diff mbox

[RESEND] scsi: lpfc: fix linking against modular NVMe support

Message ID 20170721205401.3374331-1-arnd@arndb.de (mailing list archive)
State Accepted, archived
Headers show

Commit Message

Arnd Bergmann July 21, 2017, 8:53 p.m. UTC
When LPFC is built-in but NVMe is a loadable module, we fail to
link the kernel:

drivers/scsi/built-in.o: In function `lpfc_nvme_create_localport':
(.text+0x156a82): undefined reference to `nvme_fc_register_localport'
drivers/scsi/built-in.o: In function `lpfc_nvme_destroy_localport':
(.text+0x156eaa): undefined reference to `nvme_fc_unregister_remoteport'

We can avoid this either by forcing lpfc to be a module, or by disabling
NVMe support in this case. This implements the former.

Fixes: 7d7080335f8d ("scsi: lpfc: Finalize Kconfig options for nvme")
Cc: stable@vger.kernel.org
Link: https://patchwork.kernel.org/patch/9636569/
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
I originally sent the patch in March 2013, the day after the broken
patch made it into Linux-next. Unfortunately there was some
misunderstanding about it (James Smart thought the broken patch
was the correct version), and my fix never got merged.
---
 drivers/scsi/Kconfig | 2 ++
 1 file changed, 2 insertions(+)

Comments

Johannes Thumshirn July 26, 2017, 9:08 a.m. UTC | #1
On Fri, Jul 21, 2017 at 10:53:46PM +0200, Arnd Bergmann wrote:
> I originally sent the patch in March 2013, the day after the broken
                                       ^ 2017?

> +	depends on NVME_TARGET_FC || NVME_TARGET_FC=n
> +	depends on NVME_FC || NVME_FC=n

OK Kconfig is black magic, but X || X=n ? I have no doubt you do the right
thing, I just don't understand it.

Thanks,
	Johannes
Arnd Bergmann July 26, 2017, 9:16 a.m. UTC | #2
On Wed, Jul 26, 2017 at 11:08 AM, Johannes Thumshirn <jthumshirn@suse.de> wrote:
> On Fri, Jul 21, 2017 at 10:53:46PM +0200, Arnd Bergmann wrote:
>> I originally sent the patch in March 2013, the day after the broken
>                                        ^ 2017?

Oops. Let me resend that.

>> +     depends on NVME_TARGET_FC || NVME_TARGET_FC=n
>> +     depends on NVME_FC || NVME_FC=n
>
> OK Kconfig is black magic, but X || X=n ? I have no doubt you do the right
> thing, I just don't understand it.

We use this construct in many places these days. A few years ago I added
a comment whenever I added an instance, but most others don't so I treat
it as an idiom now.

"depends on X" means that if X=m, the current symbol can only be "m" or
"n", while "X=y" means it can have any value.

So for "depends on X || X=n", the current symbol cannot be "y" when "X=m",
but otherwise can have any of the three possibilities. This is exactly what
we need here.

       Arnd
Arnd Bergmann July 26, 2017, 9:17 a.m. UTC | #3
On Wed, Jul 26, 2017 at 11:16 AM, Arnd Bergmann <arnd@arndb.de> wrote:
> On Wed, Jul 26, 2017 at 11:08 AM, Johannes Thumshirn <jthumshirn@suse.de> wrote:
>> On Fri, Jul 21, 2017 at 10:53:46PM +0200, Arnd Bergmann wrote:
>>> I originally sent the patch in March 2013, the day after the broken
>>                                        ^ 2017?
>
> Oops. Let me resend that.

Ah, it was only below the "---" line, so I'll skip the resending
unless you want me
to add a comment in the Kconfig file.

       Arnd
Johannes Thumshirn July 26, 2017, 9:18 a.m. UTC | #4
On Wed, Jul 26, 2017 at 11:16:10AM +0200, Arnd Bergmann wrote:
> We use this construct in many places these days. A few years ago I added
> a comment whenever I added an instance, but most others don't so I treat
> it as an idiom now.
> 
> "depends on X" means that if X=m, the current symbol can only be "m" or
> "n", while "X=y" means it can have any value.
> 
> So for "depends on X || X=n", the current symbol cannot be "y" when "X=m",
> but otherwise can have any of the three possibilities. This is exactly what
> we need here.

Obvious o.O
Johannes Thumshirn July 26, 2017, 9:19 a.m. UTC | #5
On Wed, Jul 26, 2017 at 11:17:09AM +0200, Arnd Bergmann wrote:
> Ah, it was only below the "---" line, so I'll skip the resending
> unless you want me
> to add a comment in the Kconfig file.

Nah no need, really.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
diff mbox

Patch

diff --git a/drivers/scsi/Kconfig b/drivers/scsi/Kconfig
index d384f4f86c26..f4538d7a3016 100644
--- a/drivers/scsi/Kconfig
+++ b/drivers/scsi/Kconfig
@@ -1230,6 +1230,8 @@  config SCSI_LPFC
 	tristate "Emulex LightPulse Fibre Channel Support"
 	depends on PCI && SCSI
 	depends on SCSI_FC_ATTRS
+	depends on NVME_TARGET_FC || NVME_TARGET_FC=n
+	depends on NVME_FC || NVME_FC=n
 	select CRC_T10DIF
 	---help---
           This lpfc driver supports the Emulex LightPulse