Hi Jian-Xin,
The check with flag_spin_file should not be necessary, since it's already
inside the phase generating spin, and hence not a taken path when running
cc1/cc1plus in normal mode. Otherwise looks okay.
Gautam
From: Jian-Xin Lai <laij...@gmail.com>
To: Gautam Chakrabarti <gautam.c...@yahoo.com>
Cc: open64-devel <open64-devel@lists.sourceforge.net>
Sent: Monday, September 19, 2011 7:38 PM
Subject: Re: [Open64-devel] Code review for bug #827 [GCC FE]
HiGautam,
I made a minor change to this patch to put the new code enclosed by #ifdef
OPEN64_SPIN and flag_spin_file to make the check only valid for emitting spin
file.
Index: osprey-gcc-4.2.0/gcc/tree.c
===================================================================
--- osprey-gcc-4.2.0/gcc/tree.c (revision 3731)
+++ osprey-gcc-4.2.0/gcc/tree.c (working copy)
@@ -10916,6 +10916,18 @@
/* bug 12598: Try to fold OBJ_TYPE_REF if it is present
under the CALL_EXPR. Code adapted from fold_stmt() . */
tree callee = get_callee_fndecl (t);
+#ifdef OPEN64_SPIN
+ if (flag_spin_file &&
+ callee && TREE_CODE(callee) == FUNCTION_DECL)
+ {
+ /* we need to emit the function be calleed, no matter
+ if the call is removed later by gcc cfg cleanup, so
+ the open64 backend wouldn't be surprised by missing
+ function definition. */
+ TREE_SYMBOL_REFERENCED (DECL_ASSEMBLER_NAME (callee)) = 1;
+ mark_decl_referenced(callee);
+ }
+#endif
if (!(callee && DECL_BUILT_IN(callee)))
{
callee = TREE_OPERAND(t,0);
Could you please take a look at the new patch? Thank you very much.
2011/9/18 Gautam Chakrabarti <gautam.c...@yahoo.com>
I am a bit cautious with this change, but please go ahead assuming the testing
turns out okay.
>
>Gautam
>
>
>
>From: Jian-Xin Lai <laij...@gmail.com>
>To: Gautam Chakrabarti <gautam.c...@yahoo.com>
>Cc: open64-devel <open64-devel@lists.sourceforge.net>
>Sent: Friday, September 16, 2011 12:27 AM
>
>Subject: Re: [Open64-devel] Code review for bug #827 [GCC FE]
>
>
>
>The flag is set when the asm code of caller is emitted. Here is the stack
>trace:
>#0 mark_referenced (id=0x55877288) at
>../../../osprey-gcc-4.2.0/gcc/varasm.c:2067
>#1 0x085062eb in assemble_name (file=0x8855788, name=0x8856a31 "_Z1fv") at
>../../../osprey-gcc-4.2.0/gcc/varasm.c:2155
>#2 0x082aa373 in output_addr_const (file=0x8855788, x=0x5586f840) at
>../../../osprey-gcc-4.2.0/gcc/final.c:3213
>#3 0x0855451f in print_operand (file=0x8855788, x=0x5586f840, code=80) at
>../../../osprey-gcc-4.2.0/gcc/config/i386/i386.c:8167
>#4 0x082aa298 in output_operand (x=0x5586f840, code=80) at
>../../../osprey-gcc-4.2.0/gcc/final.c:3177
>#5 0x082a9ff5 in output_asm_insn (template=0x86fb45c "call\t%P0",
>operands=0x880fc60) at ../../../osprey-gcc-4.2.0/gcc/final.c:3087
>#6 0x082a925b in final_scan_insn (insn=0x55741084, file=0x8855788,
>optimize=0, nopeepholes=0, seen=0xffffcfcc) at ../../../osprey-gcc-4.2.0/\
>gcc/final.c:2472
>#7 0x082a8413 in final (first=0x557470a0, file=0x8855788, optimize=0) at
>../../../osprey-gcc-4.2.0/gcc/final.c:1614
>#8 0x082ab17d in rest_of_handle_final () at
>../../../osprey-gcc-4.2.0/gcc/final.c:3920
>#9 0x0851b8c5 in execute_one_pass (pass=0x8779200) at
>../../../osprey-gcc-4.2.0/gcc/passes.c:900
>#10 0x0851ba06 in execute_pass_list (pass=0x8779200) at
>../../../osprey-gcc-4.2.0/gcc/passes.c:951
>#11 0x0851ba22 in execute_pass_list (pass=0x8779f40) at
>../../../osprey-gcc-4.2.0/gcc/passes.c:952
>#12 0x0851ba22 in execute_pass_list (pass=0x8779f00) at
>../../../osprey-gcc-4.2.0/gcc/passes.c:952
>#13 0x0815a158 in tree_rest_of_compilation (fndecl=0x55873700) at
>../../../osprey-gcc-4.2.0/gcc/tree-optimize.c:483
>#14 0x080e1468 in expand_body (fn=0x55873700) at
>../../../osprey-gcc-4.2.0/gcc/cp/semantics.c:3100
>#15 0x0857ef86 in cgraph_expand_function (node=0x55740a20) at
>../../../osprey-gcc-4.2.0/gcc/cgraphunit.c:1244
>#16 0x0857f0bb in cgraph_expand_all_functions () at
>../../../osprey-gcc-4.2.0/gcc/cgraphunit.c:1309
>#17 0x0857f8ab in cgraph_optimize () at
>../../../osprey-gcc-4.2.0/gcc/cgraphunit.c:1588
>#18 0x0809b40d in cp_finish_file () at
>../../../osprey-gcc-4.2.0/gcc/cp/decl2.c:3410
>#19 0x08049e28 in finish_file () at
>../../../osprey-gcc-4.2.0/gcc/cp/cp-lang.c:160
>#20 0x08139017 in c_common_parse_file (set_yydebug=0) at
>../../../osprey-gcc-4.2.0/gcc/c-opts.c:1193
>#21 0x084dcfc3 in compile_file () at
>../../../osprey-gcc-4.2.0/gcc/toplev.c:1077
>#22 0x084de743 in do_compile () at ../../../osprey-gcc-4.2.0/gcc/toplev.c:2094
>#23 0x084de7a8 in toplev_main (argc=11, argv=0xffffd334) at
>../../../osprey-gcc-4.2.0/gcc/toplev.c:2126
>#24 0x08143fa5 in main (argc=11, argv=0xffffd334) at
>../../../osprey-gcc-4.2.0/gcc/main.c:52
>
>mark_referenced() is only called when the compiler assemble the symbol name
>during the final asm emit phase. Since we skipped all the
>"tree_rest_of_compilation", the flag is not set in the modified front end.
>
>
>2011/9/16 Gautam Chakrabarti <gautam.c...@yahoo.com>
>
>What phase is it? Is it really not run, or just that it is run after tree->gs
>translation? If the phase is run after the translation, then there are other
>instances of this scenario, and we use gs_set_flag_value() to update the flag
>of the already created gs node.
>>
>>The proposed change in itself is small, but I wanted to be cautious because
>>
>>1. that particular place is just a (mostly faithful) translation phase from
>>tree -> gs, so it would be good to not have to update the tree there.
>>2. And in the past the matter of setting what functions need to be
>>defined/emitted has proved tricky because differences between our notion of
>>needing to emit a function and g++'s notion can cause trouble.
>>
>>Gautam
>>
>>
>>
>>From: Jian-Xin Lai <laij...@gmail.com>
>>To: Gautam Chakrabarti <gautam.c...@yahoo.com>
>>Cc: open64-devel <open64-devel@lists.sourceforge.net>
>>Sent: Thursday, September 15, 2011 1:15 AM
>>
>>Subject: Re: [Open64-devel] Code review for bug #827 [GCC FE]
>>
>>
>>
>>GCC checks this and set the flag in the later phase but we skip all those
>>phase.
>>
>>
>>2011/9/15 Gautam Chakrabarti <gautam.c...@yahoo.com>
>>
>>Hi Jian-Xin,
>>>
>>>Where does normal g++ execution set this flag? I.e., why is it not getting
>>>set for us?
>>>
>>>Thanks,
>>>Gautam
>>>
>>>
>>>
>>>From: Jian-Xin Lai <laij...@gmail.com>
>>>
>>>To: open64-devel <open64-devel@lists.sourceforge.net>
>>>Sent: Tuesday, September 13, 2011 12:49 AM
>>>Subject: Re: [Open64-devel] Code review for bug #827 [GCC FE]
>>>
>>>
>>>
>>>Hi,
>>>Could a gate keeper review this patch? Thank you very much.
>>>
>>>
>>>2011/9/7 Jian-Xin Lai <laij...@gmail.com>
>>>
>>>Hi,
>>>>
>>>>Could a gate keeper review the patch for #827? Thank you very much.
>>>>
>>>>The error message should be reported at toplev.c, line 884:
>>>> 874 if (TREE_CODE (decl) == FUNCTION_DECL
>>>> 875 && DECL_INITIAL (decl) == 0
>>>> 876 && DECL_EXTERNAL (decl)
>>>> 877 && ! DECL_ARTIFICIAL (decl)
>>>> 878 && ! TREE_NO_WARNING (decl)
>>>> 879 && ! TREE_PUBLIC (decl)
>>>> 880 && (warn_unused_function
>>>> 881 || TREE_SYMBOL_REFERENCED (DECL_ASSEMBLER_NAME (decl))))
>>>> 882 {
>>>> 883 if (TREE_SYMBOL_REFERENCED (DECL_ASSEMBLER_NAME (decl)))
>>>> 884 pedwarn ("%q+F used but never defined", decl);
>>>> 885 else
>>>> 886 warning (0, "%q+F declared %<static%> but never defined",
>>>>decl);
>>>> 887 /* This symbol is effectively an "extern" declaration now. */
>>>> 888 TREE_PUBLIC (decl) = 1;
>>>> 889 assemble_external (decl);
>>>> 890 }
>>>>
>>>>
>>>>But because the TREE_SYMBOL_REFERENCED (DECL_ASSEMBLER_NAME (decl)) is not
>>>>set for the callee in this case, the compiler doesn't report any error on
>>>>this case. The patch is to set this flag when handle the CALL_EXPR:
>>>>
>>>>Index: /open64/open64-trunk/osprey-gcc-4.2.0/gcc/tree.c
>>>>===================================================================
>>>>--- /open64/open64-trunk/osprey-gcc-4.2.0/gcc/tree.c (revision 3731)
>>>>+++ /open64/open64-trunk/osprey-gcc-4.2.0/gcc/tree.c (working copy)
>>>>@@ -10916,6 +10916,16 @@
>>>> /* bug 12598: Try to fold OBJ_TYPE_REF if it is present
>>>> under the CALL_EXPR. Code adapted from fold_stmt() . */
>>>> tree callee = get_callee_fndecl (t);
>>>>+ if (callee && TREE_CODE(callee) == FUNCTION_DECL)
>>>>+ {
>>>>+ /* we need to emit the function be calleed, no matter
>>>>+ if the call is removed later by gcc cfg cleanup, so
>>>>+ the open64 backend wouldn't be surprised by missing
>>>>+ function definition. */
>>>>+ TREE_SYMBOL_REFERENCED (DECL_ASSEMBLER_NAME (callee)) = 1;
>>>>+ mark_decl_referenced(callee);
>>>>+ }
>>>>+
>>>> if (!(callee && DECL_BUILT_IN(callee)))
>>>> {
>>>> callee = TREE_OPERAND(t,0);
>>>>
>>>>
>>>>--
>>>>Regards,
>>>>Lai Jian-Xin
>>>>
>>>
>>>
>>>--
>>>Regards,
>>>Lai Jian-Xin
>>>
>>>------------------------------------------------------------------------------
>>>BlackBerry® DevCon Americas, Oct. 18-20, San Francisco, CA
>>>Learn about the latest advances in developing for the
>>>BlackBerry® mobile platform with sessions, labs & more.
>>>See new tools and technologies. Register for BlackBerry® DevCon today!
>>>http://p.sf.net/sfu/rim-devcon-copy1
>>>_______________________________________________
>>>Open64-devel mailing list
>>>Open64-devel@lists.sourceforge.net
>>>https://lists.sourceforge.net/lists/listinfo/open64-devel
>>>
>>>
>>>
>>
>>
>>--
>>Regards,
>>Lai Jian-Xin
>>
>>
>>
>
>
>--
>Regards,
>Lai Jian-Xin
>
>
>
--
Regards,
Lai Jian-Xin------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2dcopy1
_______________________________________________
Open64-devel mailing list
Open64-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/open64-devel
