Any comments on making parsing the interface list common for all examples? This touches on https://bugs.linaro.org/show_bug.cgi?id=1079. I don't think it's time to do that yet. In fact I think we could try changing the way we pass the if names, instead of using a comma separated list we could repeat the -i option, e.g.
./odp_l2fwd -i eth0 -i eth1 -i eth2 -i eth3 We would have to limit the number of interfaces we can pass, or parse the args twice, once to count and allocate and once to save the names. But right now the examples are merely demonstration apps. On Wed, Mar 11, 2015 at 6:30 PM, Ola Liljedahl <[email protected]> wrote: > On 11 March 2015 at 17:25, Mike Holmes <[email protected]> wrote: >> >> >> >> On 11 March 2015 at 10:32, Ola Liljedahl <[email protected]> wrote: >>> >>> These examples won't leak memory continuously so why is this a problem? >>> Are we catering to e.g. bare metal environments which may not clean up >>> after applications have terminated? >> >> >> I think the tests should be clean so they can be reused in bare metal >> environment without too much trouble, but also I agree we only test on >> linux-generic. >> However on linux generic tools such as clang-scan, Coverity, Valgrind etc >> are run to catch mistakes, that is a lot easier if the code is inherently >> clean in those tools to begin with. > > I agree. I missed that this was the cause for the fixes. I don't read all > emails that carefully... delete delete delete.... > >> >> In some cases we need to mark false positives, in others add some clean up >> - this is one of those clean up items from Coverity >> https://bugs.linaro.org/show_bug.cgi?id=1117 >> >> Mike >> >> >> >>> >>> >>> >>> On 11 March 2015 at 14:52, Ciprian Barbu <[email protected]> >>> wrote: >>>> >>>> This series continues the patch I sent some while ago to replace strtok >>>> with >>>> strtok_r in odp_pktio: >>>> >>>> https://git.linaro.org/lng/odp.git/commit/273955e6db6bb220f2736d3709e4237c50d04772 >>>> >>>> The bug that this series fixes only refers to resource leaks, but the >>>> initial >>>> strtok_r fix should have been done for all examples, there is no need to >>>> use the >>>> reentrant version of strtok because it is only used from one thread, >>>> once, >>>> during parsing of args. >>>> >>>> There was a suggestion at some point to make the parsing of if_names >>>> common for >>>> all examples, I don't think it's easy to factor it out, it would >>>> introduce more >>>> code than necessary and we're only talking about a few lines of code. >>>> >>>> Ciprian Barbu (3): >>>> example: odp_l2fwd: replace strtok_r with strtok and fix leaks >>>> example: odp_ipsec: replace strtok_r with strtok and fix leaks >>>> example: generator: replace strtok_r with strtok and fix leaks >>>> >>>> example/generator/odp_generator.c | 30 ++++++++++++++++-------------- >>>> example/ipsec/odp_ipsec.c | 30 ++++++++++++++---------------- >>>> example/l2fwd/odp_l2fwd.c | 29 +++++++++++++++-------------- >>>> 3 files changed, 45 insertions(+), 44 deletions(-) >>>> >>>> -- >>>> 1.8.3.2 >>>> >>>> >>>> _______________________________________________ >>>> lng-odp mailing list >>>> [email protected] >>>> http://lists.linaro.org/mailman/listinfo/lng-odp >>> >>> >>> >>> _______________________________________________ >>> lng-odp mailing list >>> [email protected] >>> http://lists.linaro.org/mailman/listinfo/lng-odp >>> >> >> >> >> -- >> Mike Holmes >> Technical Manager - Linaro Networking Group >> Linaro.org │ Open source software for ARM SoCs >> >> > _______________________________________________ lng-odp mailing list [email protected] http://lists.linaro.org/mailman/listinfo/lng-odp
