[dpdk-dev] [EXT] [PATCH v2 2/4] trace: remove size limit on CTF event description

[Sunil Kumar Kori]

>-----Original Message-----
>From: David Marchand <david.marchand at redhat.com>
>Sent: Thursday, October 29, 2020 3:33 PM
>To: Sunil Kumar Kori <skori at marvell.com>
>Cc: dev at dpdk.org; Jerin Jacob Kollanukkaran <jerinj at marvell.com>
>Subject: Re: [EXT] [PATCH v2 2/4] trace: remove size limit on CTF event
>description
>
>On Thu, Oct 29, 2020 at 10:36 AM Sunil Kumar Kori <skori at marvell.com>
>wrote:
>> Yes but I am not able to understand that how tp->ctf_field will be
>> populated with latest memory because RTE_PER_LCORE(ctf_field) is being
>free and re-allocated at runtime.
>
>Initially, RTE_PER_LCORE(ctf_field) is NULL.
>
>Then registration for a trace point happens:
>
>__rte_trace_point_register() calls register_fn() which for each field in the trace
>point calls __rte_trace_point_emit_field().
>
>Each call to __rte_trace_point_emit_field accumulates the previous ctf_field
>with the new field.
>
>        rc = asprintf(&field, "%s        %s %s;\n",
>                RTE_PER_LCORE(ctf_field) != NULL ?
>                        RTE_PER_LCORE(ctf_field) : "",
>                datatype, in);
>        free(RTE_PER_LCORE(ctf_field));
>        RTE_PER_LCORE(ctf_field) = field;
>
>Here, RTE_PER_LCORE(ctf_field) is != NULL.
>
>Back to __rte_trace_point_register:
>
>        /* Copy the accumulated fields description and clear it for the next
>         * trace point.
>         */
>        tp->ctf_field = RTE_PER_LCORE(ctf_field);
>
>Once stored, RTE_PER_LCORE(ctf_field) is cleared again for the next trace
>point registration.
>
>        RTE_PER_LCORE(ctf_field) = NULL;
>
>
>

Acked-by: Sunil Kumar Kori <skori at mavell.com>
>--
>David Marchand