On Mon, Jan 18, 2010 at 1:40 PM, Paulo J. Matos <pocma...@gmail.com> wrote: > On Mon, Jan 18, 2010 at 10:30 AM, Richard Guenther > <richard.guent...@gmail.com> wrote: >> On Mon, Jan 18, 2010 at 11:24 AM, Paulo J. Matos <pocma...@gmail.com> wrote: >>> Hi, >>> >>> As a continuation of my previous issue, what's the difference between >>> cfun and current_function_decl and which one should I use to walk the >>> tree during TARGET_FUNCTION_OK_FOR_SIBCALL? >>> >>> [In the internals document I only found references to cfun and even >>> there was hard to understand what it actually is] >> >> It's supposed to be that cfun->decl == current_function_decl. Use >> cfun in the middle-end, current_function_decl is used by frontends >> as long as there is no struct function allocated for a function. >> > > Hello Richard, > > Thanks for your reply. I tried cfun hoping that would do something > different than current_function_decl but I was wrong. > cfun->decl during the expand pass seems to only contain parts of the > function instead of the whole thing. > > print_c_tree(stderr, cfun->decl); (in the TARGET_FUNCTION_OK_FOR_SIBCALL hook) > only prints part of the first basic block. > > My theory is that the expand pass during the conversion to RTL > destroyed the tree of the caller (cfun), and therefore you cannot > actually access it during the hooks execution. > Do you know if this is the case?
Yes. During expansion we destroy the trees. Richard.
