[PATCH v1 1/1] serial: 8250_ce4100: Fix CONFIG_SERIAL_8250=n build

[PATCH v1 1/1] serial: 8250_ce4100: Fix CONFIG_SERIAL_8250=n build

[Andy Shevchenko]

On i386, when

  CONFIG_X86_INTEL_CE=y
  # CONFIG_SERIAL_8250 is not set

will try to compile the driver and use the stub simultaneously.
This breaks the build. Fix it by making sure that the driver
compiles only when CONFIG_SERIAL_8250 is also enabled.

Fixes: acc902de05b2 ("serial: 8250: Move CE4100 quirks to a module under 8250 driver")
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Closes: https://lore.kernel.org/r/cdf4ee46-7bf8-4379-9245-fed9db72e7e8@infradead.org
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/tty/serial/8250/Makefile | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/tty/serial/8250/Makefile b/drivers/tty/serial/8250/Makefile
index e61dc3f4ca50..1d6633bbcd70 100644
--- a/drivers/tty/serial/8250/Makefile
+++ b/drivers/tty/serial/8250/Makefile
@@ -24,7 +24,9 @@ obj-$(CONFIG_SERIAL_8250_ASPEED_VUART)	+= 8250_aspeed_vuart.o
 obj-$(CONFIG_SERIAL_8250_BCM2835AUX)	+= 8250_bcm2835aux.o
 obj-$(CONFIG_SERIAL_8250_BCM7271)	+= 8250_bcm7271.o
 obj-$(CONFIG_SERIAL_8250_BOCA)		+= 8250_boca.o
+ifneq ($(CONFIG_SERIAL_8250),)
 obj-$(CONFIG_X86_INTEL_CE)		+= 8250_ce4100.o
+endif
 obj-$(CONFIG_SERIAL_8250_DFL)		+= 8250_dfl.o
 obj-$(CONFIG_SERIAL_8250_DW)		+= 8250_dw.o
 obj-$(CONFIG_SERIAL_8250_EM)		+= 8250_em.o
-- 
2.47.2

Re: [PATCH v1 1/1] serial: 8250_ce4100: Fix CONFIG_SERIAL_8250=n build

[Jiri Slaby]

On 16. 07. 25, 11:44, Andy Shevchenko wrote:
> On i386, when
> 
>    CONFIG_X86_INTEL_CE=y
>    # CONFIG_SERIAL_8250 is not set
> 
> will try to compile the driver and use the stub simultaneously.
> This breaks the build. Fix it by making sure that the driver
> compiles only when CONFIG_SERIAL_8250 is also enabled.
> 
> Fixes: acc902de05b2 ("serial: 8250: Move CE4100 quirks to a module under 8250 driver")
> Reported-by: Randy Dunlap <rdunlap@infradead.org>
> Closes: https://lore.kernel.org/r/cdf4ee46-7bf8-4379-9245-fed9db72e7e8@infradead.org
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> ---
>   drivers/tty/serial/8250/Makefile | 2 ++
>   1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/tty/serial/8250/Makefile b/drivers/tty/serial/8250/Makefile
> index e61dc3f4ca50..1d6633bbcd70 100644
> --- a/drivers/tty/serial/8250/Makefile
> +++ b/drivers/tty/serial/8250/Makefile
> @@ -24,7 +24,9 @@ obj-$(CONFIG_SERIAL_8250_ASPEED_VUART)	+= 8250_aspeed_vuart.o
>   obj-$(CONFIG_SERIAL_8250_BCM2835AUX)	+= 8250_bcm2835aux.o
>   obj-$(CONFIG_SERIAL_8250_BCM7271)	+= 8250_bcm7271.o
>   obj-$(CONFIG_SERIAL_8250_BOCA)		+= 8250_boca.o
> +ifneq ($(CONFIG_SERIAL_8250),)

Why not ifdef CONFIG_SERIAL_8250 then?

Also, what happens if 8250=m and X86_INTEL_CE=y?

>   obj-$(CONFIG_X86_INTEL_CE)		+= 8250_ce4100.o
> +endif
>   obj-$(CONFIG_SERIAL_8250_DFL)		+= 8250_dfl.o
>   obj-$(CONFIG_SERIAL_8250_DW)		+= 8250_dw.o
>   obj-$(CONFIG_SERIAL_8250_EM)		+= 8250_em.o


-- 
js
suse labs

Re: [PATCH v1 1/1] serial: 8250_ce4100: Fix CONFIG_SERIAL_8250=n build

[Andy Shevchenko]

On Wed, Jul 16, 2025 at 12:14:23PM +0200, Jiri Slaby wrote:
> On 16. 07. 25, 11:44, Andy Shevchenko wrote:
> > On i386, when
> > 
> >    CONFIG_X86_INTEL_CE=y
> >    # CONFIG_SERIAL_8250 is not set
> > 
> > will try to compile the driver and use the stub simultaneously.
> > This breaks the build. Fix it by making sure that the driver
> > compiles only when CONFIG_SERIAL_8250 is also enabled.

...

> > +ifneq ($(CONFIG_SERIAL_8250),)
> 
> Why not ifdef CONFIG_SERIAL_8250 then?

$ git grep -n '^ifn\?eq .*CONFIG_' | wc -l
427
$ git grep -n '^ifdef CONFIG_' | wc -l
431

Is there a preference in serial drivers?

> Also, what happens if 8250=m and X86_INTEL_CE=y?

So, in such a case if somebody wants to use UART, it will go crazy due to
missed workaround applied. BUT, this is preexisted issue and not related
(directly) to this fix. Perhaps we can combine both with

ifeq ($(CONFIG_SERIAL_8250),y)

and add two Fixes tags.

> >   obj-$(CONFIG_X86_INTEL_CE)		+= 8250_ce4100.o
> > +endif

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH v1 1/1] serial: 8250_ce4100: Fix CONFIG_SERIAL_8250=n build

[Jiri Slaby]

On 16. 07. 25, 12:28, Andy Shevchenko wrote:
> On Wed, Jul 16, 2025 at 12:14:23PM +0200, Jiri Slaby wrote:
>> On 16. 07. 25, 11:44, Andy Shevchenko wrote:
>>> On i386, when
>>>
>>>     CONFIG_X86_INTEL_CE=y
>>>     # CONFIG_SERIAL_8250 is not set
>>>
>>> will try to compile the driver and use the stub simultaneously.
>>> This breaks the build. Fix it by making sure that the driver
>>> compiles only when CONFIG_SERIAL_8250 is also enabled.
> 
> ...
> 
>>> +ifneq ($(CONFIG_SERIAL_8250),)
>>
>> Why not ifdef CONFIG_SERIAL_8250 then?
> 
> $ git grep -n '^ifn\?eq .*CONFIG_' | wc -l
> 427
> $ git grep -n '^ifdef CONFIG_' | wc -l
> 431
> 
> Is there a preference in serial drivers?

The simpler to read/understand -- the latter.

>> Also, what happens if 8250=m and X86_INTEL_CE=y?
> 
> So, in such a case if somebody wants to use UART, it will go crazy due to
> missed workaround applied. BUT, this is preexisted issue and not related
> (directly) to this fix. Perhaps we can combine both with
> 
> ifeq ($(CONFIG_SERIAL_8250),y)
> 
> and add two Fixes tags.

+1 makes sense to me. Or introduce a Kconfig entry to handle the deps...

thanks,
-- 
js
suse labs