[dpdk-dev] [PATCH v10 07/10] app/testpmd: replace POSIX specific code

[Jie Zhou]

On Wed, May 05, 2021 at 09:09:20AM -0700, Jie Zhou wrote:
> On Wed, May 05, 2021 at 08:34:34AM +0000, Tal Shnaiderman wrote:
> > > Subject: [PATCH v10 07/10] app/testpmd: replace POSIX specific code
> > > 
> > > External email: Use caution opening links or attachments
> > > 
> > > 
> > >  - Make printf format OS independent
> > >  - Replace htons with RTE_BE16
> > >  - Replace POSIX specific inet_aton with OS independent inet_pton
> > >  - Replace sleep with rte_delay_us_sleep
> > >  - Repalce random with rte_rand
> > >  - #ifndef mman related code for now
> > 
> > Do you think we can reduce the number of ifndefs by using the functions rte_mem_map, rte_mem_lock?
> 
> I used rte_mem_xxx APIs in V3. DmitryK pointed out that Apps except unit tests should not use these internal APIs, and also rte_mem_map does not support hugepage with different sized. Given that, decided to work on the mman related part in a future separate patch.
> 
> > 
> > > 
> > > Signed-off-by: Jie Zhou <jizh at microsoft.com>
> > > Signed-off-by: Jie Zhou <jizh at linux.microsoft.com>
> > > ---
> > >  app/test-pmd/cmdline.c     |  3 +--
> > >  app/test-pmd/csumonly.c    |  2 +-
> > >  app/test-pmd/icmpecho.c    |  4 ++--
> > >  app/test-pmd/ieee1588fwd.c |  8 ++++----  app/test-pmd/parameters.c  |
> > > 10 +++++++---
> > >  app/test-pmd/testpmd.c     | 21 ++++++++++++++++++++-
> > >  app/test-pmd/testpmd.h     |  2 +-
> > >  7 files changed, 36 insertions(+), 14 deletions(-)
> > > 
> > 
> > <snip>
> > 
> > > @@ mbuf_pool_create(uint16_t mbuf_seg_size, unsigned nb_mbuf,
> > >                 }
> > >         }
> > > 
> > > +#ifndef RTE_EXEC_ENV_WINDOWS
> > >  err:
> > >         if (rte_mp == NULL) {
> > >                 rte_exit(EXIT_FAILURE,
> > > @@ -1065,6 +1077,7 @@ mbuf_pool_create(uint16_t mbuf_seg_size,
> > > unsigned nb_mbuf,
> > >         } else if (verbose_level > 0) {
> > >                 rte_mempool_dump(stdout, rte_mp);
> > >         }
> > > +#endif
> > 
> > Why do we need the err flow ifndef'ed?  Isn't it needed in the MP_ALLOC_NATIVE case?
> 

You are right. Will fix in V11. To avoid the unused err: lable warning, just need to ifndef the err: lable instead of the whole err flow. My mistake.

> 
> > >         return rte_mp;
> > >  }
> > >