All of lore.kernel.org
 help / color / mirror / Atom feed
From: Alan Maguire <almagui@kernel.org>
To: dwarves@vger.kernel.org
Cc: Alan Maguire <almagui@kernel.org>
Subject: [PATCH dwarves 4/6] tests: Expect no DWARF loader threads in reproducible_build.sh when CUs are merged
Date: Mon, 16 Nov 2026 11:20:41 +0000	[thread overview]
Message-ID: <20261116112043.3831978-5-almagui@kernel.org> (raw)
In-Reply-To: <20261116112043.3831978-1-almagui@kernel.org>

reproducible_build.sh checks that "pahole -jN" starts N threads.  When
pahole merges all CUs into one, as it does for LTO kernels, DWARF is
loaded serially and no worker threads are started, so the test fails
on a clang ThinLTO vmlinux, as seen in the CI clang-lto-thin x86_64
job:

  ERROR: pahole asked to start 1 encoding threads, started 0

Add a dwarf_cus_merged() helper to test_lib.sh mirroring
cus__merging_cu(): the kernel's LINUX_ELFNOTE_LTO_INFO note is set, or
the DWARF uses DW_FORM_ref_addr.  When CUs are merged, expect zero
worker threads; the serial vs parallel BTF comparison is still done.

The helper walks the .notes section itself using the 4-byte layout of
the kernel's ELFNOTE() notes rather than relying on readelf -n.  On
kernels whose linker script does not discard .note.gnu.property (it is
discarded since kernel commit f7ba52f302fd ("vmlinux.lds.h: Discard
.note.gnu.property section")), .notes picks up the 8-byte alignment of
that section and readelf -n misparses the 4-byte aligned notes.

Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Alan Maguire <almagui@kernel.org>
---
 tests/reproducible_build.sh | 17 +++++++++++++++--
 tests/test_lib.sh           | 36 ++++++++++++++++++++++++++++++++++++
 2 files changed, 51 insertions(+), 2 deletions(-)

diff --git a/tests/reproducible_build.sh b/tests/reproducible_build.sh
index 532b320..8b7b8eb 100755
--- a/tests/reproducible_build.sh
+++ b/tests/reproducible_build.sh
@@ -27,6 +27,16 @@ export PAHOLE_LANG_EXCLUDE=rust
 pahole --btf_features=default --btf_encode_detached=$outdir/vmlinux.btf.serial $vmlinux
 bpftool btf dump file $outdir/vmlinux.btf.serial > $outdir/bpftool.output.vmlinux.btf.serial
 
+# When CUs are merged (e.g. LTO kernels) pahole loads DWARF serially and
+# starts no loader threads whatever -j says; the encoding must still be
+# reproducible.
+if dwarf_cus_merged $vmlinux ; then
+	verbose_log "CUs are merged (LTO?); DWARF loading is serial, expecting no worker threads"
+	merged_cus=1
+else
+	merged_cus=0
+fi
+
 nr_proc=$(getconf _NPROCESSORS_ONLN)
 # Clamp to 1 so single-core machines don't produce -j0
 half_proc=$(( nr_proc / 2 > 0 ? nr_proc / 2 : 1 ))
@@ -43,8 +53,11 @@ for threads in $thread_list ; do
 	nr_threads_started=$(ls /proc/$pahole/task 2>/dev/null | wc -l)
 	nr_threads_started=$((nr_threads_started - 1)) # subtract main thread
 
-	if [ $threads != $nr_threads_started ] ; then
-		error_log "ERROR: pahole asked to start $threads encoding threads, started $nr_threads_started"
+	expected_threads=$threads
+	[ $merged_cus = 1 ] && expected_threads=0
+
+	if [ $expected_threads != $nr_threads_started ] ; then
+		error_log "ERROR: pahole asked to start $threads encoding threads, expected $expected_threads, started $nr_threads_started"
 		test_fail
 	fi
 
diff --git a/tests/test_lib.sh b/tests/test_lib.sh
index e2d0b0c..bd3784e 100755
--- a/tests/test_lib.sh
+++ b/tests/test_lib.sh
@@ -464,6 +464,42 @@ get_perf_with_debug()
 	fi
 }
 
+# Return success if pahole's DWARF loader will merge all CUs of $1 into
+# one and load them serially rather than with -j worker threads.  This
+# mirrors cus__merging_cu() in dwarf_loader.c: the kernel's
+# LINUX_ELFNOTE_LTO_INFO note says it was built with LTO, or the DWARF
+# uses DW_FORM_ref_addr (references across CUs, as with clang LTO).
+#
+# readelf -n gives up on the kernel's .notes section (4-byte aligned
+# notes in an 8-byte aligned section), so walk the notes directly.
+dwarf_cus_merged()
+{
+	local obj=$1 notes
+
+	notes=$(mktemp)
+	if objcopy -O binary --only-section=.notes "$obj" "$notes" 2>/dev/null &&
+	   od -An -v -tu4 -w4 "$notes" | awk '
+		# namesz, descsz, type, name (padded to 4), desc (padded to 4)
+		{ w[n++] = $1 }
+		END {
+			for (i = 0; i + 3 <= n; ) {
+				namesz = w[i]; descsz = w[i+1]; type = w[i+2]
+				i += 3 + int((namesz + 3) / 4)
+				# LINUX_ELFNOTE_LTO_INFO, owner "Linux"
+				if (type == 257 && namesz == 6 && descsz == 4 && w[i] != 0)
+					exit 0
+				i += int((descsz + 3) / 4)
+			}
+			exit 1
+		}'; then
+		rm -f "$notes"
+		return 0
+	fi
+	rm -f "$notes"
+
+	readelf --debug-dump=abbrev "$obj" 2>/dev/null | grep -q DW_FORM_ref_addr
+}
+
 make_tmpdir()
 {
 	# Ensure master test directory exists
-- 
2.56.0


  parent reply	other threads:[~2026-11-16 11:20 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-16 11:20 [PATCH dwarves 0/6] Add support for LTO kernel builds, other CI fixes Alan Maguire
2026-11-16 11:20 ` [PATCH dwarves 1/6] CI: Make kernel source fetch tolerant of git host failures Alan Maguire
2026-11-16 11:20 ` [PATCH dwarves 2/6] tests: Match BTF and DWARF functions with set operations in btf_functions.sh Alan Maguire
2026-11-16 11:20 ` [PATCH dwarves 3/6] CI: Scope workflow concurrency groups to the branch for push runs Alan Maguire
2026-11-16 11:20 ` Alan Maguire [this message]
2026-11-16 11:20 ` [PATCH dwarves 5/6] CI: Fix ondemand workflow inputs and LLVM version in function comparison Alan Maguire
2026-11-16 11:20 ` [PATCH dwarves 6/6] CI: Add clang ThinLTO kernel build to test matrix Alan Maguire

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=20261116112043.3831978-5-almagui@kernel.org \
    --to=almagui@kernel.org \
    --cc=dwarves@vger.kernel.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.