BPF List
 help / color / mirror / Atom feed
From: Yonghong Song <yonghong.song@linux.dev>
To: Alan Maguire <alan.maguire@oracle.com>,
	Arnaldo Carvalho de Melo <arnaldo.melo@gmail.com>,
	dwarves@vger.kernel.org
Cc: Alexei Starovoitov <ast@kernel.org>,
	Andrii Nakryiko <andrii@kernel.org>,
	bpf@vger.kernel.org, kernel-team@fb.com
Subject: [PATCH dwarves v9 5/6] btf_encoder: Avoid comparing inlined and out-of-line function prototypes
Date: Tue, 23 Jun 2026 22:26:19 -0700	[thread overview]
Message-ID: <20260624052619.3143003-1-yonghong.song@linux.dev> (raw)
In-Reply-To: <20260624052553.3139112-1-yonghong.song@linux.dev>

When true_signature is enabled, the same function can appear as two kinds
of saved states: an inlined (abstract) instance that carries the original
source signature with no address, and an out-of-line instance that carries
the reconstructed true signature. These two legitimately differ, so
comparing them in saved_functions_combine() wrongly flags the function as
having an inconsistent prototype and drops it from BTF.

Record which kind a state is by adding an 'inlined' bit to
btf_encoder_func_state, set from function__inlined() in
btf_encoder__save_func().

Use it in two places:
 - saved_functions_cmp() breaks ties by 'inlined' so that out-of-line
   (real) instances sort before inlined (abstract) ones. Since
   saved_functions_combine() compares every state in a group against the
   first (anchor) state, keeping a real instance as the anchor ensures real
   instances are still compared against each other for prototype
   consistency.
 - saved_functions_combine() only runs the prototype-consistency check when
   both states are of the same kind (a->inlined == b->inlined). The true
   signature always comes from the out-of-line instance, which is the one
   picked by btf_encoder__select_canonical_state().

Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
---
 btf_encoder.c | 27 +++++++++++++++++++++++++--
 1 file changed, 25 insertions(+), 2 deletions(-)

diff --git a/btf_encoder.c b/btf_encoder.c
index 38455a4..81bfda3 100644
--- a/btf_encoder.c
+++ b/btf_encoder.c
@@ -96,6 +96,7 @@ struct btf_encoder_func_state {
 	uint8_t uncertain_parm_loc:1;
 	uint8_t reordered_parm:1;
 	uint8_t ambiguous_addr:1;
+	uint8_t inlined:1;
 	int ret_type_id;
 	struct btf_encoder_func_parm *parms;
 	struct btf_encoder_func_annot *annots;
@@ -1295,6 +1296,7 @@ static int32_t btf_encoder__save_func(struct btf_encoder *encoder, struct functi
 	state->optimized_parms = ftype->optimized_parms;
 	state->uncertain_parm_loc = ftype->uncertain_parm_loc;
 	state->reordered_parm = ftype->reordered_parm;
+	state->inlined = function__inlined(fn);
 	ftype__for_each_parameter(ftype, param) {
 		const char *name;
 		char *final_name = NULL;
@@ -1489,8 +1491,19 @@ static int saved_functions_cmp(const void *_a, const void *_b)
 {
 	const struct btf_encoder_func_state *a = _a;
 	const struct btf_encoder_func_state *b = _b;
+	int ret;
 
-	return elf_function__name_cmp(a->elf, b->elf);
+	ret = elf_function__name_cmp(a->elf, b->elf);
+	if (ret)
+		return ret;
+
+	/* For the same function, sort the out-of-line (real) instances before
+	 * the inlined (abstract) ones. saved_functions_combine() compares every
+	 * state in a group against the first (anchor) one, so keeping a real
+	 * instance as the anchor ensures real instances are still compared
+	 * against each other for prototype consistency.
+	 */
+	return a->inlined - b->inlined;
 }
 
 static int saved_functions_combine(struct btf_encoder *encoder,
@@ -1507,7 +1520,17 @@ static int saved_functions_combine(struct btf_encoder *encoder,
 	inconsistent = a->inconsistent_proto | b->inconsistent_proto;
 	uncertain_parm_loc = a->uncertain_parm_loc | b->uncertain_parm_loc;
 	reordered_parm = a->reordered_parm | b->reordered_parm;
-	if (!unexpected && !inconsistent && !reordered_parm && !funcs__match(encoder, a, b))
+
+	/* Only compare prototypes that are of the same kind. An inlined
+	 * (abstract) instance carries the original signature while the
+	 * out-of-line instance carries the true signature, so the two will
+	 * legitimately differ; comparing them would wrongly flag the function
+	 * as having an inconsistent prototype and drop it from BTF. The true
+	 * signature always comes from the out-of-line (real) instance, which is
+	 * the one selected by btf_encoder__select_canonical_state().
+	 */
+	if (a->inlined == b->inlined &&
+	    !unexpected && !inconsistent && !reordered_parm && !funcs__match(encoder, a, b))
 		inconsistent = 1;
 	a->optimized_parms = b->optimized_parms = optimized;
 	a->unexpected_reg = b->unexpected_reg = unexpected;
-- 
2.53.0-Meta


  parent reply	other threads:[~2026-06-24  5:26 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-06-24  5:25 [PATCH dwarves v9 0/6] pahole: Encode true signatures in kernel BTF Yonghong Song
2026-06-24  5:25 ` [PATCH dwarves v9 1/6] dwarf_loader: Detect aggregate ABI register usage and signature changes Yonghong Song
2026-06-24  5:26 ` [PATCH dwarves v9 2/6] dwarf_loader: Collect per-parameter information Yonghong Song
2026-06-24  5:26 ` [PATCH dwarves v9 3/6] dwarf_loader: Analyze per-parameter information for true signatures Yonghong Song
2026-06-24  5:26 ` [PATCH dwarves v9 4/6] btf_encoder: Emit true function signatures Yonghong Song
2026-06-24  5:26 ` Yonghong Song [this message]
2026-06-24  5:26 ` [PATCH dwarves v9 6/6] tests: Add BTF true_signature encoding tests Yonghong Song

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260624052619.3143003-1-yonghong.song@linux.dev \
    --to=yonghong.song@linux.dev \
    --cc=alan.maguire@oracle.com \
    --cc=andrii@kernel.org \
    --cc=arnaldo.melo@gmail.com \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=dwarves@vger.kernel.org \
    --cc=kernel-team@fb.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox