[PATCH 6.7 225/713] tools/resolve_btfids: Fix cross-compilation to non-host endianness

From: Sasha Levin
Date: Sun Mar 24 2024 - 21:22:41 EST


From: Viktor Malik <vmalik@xxxxxxxxxx>

[ Upstream commit 903fad4394666bc23975c93fb58f137ce64b5192 ]

The .BTF_ids section is pre-filled with zeroed BTF ID entries during the
build and afterwards patched by resolve_btfids with correct values.
Since resolve_btfids always writes in host-native endianness, it relies
on libelf to do the translation when the target ELF is cross-compiled to
a different endianness (this was introduced in commit 61e8aeda9398
("bpf: Fix libelf endian handling in resolv_btfids")).

Unfortunately, the translation will corrupt the flags fields of SET8
entries because these were written during vmlinux compilation and are in
the correct endianness already. This will lead to numerous selftests
failures such as:

$ sudo ./test_verifier 502 502
#502/p sleepable fentry accept FAIL
Failed to load prog 'Invalid argument'!
bpf_fentry_test1 is not sleepable
verification time 34 usec
stack depth 0
processed 0 insns (limit 1000000) max_states_per_insn 0 total_states 0 peak_states 0 mark_read 0
Summary: 0 PASSED, 0 SKIPPED, 1 FAILED

Since it's not possible to instruct libelf to translate just certain
values, let's manually bswap the flags (both global and entry flags) in
resolve_btfids when needed, so that libelf then translates everything
correctly.

Fixes: ef2c6f370a63 ("tools/resolve_btfids: Add support for 8-byte BTF sets")
Signed-off-by: Viktor Malik <vmalik@xxxxxxxxxx>
Signed-off-by: Andrii Nakryiko <andrii@xxxxxxxxxx>
Link: https://lore.kernel.org/bpf/7b6bff690919555574ce0f13d2a5996cacf7bf69.1707223196.git.vmalik@xxxxxxxxxx
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
tools/bpf/resolve_btfids/main.c | 35 +++++++++++++++++++++++++++++++++
1 file changed, 35 insertions(+)

diff --git a/tools/bpf/resolve_btfids/main.c b/tools/bpf/resolve_btfids/main.c
index 32634f00abba4..d9520cb826b31 100644
--- a/tools/bpf/resolve_btfids/main.c
+++ b/tools/bpf/resolve_btfids/main.c
@@ -90,6 +90,14 @@

#define ADDR_CNT 100

+#if __BYTE_ORDER == __LITTLE_ENDIAN
+# define ELFDATANATIVE ELFDATA2LSB
+#elif __BYTE_ORDER == __BIG_ENDIAN
+# define ELFDATANATIVE ELFDATA2MSB
+#else
+# error "Unknown machine endianness!"
+#endif
+
struct btf_id {
struct rb_node rb_node;
char *name;
@@ -117,6 +125,7 @@ struct object {
int idlist_shndx;
size_t strtabidx;
unsigned long idlist_addr;
+ int encoding;
} efile;

struct rb_root sets;
@@ -320,6 +329,7 @@ static int elf_collect(struct object *obj)
{
Elf_Scn *scn = NULL;
size_t shdrstrndx;
+ GElf_Ehdr ehdr;
int idx = 0;
Elf *elf;
int fd;
@@ -351,6 +361,13 @@ static int elf_collect(struct object *obj)
return -1;
}

+ if (gelf_getehdr(obj->efile.elf, &ehdr) == NULL) {
+ pr_err("FAILED cannot get ELF header: %s\n",
+ elf_errmsg(-1));
+ return -1;
+ }
+ obj->efile.encoding = ehdr.e_ident[EI_DATA];
+
/*
* Scan all the elf sections and look for save data
* from .BTF_ids section and symbols.
@@ -681,6 +698,24 @@ static int sets_patch(struct object *obj)
*/
BUILD_BUG_ON(set8->pairs != &set8->pairs[0].id);
qsort(set8->pairs, set8->cnt, sizeof(set8->pairs[0]), cmp_id);
+
+ /*
+ * When ELF endianness does not match endianness of the
+ * host, libelf will do the translation when updating
+ * the ELF. This, however, corrupts SET8 flags which are
+ * already in the target endianness. So, let's bswap
+ * them to the host endianness and libelf will then
+ * correctly translate everything.
+ */
+ if (obj->efile.encoding != ELFDATANATIVE) {
+ int i;
+
+ set8->flags = bswap_32(set8->flags);
+ for (i = 0; i < set8->cnt; i++) {
+ set8->pairs[i].flags =
+ bswap_32(set8->pairs[i].flags);
+ }
+ }
}

pr_debug("sorting addr %5lu: cnt %6d [%s]\n",
--
2.43.0