Le 06/05/2020 à 19:58, Segher Boessenkool a écrit :
On Wed, May 06, 2020 at 10:58:55AM +1000, Michael Ellerman wrote:
The "m<>" here is breaking GCC 4.6.3, which we allegedly still support.

[ You shouldn't use 4.6.3, there has been 4.6.4 since a while.  And 4.6
   is nine years old now.  Most projects do not support < 4.8 anymore, on
   any architecture.  ]

Moving up to 4.6.4 wouldn't actually help with this though would it?

Nope.  But 4.6.4 is a bug-fix release, 91 bugs fixed since 4.6.3, so you
should switch to it if you can :-)

Also I have 4.6.3 compilers already built, I don't really have time to
rebuild them for 4.6.4.

The kernel has a top-level minimum version, which I'm not in charge of, see:


Yes, I know.  And it is much preferred not to have stricter requirements
for Power, I know that too.  Something has to give though :-/

There were discussions about making 4.8 the minimum, but I'm not sure
where they got to.

Yeah, just petered out I think?

All significant distros come with a 4.8 as system compiler.

Plain "m" works, how much does the "<>" affect code gen in practice?

A quick diff here shows no difference from removing "<>".

It will make it impossible to use update-form instructions here.  That
probably does not matter much at all, in this case.

If you remove the "<>" constraints, also remove the "%Un" output modifier?

So like this?

diff --git a/arch/powerpc/include/asm/uaccess.h 
index 62cc8d7640ec..ca847aed8e45 100644
--- a/arch/powerpc/include/asm/uaccess.h
+++ b/arch/powerpc/include/asm/uaccess.h
@@ -207,10 +207,10 @@ do {                                                      
#define __put_user_asm_goto(x, addr, label, op) \
        asm volatile goto(                                      \
-               "1:        " op "%U1%X1 %0,%1    # put_user\n"      \
+               "1:        " op "%X1 %0,%1       # put_user\n"              \
                EX_TABLE(1b, %l2)                               \
                :                                               \
-               : "r" (x), "m<>" (*addr)                              \
+               : "r" (x), "m" (*addr)                              \
                :                                               \
                : label)

Like that.  But you will have to do that to *all* places we use the "<>"
constraints, or wait for more stuff to fail?  And, there probably are
places we *do* want update form insns used (they do help in some loops,
for example)?

AFAICT, git grep "m<>" provides no result.

However many places have %Ux:

arch/powerpc/boot/io.h: __asm__ __volatile__("lbz%U1%X1 %0,%1; twi 0,%0,0; isync"
arch/powerpc/boot/io.h: __asm__ __volatile__("stb%U0%X0 %1,%0; sync"
arch/powerpc/boot/io.h: __asm__ __volatile__("lhz%U1%X1 %0,%1; twi 0,%0,0; isync"
arch/powerpc/boot/io.h: __asm__ __volatile__("sth%U0%X0 %1,%0; sync"
arch/powerpc/boot/io.h: __asm__ __volatile__("lwz%U1%X1 %0,%1; twi 0,%0,0; isync"
arch/powerpc/boot/io.h: __asm__ __volatile__("stw%U0%X0 %1,%0; sync"
arch/powerpc/include/asm/atomic.h: __asm__ __volatile__("lwz%U1%X1 %0,%1" : "=r"(t) : "m"(v->counter)); arch/powerpc/include/asm/atomic.h: __asm__ __volatile__("stw%U0%X0 %1,%0" : "=m"(v->counter) : "r"(i)); arch/powerpc/include/asm/atomic.h: __asm__ __volatile__("ld%U1%X1 %0,%1" : "=r"(t) : "m"(v->counter)); arch/powerpc/include/asm/atomic.h: __asm__ __volatile__("std%U0%X0 %1,%0" : "=m"(v->counter) : "r"(i));
arch/powerpc/include/asm/book3s/32/pgtable.h:           stw%U0%X0 %2,%0\n\
arch/powerpc/include/asm/book3s/32/pgtable.h:           stw%U0%X0 %L2,%1"
arch/powerpc/include/asm/io.h: __asm__ __volatile__("sync;"#insn"%U1%X1 %0,%1;twi 0,%0,0;isync"\ arch/powerpc/include/asm/io.h: __asm__ __volatile__("sync;"#insn"%U0%X0 %1,%0" \
arch/powerpc/include/asm/nohash/pgtable.h:                      stw%U0%X0 
arch/powerpc/include/asm/nohash/pgtable.h:                      stw%U0%X0 
arch/powerpc/kvm/powerpc.c: asm ("lfs%U1%X1 0,%1; stfd%U0%X0 0,%0" : "=m" (fprd) : "m" (fprs) arch/powerpc/kvm/powerpc.c: asm ("lfd%U1%X1 0,%1; stfs%U0%X0 0,%0" : "=m" (fprs) : "m" (fprd)


Reply via email to