On Wed, Dec 20, 2023 at 4:37 PM David Marchand
<david.march...@redhat.com> wrote:
>
> Getting readable and consistent logs is important when running a DPDK
> application, especially when troubleshooting.
> A common issue with logs is when a DPDK change do not add (or on the
> contrary add too many \n) in the format string.
>
> This issue would only get noticed when actually hitting this log (which
> may be a situation hard to reach).
>
> This series proposes to introduce a new RTE_LOG_LINE helper that is
> responsible for logging a one line message and spews a build error (with
> gcc) if any \n is part of the format string.
>
>
> Since the v1 discussion on the cover letter, I changed my mind, and made the
> choice to break existing logging helpers exported in the public API.
> The reasoning is that those should not be used in the first place:
> logs should be produced only by the library that registers the logtype.
>
> Some multiline logging for debugging and the test assert macros are
> still present, but in this case an explicit call to RTE_LOG() is done.
> This can be checked with a simple:
> $ git grep -E 'RTE_LOG(_DP)?\(' -- lib/ :^lib/log/
> lib/acl/acl_bld.c: RTE_LOG(DEBUG, ACL, "Build phase for ACL \"%s\":\n"
> lib/acl/acl_gen.c: RTE_LOG(DEBUG, ACL, "Gen phase for ACL \"%s\":\n"
> lib/bpf/bpf_validate.c: RTE_LOG(DEBUG, BPF, "%s(%p) stats:\n"
> lib/bpf/bpf_validate.c: RTE_LOG(DEBUG, BPF,
> lib/eal/common/eal_common_debug.c: RTE_LOG(CRIT, EAL, "Error -
> exiting with code: %d\n"
> lib/eal/include/rte_test.h: RTE_LOG(ERR, EAL, "Test assert %s
> line %d failed: " \
> lib/ip_frag/ip_frag_common.h:#define IP_FRAG_LOG(lvl, fmt, args...)
> RTE_LOG(lvl, IPFRAG, fmt, ##args)
> lib/sched/rte_sched.c: RTE_LOG(DEBUG, SCHED, "Low level config for pipe
> profile %u:\n"
> lib/sched/rte_sched.c: RTE_LOG(DEBUG, SCHED, "Low level config for subport
> profile %u:\n"
> lib/vhost/vhost.h: RTE_LOG_DP(DEBUG, VHOST_DATA, "VHOST_DATA: (%s) %s",
> dev->ifname, packet); \
Series applied, thanks for the reviews.
--
David Marchand
