On 6/1/22 04:40, Arnd Bergmann wrote:
lib/test_printf.c:215: warning: "PTR" redefined
215 | #define PTR ((void *)0xffff0123456789abUL)
|
In file included from /git/arm-soc/arch/loongarch/include/asm/vdso/vdso.h:9,
from
/git/arm-soc/arch/loongarch/include/asm/vdso/gettimeofday.h:13,
from /git/arm-soc/include/vdso/datapage.h:137,
from /git/arm-soc/arch/loongarch/include/asm/vdso.h:11,
from /git/arm-soc/arch/loongarch/include/asm/elf.h:13,
from /git/arm-soc/include/linux/elf.h:6,
from /git/arm-soc/include/linux/module.h:19,
from /git/arm-soc/lib/test_printf.c:10:
/git/arm-soc/arch/loongarch/include/asm/asm.h:182: note: this is the
location of the previous definition
182 | #define PTR .dword
|
Not sure what the best fix is for this, maybe the contents of asm/asm.h could
just be hidden in an "#idef __ASSEMBLER__" check. This can be a follow-up
patch when the branch is merged.
Ah, the dreaded PTR... This has plagued Loongson users since antiquity
(i.e. the MIPS era).
It must have been the case that the arch/loongarch was based on an
earlier version of arch/mips, that didn't have the commit fa62f39dc7e25
("MIPS: Fix build error due to PTR used in more places"). So the fix
would be simple: just rename the PTR to something else. MIPS changed
that to PTR_WD and maybe we could re-use that name.
But I agree that wrapping the whole asm/asm.h with an #ifdef
__ASSEMBLY__ is very reasonable regardless. Maybe both could be done.
