While working on PR 83335 I proposed a change to a test case that used __builtin_extend_pointer and Richared Earnshaw pointed out that this builtin is not documented. Since I could not find any other (reasonable) way to generate an extended address in inline assembly other than this builtin I would like to document it for use.
Here is a proposed patch, the one problem I found was the return type of the builtin. I don't know how to describe it other than Pmode, but that is not a user visible type. See https://gcc.gnu.org/ml/gcc-patches/2018-02/msg01051.html for my PR 83335 patch and follow up comments. Should I go ahead and add this documentation? 2018-02-20 Steve Ellcey <sell...@cavium.com> * doc/extend.texi (__builtin_extend_pointer): Document builtin. diff --git a/gcc/doc/extend.texi b/gcc/doc/extend.texi index d38840e..94e47aa 100644 --- a/gcc/doc/extend.texi +++ b/gcc/doc/extend.texi @@ -11042,6 +11042,7 @@ the built-in function returns -1. @findex __builtin_alloca_with_align @findex __builtin_alloca_with_align_and_max @findex __builtin_call_with_static_chain +@findex __builtin_extend_pointer @findex __builtin_fpclassify @findex __builtin_isfinite @findex __builtin_isnormal @@ -12419,6 +12420,15 @@ Similar to @code{__builtin_bswap32}, except the argument and return types are 64 bit. @end deftypefn +@deftypefn {Built-in Function} Pmode __builtin_extend_pointer (void * x) +On targets where the user visible pointer size is different than the size +of an actual hardware address this function returns the extended user +pointer. Targets where this is true included ILP32 mode on x86_64 or +Aarch64. This function is mainly useful when writing inline assembly +code. +@var{addr} +@end deftypefn + @node Target Builtins @section Built-in Functions Specific to Particular Target Machines
