On Sun, 4 Mar 2018, Earnie wrote:
> supplied. As you say "it's a userland-only architecture" then it is up
> to userland to specify and not rely on the conveniences. Allowing the
… eh, no. Also, I don’t think you understand the problem.
> precedent here will open up a maintenance nightmare.
M
On 3/3/2018 10:13 PM, Thorsten Glaser wrote:
> On Thu, 1 Mar 2018, James Clarke wrote:
>
>> x32, so it's not a *pure* system, but build=host=target=x32.
>
> Well, “pure” x32 systems don’t exist, as it’s a userland-only
> architecture, the kernel is simply amd64 for example. Same for
> arm64ilp32
On Thu, 1 Mar 2018, James Clarke wrote:
> x32, so it's not a *pure* system, but build=host=target=x32.
Well, “pure” x32 systems don’t exist, as it’s a userland-only
architecture, the kernel is simply amd64 for example. Same for
arm64ilp32 and arm64/aarch64 and perhaps(?) MIPS n32/n64.
Worse, x32
On Sat, 3 Mar 2018, Earnie wrote:
> Am I understanding correctly; your compiler is a 32bit compiler on a
> 64bit OS whose uname returns the 64bit OS system?
No, you aren’t.
My compiler is an x32 (amd64ilp32) compiler on an x32 host system.
The distinction in “32-bit” and “64-bit” didn’t work on
On 2/28/2018 7:53 PM, James Clarke wrote:
> On 1 Mar 2018, at 00:50, Ben Elliston wrote:
>>
>> On Wed, Feb 28, 2018 at 03:55:57PM +0100, Thorsten Glaser wrote:
>>
>>> I really have no other idea that???s in scope. After all,
>>> CC_FOR_BUILD is the *only* tool guaranteed to correspond to the
>>> t
On Wed, Feb 28, 2018 at 03:55:57PM +0100, Thorsten Glaser wrote:
> I really have no other idea that???s in scope. After all,
> CC_FOR_BUILD is the *only* tool guaranteed to correspond to the
> target (in FreeWRT speak; --build= in GNU autotools speak) system.
We're trying to guess the build syste
On 1 Mar 2018, at 00:50, Ben Elliston wrote:
>
> On Wed, Feb 28, 2018 at 03:55:57PM +0100, Thorsten Glaser wrote:
>
>> I really have no other idea that???s in scope. After all,
>> CC_FOR_BUILD is the *only* tool guaranteed to correspond to the
>> target (in FreeWRT speak; --build= in GNU autotoo
On Wed, 28 Feb 2018, James Clarke wrote:
> This is what I proposed to upstream initially[0], but I was told that
> CC_FOR_BUILD was deprecated and being phased out, and that I should be relying
Ouch!
> on standard system utilities instead. I chose objdump /bin/sh as another
> architecture was al
On 28 Feb 2018, at 10:16, Thorsten Glaser wrote:
>
> On Wed, 28 Feb 2018, root wrote:
>
>> autotools-dev (20180224.1) unstable; urgency=medium
>>
>> * Sync to upstream git 2018-02-24
>>[commit bd9626458c30d7faec17d7dfbd85a80617b10007]
>>+ Add detection of x32 ABI for x86_64-*-linux-gnu
On Wed, 28 Feb 2018, root wrote:
> autotools-dev (20180224.1) unstable; urgency=medium
>
> * Sync to upstream git 2018-02-24
> [commit bd9626458c30d7faec17d7dfbd85a80617b10007]
> + Add detection of x32 ABI for x86_64-*-linux-gnu (closes: #891281)
Thanks to klibc lacking x32 support, /bi
10 matches
Mail list logo
