Some work in progress patches I've been working on to add rseq support to sparc. It currently works with one minor hack on sparc64. The issue is that __thread object alignment isn't working properly in the test cases so I disable the alignment check in the rseq syscall implementation whilst running the tests. All the tests pass on sparc64 with this hack. I plan to look into why glibc isn't aligning TLS objects in threads properly. Sparc32 needs some fleshing out. First, sparc32 needs HAVE_REGS_AND_STACK_ACCESS_API support added, which isn't much work. Second, the rseq testcase sparc code needs to be made wordsize agnostic. I just rushed the implementation to test the sparc64 kernel bits out quickly. Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>
