Re: [dpdk-dev] [PATCH v3 3/4] app/test-eventdev: add Tx adapter support

[Jerin Jacob <jerin.jacob@caviumnetworks.com>]

-----Original Message-----
> Date: Mon, 24 Sep 2018 10:30:30 +0200
> From: Andrzej Ostruszka <amo@semihalf.com>
> To: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v3 3/4] app/test-eventdev: add Tx adapter
>  support
> User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
>  Thunderbird/52.9.1
> 
> 
> On 23.09.2018 13:35, Jerin Jacob wrote:
> > -----Original Message-----
> >> Date: Thu, 20 Sep 2018 03:52:34 +0530
> >> From: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
> [...]
> >> -    struct rte_event_dev_info info;
> >> -    struct test_pipeline *t = evt_test_priv(test);
> >> -    uint8_t tx_evqueue_id = 0;
> >> +    uint8_t tx_evqueue_id[RTE_MAX_ETHPORTS] = {0};
> >
> > Some old compiler throws error with this scheme. Please change to memset.
> 
> Really?  Could you give an example?
> 
> That is perfectly legal C (since "forever"?) and I find it more readable
> than memset.  Don't treat it as a request to keep the original version -
> if I were Pavan I would object this particular request since I prefer
> direct initialization, however here I'm more interested in learning more
> about your statement about compilers not supporting zero initialization
> of array members after the last initializer.  And maybe also about to
> what extent we should be supporting old/non compliant compilers (the doc
> suggest to use gcc 4.9+).

Clang don't like this kind of zero-initialization depending on which 
type of parameter comes first in the structure. An array of uint8_t
should be OK. I thought of keeping safe here as it was going for next
revision.

Unofficially, people used to test with old compiler such as gcc 4.7 etc.

http://patches.dpdk.org/patch/40750/ // Search clang here.
http://patches.dpdk.org/patch/38189/



> 
> Best regards
> Andrzej