Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CA+fZqCWZFeB1bNj+pV8B1AKw0od9m9NJyNvd6bd-M7fE++ks5Q@mail.gmail.com>
Date: Wed, 6 Dec 2017 20:33:54 +0100
From: ardi <ardillasdelmonte@...il.com>
To: musl@...ts.openwall.com
Subject: Re: Feasability of patching libm with OS X 10.5.8 libc for IBM
 double-double in PPC

On Wed, Dec 6, 2017 at 6:53 PM, Szabolcs Nagy <nsz@...t70.net> wrote:
> * ardi <ardillasdelmonte@...il.com> [2017-12-06 18:23:33 +0100]:
>> Said this, I need IBM double-double in PowerPC targets (using clang as
>> compiler), and I'd like to use musl because of its well written code.
>
> is this because of some abi compat issue?
> or some source code depends on double-double semantics?

It's because I have code that was very well tuned for double-double
precision. Of course I'd be happy to be able to work with an IEEE
128bit long double, and I'd use it from the first day if the LLVM
compiler team implements it with hardware optimizations for PPC64, but
I'd like to have the choice of using double-double if I need to check
or compare at some point.

>From the comments by Rich, I think maybe it's not hard to get it
working. If I find I can get it by patching a few files (maybe
conditionally using the code for FP to/from string conversions in
Apple libc in my custom musl build), then I'll do it. OTOH, if it
requires a lot of editing, I won't doing it, because I want to have an
easy procedure for updating musl to newer versions. If I have to apply
a huge patchset whenever I update, it wouldn't be reasonable.

Thanks!

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.