Re: [PATCH] Fix PR middle-end/24003

[Ian Lance Taylor <ian at airs dot com>]

Eric Botcazou <ebotcazou@adacore.com> writes:

> Pretty sophisticated. :-)  Do we really have to work so hard at RTL generation 
> time?  I mean, the head comment of expand_call reads:
> 
> /* Generate all the code for a function call
>    and return an rtx for its value.
>    Store the value in TARGET (specified as an rtx) if convenient.
>    If the value is stored in TARGET then TARGET is returned.
>    If IGNORE is nonzero, then we ignore the value of the function call.  */
> 
> so can't we simply get away with forcing TARGET to a register if it is a MEM 
> and there is at least one argument to restore?

That's a good point.  The RTL optimizers are quite likely to sort it
out.  Please try the simpler patch.

Ian