[PATCH] serial: 8250: Fix section mismatch with CONFIG_MODULES=n

[PATCH] serial: 8250: Fix section mismatch with CONFIG_MODULES=n

[Danish Khateeb]

Without CONFIG_MODULES, __init_or_module means __init, so
serial8250_pnp_exit() and serial8250_hub6_exit() end up in .init.text.
They are also called from serial8250_exit(), which is __exit, and
modpost warns:

  WARNING: modpost: vmlinux: section mismatch in reference: serial8250_exit+0x2a (section: .exit.text) -> serial8250_pnp_exit (section: .init.text)

The exit code never runs in such a kernel, so nothing breaks at runtime.
But a function that is called from both __init and __exit code can't
carry either annotation, so drop it.

Fixes: a300273c458c ("serial: 8250: hub6: Add cleanup code")
Fixes: 4cf187f9dea9 ("serial: 8250: pnp: Annotate init and exit functions for conditional discarding")
Assisted-by: LLM
Signed-off-by: Danish Khateeb <danishkhateeb03@gmail.com>
---

Notes:
    Tested on tty-next bf961847813d:
    - x86_64 defconfig with MODULES=n warns for serial8250_pnp_exit(), and
      i386 defconfig with MODULES=n, ISA and SERIAL_8250_HUB6 also warns for
      serial8250_hub6_exit(). Neither warns before a300273c458c. With this
      patch both build without warnings, and so does x86_64 defconfig.
    - Both MODULES=n kernels still register ttyS0 through 8250_pnp in QEMU.

 drivers/tty/serial/8250/8250_hub6.c | 2 +-
 drivers/tty/serial/8250/8250_pnp.c  | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/tty/serial/8250/8250_hub6.c b/drivers/tty/serial/8250/8250_hub6.c
index 4d4c54c951cb..848d9723ac56 100644
--- a/drivers/tty/serial/8250/8250_hub6.c
+++ b/drivers/tty/serial/8250/8250_hub6.c
@@ -54,7 +54,7 @@ int __init serial8250_hub6_init(void)
 	return platform_device_register(&hub6_device);
 }
 
-void __init_or_module serial8250_hub6_exit(void)
+void serial8250_hub6_exit(void)
 {
 	platform_device_unregister(&hub6_device);
 }
diff --git a/drivers/tty/serial/8250/8250_pnp.c b/drivers/tty/serial/8250/8250_pnp.c
index 7decf62400ab..a6257cd6e246 100644
--- a/drivers/tty/serial/8250/8250_pnp.c
+++ b/drivers/tty/serial/8250/8250_pnp.c
@@ -531,7 +531,7 @@ int __init serial8250_pnp_init(void)
 	return pnp_register_driver(&serial_pnp_driver);
 }
 
-void __init_or_module serial8250_pnp_exit(void)
+void serial8250_pnp_exit(void)
 {
 	pnp_unregister_driver(&serial_pnp_driver);
 }

base-commit: bf961847813d9edcb5a9c089714069428d141fa9
-- 
2.55.0

Re: [PATCH] serial: 8250: Fix section mismatch with CONFIG_MODULES=n

[Uwe Kleine-König]

[-- Attachment #1: Type: text/plain, Size: 884 bytes --]

Hello Danish,

On Sat, Sep 26, 2026 at 11:35:06AM -0500, Danish Khateeb wrote:
> Without CONFIG_MODULES, __init_or_module means __init, so
> serial8250_pnp_exit() and serial8250_hub6_exit() end up in .init.text.
> They are also called from serial8250_exit(), which is __exit, and
> modpost warns:
> 
>   WARNING: modpost: vmlinux: section mismatch in reference: serial8250_exit+0x2a (section: .exit.text) -> serial8250_pnp_exit (section: .init.text)

Isn't that a false positive because with CONFIG_MODULES=n .exit is
discarded and so the section mismatch doesn't matter?

> The exit code never runs in such a kernel, so nothing breaks at runtime.

Ah, you found that, too.

> But a function that is called from both __init and __exit code can't
> carry either annotation, so drop it.

I thought __init_or_module was the right annotation ...

Best regards
Uwe

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH] serial: 8250: Fix section mismatch with CONFIG_MODULES=n

[Danish Khateeb]

Hello Uwe,

On Sat, Sep 26, 2026 at 07:40:03PM +0200, Uwe Kleine-König wrote:
> Isn't that a false positive because with CONFIG_MODULES=n .exit is
> discarded and so the section mismatch doesn't matter?

Yes, it's harmless at runtime, but modpost still warns on every
CONFIG_MODULES=n build with SERIAL_8250_PNP (e.g. kunit.py --alltests),
and the build fails with CONFIG_SECTION_MISMATCH_WARN_ONLY=n.

> I thought __init_or_module was the right annotation ...

It is, modpost just doesn't know that. Without it the two helpers stay in
memory after boot, 16 bytes each on x86_64. The alternative is teaching
modpost to skip exit -> init references in vmlinux when CONFIG_MODULES=n.

Thanks
Danish