On Fri, Jan 16, 2026 at 6:17 AM Leon Hwang <[email protected]> wrote:
>
>
>
> On 2026/1/16 09:03, Andrii Nakryiko wrote:
> > On Mon, Jan 12, 2026 at 6:59 AM Leon Hwang <[email protected]> wrote:
> >>
> >> With the previous commit adding common attribute support for
> >> BPF_MAP_CREATE, users can now retrieve detailed error messages when map
> >> creation fails via the log_buf field.
> >>
> >> Introduce struct bpf_syscall_common_attr_opts with the following fields:
> >> log_buf, log_size, log_level, and log_true_size.
> >>
> >> Extend bpf_map_create_opts with a new field common_attr_opts, allowing
> >> users to capture and inspect log messages on map creation failures.
> >>
> >> Signed-off-by: Leon Hwang <[email protected]>
> >> ---
> >> tools/lib/bpf/bpf.c | 15 ++++++++++++++-
> >> tools/lib/bpf/bpf.h | 17 ++++++++++++++++-
> >> 2 files changed, 30 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
> >> index d44e667aaf02..d65df1b7b2be 100644
> >> --- a/tools/lib/bpf/bpf.c
> >> +++ b/tools/lib/bpf/bpf.c
> >> @@ -207,6 +207,9 @@ int bpf_map_create(enum bpf_map_type map_type,
> >> const struct bpf_map_create_opts *opts)
> >> {
> >> const size_t attr_sz = offsetofend(union bpf_attr,
> >> excl_prog_hash_size);
> >> + const size_t common_attr_sz = sizeof(struct bpf_common_attr);
> >> + struct bpf_syscall_common_attr_opts *common_attr_opts;
> >> + struct bpf_common_attr common_attr;
> >> union bpf_attr attr;
> >> int fd;
> >>
> >> @@ -240,7 +243,17 @@ int bpf_map_create(enum bpf_map_type map_type,
> >> attr.excl_prog_hash = ptr_to_u64(OPTS_GET(opts, excl_prog_hash,
> >> NULL));
> >> attr.excl_prog_hash_size = OPTS_GET(opts, excl_prog_hash_size, 0);
> >>
> >> - fd = sys_bpf_fd(BPF_MAP_CREATE, &attr, attr_sz);
> >> + common_attr_opts = OPTS_GET(opts, common_attr_opts, NULL);
> >> + if (common_attr_opts && feat_supported(NULL,
> >> FEAT_EXTENDED_SYSCALL)) {
> >> + memset(&common_attr, 0, common_attr_sz);
> >> + common_attr.log_buf =
> >> ptr_to_u64(OPTS_GET(common_attr_opts, log_buf, NULL));
> >> + common_attr.log_size = OPTS_GET(common_attr_opts,
> >> log_size, 0);
> >> + common_attr.log_level = OPTS_GET(common_attr_opts,
> >> log_level, 0);
> >> + fd = sys_bpf_ext_fd(BPF_MAP_CREATE, &attr, attr_sz,
> >> &common_attr, common_attr_sz);
> >> + OPTS_SET(common_attr_opts, log_true_size,
> >> common_attr.log_true_size);
> >> + } else {
> >> + fd = sys_bpf_fd(BPF_MAP_CREATE, &attr, attr_sz);
> >
> > OPTS_SET(log_true_size) to zero here, maybe?
> >
>
> Unnecessary, but ok to do it.
>
> >> + }
> >> return libbpf_err_errno(fd);
> >> }
> >>
> >> diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h
> >> index 2c8e88ddb674..c4a26e6b71ea 100644
> >> --- a/tools/lib/bpf/bpf.h
> >> +++ b/tools/lib/bpf/bpf.h
> >> @@ -37,6 +37,18 @@ extern "C" {
> >>
> >> LIBBPF_API int libbpf_set_memlock_rlim(size_t memlock_bytes);
> >>
> >> +struct bpf_syscall_common_attr_opts {
> >> + size_t sz; /* size of this struct for forward/backward
> >> compatibility */
> >> +
> >> + char *log_buf;
> >> + __u32 log_size;
> >> + __u32 log_level;
> >> + __u32 log_true_size;
> >> +
> >> + size_t :0;
> >> +};
> >> +#define bpf_syscall_common_attr_opts__last_field log_true_size
> >
> > see below, let's drop this struct and just add these 4 fields directly
> > to bpf_map_create_opts
> >
> >> +
> >> struct bpf_map_create_opts {
> >> size_t sz; /* size of this struct for forward/backward
> >> compatibility */
> >>
> >> @@ -57,9 +69,12 @@ struct bpf_map_create_opts {
> >>
> >> const void *excl_prog_hash;
> >> __u32 excl_prog_hash_size;
> >> +
> >> + struct bpf_syscall_common_attr_opts *common_attr_opts;
> >
> > maybe let's just add those log_xxx fields here directly? This whole
> > extra bpf_syscall_common_attr_opts pointer and struct seems like a
> > cumbersome API.
> >
>
> Oops... This struct was suggested by the v3 discussion [1].
>
> This struct was used to report 'log_true_size' without changing
> 'bpf_map_create()' API.
>
Ah, I already forgot. log_true_size being an output parameter here...
Sigh. I don't like the verboseness of bpf_syscall_common_attr_opts and
"common_attr_opts" and all that stuff...
What if we make it struct bpf_log_opts {} and keep it log-specific?
> Links
> [1]
> https://lore.kernel.org/bpf/CAEf4Bzaw9cboFSf1OXmD84S7pKaeyj=bcqg_diuzgwakfsj...@mail.gmail.com/
>
> Thanks,
> Leon
>
> >> +
> >> size_t :0;
> >> };
> >> -#define bpf_map_create_opts__last_field excl_prog_hash_size
> >> +#define bpf_map_create_opts__last_field common_attr_opts
> >>
> >> LIBBPF_API int bpf_map_create(enum bpf_map_type map_type,
> >> const char *map_name,
> >> --
> >> 2.52.0
> >>
>
