libbpf: support weak typed ksyms.
Subsystems (6)
PATCH v3 1/1 libbpf: support weak typed ksyms.
Currently weak typeless ksyms have default value zero, when they don't exist in the kernel. However, weak typed ksyms are rejected by libbpf if they can not be resolved. This means that if a bpf object contains the declaration of a nonexistent weak typed ksym, it will be rejected even if there is no program that references the symbol.
Nonexistent weak typed ksyms can also default to zero just like typeless ones. This allows programs that access weak typed ksyms to be accepted by verifier, if the accesses are guarded. For example,
extern const int bpf_link_fops3 __ksym __weak;
/* then in BPF program */
if (&bpf_link_fops3) {
/* use bpf_link_fops3 */
}
If actual use of nonexistent typed ksym is not guarded properly, verifier would see that register is not PTR_TO_BTF_ID and wouldn't allow to use it for direct memory reads or passing it to BPF helpers.
---
Changes since v2:
- Move special handling and warning from find_ksym_btf_id() to bpf_object__resolve_ksym_var_btf_id().
- Removed bpf_link_fops3 from tests since it's not used.
- Separated variable declaration and statements.
Changes since v1:
- Weak typed symbols default to zero, as suggested by Andrii.
- Use ASSERT_XXX() for tests.
tools/lib/bpf/libbpf.c | 16 +++--- .../selftests/bpf/prog_tests/ksyms_btf.c | 31 ++++++++++ .../selftests/bpf/progs/test_ksyms_weak.c | 56 +++++++++++++++++++ 3 files changed, 96 insertions(+), 7 deletions(-) create mode 100644 tools/testing/selftests/bpf/progs/test_ksyms_weak.c
tools/lib/bpf/libbpf.c +9 -7
@@ -5277,11 +5277,11 @@ bpf_object__relocate_data(struct bpf_object *obj, struct bpf_program *prog) } insn[1 ].imm = ext->kcfg.data_off; }else /* EXT_KSYM */ { -if (ext->ksym.type_id) {/* typed ksyms */ +if (ext->ksym.type_id && ext->is_set) {/* typed ksyms */ insn[0 ].src_reg =BPF_PSEUDO_BTF_ID ; insn[0 ].imm = ext->ksym.kernel_btf_id; insn[1 ].imm = ext->ksym.kernel_btf_obj_fd; - }else {/* typeless ksyms */ + }else {/* typeless ksyms or unresolved typed ksyms */ insn[0 ].imm = (__u32 )ext->ksym.addr; insn[1 ].imm = ext->ksym.addr >>32 ; }
@@ -6608,11 +6608,8 @@ static int find_ksym_btf_id(struct bpf_object *obj, const char *ksym_name,break ; } } -if (id <=0 ) { -pr_warn ("extern (%s ksym) '%s': failed to find BTF ID in kernel BTF(s).\n" , -__btf_kind_str (kind), ksym_name); +if (id <=0 )return -ESRCH ; - } *res_btf = btf; *res_btf_fd = btf_fd;
@@ -6629,8 +6626,13 @@ static int bpf_object__resolve_ksym_var_btf_id(struct bpf_object *obj,struct btf *btf =NULL ; id =find_ksym_btf_id (obj, ext->name,BTF_KIND_VAR , &btf, &btf_fd); -if (id <0 ) +if (id == -ESRCH && ext->is_weak) { +return 0 ; + }else if (id <0 ) { +pr_warn ("extern (var ksym) '%s': not found in kernel BTF\n" , + ext->name);return id; + }/* find local type_id */ local_type_id = ext->ksym.type_id;
tools/testing/selftests/bpf/prog_tests/ksyms_btf.c +31 -0
@@ -6,6 +6,7 @@#include <bpf/btf.h> #include "test_ksyms_btf.skel.h" #include "test_ksyms_btf_null_check.skel.h" +#include "test_ksyms_weak.skel.h" static int duration;
@@ -81,6 +82,33 @@ static void test_null_check(void)test_ksyms_btf_null_check__destroy (skel); } +static void test_weak_syms (void ) +{ +struct test_ksyms_weak *skel; +struct test_ksyms_weak__data *data; +int err; + + skel =test_ksyms_weak__open_and_load (); +if (CHECK (!skel,"test_ksyms_weak__open_and_load" ,"failed\n" )) +return ; + + err =test_ksyms_weak__attach (skel); +if (CHECK (err,"test_ksyms_weak__attach" ,"skeleton attach failed: %d\n" , err)) + goto cleanup; + +/* trigger tracepoint */ +usleep (1 ); + + data = skel->data; +ASSERT_EQ (data->out__existing_typed,0 ,"existing typed ksym" ); +ASSERT_NEQ (data->out__existing_typeless,-1 ,"existing typeless ksym" ); +ASSERT_EQ (data->out__non_existent_typeless,0 ,"nonexistent typeless ksym" ); +ASSERT_EQ (data->out__non_existent_typed,0 ,"nonexistent typed ksym" ); + +cleanup: +test_ksyms_weak__destroy (skel); +} +void test_ksyms_btf (void ) {int percpu_datasec;
@@ -105,4 +133,7 @@ void test_ksyms_btf(void)if (test__start_subtest ("null_check" ))test_null_check (); + +if (test__start_subtest ("weak_ksyms" )) +test_weak_syms (); }
tools/testing/selftests/bpf/progs/test_ksyms_weak.c new file +56 -0
56 lines not on this page · raw message
Signature · 1 line
2.32.0.605.g8dce9f2422-goog
Quoted 19 lines · Hao Luo
Currently weak typeless ksyms have default value zero, when they don't exist in the kernel. However, weak typed ksyms are rejected by libbpf if they can not be resolved. This means that if a bpf object contains the declaration of a nonexistent weak typed ksym, it will be rejected even if there is no program that references the symbol.
Nonexistent weak typed ksyms can also default to zero just like typeless ones. This allows programs that access weak typed ksyms to be accepted by verifier, if the accesses are guarded. For example,
extern const int
bpf_link_fops3__ksym__weak;/* then in BPF program */
if (&bpf_link_fops3) { /* use bpf_link_fops3 */ }
Quoted 6 lines
If actual use of nonexistent typed ksym is not guarded properly, verifier would see that register is not
PTR_TO_BTF_IDand wouldn't allow to use it for direct memory reads or passing it to BPF helpers.Signed-off-by Hao Luo <[email protected]>---
Looks good, applied to bpf-next. For the future, please split libbpf and selftests changes into separate patches, it's nicer to have those logically separate.
At some point we should probably also improve libbpf error reporting for such situations, for better user experience. We have a similar problem with CO-RE relocation, verifier doesn't know about those concepts, so verifier log is not very helpful, but libbpf can make sense out of it with some extra BPF verifier log parsing.
Quoted 16 lines
Changes since v2:
- Move special handling and warning fromfind_ksym_btf_id()tobpf_object__resolve_ksym_var_btf_id().
- Removedbpf_link_fops3from tests since it's not used.
- Separated variable declaration and statements.Changes since v1:
- Weak typed symbols default to zero, as suggested by Andrii.
- UseASSERT_XXX()for tests.tools/lib/bpf/libbpf.c | 16 +++--- .../selftests/bpf/prog_tests/ksyms_btf.c | 31 ++++++++++ .../selftests/bpf/progs/test_ksyms_weak.c | 56 +++++++++++++++++++ 3 files changed, 96 insertions(+), 7 deletions(-) create mode 100644 tools/testing/selftests/bpf/progs/test_ksyms_weak.c
[...]
Quoted 32 lines · Andrii Nakryiko
Quoted 19 lines · Hao Luo
Currently weak typeless ksyms have default value zero, when they don't exist in the kernel. However, weak typed ksyms are rejected by libbpf if they can not be resolved. This means that if a bpf object contains the declaration of a nonexistent weak typed ksym, it will be rejected even if there is no program that references the symbol.
Nonexistent weak typed ksyms can also default to zero just like typeless ones. This allows programs that access weak typed ksyms to be accepted by verifier, if the accesses are guarded. For example,
extern const int
bpf_link_fops3__ksym__weak;/* then in BPF program */
if (&bpf_link_fops3) { /* use bpf_link_fops3 */ }Quoted 6 lines
If actual use of nonexistent typed ksym is not guarded properly, verifier would see that register is not
PTR_TO_BTF_IDand wouldn't allow to use it for direct memory reads or passing it to BPF helpers.Signed-off-by Hao Luo <[email protected]>---
Looks good, applied to bpf-next. For the future, please split libbpf and selftests changes into separate patches, it's nicer to have those logically separate.
Quoted 6 lines
At some point we should probably also improve libbpf error reporting for such situations, for better user experience. We have a similar problem with CO-RE relocation, verifier doesn't know about those concepts, so verifier log is not very helpful, but libbpf can make sense out of it with some extra BPF verifier log parsing.
Thanks. Will split the tests in future.
Quoted 18 lines
Quoted 16 lines
Changes since v2:
- Move special handling and warning fromfind_ksym_btf_id()tobpf_object__resolve_ksym_var_btf_id().
- Removedbpf_link_fops3from tests since it's not used.
- Separated variable declaration and statements.Changes since v1:
- Weak typed symbols default to zero, as suggested by Andrii.
- UseASSERT_XXX()for tests.tools/lib/bpf/libbpf.c | 16 +++--- .../selftests/bpf/prog_tests/ksyms_btf.c | 31 ++++++++++ .../selftests/bpf/progs/test_ksyms_weak.c | 56 +++++++++++++++++++ 3 files changed, 96 insertions(+), 7 deletions(-) create mode 100644 tools/testing/selftests/bpf/progs/test_ksyms_weak.c[...]