Source: python3.14
Version: 3.14.0-2
Severity: normal
Tags: patch upstream
User: debian-68k@lists.debian.org
Usertags: m68k
X-Debbugs-Cc: debian-68k@lists.debian.org
Hello,
the previously reported FTBFS regression with GCC 15 turned out to be a bug
in the m68k-specific assembler code in the Python interpretor, see [1].
The attached patch fixes the problem. I will send the patch upstream.
Adrian

[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122211

--
.''`. John Paul Adrian Glaubitz
: :' : Debian Developer
`. `' Physicist
`- GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913

Index: python3.14-3.14.0/Include/internal/pycore_pymath.h ===================================================================
--- python3.14-3.14.0.orig/Include/internal/pycore_pymath.h
+++ python3.14-3.14.0/Include/internal/pycore_pymath.h
@@ -146,17 +146,17 @@ extern void _Py_set_387controlword(unsig
unsigned int old_fpcr, new_fpcr
#define _Py_SET_53BIT_PRECISION_START \
do { \
- __asm__ ("fmove.l %%fpcr,%0" : "=g" (old_fpcr)); \
+ __asm__ ("fmove.l %%fpcr,%0" : "=dm" (old_fpcr)); \
/* Set double precision / round to nearest. */ \
new_fpcr = (old_fpcr & ~0xf0) | 0x80; \
if (new_fpcr != old_fpcr) { \
- __asm__ volatile ("fmove.l %0,%%fpcr" : : "g" (new_fpcr));\
+ __asm__ volatile ("fmove.l %0,%%fpcr" : : "dm" (new_fpcr)); \
} \
} while (0)
#define _Py_SET_53BIT_PREC