Re: Make max_align_t respect _Float128 [version 2]

[Bernd Edlinger <bernd.edlinger@hotmail.de>]

Hi,

I'd like to share some of my thoughts about the max_align_t,
if you don't mind.

I think that for _Float128 to become a fundamental data type,
it is *not* sufficient that gcc supports it alone, because glibc
needs to support also all the necessary math library functions.

When that is done, glibc can also increase the malloc alignment.
That will mean that programs built for glibc-2.24 will not
run on old glibc installations.

I think that glibc-2.24 will still have to support older
gcc versions, and should make malloc alignment be fixed,
even if an application is built using an old gcc version,
or am I wrong, which is possible here?

I have some doubt that it will work to change max_align_t's
alignment only based on the gcc version, I mean
there should be a way to allow the max_align_t keep at
8 bytes if the existing glibc will not support 16 bytes.

I *think*, there should be some kind of hand-shake in the
max_align_t to enable __float128 when gcc supports that,
*and* glibc supports that at the same time.

Like, maybe including some glibc-header, that defines some
macro, what the real malloc alignment will be.  And probably
some builtin-macro, if gcc supports __float128, independently
of the i386, that could probably also be helpful for writing
portable applications.



Bernd.