[PATCH] Do not fold addressable operands of m into non-addressable (PR inline-asm/50571)

2011-09-30 Thread Jakub Jelinek
Hi! GCC on the following testcase warns warning: use of memory input without lvalue in asm operand 0 is deprecated [enabled by default] starting with 4.6, but the source actually had an lvalue there (I don't think we should forbid for input operands const qualified memory). On m (1) in the

Re: [PATCH] Do not fold addressable operands of m into non-addressable (PR inline-asm/50571)

2011-09-30 Thread Richard Guenther
On Fri, Sep 30, 2011 at 8:17 AM, Jakub Jelinek ja...@redhat.com wrote: Hi! GCC on the following testcase warns warning: use of memory input without lvalue in asm operand 0 is deprecated [enabled by default] starting with 4.6, but the source actually had an lvalue there (I don't think we

[PATCH] Do not fold addressable operands of m into non-addressable (PR inline-asm/50571, take 2)

2011-09-30 Thread Jakub Jelinek
Hi! On Fri, Sep 30, 2011 at 10:34:28AM +0200, Richard Guenther wrote: Hmm, I don't think this change is ok. We rely on maybe_fold_reference to re-fold mem-refs to valid gimple form (from propagating say a.b.c to MEM[p, 4] which first gives the invalid MEM[a.b.c, 4] and then the folding

Re: [PATCH] Do not fold addressable operands of m into non-addressable (PR inline-asm/50571, take 2)

2011-09-30 Thread Richard Guenther
On Fri, Sep 30, 2011 at 10:56 AM, Jakub Jelinek ja...@redhat.com wrote: Hi! On Fri, Sep 30, 2011 at 10:34:28AM +0200, Richard Guenther wrote: Hmm, I don't think this change is ok.  We rely on maybe_fold_reference to re-fold mem-refs to valid gimple form (from propagating say a.b.c to MEM[p,