Thread (9 messages) 9 messages, 2 authors, 2026-05-19

Re: [PATCH v5 3/6] selftest/bpf: Fixing powerpc JIT disassembly failure

From: bot+bpf-ci@kernel.org
Date: 2026-05-19 20:14:04
Also in: bpf, linux-kselftest, stable

quoted hunk ↗ jump to hunk
diff --git a/tools/testing/selftests/bpf/jit_disasm_helpers.c b/tools/testing/selftests/bpf/jit_disasm_helpers.c
index 364c557c5115..9e6613479145 100644
--- a/tools/testing/selftests/bpf/jit_disasm_helpers.c
+++ b/tools/testing/selftests/bpf/jit_disasm_helpers.c
@@ -170,9 +170,11 @@ int get_jited_program_text(int fd, char *text, size_t text_sz)
 	struct bpf_prog_info info = {};
 	__u32 info_len = sizeof(info);
 	__u32 jited_funcs, len, pc;
+	__u32 trunc_len = 0;
 	__u32 *func_lens = NULL;
 	FILE *text_out = NULL;
 	uint8_t *image = NULL;
+	char *triple = NULL;
 	int i, err = 0;

 	if (!llvm_initialized) {
@@ -216,9 +218,21 @@ int get_jited_program_text(int fd, char *text, size_t text_sz)
 	if (!ASSERT_OK(err, "bpf_prog_get_info_by_fd #2"))
 		goto out;

+	/*
+	 * last 8 bytes contains dummy_trampoline address in JIT
+	 * output for 64-bit and 32-bit powerpc, which can't
+	 * disassemble a to valid instruction.
+	 */
This isn't a bug, but there's a grammar error here: "disassemble a to
valid instruction" should be "disassemble to a valid instruction".
+	triple = LLVMGetDefaultTargetTriple();
+	if (triple) {
+		if (strstr(triple, "powerpc"))
+			trunc_len = 8;
Will this remove a valid instruction on 32-bit PowerPC targets?

The dummy trampoline address is stored as an unsigned long, which
occupies 8 bytes on 64-bit PowerPC, but only 4 bytes on 32-bit
PowerPC.

Because the long branch stub ends with a 4-byte bctr instruction
followed by the dummy address, unconditionally truncating 8 bytes on
32-bit PowerPC would remove both the 4-byte dummy address and the
valid 4-byte bctr instruction from the disassembly.

Does the truncation length need to be determined dynamically based on
whether the target is 32-bit or 64-bit?

(Note: This concern was raised by sashiko-bot@kernel.org in v4:
https://lore.kernel.org/bpf/20260517183846.CD7C2C2BCB0@smtp.kernel.org (local))
+		LLVMDisposeMessage(triple);
+	}
+
 	for (pc = 0, i = 0; i < jited_funcs; ++i) {
 		fprintf(text_out, "func #%d:\n", i);
-		disasm_one_func(text_out, image + pc, func_lens[i]);
+		disasm_one_func(text_out, image + pc, func_lens[i] - trunc_len);
 		fprintf(text_out, "\n");
 		pc += func_lens[i];
 	}
---
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/26121409898
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help