[PATCH] correct sparcv9 defines

[PATCH] correct sparcv9 defines

[john.levon]

From: Toomas Soome <tsoome@me.com>

The SPARCV9 compile check needs to look for __sparcv9 on some systems,
and should also define "sparc".
---
 lib.c     | 13 ++++++++-----
 machine.h |  2 +-
 2 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/lib.c b/lib.c
index 711e8fbe..14e51979 100644
--- a/lib.c
+++ b/lib.c
@@ -1587,15 +1587,18 @@ static void predefined_macros(void)
 		predefine("__s390__", 1, "1");
 		break;
 	case MACH_SPARC64:
-		predefine("__sparc_v9__", 1, "1");
-		predefine("__sparcv9__", 1, "1");
-		predefine("__sparcv9", 1, "1");
-		predefine("__sparc64__", 1, "1");
-		predefine("__arch64__", 1, "1");
+		if (arch_m64 != ARCH_LP32) {
+			predefine("__sparc_v9__", 1, "1");
+			predefine("__sparcv9__", 1, "1");
+			predefine("__sparcv9", 1, "1");
+			predefine("__sparc64__", 1, "1");
+			predefine("__arch64__", 1, "1");
+		}
 		/* fall-through */
 	case MACH_SPARC32:
 		predefine("__sparc__", 1, "1");
 		predefine("__sparc", 1, "1");
+		predefine("sparc", 1, "1");
 		break;
 	case MACH_X86_64:
 		if (arch_m64 != ARCH_LP32) {
diff --git a/machine.h b/machine.h
index 6eba448c..3389579c 100644
--- a/machine.h
+++ b/machine.h
@@ -56,7 +56,7 @@ enum machine {
 #define	MACH_NATIVE	MACH_RISCV64
 #elif defined(__riscv) && (__riscv_xlen == 32)
 #define	MACH_NATIVE	MACH_RISCV32
-#elif defined(__sparc_v9__)
+#elif defined(__sparc_v9__) || defined(__sparcv9)
 #define	MACH_NATIVE	MACH_SPARC64
 #elif defined(__sparc__) || defined(__sparc)
 #define	MACH_NATIVE	MACH_SPARC32
-- 
2.23.0

Re: [PATCH] correct sparcv9 defines

[Luc Van Oostenryck]

On Mon, Jan 20, 2020 at 02:38:24AM -0800, john.levon@joyent.com wrote:
> From: Toomas Soome <tsoome@me.com>
> 
> The SPARCV9 compile check needs to look for __sparcv9 on some systems,
> and should also define "sparc".

Hi.

Thanks for the patch.
Is it possible to have Toomas' Signed-off-by for this patch?
No need to resend the patch, just a reply to this email
saying it's OK is enough.

>  	case MACH_SPARC32:
>  		predefine("__sparc__", 1, "1");
>  		predefine("__sparc", 1, "1");
> +		predefine("sparc", 1, "1");

I'm wondering if there is a real need for this one or if
it's just to reflect what GCC do? I'm a bit reluctant to
add it since it lies in the user's namespace. But I'm fine
to add it if there is a real use case for it (in which case
I'll just slightly change the patch to not predefine it if
one of the ISO dialect is selected (via --std=c99, ...)).

-- Luc

Re: [PATCH] correct sparcv9 defines

[Luc Van Oostenryck]

On Wed, Jan 22, 2020 at 12:53:58PM +0200, Toomas Soome wrote:
> > On 21. Jan 2020, at 02:13, Luc Van Oostenryck <luc.vanoostenryck@gmail.com> wrote:
> > 
> > Thanks for the patch.
> > Is it possible to have Toomas' Signed-off-by for this patch?
> > No need to resend the patch, just a reply to this email
> > saying it's OK is enough.
> 
> OK by me:)
> 
> > 
> >> 	case MACH_SPARC32:
> >> 		predefine("__sparc__", 1, "1");
> >> 		predefine("__sparc", 1, "1");
> >> +		predefine("sparc", 1, "1");
> > 
> > I'm wondering if there is a real need for this one or if
> > it's just to reflect what GCC do? I'm a bit reluctant to
> > add it since it lies in the user's namespace. But I'm fine
> > to add it if there is a real use case for it (in which case
> > I'll just slightly change the patch to not predefine it if
> > one of the ISO dialect is selected (via --std=c99, ...)).
> > 
> 
> 
> The problem is that gcc does define those symbols and the code is using them, hence smatch should behave the same:

OK, fine for me if the symbol is effectively used.
This patch and John's are applied and pushed with some adaptations
for a related pending branch I merged first.

Thanks,
-- Luc