| Message ID | b73550a69ea8c02fd93c862f9cfe38f7e1813e7a.1644970147.git.delyank@fb.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | BPF |
| Headers | show |
| Series | Avoid size mismatches in skeletons | expand |
| Context | Check | Description |
|---|---|---|
| netdev/tree_selection | success | Clearly marked for bpf-next |
| netdev/apply | fail | Patch does not apply to bpf-next |
| bpf/vmtest-bpf-next-PR | success | PR summary |
| bpf/vmtest-bpf-next | success | VM_Test |
On Tue, Feb 15, 2022 at 4:12 PM Delyan Kratunov <delyank@fb.com> wrote: > > When emitting type declarations in skeletons, bpftool will now also emit > static assertions on the size of the data/bss/rodata/etc fields. This > ensures that in situations where userspace and kernel types have the same > name but differ in size we do not silently produce incorrect results but > instead break the build. > > This was reported in [1] and as expected the repro in [2] fails to build > on the new size assert after this change. > > [1]: Closes: https://github.com/libbpf/libbpf/issues/433 > [2]: https://github.com/fuweid/iovisor-bcc-pr-3777 > > Signed-off-by: Delyan Kratunov <delyank@fb.com> > --- LGTM with a trivial styling nits. But this doesn't apply cleanly to bpf-next (see [0]). Can you please rebase and resend. Also for single-patch submissions we don't require cover letter, to please just put all the description into one patch without cover letter. [0] https://github.com/kernel-patches/bpf/pull/2563#issuecomment-1040929960 > tools/bpf/bpftool/gen.c | 134 +++++++++++++++++++++++++++++++++------- > 1 file changed, 112 insertions(+), 22 deletions(-) [...] > + > + bpf_object__for_each_map(map, obj) { > + if (!bpf_map__is_internal(map)) > + continue; > + if (!(bpf_map__map_flags(map) & BPF_F_MMAPABLE)) > + continue; > + if (!get_map_ident(map, map_ident, sizeof(map_ident))) > + continue; > + > + sec = find_type_for_map(obj, map_ident); > + nit: unnecessary empty line between assignment and "error checking" > + if (!sec) { > + /* best effort, couldn't find the type for this map */ > + continue; > + } > + > + sec_var = btf_var_secinfos(sec); > + vlen = btf_vlen(sec); > + > + for (i = 0; i < vlen; i++, sec_var++) { > + const struct btf_type *var = btf__type_by_id(btf, sec_var->type); > + const char *var_name = btf__name_by_offset(btf, var->name_off); > + __u32 var_type_id = var->type; > + __s64 var_size = btf__resolve_size(btf, var_type_id); > + > + if (var_size < 0) > + continue; > + > + /* static variables are not exposed through BPF skeleton */ > + if (btf_var(var)->linkage == BTF_VAR_STATIC) > + continue; > + > + var_ident[0] = '\0'; > + strncat(var_ident, var_name, sizeof(var_ident) - 1); > + sanitize_identifier(var_ident); > + > + printf("\t_Static_assert("); > + printf("sizeof(s->%1$s->%2$s) == %3$lld, ", > + map_ident, var_ident, var_size); > + printf("\"unexpected size of '%1$s'\");\n", var_ident); nit: I'd keep this as one printf, it makes it a bit easier to follow. > + } > + } [...]
On 2/16/22 09:19, Andrii Nakryiko wrote: > On Tue, Feb 15, 2022 at 4:12 PM Delyan Kratunov <delyank@fb.com> wrote: >> >> When emitting type declarations in skeletons, bpftool will now also emit >> static assertions on the size of the data/bss/rodata/etc fields. This >> ensures that in situations where userspace and kernel types have the same >> name but differ in size we do not silently produce incorrect results but >> instead break the build. >> >> This was reported in [1] and as expected the repro in [2] fails to build >> on the new size assert after this change. >> >> [1]: Closes: https://github.com/libbpf/libbpf/issues/433 >> [2]: https://github.com/fuweid/iovisor-bcc-pr-3777 >> >> Signed-off-by: Delyan Kratunov <delyank@fb.com> >> --- > > LGTM with a trivial styling nits. But this doesn't apply cleanly to > bpf-next (see [0]). Can you please rebase and resend. Also for > single-patch submissions we don't require cover letter, to please just > put all the description into one patch without cover letter. > > [0] https://github.com/kernel-patches/bpf/pull/2563#issuecomment-1040929960 > >> tools/bpf/bpftool/gen.c | 134 +++++++++++++++++++++++++++++++++------- >> 1 file changed, 112 insertions(+), 22 deletions(-) > > [...] > >> + >> + bpf_object__for_each_map(map, obj) { >> + if (!bpf_map__is_internal(map)) >> + continue; >> + if (!(bpf_map__map_flags(map) & BPF_F_MMAPABLE)) >> + continue; >> + if (!get_map_ident(map, map_ident, sizeof(map_ident))) >> + continue; >> + >> + sec = find_type_for_map(obj, map_ident); >> + > > nit: unnecessary empty line between assignment and "error checking" > >> + if (!sec) { >> + /* best effort, couldn't find the type for this map */ >> + continue; >> + } >> + >> + sec_var = btf_var_secinfos(sec); >> + vlen = btf_vlen(sec); >> + >> + for (i = 0; i < vlen; i++, sec_var++) { >> + const struct btf_type *var = btf__type_by_id(btf, sec_var->type); >> + const char *var_name = btf__name_by_offset(btf, var->name_off); >> + __u32 var_type_id = var->type; >> + __s64 var_size = btf__resolve_size(btf, var_type_id); >> + >> + if (var_size < 0) >> + continue; >> + >> + /* static variables are not exposed through BPF skeleton */ >> + if (btf_var(var)->linkage == BTF_VAR_STATIC) >> + continue; >> + >> + var_ident[0] = '\0'; >> + strncat(var_ident, var_name, sizeof(var_ident) - 1); >> + sanitize_identifier(var_ident); >> + >> + printf("\t_Static_assert("); >> + printf("sizeof(s->%1$s->%2$s) == %3$lld, ", >> + map_ident, var_ident, var_size); >> + printf("\"unexpected size of '%1$s'\");\n", var_ident); > > nit: I'd keep this as one printf, it makes it a bit easier to follow. > >> + } >> + } > > [...] Feel free to add: Acked-by: Hengqi Chen <hengqi.chen@gmail.com> Tested-by: Hengqi Chen <hengqi.chen@gmail.com>
diff --git a/tools/bpf/bpftool/gen.c b/tools/bpf/bpftool/gen.c index 6f2e20be0c62..c1440c0d60b5 100644 --- a/tools/bpf/bpftool/gen.c +++ b/tools/bpf/bpftool/gen.c @@ -208,15 +208,38 @@ static int codegen_datasec_def(struct bpf_object *obj, return 0; } +static const struct btf_type *find_type_for_map(struct bpf_object *obj, + const char *map_ident) +{ + struct btf *btf = bpf_object__btf(obj); + int n = btf__type_cnt(btf), i; + char sec_ident[256]; + + for (i = 1; i < n; i++) { + const struct btf_type *t = btf__type_by_id(btf, i); + const char *name; + + if (!btf_is_datasec(t)) + continue; + + name = btf__str_by_offset(btf, t->name_off); + if (!get_datasec_ident(name, sec_ident, sizeof(sec_ident))) + continue; + + if (strcmp(sec_ident, map_ident) == 0) + return t; + } + return NULL; +} + static int codegen_datasecs(struct bpf_object *obj, const char *obj_name) { struct btf *btf = bpf_object__btf(obj); - int n = btf__type_cnt(btf); struct btf_dump *d; struct bpf_map *map; const struct btf_type *sec; - char sec_ident[256], map_ident[256]; - int i, err = 0; + char map_ident[256]; + int err = 0; d = btf_dump__new(btf, codegen_btf_dump_printf, NULL, NULL); err = libbpf_get_error(d); @@ -233,23 +256,7 @@ static int codegen_datasecs(struct bpf_object *obj, const char *obj_name) if (!get_map_ident(map, map_ident, sizeof(map_ident))) continue; - sec = NULL; - for (i = 1; i < n; i++) { - const struct btf_type *t = btf__type_by_id(btf, i); - const char *name; - - if (!btf_is_datasec(t)) - continue; - - name = btf__str_by_offset(btf, t->name_off); - if (!get_datasec_ident(name, sec_ident, sizeof(sec_ident))) - continue; - - if (strcmp(sec_ident, map_ident) == 0) { - sec = t; - break; - } - } + sec = find_type_for_map(obj, map_ident); /* In some cases (e.g., sections like .rodata.cst16 containing * compiler allocated string constants only) there will be @@ -362,6 +369,81 @@ static size_t bpf_map_mmap_sz(const struct bpf_map *map) return map_sz; } +/* Emit type size asserts for all top-level fields in memory-mapped internal maps. + */ +static void codegen_asserts(struct bpf_object *obj, const char *obj_name) +{ + struct btf *btf = bpf_object__btf(obj); + struct bpf_map *map; + struct btf_var_secinfo *sec_var; + int i, vlen; + const struct btf_type *sec; + char map_ident[256], var_ident[256]; + + codegen("\ + \n\ + \n\ + #ifdef __cplusplus \n\ + #define _Static_assert static_assert \n\ + #endif \n\ + \n\ + __attribute__((unused)) static void \n\ + %1$s__type_asserts(struct %1$s *s) \n\ + { \n\ + ", obj_name); + + bpf_object__for_each_map(map, obj) { + if (!bpf_map__is_internal(map)) + continue; + if (!(bpf_map__map_flags(map) & BPF_F_MMAPABLE)) + continue; + if (!get_map_ident(map, map_ident, sizeof(map_ident))) + continue; + + sec = find_type_for_map(obj, map_ident); + + if (!sec) { + /* best effort, couldn't find the type for this map */ + continue; + } + + sec_var = btf_var_secinfos(sec); + vlen = btf_vlen(sec); + + for (i = 0; i < vlen; i++, sec_var++) { + const struct btf_type *var = btf__type_by_id(btf, sec_var->type); + const char *var_name = btf__name_by_offset(btf, var->name_off); + __u32 var_type_id = var->type; + __s64 var_size = btf__resolve_size(btf, var_type_id); + + if (var_size < 0) + continue; + + /* static variables are not exposed through BPF skeleton */ + if (btf_var(var)->linkage == BTF_VAR_STATIC) + continue; + + var_ident[0] = '\0'; + strncat(var_ident, var_name, sizeof(var_ident) - 1); + sanitize_identifier(var_ident); + + printf("\t_Static_assert("); + printf("sizeof(s->%1$s->%2$s) == %3$lld, ", + map_ident, var_ident, var_size); + printf("\"unexpected size of '%1$s'\");\n", var_ident); + } + } + codegen("\ + \n\ + } \n\ + \n\ + #ifdef __cplusplus \n\ + #undef _Static_assert \n\ + #endif \n\ + "); +} + + static void codegen_attach_detach(struct bpf_object *obj, const char *obj_name) { struct bpf_program *prog; @@ -640,6 +722,8 @@ static int gen_trace(struct bpf_object *obj, const char *obj_name, const char *h } \n\ ", obj_name); + codegen_asserts(obj, obj_name); + codegen("\ \n\ \n\ @@ -1024,8 +1108,14 @@ static int do_skeleton(int argc, char **argv) \n\ \"; \n\ } \n\ - \n\ - #endif /* %s */ \n\ + "); + + codegen_asserts(obj, obj_name); + + codegen("\ + \n\ + \n\ + #endif /* %s */ \n\ ", header_guard); err = 0;
When emitting type declarations in skeletons, bpftool will now also emit static assertions on the size of the data/bss/rodata/etc fields. This ensures that in situations where userspace and kernel types have the same name but differ in size we do not silently produce incorrect results but instead break the build. This was reported in [1] and as expected the repro in [2] fails to build on the new size assert after this change. [1]: Closes: https://github.com/libbpf/libbpf/issues/433 [2]: https://github.com/fuweid/iovisor-bcc-pr-3777 Signed-off-by: Delyan Kratunov <delyank@fb.com> --- tools/bpf/bpftool/gen.c | 134 +++++++++++++++++++++++++++++++++------- 1 file changed, 112 insertions(+), 22 deletions(-)