[PATCH v2 1/3] include: Add support for byteorder/endianness conversion

[Anup Patel]

On Thu, Feb 9, 2023 at 5:58 AM Samuel Holland <samuel at sholland.org> wrote:
>
> On 2/1/23 22:44, Rahul Pathak wrote:
> > Define macros general byteorder conversion
> > Define functions for endianness conversion
> > from general byteorder conversion macros
> >
> > Signed-off-by: Rahul Pathak <rpathak at ventanamicro.com>
> > Reviewed-by: Xiang W <wxjstz at 126.com>
> > Reviewed-by: Sergey Matyukevich <sergey.matyukevich at syntacore.com>
> > ---
> >  include/sbi/sbi_byteorder.h | 57 +++++++++++++++++++++++++++++++++++++
> >  1 file changed, 57 insertions(+)
> >  create mode 100644 include/sbi/sbi_byteorder.h
> >
> > diff --git a/include/sbi/sbi_byteorder.h b/include/sbi/sbi_byteorder.h
> > new file mode 100644
> > index 000000000000..680710fe6d91
> > --- /dev/null
> > +++ b/include/sbi/sbi_byteorder.h
> > @@ -0,0 +1,57 @@
> > +/*
> > + * SPDX-License-Identifier: BSD-2-Clause
> > + *
> > + * Copyright (c) 2023 Ventana Micro Systems Inc.
> > + */
> > +
> > +#ifndef __SBI_BYTEORDER_H__
> > +#define __SBI_BYTEORDER_H__
> > +
> > +#include <sbi/sbi_types.h>
> > +
> > +#define EXTRACT_BYTE(x, n)   ((unsigned long long)((uint8_t *)&x)[n])
> > +
> > +#define BSWAP16(x)   ((EXTRACT_BYTE(x, 0) << 8) | EXTRACT_BYTE(x, 1))
> > +#define BSWAP32(x)   ((EXTRACT_BYTE(x, 0) << 24) | (EXTRACT_BYTE(x, 1) << 16) | \
> > +                      (EXTRACT_BYTE(x, 2) << 8) | EXTRACT_BYTE(x, 3))
> > +#define BSWAP64(x)   ((EXTRACT_BYTE(x, 0) << 56) | (EXTRACT_BYTE(x, 1) << 48) | \
> > +                      (EXTRACT_BYTE(x, 2) << 40) | (EXTRACT_BYTE(x, 3) << 32) | \
> > +                      (EXTRACT_BYTE(x, 4) << 24) | (EXTRACT_BYTE(x, 5) << 16) | \
> > +                      (EXTRACT_BYTE(x, 6) << 8) | EXTRACT_BYTE(x, 7))
>
> This implementation is wrong. It only does any swapping on little-endian
> systems. That was the desired behavior for CPU_TO_FDT*, but these macros
> need to swap bytes regardless of the native endianness.
>
> (Somehow, none of the other three reviewers noticed this?)

Argh, this was totally missed. Thanks for catching.

The EXTRACT_BYTE() macro is the culprit and does not do the
right thing for a big-endian host. Also, this macro will fail if "x" is
a constant.

I will send a fix ASAP.

Regards,
Anup

>
> Regards,
> Samuel
>
> > +#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__  /* CPU(little-endian) */
> > +#define cpu_to_be16(x)               ((uint16_t)BSWAP16(x))
> > +#define cpu_to_be32(x)               ((uint32_t)BSWAP32(x))
> > +#define cpu_to_be64(x)               ((uint64_t)BSWAP64(x))
> > +
> > +#define be16_to_cpu(x)               ((uint16_t)BSWAP16(x))
> > +#define be32_to_cpu(x)               ((uint32_t)BSWAP32(x))
> > +#define be64_to_cpu(x)               ((uint64_t)BSWAP64(x))
> > +
> > +#define cpu_to_le16(x)               ((uint16_t)(x))
> > +#define cpu_to_le32(x)               ((uint32_t)(x))
> > +#define cpu_to_le64(x)               ((uint64_t)(x))
> > +
> > +#define le16_to_cpu(x)               ((uint16_t)(x))
> > +#define le32_to_cpu(x)               ((uint32_t)(x))
> > +#define le64_to_cpu(x)               ((uint64_t)(x))
> > +#else /* CPU(big-endian) */
> > +#define cpu_to_be16(x)               ((uint16_t)(x))
> > +#define cpu_to_be32(x)               ((uint32_t)(x))
> > +#define cpu_to_be64(x)               ((uint64_t)(x))
> > +
> > +#define be16_to_cpu(x)               ((uint16_t)(x))
> > +#define be32_to_cpu(x)               ((uint32_t)(x))
> > +#define be64_to_cpu(x)               ((uint64_t)(x))
> > +
> > +#define cpu_to_le16(x)               ((uint16_t)BSWAP16(x))
> > +#define cpu_to_le32(x)               ((uint32_t)BSWAP32(x))
> > +#define cpu_to_le64(x)               ((uint64_t)BSWAP64(x))
> > +
> > +#define le16_to_cpu(x)               ((uint16_t)BSWAP16(x))
> > +#define le32_to_cpu(x)               ((uint32_t)BSWAP32(x))
> > +#define le64_to_cpu(x)               ((uint64_t)BSWAP64(x))
> > +#endif
> > +
> > +#endif /* __SBI_BYTEORDER_H__ */
>
>
> --
> opensbi mailing list
> opensbi at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/opensbi