Run a Travis test on s390x (big endian) #696

pull sipa wants to merge 1 commits into bitcoin-core:master from sipa:201911_s390x changing 1 files +4 −0

sipa commented at 7:27 PM on November 28, 2019: contributor

No description provided.
sipa commented at 7:33 PM on November 28, 2019: contributor

checking whether byte ordering is bigendian... yes

\o/
elichai commented at 7:35 PM on November 28, 2019: contributor

It worked on core :P But still satisfying to see. There's nothing more satisfying than seeing that your code works on an edge case you never tested before lol
jonasnick commented at 7:54 AM on November 29, 2019: contributor

Cool! What's missing for un-WIP'ing? How about turning on the ECDH and recovery module?
real-or-random commented at 12:14 PM on July 21, 2020: contributor

Cool! What's missing for un-WIP'ing? How about turning on the ECDH and recovery module? @sipa Could you do this, and maybe also add a comment to the Travis file that this is a big-endian host?

in .travis.yml:39 in 4f682a4e40 outdated

  34 | @@ -35,6 +35,9 @@ env:
  35 |  matrix:
  36 |    fast_finish: true
  37 |    include:
  38 | +    - compiler: gcc
  39 | +      env: HOST=s390x-unknown-linux-gnu DOCKER_NAME_TAG=s390x/ubuntu:18.04

maflcko commented at 8:16 AM on July 28, 2020:

What does DOCKER_NAME_TAG do?

      env: HOST=s390x-unknown-linux-gnu

sipa commented at 12:36 AM on July 29, 2020:

I have no idea anymore, I assume I copied it from somewhere.

sipa force-pushed on Jul 29, 2020
sipa renamed this:
~~[WIP] Let's see if s390x testing works~~
Run a Travis test on s390x (big endian)
on Jul 29, 2020
sipa force-pushed on Jul 29, 2020
sipa commented at 1:50 AM on July 29, 2020: contributor
Rebased, and added ECDH/recovery.

I did notice valgrind ctime test failure on this line in src/scalar_4x64_impl.h:
```
    no |= (a->d[1] < SECP256K1_N_1);
```
so I've disabled the ctime test for now.

in .travis.yml:88 in a940a7127e outdated

  82 | @@ -83,6 +83,10 @@ matrix:
  83 |              - valgrind
  84 |              - libtool-bin
  85 |              - libc6-dbg:i386
  86 | +    # S390x build (big endian system)
  87 | +    - compiler: gcc
  88 | +      env: HOST=s390x-unknown-linux-gnu ECDH=yes RECOVERY=yes EXPERIMENTAL=yes CTIMETEST=no

real-or-random commented at 5:50 AM on July 29, 2020:

travis still doesn't like you

      env: HOST=s390x-unknown-linux-gnu ECDH=yes RECOVERY=yes EXPERIMENTAL=yes CTIMETEST=

I admit that this is not intuitive...

real-or-random commented at 5:51 AM on July 29, 2020: contributor
no |= (a->d[1] < SECP256K1_N_1);
and two lines below too:

https://github.com/bitcoin-core/secp256k1/blob/master/src/scalar_4x64_impl.h#L65

so I've disabled the ctime test for now.

makes sense.
real-or-random changes_requested
real-or-random commented at 6:07 AM on July 30, 2020: contributor

ACK c5e43bb8d6c60c5f8ee38c89ac4d0d755e6515d0 Travis works and says it's big endian

The travis failure is a randomness test again (#68), see https://travis-ci.org/github/bitcoin-core/secp256k1/jobs/713125311#L497

If you want, you can squash this. Otherwise I'll just merge it in the next days if noone objects.
Test travis s390x (big endian) 39295362cf
real-or-random force-pushed on Aug 7, 2020
real-or-random approved
real-or-random commented at 10:49 AM on August 7, 2020: contributor

ACK 39295362cfc856aae1c37cc1194c2f6d53fd6f25 Travis works and says it's big endian

(I squashed this for merging).
real-or-random merged this on Aug 7, 2020
real-or-random closed this on Aug 7, 2020
jasonbcox referenced this in commit f7cc6a9fb9 on Sep 29, 2020
deadalnix referenced this in commit a894289a5a on Sep 30, 2020

Contributors

Linked (view graph)

#707 General CI discussion #782 Check if variable=yes instead of if var is set in travis.sh