static g++: clarify pthread problems further

README.adoc

@@ -3899,7 +3899,10 @@ it fails with:
 ld: cannot find -lopenblas
 ....
 
-`g++` and pthreads also causes issues: https://stackoverflow.com/questions/35116327/when-g-static-link-pthread-cause-segmentation-fault-why
+`g++` and pthreads also causes issues:
+
+* https://stackoverflow.com/questions/35116327/when-g-static-link-pthread-cause-segmentation-fault-why
+* https://stackoverflow.com/questions/58848694/gcc-whole-archive-recipe-for-static-linking-to-pthread-stopped-working-in-rec
 
 As a consequence, the following fails:
 
@@ -3916,13 +3919,13 @@ qemu-x86_64: /path/to/linux-kernel-module-cheat/submodules/qemu/accel/tcg/cpu-ex
 
 and if we manually build and run natively on host it segfaults.
 
-If we hack the compilation command to do instead:
+The workaround:
 
 ....
 -pthread -Wl,--whole-archive -lpthread -Wl,--no-whole-archive
 ....
 
-then it works. We should automate that at some point.
+fixes some of the problems, but not all, so we are just skipping those tests for now.
 
 === gem5 syscall emulation mode
 

path_properties.py

@@ -561,7 +561,10 @@ path_properties_tuples = (
                     },
                     {
                         'abort.c': {'signal_received': signal.Signals.SIGABRT},
-                        'atomic.c': {'baremetal': False},
+                        'atomic.c': {
+                            'baremetal': False,
+                            'test_run_args': {'cpus': 3},
+                        },
                         'assert_fail.c': {'signal_received': signal.Signals.SIGABRT},
                         # This has complex failure modes, too hard to assert.
                         'smash_stack.c': {'skip_run_unclassified': True},