks
> Shally