[PATCH] RISC-V: selftests: cbo: Use exported __cpu_to_le32() with uapi header

Andrew Jones ajones at ventanamicro.com
Fri Mar 22 01:40:05 PDT 2024


On Thu, Mar 21, 2024 at 08:14:34PM +0800, Woodrow Shen wrote:
> Short background: while I'm building the riscv/hwprobe tests, it seems
> I always hit the "error: impossible constraint in 'asm'" even though
> the commit https://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux.git/commit/tools/testing/selftests/riscv/hwprobe/cbo.c?id=0de65288d75ff96c30e216557d979fb9342c4323
> is applied. I'm not sure if this only happens to me as I tried to
> build them on a clean debian chroot. To apply this patch makes the
> build happy, but if there is no need to change, please ignore this
> patch.

It looks like I managed to mess up my fix to my mess up. Clearly I
shouldn't still have a potential function call in a macro which I want
to always resolve to a literal. Instead of using cpu_to_le32 I should
use something that results in ___constant_swab32, like your suggestion
of __cpu_to_le32. I'm not sure we want to depend on Linux UAPI that isn't
in the tools directory though, so maybe we should just reproduce it here.
Something like

#if __BYTE_ORDER == __BIG_ENDIAN
# define constant_swab32(_x)                     \
    ((((_x) & 0x000000ffU) << 24) |              \
     (((_x) & 0x0000ff00U) <<  8) |              \
     (((_x) & 0x00ff0000U) >>  8) |              \
     (((_x) & 0xff000000U) >> 24))
#else
# constant_swab32(_x) (_x)
#endif

#define MK_CBO(fn) constant_swab32((uint32_t)(fn) << 20 | 10 << 15 | 2 << 12 | 0 << 7 | 15)

Thanks,
drew

> 
> And sorry to forget cc more maintainers for reviewing.
> 
> Cheers,
> Woodrow
> 
> 
> On Thu, Mar 21, 2024 at 7:52 PM <woodrow.shen at sifive.com> wrote:
> >
> > From: Hsieh-Tseng Shen <woodrow.shen at sifive.com>
> >
> > cpu_to_le32 is not defined in uapi headers, and it could cause an error
> > of impossible constraint in 'asm' during compilation. However,
> > the reason is due to undefined reference to cpu_to_le32.
> > __cpu_to_le32() defined from byteorder.h should be used instead.
> >
> > Signed-off-by: Hsieh-Tseng Shen <woodrow.shen at sifive.com>
> > ---
> >  tools/testing/selftests/riscv/hwprobe/cbo.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/tools/testing/selftests/riscv/hwprobe/cbo.c b/tools/testing/selftests/riscv/hwprobe/cbo.c
> > index c537d52fafc5..b2dd18b3e360 100644
> > --- a/tools/testing/selftests/riscv/hwprobe/cbo.c
> > +++ b/tools/testing/selftests/riscv/hwprobe/cbo.c
> > @@ -15,11 +15,12 @@
> >  #include <linux/compiler.h>
> >  #include <linux/kernel.h>
> >  #include <asm/ucontext.h>
> > +#include <asm/byteorder.h>
> >
> >  #include "hwprobe.h"
> >  #include "../../kselftest.h"
> >
> > -#define MK_CBO(fn) cpu_to_le32((fn) << 20 | 10 << 15 | 2 << 12 | 0 << 7 | 15)
> > +#define MK_CBO(fn) __cpu_to_le32((fn) << 20 | 10 << 15 | 2 << 12 | 0 << 7 | 15)
> >
> >  static char mem[4096] __aligned(4096) = { [0 ... 4095] = 0xa5 };
> >
> > --
> > 2.34.1
> >



More information about the linux-riscv mailing list