[dpdk-dev] [PATCH v3] efd: support lookup using neon intrinsics

[Jianbo Liu]

On 2 May 2017 at 14:41, Jerin Jacob <jerin.jacob at caviumnetworks.com> wrote:
> -----Original Message-----
>> Date: Mon,  1 May 2017 22:59:53 -0700
>> From: Ashwin Sekhar T K <ashwin.sekhar at caviumnetworks.com>
>> To: byron.marohn at intel.com, pablo.de.lara.guarch at intel.com,
>>  jerin.jacob at caviumnetworks.com, jianbo.liu at linaro.org
>> Cc: dev at dpdk.org, Ashwin Sekhar T K <ashwin.sekhar at caviumnetworks.com>
>> Subject: [dpdk-dev] [PATCH v3] efd: support lookup using neon intrinsics
>> X-Mailer: git-send-email 2.13.0.rc1
>>
>> * Added file lib/librte_efd/rte_efd_arm64.h to hold arm64
>>   specific definitions
>> * Verified the changes with efd_autotest unit test case
>>
>> Signed-off-by: Ashwin Sekhar T K <ashwin.sekhar at caviumnetworks.com>
>> ---
>> v2:
>> * Slightly modified the content of the commit message body
>> * Added prefix [dpdk-dev] to the email subject line
>>
>> v3:
>> * Moved enum 'EFD_LOOKUP_NEON' under '#if defined(RTE_ARCH_ARM64)'
>>
>>  MAINTAINERS                    |  1 +
>>  lib/librte_efd/rte_efd.c       | 24 +++++++++++++
>>  lib/librte_efd/rte_efd_arm64.h | 76 ++++++++++++++++++++++++++++++++++++++++++
>>  3 files changed, 101 insertions(+)
>>  create mode 100644 lib/librte_efd/rte_efd_arm64.h
>>
>> diff --git a/MAINTAINERS b/MAINTAINERS
>> index b6495d2..7d708ae 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -147,6 +147,7 @@ F: lib/librte_eal/common/include/arch/arm/*_64.h
>>  F: lib/librte_acl/acl_run_neon.*
>>  F: lib/librte_lpm/rte_lpm_neon.h
>>  F: lib/librte_hash/rte*_arm64.h
>> +F: lib/librte_efd/rte*_arm64.h
>>  F: drivers/net/ixgbe/ixgbe_rxtx_vec_neon.c
>>  F: drivers/net/i40e/i40e_rxtx_vec_neon.c
>>  F: drivers/net/virtio/virtio_rxtx_simple_neon.c
>> diff --git a/lib/librte_efd/rte_efd.c b/lib/librte_efd/rte_efd.c
>> index f601d62..5cc6283 100644
>> --- a/lib/librte_efd/rte_efd.c
>> +++ b/lib/librte_efd/rte_efd.c
>> @@ -53,6 +53,8 @@
>>  #include "rte_efd.h"
>>  #if defined(RTE_ARCH_X86)
>>  #include "rte_efd_x86.h"
>> +#elif defined(RTE_ARCH_ARM64)
>> +#include "rte_efd_arm64.h"
>>  #endif
>>
>>  #define EFD_KEY(key_idx, table) (table->keys + ((key_idx) * table->key_len))
>> @@ -103,6 +105,9 @@ allocated memory
>>  enum efd_lookup_internal_function {
>>       EFD_LOOKUP_SCALAR = 0,
>>       EFD_LOOKUP_AVX2,
>> +#if defined(RTE_ARCH_ARM64)
>> +     EFD_LOOKUP_NEON,
>> +#endif
>
> I think, we can remove this ifdef to
> - Make code looks clean
> - In future, in some case a new enum value gets added then the value
> will be different for each build.
>

But the enum items are same for each ARCH.
Besides, the ifdef could be considered as explanation to that enum. If
someone knows nothing about arm/neon, he can ignore it totally after
see the ifdef.

> Any valid point to keep under RTE_ARCH_ARM64?
>
>>       EFD_LOOKUP_NUM
>>  };