<div dir="ltr">If this is wrong and just caught now, should it also be applied to the 5 branch?<div><br></div><div>--joel</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Sun, Jul 5, 2020 at 11:00 AM Sebastian Huber <<a href="mailto:sebh@rtems.org">sebh@rtems.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Module:    rtems<br>
Branch:    master<br>
Commit:    d0c1ce658ed1e1510f54279f662fc8a096cae6f7<br>
Changeset: <a href="http://git.rtems.org/rtems/commit/?id=d0c1ce658ed1e1510f54279f662fc8a096cae6f7" rel="noreferrer" target="_blank">http://git.rtems.org/rtems/commit/?id=d0c1ce658ed1e1510f54279f662fc8a096cae6f7</a><br>
<br>
Author:    Sebastian Huber <<a href="mailto:sebastian.huber@embedded-brains.de" target="_blank">sebastian.huber@embedded-brains.de</a>><br>
Date:      Tue Apr 21 10:17:00 2020 +0200<br>
<br>
i386: Replace fpcr clobber with memory clobber<br>
<br>
Update #3943.<br>
<br>
---<br>
<br>
 cpukit/score/cpu/i386/include/rtems/score/cpu.h | 4 ++--<br>
 cpukit/score/cpu/i386/sse_test.c                | 2 +-<br>
 2 files changed, 3 insertions(+), 3 deletions(-)<br>
<br>
diff --git a/cpukit/score/cpu/i386/include/rtems/score/cpu.h b/cpukit/score/cpu/i386/include/rtems/score/cpu.h<br>
index 225b728..632bf5f 100644<br>
--- a/cpukit/score/cpu/i386/include/rtems/score/cpu.h<br>
+++ b/cpukit/score/cpu/i386/include/rtems/score/cpu.h<br>
@@ -621,7 +621,7 @@ void _CPU_Context_save_fp(<br>
     __asm__ __volatile__(                      \<br>
       "fldcw %0"                               \<br>
       ::"m"((*(fp_context_pp))->fpucw)         \<br>
-      :"fpcr"                                  \<br>
+      :"memory"                                \<br>
     );                                         \<br>
     __builtin_ia32_ldmxcsr(_Thread_Executing->fp_context->mxcsr);  \<br>
   } while (0)<br>
@@ -640,7 +640,7 @@ void _CPU_Context_restore_fp(<br>
       :                                               \<br>
       :"st","st(1)","st(2)","st(3)",                  \<br>
        "st(4)","st(5)","st(6)","st(7)",               \<br>
-       "fpsr","fpcr"                                  \<br>
+       "fpsr","memory"                                \<br>
     );                                                \<br>
        if ( _Thread_Executing->fp_context ) {            \<br>
          _CPU_Context_restore_fp(&_Thread_Executing->fp_context); \<br>
diff --git a/cpukit/score/cpu/i386/sse_test.c b/cpukit/score/cpu/i386/sse_test.c<br>
index dc87371..14e7dd9 100644<br>
--- a/cpukit/score/cpu/i386/sse_test.c<br>
+++ b/cpukit/score/cpu/i386/sse_test.c<br>
@@ -229,7 +229,7 @@ fp_ld(Context_Control_sse *p_ctxt, int i)<br>
 #define FPUCLOBBER \<br>
        "st","st(1)","st(2)","st(3)",   \<br>
        "st(4)","st(5)","st(6)","st(7)",\<br>
-       "fpsr","fpcr"<br>
+       "fpsr","memory"<br>
<br>
 /* There seems to be no way to say that mxcsr was clobbered */<br>
<br>
<br>
_______________________________________________<br>
vc mailing list<br>
<a href="mailto:vc@rtems.org" target="_blank">vc@rtems.org</a><br>
<a href="http://lists.rtems.org/mailman/listinfo/vc" rel="noreferrer" target="_blank">http://lists.rtems.org/mailman/listinfo/vc</a><br>
</blockquote></div>