Building libgpg-error for powerpc64-e5500-linux-gnu

Grzegorz Borowiak GBorowiak at advaoptical.com
Tue Apr 28 17:55:48 CEST 2015


Thank you for so quick answer.
I will try how it works.
I'm almost sure the ABI is the same, but I must check. 

And is there an architecture-independent and ABI-independent way of building libgpg-error?

________________________________________
From: Werner Koch <wk at gnupg.org>
Sent: 28 April 2015 17:48
To: Grzegorz Borowiak
Cc: gnupg-users at gnupg.org
Subject: Re: Building libgpg-error for powerpc64-e5500-linux-gnu

On Tue, 28 Apr 2015 14:32, GBorowiak at advaoptical.com said:

> Can I somehow convince it to recognise powerpc64-e5500-linux-gnu as
> powerpc64-unknown-linux-gnu?

If both systems use the same ABI config.sub should have returned a
canonicalized versions.  If not we can use a new mechanism available in
1.19.  You need to change the code, though: In src/mkheader.c find the
function

--8<---------------cut here---------------start------------->8---
static char *
canon_host_triplet (const char *triplet)
{
  struct {
    const char *name;
    const char *alias;
  } tbl[] = {
    {"i486-pc-linux-gnu", "i686-pc-linux-gnu" },
    {"i586-pc-linux-gnu" },

    { NULL }
  };
--8<---------------cut here---------------end--------------->8---

and add a line

    {"powerpc64-e5500-linux-gnu", "powerpc64-unknown-linux-gnu" },

before the {NULL} marker.  But do that only if you are sure both use the
same ABI.

Let me know if this works and I'll add it for 1.20.


Salam-Shalom,

   Werner

--
Die Gedanken sind frei.  Ausnahmen regelt ein Bundesgesetz.




More information about the Gnupg-users mailing list