Dmitry Eremin-Solenikov(lumag) replied on github web page: platform/linux-generic/m4/odp_pcapng.m4 line 17 @@ -0,0 +1,19 @@ +########################################################################## +# Enable PCAPNG support +########################################################################## +have_pcapng=no +AC_ARG_ENABLE([pcapng-support], + [AS_HELP_STRING([--enable-pcapng-support], + [enable experimental tcpdump for pktios])], + have_pcapng=$enableval) + +if test x$have_pcapng = xyes +then + AC_DEFINE([ODP_PCAPNG], [1], + [Define to 1 to enable pcapng support]) +else + AC_DEFINE([ODP_PCAPNG], [0], + [Define to 0 to disable pcapng support]) +fi
Comment: Anyway, just call AC_DEFINE once. > Ilias Apalodimas(apalos) wrote: > will rename >> Ilias Apalodimas(apalos) wrote: >> will fix >>> Ilias Apalodimas(apalos) wrote: >>> no the extra * will be removed on v2 >>>> Ilias Apalodimas(apalos) wrote: >>>> Cuinit caught actual bugs in the original implementation, i'd like to keep >>>> that >>>>> Ilias Apalodimas(apalos) wrote: >>>>> We did that in net_mdev and the "if" was corrected to check the value was >>>>> equal to 1 >>>>>> Ilias Apalodimas(apalos) wrote: >>>>>> This is how dpdk zero-copy was introduced. I think we need to keep that >>>>>> for the same reasons @Bill-Fischofer-Linaro mentioned >>>>>>> Bill Fischofer(Bill-Fischofer-Linaro) wrote: >>>>>>> @lumag This style ensures that the code is always syntax-checked since >>>>>>> it is compiled, but then optimized away if it is not active. I've found >>>>>>> this to be a useful idiom. >>>>>>>> Bill Fischofer(Bill-Fischofer-Linaro) wrote: >>>>>>>> Having this as a section of the User Guide is fine, but you need to >>>>>>>> add this file to the list in `doc/users-guide/Makefile.am` so that >>>>>>>> changes to it will cause the User Guide to be regenerated. >>>>>>>>> Bill Fischofer(Bill-Fischofer-Linaro) wrote: >>>>>>>>> This formats as __\<odp global pid\>-\<NIC name\>-flow-\<queue >>>>>>>>> *number\>__ >>>>>>>>> >>>>>>>>> Is that what you wanted? >>>>>>>>>> Bill Fischofer(Bill-Fischofer-Linaro) wrote: >>>>>>>>>> @lumag Unfortunately flexibility comes at a price. We can't have >>>>>>>>>> untested features and if they are selectable at `configure` time >>>>>>>>>> then I don't see how to avoid this. >>>>>>>>>>> Dmitry Eremin-Solenikov(lumag) wrote: >>>>>>>>>>> I'm thinking about limiting variety of build options here. Not in >>>>>>>>>>> your PR, but there are just way tooo many >>>>>>>>>>>> Dmitry Eremin-Solenikov(lumag) wrote: >>>>>>>>>>>> And also it should be `_odp_pcapng_FOO` >>>>>>>>>>>>> Dmitry Eremin-Solenikov(lumag) wrote: >>>>>>>>>>>>> `_odp_smth...` >>>>>>>>>>>>>> Dmitry Eremin-Solenikov(lumag) wrote: >>>>>>>>>>>>>> Neat idea, but just put pcapng_prepare inside header and define >>>>>>>>>>>>>> it to empty val if ODP_PCAPNG is undefined. >>>>>>>>>>>>>> Also if it returns an error, it should not require one to call >>>>>>>>>>>>>> pcapng_destroy() afterwards. Just make it clean up after itself >>>>>>>>>>>>>> in case of an error. >>>>>>>>>>>>>>> Dmitry Eremin-Solenikov(lumag) wrote: >>>>>>>>>>>>>>> Just use one AC_DEFINE and check using `#ifdef` inside. https://github.com/Linaro/odp/pull/488#discussion_r169274885 updated_at 2018-02-20 10:26:48