[Bug tree-optimization/44063] New: [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[Bug tree-optimization/44063] New: [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[hp at gcc dot gnu dot org]

With revision 159199 the trunk built for cris-elf.
With revision 159204 and on, including at least 159229, build was broken as
follows:

/tmp/hpautotest-gcc1/cris-elf/gccobj/./gcc/xgcc
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/./gcc/ -nostdinc
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/newlib/ -isystem
/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/newlib/targ-include -isystem
/tmp/hpautotest-gcc1/gcc/newlib/libc/include
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/libgloss/cris
-L/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/libgloss/libnosys
-L/tmp/hpautotest-gcc1/gcc/libgloss/cris
-B/tmp/hpautotest-gcc1/cris-elf/pre/cris-elf/bin/
-B/tmp/hpautotest-gcc1/cris-elf/pre/cris-elf/lib/ -isystem
/tmp/hpautotest-gcc1/cris-elf/pre/cris-elf/include -isystem
/tmp/hpautotest-gcc1/cris-elf/pre/cris-elf/sys-include    -g -O2 -march=v10
-mbest-lib-options -O2  -g -O2 -DIN_GCC -DCROSS_DIRECTORY_STRUCTURE  -W -Wall
-Wwrite-strings -Wcast-qual -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition  -isystem ./include   -g  -DIN_LIBGCC2
-D__GCC_FLOAT_NOT_NEEDED -Dinhibit_libc  -I. -I. -I../../.././gcc
-I/tmp/hpautotest-gcc1/gcc/libgcc -I/tmp/hpautotest-gcc1/gcc/libgcc/.
-I/tmp/hpautotest-gcc1/gcc/libgcc/../gcc
-I/tmp/hpautotest-gcc1/gcc/libgcc/../include  -DHAVE_CC_TLS -DUSE_EMUTLS -o
_mul_sf.o -MT _mul_sf.o -MD -MP -MF _mul_sf.dep -DFINE_GRAINED_LIBRARIES
-DL_mul_sf -c ../../.././gcc/tmplibgcc_fp_bit.c 
../../.././gcc/tmplibgcc_fp_bit.c: In function '__mulsf3':
../../.././gcc/tmplibgcc_fp_bit.c:959:1: internal compiler error: in
cgraph_estimate_size_after_inlining, at ipa-inline.c:208

Author of suspect patch in revision range CC:ed.
Preprocessed code attached momentarily.


-- 
           Summary: [4.6 Regression]: build broken for libgcc cris-elf, ICE
                    in cgraph_estimate_size_after_inlining, at ipa-inline
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code, build
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: hp at gcc dot gnu dot org
  GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: cris-axis-elf


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[hp at gcc dot gnu dot org]

------- Comment #1 from hp at gcc dot gnu dot org  2010-05-10 16:35 -------
Created an attachment (id=20621)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20621&action=view)
Preprocessed code, slightly reduced.

Compile with "./cc1 -O2 t.i", observe ICE as in the description.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[astrange at ithinksw dot com]

------- Comment #2 from astrange at ithinksw dot com  2010-05-11 03:38 -------
Created an attachment (id=20623)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20623&action=view)
testcase

This happens building ffmpeg on x86-64 now. Minimal-ish testcase attached.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[laurent at guerby dot net]

------- Comment #3 from laurent at guerby dot net  2010-05-11 08:15 -------
Native bootstrap on mips-linux fails with the same ICE:

...
/n/42/guerby/build/./gcc/xgcc -B/n/42/guerby/build/./gcc/
-B/n/42/guerby/install-trunk-159243/mips64el-unknown-linux-gnu/bin/
-B/n/42/guerby/insta\
ll-trunk-159243/mips64el-unknown-linux-gnu/lib/ -isystem
/n/42/guerby/install-trunk-159243/mips64el-unknown-linux-gnu/include -isystem
/n/42/guerb\
y/install-trunk-159243/mips64el-unknown-linux-gnu/sys-include    -g -O2
-minterlink-mips16 -mabi=32 -O2  -g -O2 -minterlink-mips16 -DIN_GCC   -W -\
Wall -Wwrite-strings -Wcast-qual -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition  -isystem ./include  -fPIC -g -DHAVE_GTHR_DEFAULT\
 -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED   -I. -I. -I../../.././gcc
-I../../../../trunk/libgcc -I../../../../trunk/libgcc/.
-I../../../../trunk/libg\
cc/../gcc -I../../../../trunk/libgcc/../include  -DHAVE_CC_TLS  -o _mul_sf.o
-MT _mul_sf.o -MD -MP -MF _mul_sf.dep -DFINE_GRAINED_LIBRARIES -DL_mu\
l_sf -c ../../.././gcc/fp-bit.c -fvisibility=hidden -DHIDE_EXPORTS
../../.././gcc/fp-bit.c: In function '__mulsf3':
../../.././gcc/fp-bit.c:962:1: internal compiler error: in
cgraph_estimate_size_after_inlining, at ipa-inline.c:208
Please submit a full bug report,


-- 

laurent at guerby dot net changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |laurent at guerby dot net
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2010-05-11 08:15:33
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[rguenth at gcc dot gnu dot org]

------- Comment #4 from rguenth at gcc dot gnu dot org  2010-05-11 08:43 -------
*** Bug 44070 has been marked as a duplicate of this bug. ***


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |joel at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.6.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[laurent at guerby dot net]

------- Comment #5 from laurent at guerby dot net  2010-05-11 08:47 -------
backtrace:

(gdb) bt
#0  internal_error (gmsgid=0x13b2f720 "in %s, at %s:%d") at
../../trunk/gcc/diagnostic.c:755
#1  0x10837bd4 in fancy_abort (file=0x13cac5e8 "../../trunk/gcc/ipa-inline.c",
line=208, function=0x13cac5c0 "cgraph_estimate_size_after_inlining") at
../../trunk/gcc/diagnostic.c:811
#2  0x12c6716c in cgraph_estimate_size_after_inlining (times=1, to=0x2adf2680,
what=0x2adf3560) at ../../trunk/gcc/ipa-inline.c:208
#3  0x12c6bf28 in cgraph_decide_inlining_incrementally (node=0x2adf2680,
mode=INLINE_SIZE_NORECURSIVE) at ../../trunk/gcc/ipa-inline.c:1587
#4  0x12c6c58c in cgraph_early_inlining () at ../../trunk/gcc/ipa-inline.c:1686
#5  0x111310a0 in execute_one_pass (pass=0x13eb8c70) at
../../trunk/gcc/passes.c:1590
#6  0x1113139c in execute_pass_list (pass=0x13eb8c70) at
../../trunk/gcc/passes.c:1645
#7  0x1112fca8 in do_per_function_toporder (callback=0x1113132c
<execute_pass_list>, data=0x13dcbf60) at ../../trunk/gcc/passes.c:1156
#8  0x111323e0 in execute_ipa_pass_list (pass=0x13dcbe48) at
../../trunk/gcc/passes.c:1930
#9  0x12be4e28 in ipa_passes () at ../../trunk/gcc/cgraphunit.c:1849
#10 0x12be50a8 in cgraph_optimize () at ../../trunk/gcc/cgraphunit.c:1918
#11 0x12be253c in cgraph_finalize_compilation_unit () at
../../trunk/gcc/cgraphunit.c:1166
#12 0x10126460 in c_write_global_declarations () at
../../trunk/gcc/c-decl.c:9573
#13 0x1165bb90 in compile_file () at ../../trunk/gcc/toplev.c:1065
#14 0x1165f14c in do_compile () at ../../trunk/gcc/toplev.c:2423
#15 0x1165f2ac in toplev_main (argc=68, argv=0x7fff6d44) at
../../trunk/gcc/toplev.c:2465
#16 0x103df340 in main (argc=68, argv=0x7fff6d44) at ../../trunk/gcc/main.c:35


-- 

laurent at guerby dot net changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.6.0                       |---


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[laurent at guerby dot net]

------- Comment #6 from laurent at guerby dot net  2010-05-11 08:50 -------
(gdb) l
203     static int
204     cgraph_estimate_size_after_inlining (int times, struct cgraph_node *to,
205                                          struct cgraph_node *what)
206     {
207       int size = (what->global.size - inline_summary
(what)->size_inlining_benefit) * times + to->global.size;
208       gcc_assert (size >= 0);
209       return size;
210     }
211     
212     /* Scale frequency of NODE edges by FREQ_SCALE and increase loop nest
(gdb) p size
$1 = -2
(gdb) p what->global.size
$2 = 1
(gdb) p times
$3 = 1
(gdb) p to->global.size
$4 = 0
(gdb) p inline_summary (what)->size_inlining_benefit
$5 = 3


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[hubicka at gcc dot gnu dot org]

------- Comment #7 from hubicka at gcc dot gnu dot org  2010-05-11 08:52 -------
Can you please check if always_inline is involved in your testcase and if to
node has disreagard_inline_limits set.

Honza


-- 

hubicka at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.6.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[laurent at guerby dot net]

------- Comment #8 from laurent at guerby dot net  2010-05-11 09:18 -------
(gdb) p *to
$4 = {decl = 0x2ad21800, callees = 0x2ac7e280, callers = 0x0, next =
0x2adf38e0, previous = 0x2adf2760, indirect_calls = 0x0, origin = 0x0, nested =
0x0, next_nested = 0x0, next_needed = 0x0, next_sibling_clone = 0x0, 
  prev_sibling_clone = 0x0, clones = 0x0, clone_of = 0x0, same_body = 0x0,
same_comdat_group = 0x0, call_site_hash = 0x0, aux = 0x0,
ipa_transforms_to_apply = 0x0, ref_list = {references = 0x0, refering = 0x0},
local = {
    lto_file_data = 0x0, inline_summary = {estimated_self_stack_size = 56,
self_size = 16, size_inlining_benefit = 4, self_time = 52,
time_inlining_benefit = 13}, local = 0, externally_visible = 1, finalized = 1,
inlinable = 1, 
    disregard_inline_limits = 0, redefined_extern_inline = 0, vtable_method =
0}, global = {estimated_stack_size = 56, stack_frame_offset = 0, inlined_to =
0x0, time = 0, size = 0, estimated_growth = -2147483648}, rtl = {
    preferred_incoming_stack_boundary = 0}, clone = {tree_map = 0x0,
args_to_skip = 0x0, combined_args_to_skip = 0x0}, thunk = {fixed_offset = 0,
virtual_value = 0, alias = 0x0, this_adjusting = 0 '\0', virtual_offset_p = 0
'\0', 
    thunk_p = 0 '\0'}, count = 0, uid = 7, order = 7, pid = 7, needed = 1,
address_taken = 0, abstract_and_needed = 0, reachable = 1,
reachable_from_other_partition = 0, lowered = 1, analyzed = 1,
in_other_partition = 0, process = 0, 
  alias = 0, finalized_by_frontend = 1, same_body_alias = 0, frequency =
NODE_FREQUENCY_NORMAL}


-- 

laurent at guerby dot net changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.6.0                       |---


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[laurent at guerby dot net]

------- Comment #9 from laurent at guerby dot net  2010-05-11 09:23 -------
For __always_inline the answer seems to be yes if I don't misinterpret the
source:

 res = _fpmul_parts (&a, &b, &tmp);

static inline __attribute__ ((__always_inline__)) const fp_number_type *        
_fpmul_parts ( fp_number_type *  a,                                             
               fp_number_type *  b,                                             
               fp_number_type * tmp)                                            
{


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[joel at gcc dot gnu dot org]

------- Comment #10 from joel at gcc dot gnu dot org  2010-05-11 11:50 -------
FWIW also seen on sparc-rtems, powerpc-rtems, and i386-rtems.

This did not happen building mips-rtems.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.6.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[hubicka at gcc dot gnu dot org]

------- Comment #11 from hubicka at gcc dot gnu dot org  2010-05-11 15:16 -------
Subject: Bug 44063

Author: hubicka
Date: Tue May 11 15:15:48 2010
New Revision: 159273

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=159273
Log:

        PR tree-optimize/44063
        * ipa-inline.c (cgraph_edge_badness): Move always inlines to top of
queue.
        (cgraph_decide_inlining_of_small_function): Skip check when disrgarding
        limits.
        (estimate_function_body_sizes): Compute sizes even when disregarding.
        * gcc.c-torture/compile/pr44063.c: New testcase.

Added:
    trunk/gcc/testsuite/gcc.c-torture/compile/pr44063.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/ipa-inline.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[hp at gcc dot gnu dot org]

------- Comment #12 from hp at gcc dot gnu dot org  2010-05-11 21:33 -------
The committed patch fixed the build issue (thanks), so closing.
If some of the issues in the follow-up comments remains, please clone.


-- 

hp at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063

[Bug tree-optimization/44063] [4.6 Regression]: build broken for libgcc cris-elf, ICE in cgraph_estimate_size_after_inlining, at ipa-inline

[hubicka at gcc dot gnu dot org]

------- Comment #13 from hubicka at gcc dot gnu dot org  2010-06-27 22:40 -------
Subject: Bug 44063

Author: hubicka
Date: Sun Jun 27 22:40:10 2010
New Revision: 161475

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=161475
Log:

        Backport inliner heruistic fixes from mainline
        2010-05-27  Jan Hubicka  <jh@suse.cz>

        * ipa-inline.c (cgraph_estimate_size_after_inlining): Make inline.
        (update_caller_keys): Return early if there are no callers;
        only update fibheap when decresing the key.
        (update_callee_keys): Avoid recursion.
        (decide_inlining_of_small_functions): When badness does not match;
        re-insert into fibheap.

        2010-05-11  Jan Hubicka  <jh@suse.cz>

        PR tree-optimize/44063
        * ipa-inline.c (cgraph_edge_badness): Move always inlines to top of
        queue.

        2010-04-13  Jan Hubicka  <jh@suse.cz>

        * ipa-inline.c (cgraph_mark_inline_edge): Avoid double accounting
        of optimized out static functions.
        (cgraph_edge_badness): Add DUMP parameter and dump reasons for the
        cost computation.  Also sanity check for overflows.
        (update_caller_keys): Update cgraph_edge_badness call; properly
        update fibheap and sanity check that it is up to date.
        (add_new_edges_to_heap): Update cgraph_edge_badness.
        (cgraph_decide_inlining_of_small_function): Likewise;
        add sanity checking that badness in heap is up to date;
        improve dumping of reason; Update badness of calls to the
        offline copy of function currently inlined; dump badness
        of functions not inlined because of unit growth limits.

Modified:
    branches/gcc-4_5-branch/gcc/ChangeLog
    branches/gcc-4_5-branch/gcc/ipa-inline.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44063