https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61929
Bug ID: 61929
Summary: Graphite: Banerjee and Omega differ on distance
vectors (internal compiler error in
compute_affine_dependence)
Product: gcc
Version: unknown
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: gcc at crg4 dot com
-fcheck-data-deps finds a discrepancy between Banerjee and Omega in gcc 4.9
when compiling the latest RIES source. The error message is:
(Number of distance vectors differ: Banerjee has 1, Omega has 2.
Banerjee dist vectors:
0
Omega dist vectors:
0
1
data dependence relation:
(Data Dep:
#(Data Ref:
# bb: 44
# stmt: debug_L.555_92 = debug_L;
# ref: debug_L
# base_object: MEM[(short int *)&debug_L];
#)
#(Data Ref:
# bb: 44
# stmt: debug_L = debug_L.568_93;
# ref: debug_L
# base_object: MEM[(short int *)&debug_L];
#)
inner loop index: 0
loop nest: (1 )
distance_vector: 0
distance_vector: 1
direction_vector: =
direction_vector: +
)
)
ries.c: In function ‘set_debug_opts’:
ries.c:10391:6: internal compiler error: in compute_affine_dependence, at
tree-data-ref.c:4233
void set_debug_opts(char * str)
^
Please submit a full bug report,
with preprocessed source if appropriate.
----
The RIES source is available at
http://www.mrob.com/pub/ries/src/ries.c.txt
and is self-contained (no .h files etc.).
I can cause the error with
gcc -o ries ries.c -lm -O1 -fcheck-data-deps
but not
gcc -o ries ries.c -lm -O0 -fcheck-data-deps
however I cannot narrow this down to a particular optimization causing the
difficulty since
gcc -o ries ries.c -lm -O1 -fcheck-data-deps -fno-branch-count-reg
-fno-combine-stack-adjustments -fno-compare-elim -fno-cprop-registers
-fno-defer-pop -fno-forward-propagate -fno-guess-branch-probability
-fno-if-conversion -fno-if-conversion2 -fno-inline-functions-called-once
-fno-ipa-profile -fno-ipa-pure-const -fno-ipa-reference -fno-merge-constants
-fno-move-loop-invariants -fno-shrink-wrap -fno-split-wide-types
-fno-tree-bit-ccp -fno-tree-ccp -fno-tree-ch -fno-tree-copy-prop
-fno-tree-copyrename -fno-tree-dce -fno-tree-dominator-opts -fno-tree-dse
-fno-tree-fre -fno-tree-pta -fno-tree-sink -fno-tree-slsr -fno-tree-sra
-fno-tree-ter
still causes the error. It appears that at least one of the Omega and Banerjee
algorithms are incorrect.
I am running
> gcc -v
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-linux-gnu/4.9/lto-wrapper
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Ubuntu
4.9.1-1ubuntu2~14.04.3' --with-bugurl=file:///usr/share/doc/gcc-4.9/README.Bugs
--enable-languages=c,c++,java,go,d,fortran,objc,obj-c++ --prefix=/usr
--program-suffix=-4.9 --enable-shared --enable-linker-build-id
--libexecdir=/usr/lib --without-included-gettext --enable-threads=posix
--with-gxx-include-dir=/usr/include/c++/4.9 --libdir=/usr/lib --enable-nls
--with-sysroot=/ --enable-clocale=gnu --enable-libstdcxx-debug
--enable-libstdcxx-time=yes --enable-gnu-unique-object --disable-libmudflap
--disable-vtable-verify --enable-plugin --with-system-zlib
--disable-browser-plugin --enable-java-awt=gtk --enable-gtk-cairo
--with-java-home=/usr/lib/jvm/java-1.5.0-gcj-4.9-amd64/jre --enable-java-home
--with-jvm-root-dir=/usr/lib/jvm/java-1.5.0-gcj-4.9-amd64
--with-jvm-jar-dir=/usr/lib/jvm-exports/java-1.5.0-gcj-4.9-amd64
--with-arch-directory=amd64 --with-ecj-jar=/usr/share/java/eclipse-ecj.jar
--enable-objc-gc --enable-multiarch --disable-werror --with-arch-32=i686
--with-abi=m64 --with-multilib-list=m32,m64,mx32 --enable-multilib
--with-tune=generic --enable-checking=release --build=x86_64-linux-gnu
--host=x86_64-linux-gnu --target=x86_64-linux-gnu
Thread model: posix
gcc version 4.9.1 (Ubuntu 4.9.1-1ubuntu2~14.04.3)
> uname -a
Linux zed 3.2.0-56-generic #86-Ubuntu SMP Wed Oct 23 09:20:45 UTC 2013 x86_64
x86_64 x86_64 GNU/Linux
I have no knowledge of either Graphite or RIES; I am merely reporting what
appears to be a bug (indeed, as instructed by gcc itself).
