Failure to build librseq on ppc

[Segher Boessenkool]

On Wed, Jul 08, 2020 at 10:32:20AM -0400, Mathieu Desnoyers wrote:
> > As far as I can see, %U is mentioned in
> > https://gcc.gnu.org/onlinedocs/gcc/Machine-Constraints.html in the
> > powerpc subpart, at the "m" constraint.
> 
> Yep, I did notice it, but mistakenly thought it was only needed for "m<>" operand,
> not "m".

Historically, "m" meant what "m<>" does now (in inline asm).  Too many
people couldn't get it right ever (on other targets -- not that the
situation was great for PowerPC, heh), so in inline asm "m" now means
"no pre-modify or post-modify".


Segher