Re: revert back the few changes to make OMAP H4 v5.0 work

["Andy Shevchenko" <andy.shevchenko%gmail.com@localhost>]

Sorry for typo in patch. 
There is attached new fixed patch.

On Wed, Apr 9, 2008 at 2:28 PM, Andy Shevchenko <andy.shevchenko%gmail.com@localhost> wrote:

Hello!

The attached patch fixes the current NetBSD (matt-armv6) sources to make work on OMAP H4 v5.0.
Actually I havn't tested in-depth, but after applying the board starts work correctly.

--
With Best Regards,
Andy Shevchenko

-- 
With Best Regards,
Andy Shevchenko

This patch contains the two reverts from the Matt's changeset for OMAP H4 v5.0.
The recent NetBSD userland works with incorrect behavior, f.e. /bin/sh just
produces the core dump and so on.

The patch isn't tested in-depth, but after applying the shells like sh or csh
and the system scripts are working fine.

--- matt-armv6-loc/lib/libc/arch/arm/gen/setjmp.S.orig  2007-08-28 
20:36:30.000000000 +0300
+++ matt-armv6-loc/lib/libc/arch/arm/gen/setjmp.S       2008-04-08 
17:42:00.000000000 +0300
@@ -87,7 +87,7 @@
        ldmfd   sp!, {r0-r2, r14}
 
        ldr     r1, .Lsetjmp_magic
-       str     r1, [r0]
+       str     r1, [r0], #4
 
 #ifdef SOFTFLOAT
        add     r0, r0, #52
--- matt-armv6-loc/lib/libc/arch/arm/gen/_setjmp.S.orig 2007-08-28 
20:36:29.000000000 +0300
+++ matt-armv6-loc/lib/libc/arch/arm/gen/_setjmp.S      2008-04-08 
10:56:21.000000000 +0300
@@ -68,6 +68,7 @@
         stmia  r0!, {r4-r6}
 
         mov    r0, #0x00000000
+        pop    {r4-r7}
 #else
        ldr     r1, .L_setjmp_magic
        str     r1, [r0], #4
@@ -85,7 +86,6 @@
 
         mov    r0, #0x00000000
 #endif /* __thumb__ */
-       XPOP    {r4-r7}
         RET
 
 ENTRY(_longjmp)