Skip to contentSkip to conversation

libbpf: Ignore exception frame section

PATCH Superseded Tw 4 participants 4 messages Aug 24

Subsystems (2)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Tw

author
#0 lore raw

PATCH v2 1/1 libbpf: Ignore exception frame section

Exception handling in BPF program is not supported right now, let's skip it. BTW, I found this when building BPF with Zig compiler[1].

[1]: https://github.com/tw4452852/bb

---
v2:
Revise the commit message.

 tools/lib/bpf/libbpf.c | 3 ++-
 tools/lib/bpf/linker.c | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)
tools/lib/bpf/libbpf.c +2 -1
index b749c0174..05727c418 100644
@@ -3867,7 +3867,8 @@ static Elf64_Rel *elf_rel_by_idx(Elf_Data *data, size_t idx)
 static bool is_sec_name_dwarf(const char *name)
 {
 	/* approximation, but the actual list is too long */
-	return str_has_pfx(name, ".debug_");
+	return str_has_pfx(name, ".debug_") ||
+		   strcmp(name, ".eh_frame") == 0;
 }
 
 static bool ignore_elf_section(Elf64_Shdr *hdr, const char *name)
tools/lib/bpf/linker.c +2 -1
index 78f92c392..1fb0efdbf 100644
@@ -599,7 +599,8 @@ int bpf_linker__add_buf(struct bpf_linker *linker, void *buf, size_t buf_sz,
 static bool is_dwarf_sec_name(const char *name)
 {
 	/* approximation, but the actual list is too long */
-	return strncmp(name, ".debug_", sizeof(".debug_") - 1) == 0;
+	return strncmp(name, ".debug_", sizeof(".debug_") - 1) == 0 ||
+		   strcmp(name, ".eh_frame") == 0;
 }
 
 static bool is_ignored_sec(struct src_sec *sec)
Signature · 1 line
2.54.0
#1 lore raw
Quoted 6 lines · Tw

Exception handling in BPF program is not supported right now, let's skip it. BTW, I found this when building BPF with Zig compiler[1].

[1]: https://github.com/tw4452852/bb

You need to have your full real name in the SoB tag. Patches cannot be accepted otherwise.

pw-bot: cr

Quoted 36 lines

---
v2:
Revise the commit message.

 tools/lib/bpf/libbpf.c | 3 ++-
 tools/lib/bpf/linker.c | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)
tools/lib/bpf/libbpf.c +2 -1
index b749c0174..05727c418 100644
@@ -3867,7 +3867,8 @@ static Elf64_Rel *elf_rel_by_idx(Elf_Data *data, size_t idx)
 static bool is_sec_name_dwarf(const char *name)
 {
 	/* approximation, but the actual list is too long */
-	return str_has_pfx(name, ".debug_");
+	return str_has_pfx(name, ".debug_") ||
+		   strcmp(name, ".eh_frame") == 0;
 }
 
 static bool ignore_elf_section(Elf64_Shdr *hdr, const char *name)
tools/lib/bpf/linker.c +2 -1
index 78f92c392..1fb0efdbf 100644
@@ -599,7 +599,8 @@ int bpf_linker__add_buf(struct bpf_linker *linker, void *buf, size_t buf_sz,
 static bool is_dwarf_sec_name(const char *name)
 {
 	/* approximation, but the actual list is too long */
-	return strncmp(name, ".debug_", sizeof(".debug_") - 1) == 0;
+	return strncmp(name, ".debug_", sizeof(".debug_") - 1) == 0 ||
+		   strcmp(name, ".eh_frame") == 0;
 }
 
 static bool is_ignored_sec(struct src_sec *sec)
#2 lore raw
In reply to Tw
Kumar Kartikeya Dwivedi pointed out in both v1 and v2 reviews that the Signed-off-by tag must contain the author's full real name, not just 'Tw'. The v2 review explicitly stated: 'You need to have…
Quoted 8 lines

libbpf: Ignore exception frame section

Exception handling in BPF program is not supported right now, let's skip it. BTW, I found this when building BPF with Zig compiler[1].

[1]: https://github.com/tw4452852/bb

Kumar Kartikeya Dwivedi pointed out in both v1 and v2 reviews that the Signed-off-by tag must contain the author's full real name, not just 'Tw'. The v2 review explicitly stated: 'You need to have your full real name in the SoB tag. Patches cannot be accepted otherwise.'

Could you update the Signed-off-by to include your full name?

tools/lib/bpf/libbpf.c static Elf64_Rel *elf_rel_by_idx(Elf_Data *data, size_t idx) · line 3871

Quoted 11 lines
tools/lib/bpf/libbpf.c +2 -1
index b749c01742ee..05727c418c40 100644
@@ -3867,7 +3867,8 @@ static Elf64_Rel *elf_rel_by_idx(Elf_Data *data, size_t idx)
 static bool is_sec_name_dwarf(const char *name)
 {
 	/* approximation, but the actual list is too long */
-	return str_has_pfx(name, ".debug_");
+	return str_has_pfx(name, ".debug_") ||
+		   strcmp(name, ".eh_frame") == 0;

This isn't a bug, but could the continuation line be aligned under the first operand (tab + 7 spaces) so the two operands line up? Every other wrapped return statement in libbpf.c uses tab plus seven spaces to align the continuation, like in bpf_map__is_ringbuf():

    return map->def.type == BPF_MAP_TYPE_RINGBUF ||
           map->def.type == BPF_MAP_TYPE_USER_RINGBUF;
Quoted 8 lines
tools/lib/bpf/libbpf.c +0 -0
 }
 
 static bool ignore_elf_section(Elf64_Shdr *hdr, const char *name)
tools/lib/bpf/linker.c +0 -0
index 78f92c39290a..1fb0efdbfae7 100644

tools/lib/bpf/linker.c int bpf_linker__add_buf(struct bpf_linker *linker, void *buf, size_t buf_sz, · line 603

Quoted 6 lines
tools/lib/bpf/linker.c +2 -1
@@ -599,7 +599,8 @@ int bpf_linker__add_buf(struct bpf_linker *linker, void *buf, size_t buf_sz,
 static bool is_dwarf_sec_name(const char *name)
 {
 	/* approximation, but the actual list is too long */
-	return strncmp(name, ".debug_", sizeof(".debug_") - 1) == 0;
+	return strncmp(name, ".debug_", sizeof(".debug_") - 1) == 0 ||
+		   strcmp(name, ".eh_frame") == 0;

This isn't a bug, but since the same .eh_frame check now has to be extended in both is_sec_name_dwarf() and is_dwarf_sec_name(), and their callers ignore_elf_section() and is_ignored_sec() are also near-verbatim copies including the same .rel prefix-stripping and the same BTF section exclusions, would it be worth moving the shared predicate into libbpf_internal.h so future additions only need one edit?

tools/lib/bpf/linker.c int bpf_linker__add_buf(struct bpf_linker *linker, void *buf, size_t buf_sz, · line 606

Quoted 3 lines
tools/lib/bpf/linker.c +0 -0
 }
 
 static bool is_ignored_sec(struct src_sec *sec)

---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

CI run summary: https://github.com/kernel-patches/bpf/actions/runs/32731757949

#3 lore raw
In reply to Tw
Quoted 4 lines · Tw

Exception handling in BPF program is not supported right now, let's skip it. BTW, I found this when building BPF with Zig compiler[1].

Instruct zig compiler to not generate .eh_frame for BPF object file, which is not supported anyways? That's been a solution for C/C++ compilers so far.

Quoted 43 lines

[1]: https://github.com/tw4452852/bb

---
v2:
Revise the commit message.

 tools/lib/bpf/libbpf.c | 3 ++-
 tools/lib/bpf/linker.c | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)
tools/lib/bpf/libbpf.c +2 -1
index b749c0174..05727c418 100644
@@ -3867,7 +3867,8 @@ static Elf64_Rel *elf_rel_by_idx(Elf_Data *data, size_t idx)
 static bool is_sec_name_dwarf(const char *name)
 {
        /* approximation, but the actual list is too long */
-       return str_has_pfx(name, ".debug_");
+       return str_has_pfx(name, ".debug_") ||
+                  strcmp(name, ".eh_frame") == 0;
 }
 
 static bool ignore_elf_section(Elf64_Shdr *hdr, const char *name)
tools/lib/bpf/linker.c +2 -2
index 78f92c392..1fb0efdbf 100644
@@ -599,7 +599,8 @@ int bpf_linker__add_buf(struct bpf_linker *linker, void *buf, size_t buf_sz,
 static bool is_dwarf_sec_name(const char *name)
 {
        /* approximation, but the actual list is too long */
-       return strncmp(name, ".debug_", sizeof(".debug_") - 1) == 0;
+       return strncmp(name, ".debug_", sizeof(".debug_") - 1) == 0 ||
+                  strcmp(name, ".eh_frame") == 0;
 }
 
 static bool is_ignored_sec(struct src_sec *sec)
--

2.54.0