Re: live insns deleted by delete_trivially_dead_insns()

[Paolo Bonzini]

Paolo Bonzini wrote:
Even if CONST_OR_PURE_CALL_P is set to 1??? :-(

I don't know, but in any case __tls_get_addr is neither pure nor const

rs6000 says it is, I should try doing the same for x86-64 and see what 
happens.

See the patches I've posted; with those I can remove 
maybe_encapsulate_block from emit_libcall_block and still get your 
testcase optimized.  One would need to perform more testing of course, 
but it's a good start.

However, there is another prerequisite for removing libcall notes, which 
is no-conflict notes.  Until Ian completes his reload-fu exercise, I 
guess we cannot remove the ~1000 lines that deal with libcall notes.

Paolo