Hi David,
Qemu in Windows should be doable, but it will be painfully slow.

I have spun up a PowerPC VM for you; please mail me your SSH key(s) and desired 
username, and I will give you the login details.

Issues I'm seeing:

1. libffi/src/powerpc/ffi_powerpc.h is missing (should have been added as part 
of f1f5ede), but the Debian package is building with --with-system-libffi, so 
you'll need to use that

2. Running "./configure --build=powerpc-linux-gnu --with-system-libffi && make 
compiler && make compiler && ./poly --script Tests/Succeed/Test166.ML" hangs on 
the test
   (You need to manually give --build since config.guess sees the 64-bit kernel 
and assumes it's building for 64-bit PowerPC)

3. Some 64-bit big endian builds have been segfaulting on the first "make 
compiler":

   Making VALUEOPSSIG
   /bin/bash: line 1: 15571 Segmentation fault      ./poly --error-exit < 
mlsource/BuildExport.sml

   I'm setting up a ppc64 chroot now to see if that can reproduced on this 
particular machine.

Regards,
James

> On 21 Mar 2017, at 13:54, David Matthews <[email protected]> 
> wrote:
> 
> Hi James,
> I have access to a few machines but I don't think there's a PowerPC among 
> them.  I don't know if it's possible to install qemu in Windows and then run 
> a PowerPC virtual machine in that.  I currently use VirtualBox to run various 
> flavours of Linux and FreeBSD but of course that only works for X86.  If you 
> can get me access to a real machine that would certainly help.
> 
> I've already prepared a release of 5.7 and I'm inclined to go ahead with that 
> anyway.  There will almost certainly be other bug fixes to include so it's 
> likely that 5.7.1 won't be that far ahead.
> 
> Regards,
> David
> 
> On 21/03/2017 13:08, James Clarke wrote:
>> Hi David, I can reproduce this (or at least a very similar issue)
>> seemingly every time on 32-bit PowerPC. All three of my uploads to
>> Debian failed to build on it (across different machines) and I can
>> reproduce it on another box. It happens with both jessie and
>> unstable; below is the state of all the threads when I built it on
>> unstable with two calls to "make compiler" followed by a "make check"
>> (though ./poly --script Tests/Succeed/Test166.ML also hangs, so the
>> test harness and running other tests first is irrelevant). Do you
>> have access to a PowerPC machine you can put Debian on? If not I can
>> probably arrange getting you remote access to one.
>> 
>> Regards, James

_______________________________________________
polyml mailing list
[email protected]
http://lists.inf.ed.ac.uk/mailman/listinfo/polyml

Reply via email to