On Thu, Jun 13, 2013 at 10:19 AM, Jan Hubicka <hubi...@ucw.cz> wrote:
>> On Thu, Jun 13, 2013 at 10:07 AM, Jan Hubicka <hubi...@ucw.cz> wrote:
>> >> Can you create a helper function to flag the error and perhaps also
>> >> put that check inside can_inline_edge_p ?
>> >>
>> >> David
>> >>
>> >>
>> >> On Wed, Jun 12, 2013 at 6:00 PM, Sriraman Tallam <tmsri...@google.com>
>> >> wrote:
>> >> > Hi Honza,
>> >> >
>> >> > I have isolated the ipa-inline.c part into a separate patch with a
>> >> > test and attached it here. The patch is simple. Could you please take
>> >> > a look?
>> >> >
>> >> > * ipa-inline.c (can_early_inline_edge_p): Flag an error when
>> >> > the function that cannot be inlined is target specific.
>> >> > * gcc.target/i386/inline_error.c: New test.
>> >
>> > Sorry for taking ages to look at the patch, I was too hooked into other
>> > problems.
>> > I also think can_early_inline_edge_p should not produce diagnostic - it is
>> > supposed
>> > to be predicate.
>> >
>> > So your problem is that the hard worker in tree-inline is not called at -O0
>> > and thus errors are not output? I would suggest arranging
>> > inline_always_inline_functions
>> > to return true even if inlining failed and thus making inline_calls to be
>> > called.
>>
>> Thanks Honza! Yes, that is the problem. Should I just make it return
>> true for these special conditions (TARGET_MISMATCH + gnu_inline +
>> always_inline + ...)?
>
> Can't it just return true if there is any alwaysinline call? I think if
> inline fails,
> we always want to error, right?
Ok, patch attached that does this. Please let me know what you think.
Thanks
Sri
>
> Honza
* tree-inline.c (expand_call_inline): Allow the error to be flagged
in early inline pass.
* ipa-inline.c (inline_always_inline_functions): Pretend always_inline
functions are inlined during failures to flag an error.
* gcc.target/i386/inline_error.c: New test.
Index: tree-inline.c
===================================================================
--- tree-inline.c (revision 200034)
+++ tree-inline.c (working copy)
@@ -3905,8 +3905,6 @@ expand_call_inline (basic_block bb, gimple stmt, c
for inlining, but we can't do that because frontends overwrite
the body. */
&& !cg_edge->callee->local.redefined_extern_inline
- /* Avoid warnings during early inline pass. */
- && cgraph_global_info_ready
/* PR 20090218-1_0.c. Body can be provided by another module. */
&& (reason != CIF_BODY_NOT_AVAILABLE || !flag_generate_lto))
{
Index: ipa-inline.c
===================================================================
--- ipa-inline.c (revision 200034)
+++ ipa-inline.c (working copy)
@@ -1911,7 +1911,15 @@ inline_always_inline_functions (struct cgraph_node
}
if (!can_early_inline_edge_p (e))
- continue;
+ {
+ /* Set inlined to true if the callee is marked "always_inline" but
+ is not inlinable. This will allow flagging an error later in
+ expand_call_inline in tree-inline.c. */
+ if (lookup_attribute ("always_inline",
+ DECL_ATTRIBUTES (callee->symbol.decl)) != NULL)
+ inlined = true;
+ continue;
+ }
if (dump_file)
fprintf (dump_file, " Inlining %s into %s (always_inline).\n",
Index: testsuite/gcc.target/i386/inline_error.c
===================================================================
--- testsuite/gcc.target/i386/inline_error.c (revision 0)
+++ testsuite/gcc.target/i386/inline_error.c (revision 0)
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-options "-O0 -mno-popcnt" } */
+
+inline int __attribute__ ((__gnu_inline__, __always_inline__,
target("popcnt")))
+foo () /* { dg-error "inlining failed in call to always_inline .* target
specific option mismatch" } */
+{
+ return 0;
+}
+
+int bar()
+{
+ return foo (); /* { dg-error "called from here" } */
+}
