This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Minimum GCC version for building glibc
- From: Torvald Riegel <triegel at redhat dot com>
- To: Roland McGrath <roland at hack dot frob dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Tue, 04 Nov 2014 21:41:11 -0600
- Subject: Re: Minimum GCC version for building glibc
- Authentication-results: sourceware.org; auth=none
- References: <Pine dot LNX dot 4 dot 64 dot 1410311352200 dot 4263 at digraph dot polyomino dot org dot uk> <20141104165815 dot GK5402 at vapier dot wh0rd dot info> <5459239B dot 4020901 at cs dot ucla dot edu> <20141104214726 dot EF16D2C244F at topped-with-meat dot com>
On Tue, 2014-11-04 at 13:47 -0800, Roland McGrath wrote:
> So I have no strong opinion between requiring 4.6 and requiring 4.7. AIUI
> the significant issue is that we believe 4.7 will make the atomics cleanups
> that Torvald is pursuing simpler. I don't really know how much simpler,
> given that the compiler's atomics in 4.7 are still less than perfect.
I think requiring 4.7 wouldn't make it much simpler, especially not
until we get closer to finishing the transition to C11 atomics.
The patches I've sent last week already have a mode in which existing
atomics are used to implement the new atomics, and this is the default
on most archs simply because I want someone to test the implementation
based on __atomic* builtins before enabling that one (and we'd need an
arch maintainer OK too, I think).
Requiring 4.7 would start to become beneficial when it allows us to
remove the then unneeded code. This would be the case for x86, I guess,
and mips to a partial extent IIRC. But I don't really know how this
would look for other archs, in particular at which GCC version the
atomics support on a particular arch becomes sufficient. I also haven't
really looked at what to do when GCC's libatomic would implement certain
atomics (e.g., when the kernel is used) -- should we just use libatomic,
or would we need to keep rolling our own.