Russell,
On 08/31/2017 09:03 AM, Russell King - ARM Linux wrote:

On Mon, Jul 24, 2017 at 04:24:54PM -0500, Dave Gerlach wrote:

quoted

diff --git a/arch/arm/kernel/asm-offsets.c b/arch/arm/kernel/asm-offsets.c
index 608008229c7d..46a7a6c41db7 100644
--- a/arch/arm/kernel/asm-offsets.c
+++ b/arch/arm/kernel/asm-offsets.c

@@ -28,6 +28,7 @@
 #include <asm/vdso_datapage.h>
 #include <asm/hardware/cache-l2x0.h>
 #include <linux/kbuild.h>
+#include <linux/ti-emif-sram.h>
 
 /*
  * Make sure that the compiler and target are compatible.

@@ -183,5 +184,8 @@ int main(void)
 #ifdef CONFIG_VDSO
   DEFINE(VDSO_DATA_SIZE,	sizeof(union vdso_data_store));
 #endif
+  BLANK();
+  ti_emif_offsets();
+
   return 0; 
 }

This doesn't make me happy - I've been thinking about this for some time,
and I think it's completely wrong for drivers to hook into asm-offsets.

Looking at the top-level Kbuild file, it looks like it would be easy
for emif to auto-generate its own offsets.h header - it looks like it's
a matter of using linux/kbuild.h to get the definitions, compiling to
assembly (using $(call if_changed_dep,cc_s_c)) and then converting the
assembly to the header file format ( using
$(call filechk,offsets,__ASM_OFFSETS_H__), replacing __ASM_OFFSETS_H__
with a more appropriate symbol.)

Could you please check whether this is possible, so we don't end up
needing to hook lots of driver-specific assembly offsets stuff into
the arch files?

I was able to try this out and was successful, I will send an updated version
that generates a local emif-asm-offsets.h. Thanks for the suggestion.

Regards,
Dave