I had a look at this. It's possible to reproduce the failure inside qemu-mipsel-static. The test set-process-filter-t is fundamentally racy: it reads output from a sub-process and compares that to an expected value. But on a slow machine it's possible to get a partial read from the child process which causes the test to fail. I don't think it's anything related to MIPS per se.
This has already been fixed upstream. See here: https://git.savannah.gnu.org/cgit/emacs.git/commit?id=aa49aa884053d0e8b33efe265f2aade19d1f3f3d Maybe we can import the above patch or mark this test as unstable?
