On 9/28/2010 3:18 AM, Rene Rebe wrote:
Hi again,
it built here:
-rw-r--r-- 1 root root 10847302 Sep 25 11:34
build/cross-mips64-9.0-trunk-generic-mips64-EB-cross-linux/var/adm/logs/1-gcc.log
While usually building 64bit binaries on 32bit system was (and should) not a
problem, there certainly can be something in that distribution that poses a
problem, ... :-(
René
Yes, Probability something specific about building on 32-bit Suse SLES-11, To move
forward, I had to roll back to GCC 4.4.3.
Eventually I would like to set up a T2 host, or at least somthing more compatable for
cross building T2, but am currently short of time & fast machines.
Looking back at the GCC 4.5.0 compiler failure back in May, that was an internal compiler
error while building e2fsprogs on Ubuntu 9.10,
GCC 4.5.0 Built OK at stage 1 in that situation.
Jan
Am Sep 28, 2010 um 12:52 AM schrieb Jan Rovins:
On 9/27/2010 6:08 PM, Rene Rebe wrote:
I think 1-gcc built on my side just fine, ... Will check when I'm back in the
office.
Do you cross build on an x86 or is that a native build on MIPS?
I am Cross Building on X86.
It's a 32-bit Suse Enterprise 11 Host.
Back in May The GCC 4.5.0 build failed on a cross build on an Ubuntu 9.10
32-bit Host
I wonder if the Cross build troubles are from building 64-bit OS on a 32 bit
build host?
Perhaps that GCC code path is not that well tested?
Jan
René
Sent abroad - http://ExactCODE.com Germany.
On Sep 27, 2010, at 23:46, Jan Rovins<[email protected]> wrote:
On 9/25/2010 7:58 AM, Rene Rebe wrote:
Hi again,
...
I added preliminary stub glue to dietlibc that let it compile for mips64, but
as I did not yet touch any assembly the resulting binaries will most likely not
work correctly (e.g. crash on startup).
At least this let's all dietlibc depending packages build for now (yay!), so
that we can concentrate on really implementing mips64 support in dietlibc
instead of marking any package depending on dietlibc to not support mips64, ...
OK, I Backed out all my -mips64 dance-a-rounds for the dietlibc dependencies.
Now, with the current svn tip I can build right up to the stage 1 GCC, but it
fails in that build with a compiler segfault. I remember that I was also
getting an internal compiler error at this stage back in May with with GCC
4.5.0. So I guess that GCC 4.5.X is still not very healthy on Mips64. I tried
upgrading some of the dependencies to GCC, to see if I could get the segfault
to go away, but that did not help Mips64 at all, but these may come in handy
for the other architectures, if you want to bump up the versions of
gmp,mpfr,ppl and mpc. Here are some patches that will do that, but they did
not fix my trouble, so I think I will have to pin the GCC version to 4.4.3
until this is sorted out.
Jan
Patch:
--------------------------------------------------------------------------------
Index: package/develop/ppl/gmp_limb_bits.patch
===================================================================
--- package/develop/ppl/gmp_limb_bits.patch (revision 0)
+++ package/develop/ppl/gmp_limb_bits.patch (revision 0)
@@ -0,0 +1,21 @@
+Common subdirectories: ppl-0.10.2/Watchdog and ppl-0.10.2.fix//Watchdog
+diff -u ppl-0.10.2/configure ppl-0.10.2.fix//configure
+--- ppl-0.10.2/configure 2010-09-24 19:23:22.000000000 -0400
++++ ppl-0.10.2.fix//configure 2010-09-24 19:23:24.000000000 -0400
+@@ -16473,7 +16473,7 @@
+ || GMP_LIMB_BITS != mp_bits_per_limb) {
+ std::cerr
+<< "GMP header (gmp.h) and library (ligmp.*) bits-per-limb mismatch:\n"
+-<< "header gives"<< __GMP_BITS_PER_MP_LIMB<< ";\n"
++<< "header gives"<< GMP_LIMB_BITS<< ";\n"
+<< "library gives"<< mp_bits_per_limb<< ".\n"
+<< "This probably means you are on a bi-arch system and\n"
+<< "you are compiling with the wrong header or linking with\n"
Index: package/scientific/mpc/mpc.desc
===================================================================
--- package/scientific/mpc/mpc.desc (revision 37710)
+++ package/scientific/mpc/mpc.desc (working copy)
@@ -31,7 +31,7 @@
[L] LGPL
[S] Stable
-[V] 0.8.1
+[V] 0.8.2
[P] X 01-3-5---9 102.250
-[D] 2996344439 mpc-0.8.1.tar.gz http://www.multiprecision.org/mpc/download/
+[D] 0 mpc-0.8.2.tar.gz http://www.multiprecision.org/mpc/download/
Index: package/scientific/mpfr/mpfr.desc
===================================================================
--- package/scientific/mpfr/mpfr.desc (revision 37710)
+++ package/scientific/mpfr/mpfr.desc (working copy)
@@ -33,7 +33,7 @@
[L] LGPL
[S] Stable
-[V] 2.4.2
+[V] 3.0.0
[P] X 01-3-5---9 102.200
-[D] 3391603398 mpfr-2.4.2.tar.bz2 http://www.mpfr.org/mpfr-current/
+[D] 3894267609 mpfr-3.0.0.tar.bz2 http://www.mpfr.org/mpfr-current/
Index: package/scientific/gmp/gmp.desc
===================================================================
--- package/scientific/gmp/gmp.desc (revision 37710)
+++ package/scientific/gmp/gmp.desc (working copy)
@@ -40,7 +40,7 @@
[L] LGPL
[S] Stable
-[V] 4.3.1
+[V] 5.0.1
[P] X 01-3-5---9 102.100
-[D] 1588644981 gmp-4.3.1.tar.gz ftp://ftp.gnu.org/pub/gnu/gmp/
+[D] 1838086893 gmp-5.0.1.tar.gz ftp://ftp.gnu.org/pub/gnu/gmp/
PS: I do not have a mips64 silicon near me at the moment, I think I already
wrote somewhere that I stored my Sgi Octane (2 IIRC) at my parent attic as it
was too loud for the office, ... Maybe someone wants to send in (as in donate)
less noisy MIPS64 hardware?
http://www.t2-project.org/hardware/workstation/Sgi/Octane/
René
-----------------------------------------------------------
If you wish to unsubscribe from this mailing, send mail to
[email protected] with a subject of: unsubscribe t2
