Alison Schofield wrote:
> When the --media-errors option was added to cxl list it inadvertently
> changed the optional libtracefs requirement into a mandatory one.
> Ndctl versions 80,81,82 no longer build without libtracefs.
>
> Remove that dependency.
>
> When libtracefs is disabled the user will see a 'Notice' level
> message, like this:
> $ cxl list -r region0 --media-errors --targets
> cxl list: cmd_list: --media-errors support disabled at build time
>
> ...followed by the region listing including the output for any other
> valid command line options, like --targets in the example above.
>
> When libtracefs is disabled the cxl-poison.sh unit test is omitted.
>
> The man page gets a note:
> The media-error option is only available with -Dlibtracefs=enabled.
>
> Reported-by: Andreas Hasenack <[email protected]>
> Fixes: d7532bb049e0 ("cxl/list: add --media-errors option to cxl list")
> Closes: https://github.com/pmem/ndctl/issues/289
> Signed-off-by: Alison Schofield <[email protected]>
> ---
>
> Changes in v2:
> - Notify and continue when --media-error info is unavailable (Dan)
>
>
> Documentation/cxl/cxl-list.txt | 2 ++
> config.h.meson | 2 +-
> cxl/json.c | 15 ++++++++++++++-
> cxl/list.c | 6 ++++++
> test/meson.build | 9 +++++++--
> 5 files changed, 30 insertions(+), 4 deletions(-)
>
> diff --git a/Documentation/cxl/cxl-list.txt b/Documentation/cxl/cxl-list.txt
> index 9a9911e7dd9b..0595638ee054 100644
> --- a/Documentation/cxl/cxl-list.txt
> +++ b/Documentation/cxl/cxl-list.txt
> @@ -425,6 +425,8 @@ OPTIONS
> "source:" is one of: External, Internal, Injected, Vendor Specific,
> or Unknown, as defined in CXL Specification v3.1 Table 8-140.
>
> +The media-errors option is only available with '-Dlibtracefs=enabled'.
> +
> ----
> # cxl list -m mem9 --media-errors -u
> {
> diff --git a/config.h.meson b/config.h.meson
> index f75db3e6360f..e8539f8d04df 100644
> --- a/config.h.meson
> +++ b/config.h.meson
> @@ -19,7 +19,7 @@
> /* ndctl test support */
> #mesondefine ENABLE_TEST
>
> -/* cxl monitor support */
> +/* cxl monitor and cxl list --media-errors support */
> #mesondefine ENABLE_LIBTRACEFS
>
> /* Define to 1 if big-endian-arch */
> diff --git a/cxl/json.c b/cxl/json.c
> index e65bd803b706..a75928bf43ed 100644
> --- a/cxl/json.c
> +++ b/cxl/json.c
> @@ -9,12 +9,15 @@
> #include <json-c/json.h>
> #include <json-c/printbuf.h>
> #include <ccan/short_types/short_types.h>
> +
> +#ifdef ENABLE_LIBTRACEFS
> #include <tracefs.h>
> +#include "../util/event_trace.h"
> +#endif
Maybe this is my kernel taste leaking through, but I am allergic to
ifdef in ".c" files. In this case you could move the include of
tracefs.h into event_trace.h and then ifdef guard all the parts of
event_trace.h that need tracefs.h.
>
> #include "filter.h"
> #include "json.h"
> #include "../daxctl/json.h"
> -#include "../util/event_trace.h"
>
> #define CXL_FW_VERSION_STR_LEN 16
> #define CXL_FW_MAX_SLOTS 4
> @@ -575,6 +578,7 @@ err_jobj:
> return NULL;
> }
>
> +#ifdef ENABLE_LIBTRACEFS
> /* CXL Spec 3.1 Table 8-140 Media Error Record */
> #define CXL_POISON_SOURCE_MAX 7
> static const char *const poison_source[] = { "Unknown", "External",
> "Internal",
> @@ -753,6 +757,15 @@ err_free:
> tracefs_instance_free(inst);
> return jpoison;
> }
> +#else
> +static struct json_object *
> +util_cxl_poison_list_to_json(struct cxl_region *region,
> + struct cxl_memdev *memdev,
> + unsigned long flags)
> +{
> + return NULL;
> +}
> +#endif
This would move a new conditionally compiled ".c" file for just these
trace helpers.
> struct json_object *util_cxl_memdev_to_json(struct cxl_memdev *memdev,
> unsigned long flags)
> diff --git a/cxl/list.c b/cxl/list.c
> index 0b25d78248d5..48bd1ebc3c0e 100644
> --- a/cxl/list.c
> +++ b/cxl/list.c
> @@ -146,6 +146,12 @@ int cmd_list(int argc, const char **argv, struct cxl_ctx
> *ctx)
> param.ctx.log_priority = LOG_DEBUG;
> }
>
> +#ifndef ENABLE_LIBTRACEFS
> + if (param.media_errors) {
> + notice(¶m, "--media-errors support disabled at build
> time\n");
> + param.media_errors = false;
> + }
> +#endif
This would be a static inline helper in a header file that optionally
reports the problem.
All that said, I am not the maintainer so go with what you want, but
leaking ifdef in ".c" is, in my opinion, a slow walk into an
increasingly unreadable code base.
