Re: [GXemul-devel] NetBSD/hpcmips install fails under GXemul 0.6.1
Status: Alpha
Brought to you by:
gavare
From: Andreas G. <gs...@gs...> - 2019-06-05 06:45:05
|
Anders Gavare wrote: > Yes, the testmips failures are known issues -- I am removing everything > MIPS related from the "new" framework, concentrating on the old framework > where MIPS things actually work. Some unit tests remain to be rewritten to > use some other architecture for testing. I removed the "test" dependency from the "install" target in the Makefile, and "make install" then worked. > The important thing now is that a full install of NetBSD/hocmips can be > done and that you can run something meaningful non-trivial in it, such as > compile programs. I have now successfully done a full hpcmips install and ATF test run. The console log output from the tests is at: http://www.gson.org/netbsd/bugs/build/hpcmips/2019/2019.06.03.15.49.04/test.log There are some 181 failing test cases, and lots of "warning: LOW reference" and "pmap_tlb_update_addr(...) returned 0" messages, but that was already the case with 0.6.0.1. I also tested pmax, and it now fails to boot after the install as it tries to mount the wrong disk as the root. A console log from 0.6.0.1 is at http://www.gson.org/netbsd/bugs/build/pmax/2019/2019.05.02.00.34.06/test.log and from trunk: http://www.gson.org/netbsd/bugs/build/pmax/2019/2019.06.03.15.49.04/test.log Comparing the two, I see the bootstring has changed from "boot 5/rz0/ -a" to "boot 5/rz1/ -a". > As for the NetBSD patches, almost all of them were applied 20190417: see > http://gavare.se/gxemul/gxemul-trunk/HISTORY.html Great, thanks. -- Andreas Gustafsson, gs...@gs... |