GMP 6.0.0a on AMD A8-6600K

[tg at gmplib.org]

Kevin Buckley <kevin.m.buckley at gmail.com> writes:

  > What I will also do, once I reboot out of the Xen environment, is to try
  > the  6.0.0a and snapshot sources against a "pure" kernel: just for
  > completeness.
  >
  > Still hoping this is something really obvious that I have missed,
  
  Dunno about "obvious" but it's obvioulsy a Xen thing.
  
This is curious; much of our own testing is done under Xen.

I've lost track, did you try a GMP snapshot in a Xen guest?  Did you try
Niels' suggested gdb commands on a failing system?  Another way which
might be easier is to re-configire with --disable-shared, then build,
then do

  $ gdb some-problem-testcase
  (gdb) run
  [expect illop]
  (gdb) x/5i
  (gdb) bt

This will show us the "illegal" instruction and the call chain which
took us there.

The goal of this exercise is to (1) properly understand the problem, and
(2) to make sure we've fixed it.

Marc suggested to use a snapshot.  He usually has very good reasons for
his suggestions...what did you have in mind, Marc?  AVX problems?

-- 
Torbjörn
Please encrypt, key id 0xC8601622