Received-SPF: pass (google.com: domain of linux-crypto-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Nathan Chancellor <nathan@kernel.org>
To:     Masahiro Yamada <masahiroy@kernel.org>
Cc:     Nick Desaulniers <ndesaulniers@google.com>,
        Tom Rix <trix@redhat.com>, Nicolas Schier <nicolas@fjasle.eu>,
        Sami Tolvanen <samitolvanen@google.com>,
        Vincent Donnefort <vdonnefort@google.com>,
        linux-kbuild@vger.kernel.org, linux-kernel@vger.kernel.org,
        llvm@lists.linux.dev, patches@lists.linux.dev,
        Nathan Chancellor <nathan@kernel.org>,
        Daniel Jordan <daniel.m.jordan@oracle.com>,
        Steffen Klassert <steffen.klassert@secunet.com>,
        linux-crypto@vger.kernel.org
Subject: [PATCH v3 1/2] padata: Mark padata_work_init() as __ref
Date:   Tue, 13 Dec 2022 11:35:28 -0700
Message-Id: <20221213183529.766630-2-nathan@kernel.org>
In-Reply-To: <20221213183529.766630-1-nathan@kernel.org>
References: <20221213183529.766630-1-nathan@kernel.org>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

When building arm64 allmodconfig + ThinLTO with clang and a proposed
modpost update to account for -ffuncton-sections, the following warning
appears:

  WARNING: modpost: vmlinux.o: section mismatch in reference: padata_work_init (section: .text.padata_work_init) -> padata_mt_helper (section: .init.text)
  WARNING: modpost: vmlinux.o: section mismatch in reference: padata_work_init (section: .text.padata_work_init) -> padata_mt_helper (section: .init.text)

LLVM has optimized padata_work_init() to include the address of
padata_mt_helper() directly because it inlined the other call to
padata_work_init() with padata_parallel_worker(), meaning the remaining
uses of padata_work_init() use padata_mt_helper() as the work_fn
argument. This optimization causes modpost to complain since
padata_work_init() is not __init, whereas padata_mt_helper() is.

Since padata_work_init() is only called from __init code when
padata_mt_helper() is passed as the work_fn argument, mark
padata_work_init() as __ref, which makes it clear to modpost that this
scenario is okay.

Suggested-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Cc: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: linux-crypto@vger.kernel.org
---
 kernel/padata.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/kernel/padata.c b/kernel/padata.c
index e5819bb8bd1d..d175cc000453 100644
--- a/kernel/padata.c
+++ b/kernel/padata.c
@@ -83,8 +83,16 @@ static struct padata_work *padata_work_alloc(void)
 	return pw;
 }
 
-static void padata_work_init(struct padata_work *pw, work_func_t work_fn,
-			     void *data, int flags)
+/*
+ * This function is marked __ref because this function may be optimized in such
+ * a way that it directly refers to work_fn's address, which causes modpost to
+ * complain when work_fn is marked __init. This scenario was observed with clang
+ * LTO, where padata_work_init() was optimized to refer directly to
+ * padata_mt_helper() because the calls to padata_work_init() with other work_fn
+ * values were eliminated or inlined.
+ */
+static void __ref padata_work_init(struct padata_work *pw, work_func_t work_fn,
+				   void *data, int flags)
 {
 	if (flags & PADATA_WORK_ONSTACK)
 		INIT_WORK_ONSTACK(&pw->pw_work, work_fn);
-- 
2.39.0