I was able to confirm this by compiling with -mthumb. Confirmed with doko that this is likely to be a testsuite problem so I'm setting the importance to low, however it does also make it more difficult to spot actual regressions when there are this many failures.
I was able to confirm this by compiling with -mthumb. Confirmed with doko that this is likely to be a testsuite problem so I'm setting the importance to low, however it does also make it more difficult to spot actual regressions when there are this many failures.