| Message ID | 20180706123528.2729378-1-arnd@arndb.de |
|---|---|
| State | Accepted |
| Commit | 4bcd8c90ac0f27d3d76fcfc50582ff3685059de9 |
| Headers | show |
| Series | omapfb: rename omap2 module to omap2fb.ko | expand |
diff --git a/drivers/video/fbdev/omap2/omapfb/Makefile b/drivers/video/fbdev/omap2/omapfb/Makefile index 602edfed09df..f54c3f56b641 100644 --- a/drivers/video/fbdev/omap2/omapfb/Makefile +++ b/drivers/video/fbdev/omap2/omapfb/Makefile @@ -2,5 +2,5 @@ obj-$(CONFIG_OMAP2_VRFB) += vrfb.o obj-y += dss/ obj-y += displays/ -obj-$(CONFIG_FB_OMAP2) += omapfb.o -omapfb-y := omapfb-main.o omapfb-sysfs.o omapfb-ioctl.o +obj-$(CONFIG_FB_OMAP2) += omap2fb.o +omap2fb-y := omapfb-main.o omapfb-sysfs.o omapfb-ioctl.o
In a kernel configuration with both CONFIG_FB_OMAP=m and CONFIG_FB_OMAP2=m, Kbuild fails to point out that we have two modules with the same name (omapfb.ko), but instead fails with a cryptic error message like: ERROR: "omapfb_register_panel" [drivers/video/fbdev/omap/lcd_osk.ko] undefined! This can now happen when building a randconfig kernel with CONFIG_ARCH_OMAP1, as the omap1 fbdev driver depends on that, whiel the omap2 fbdev driver can now be built anywhere with CONFIG_COMPILE_TEST. The solution is to rename one of the two modules, so for consistency with the directory naming I decided to rename the omap2 version to omap2fb.ko. Fixes: 7378f1149884 ("media: omap2: omapfb: allow building it with COMPILE_TEST") Signed-off-by: Arnd Bergmann <arnd@arndb.de> --- drivers/video/fbdev/omap2/omapfb/Makefile | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)