Re: [PATCH] drm: rcar-du: fix linker undefined references

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 4/23/21 2:56 PM, Randy Dunlap wrote:
On 4/23/21 2:46 PM, Laurent Pinchart wrote:
Hi Randy,

Thank you for the patch.

On Fri, Apr 23, 2021 at 02:37:27PM -0700, Randy Dunlap wrote:
When DRM_RCAR_DU=y and DRM_RCAR_LVDS=m, there are several build errors
as reported by 'kernel test robot'. These can be corrected by changing
"imply" to "select".

In looking at this, the same problem (build errors) happens when
DRM_RCAR_DU=y and DRM_RCAR_CMM=m, so again change the "imply" to
"select" here as well.

These 2 changes fix the following 8 build/link errors:

aarch64-linux-ld: drivers/gpu/drm/rcar-du/rcar_du_crtc.o: in function `rcar_du_crtc_atomic_enable': rcar_du_crtc.c:(.text+0x1be8): undefined reference to `rcar_lvds_clk_enable' aarch64-linux-ld: drivers/gpu/drm/rcar-du/rcar_du_crtc.o: in function `rcar_du_crtc_atomic_disable': rcar_du_crtc.c:(.text+0x2438): undefined reference to `rcar_lvds_clk_disable' aarch64-linux-ld: drivers/gpu/drm/rcar-du/rcar_du_drv.o: in function `rcar_du_init': rcar_du_drv.c:(.init.text+0x14): undefined reference to `rcar_du_of_init' aarch64-linux-ld: drivers/gpu/drm/rcar-du/rcar_du_encoder.o: in function `rcar_du_encoder_init': rcar_du_encoder.c:(.text+0x1d4): undefined reference to `rcar_lvds_dual_link'

aarch64-linux-ld: drivers/gpu/drm/rcar-du/rcar_du_crtc.o: in function `rcar_du_cmm_setup':
rcar_du_crtc.c:(.text+0x380): undefined reference to `rcar_cmm_setup'
aarch64-linux-ld: drivers/gpu/drm/rcar-du/rcar_du_crtc.o: in function `rcar_du_crtc_atomic_enable':
rcar_du_crtc.c:(.text+0x1c08): undefined reference to `rcar_cmm_enable'
aarch64-linux-ld: drivers/gpu/drm/rcar-du/rcar_du_crtc.o: in function `rcar_du_crtc_atomic_disable':
rcar_du_crtc.c:(.text+0x231c): undefined reference to `rcar_cmm_disable'
aarch64-linux-ld: drivers/gpu/drm/rcar-du/rcar_du_kms.o: in function `rcar_du_modeset_init':
rcar_du_kms.c:(.text+0xd08): undefined reference to `rcar_cmm_init'

Fixes: e08e934d6c28 ("drm: rcar-du: Add support for CMM")
Fixes: 02f2b30032c1 ("drm: rcar-du: lvds: Add API to enable/disable clock output")
Signed-off-by: Randy Dunlap <rdunlap@xxxxxxxxxxxxx>
Reported-by: kernel test robot <lkp@xxxxxxxxx>
Cc: Masahiro Yamada <masahiroy@xxxxxxxxxx>
Cc: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>
Cc: Kieran Bingham <kieran.bingham+renesas@xxxxxxxxxxxxxxxx>
Cc: dri-devel@xxxxxxxxxxxxxxxxxxxxx
Cc: linux-renesas-soc@xxxxxxxxxxxxxxx
Cc: Jacopo Mondi <jacopo+renesas@xxxxxxxxxx>
---
  drivers/gpu/drm/rcar-du/Kconfig |    4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)

--- linux-next-20210420.orig/drivers/gpu/drm/rcar-du/Kconfig
+++ linux-next-20210420/drivers/gpu/drm/rcar-du/Kconfig
@@ -4,8 +4,8 @@ config DRM_RCAR_DU
      depends on DRM && OF
      depends on ARM || ARM64
      depends on ARCH_RENESAS || COMPILE_TEST
-    imply DRM_RCAR_CMM
-    imply DRM_RCAR_LVDS
+    select DRM_RCAR_CMM
+    select DRM_RCAR_LVDS

No all platforms that integrate a DU have CMM and/or LVDS support, so we
shouldn't select the automatically.

Would

    depends on DRM_RCAR_CMM || DRM_RCAR_CMM=n
    depends on DRM_RCAR_LVDS || DRM_RCAR_LVDS=n


Looks good; I like it, but kconfig does not:

$ xbuild_arm64.sh  oldconfig
make[1]: Entering directory '/home/rdunlap/lnx/next/linux-next-20210420/ARM64'
   GEN     Makefile
drivers/gpu/drm/rcar-du/Kconfig:2:error: recursive dependency detected!
drivers/gpu/drm/rcar-du/Kconfig:2:    symbol DRM_RCAR_DU depends on DRM_RCAR_CMM drivers/gpu/drm/rcar-du/Kconfig:17:    symbol DRM_RCAR_CMM depends on DRM_RCAR_DU
For a resolution refer to Documentation/kbuild/kconfig-language.rst
subsection "Kconfig recursive dependency limitations"

work ? I thought that's what "imply" meant, but it seems I got it wrong.

      select DRM_KMS_HELPER
      select DRM_KMS_CMA_HELPER
      select DRM_GEM_CMA_HELPER


There is some documentation (kconfig-language.rst) that says that when a driver Kconfig is using mixed builtin and loadable modules, the driver should be using IS_REACHABLE(). The example also uses "imply", so maybe
we should/could try something like that:
If CMM is REACHBLE(), make the calls.
If LVDS is READABLE(), make the calls.

I can look into that some more if you want me to.
Or I'll be happy to let you do so. :)

Thanks.



[Index of Archives]     [Linux Samsung SOC]     [Linux Wireless]     [Linux Kernel]     [ATH6KL]     [Linux Bluetooth]     [Linux Netdev]     [Kernel Newbies]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Samba]     [Device Mapper]

  Powered by Linux