Hello,
Flavio Cruz, le dim. 12 févr. 2023 01:15:05 -0500, a ecrit:
> diff --git a/configure.ac b/configure.ac
> index e4645bd..8d52d5f 100644
> --- a/configure.ac
> +++ b/configure.ac
> @@ -55,6 +55,13 @@ AC_SUBST([MIGCOM])
> [MIG=`echo mig | sed "$program_transform_name"`]
> AC_SUBST([MIG])
>
> +AC_ARG_ENABLE(user32,
> + AS_HELP_STRING([--enable-user32],
> + [Build mig for 64 bit kernel that works with 32
> bit userland]))
> +AS_IF([test "x$enable_user32" = "xyes"], [
> + AC_DEFINE([USER32], [1], ["Build mig for 64 bit kernel that works
> with 32 bit userland"])
> +])
> +
> AC_OUTPUT([Makefile mig tests/Makefile \
> tests/good/Makefile \
> tests/generate-only/Makefile \
> diff --git a/parser.y b/parser.y
> index 03c5ec8..1ebeae5 100644
> --- a/parser.y
> +++ b/parser.y
> @@ -212,6 +212,16 @@ Subsystem : SubsystemStart
> SubsystemMods
> IsKernelUser ? ", KernelUser" : "",
> IsKernelServer ? ", KernelServer" : "");
> }
> + if (IsKernelUser || IsKernelServer) {
> + port_size = vm_offset_size;
> + port_size_in_bits = vm_offset_size_in_bits;
> +#ifdef USER32
> + /* Since we are using MIG to generate stubs for a 64 bit kernel that
> operates
> + with a 32 bit userland, then we need to consider that `word_size`
> is 4 byte aligned.
> + */
> + word_size = sizeof(uint32_t);
> +#endif
> + }
> init_type();
> }
> ;
Mmm, I thought Luca's work was already handling the message translation,
so that the stubs parts of the kernel only sees 64bit messages already
with proper alignments etc.?
At least, I believe we'd rather aim for that, to have only a 64bit mig
and a 32bit mig, and not need a separate 64/32bit mig.
Samuel
