https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93430
Bug ID: 93430
Summary: [10 Regression] ICE in final_scan_insn_1, at
final.c:3073
Product: gcc
Version: 9.0
Status: UNCONFIRMED
Keywords: ice-on-valid-code
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: marxin at gcc dot gnu.org
CC: jakub at gcc dot gnu.org
Target Milestone: ---
I see ICE in the following code snippet reduced from a polyhedron benchmark:
$ cat doduc.f90
IMPLICIT DOUBLE PRECISION(A-Z)
COMMON V0012 , DIFref0 , DIFmic0 , V00001 ,DTPre , DTPhy , DTTemp ,
DTChan , DTDis
DTTemp = V0012
DTDis = V0012
DTChan = V0012
DTPhy = V0012
sens = V00001
IF ( DABS0.GT.eps ) dtp = sens
DTPre = dtp
DATA a00 , a01 , a02 , a03 , a04 , a05 , a06 , a07 , a08 , a09 , a010 ,
a011 , a012/1.0421485616135 , &
3.7989819258871 , 1.0809520514321 , 1.7199071099232 ,
1.7003453661702 , &
1.1093079504746 , 4.9305639341012 , 3.1127728952968 ,
4.9182825188505 , &
4.4404482566015 , 1.5665047149903 , 5.6075256366923 ,
4.6890466767990/
END
FUNCTION S00093()
DOUBLE PRECISION tabpwl(32) , tabb(32)
DATA tabpwl/0.000 , 0.007 , 0.010 , 0.012 , 0.015 , 0.020 , &
0.025 , 0.030 , 0.035 , 0.040 , 0.050 , 0.060 , 0.070 , 0.080 , 0.100
, 0.120 , 0.150 , 0.200 , 0.250 , 0.300 , &
0.000452 , 0.000447 , 0.000416 , 0.000392 , 0.000384 , 0.000371 ,
0.000354 , 0.000335 , 0.000322 , 0.000307 , &
0.000287 , 0.000275/
DATA tabb/0.029 , 0.037 , 0.047 , 0.056 , 0.067 , 0.080 , 0.090 , &
0.096 , 0.108 , 0.120 , 0.132 , 0.144 , 0.156 , 0.168 , 0.184 , 0.200
, 0.220 , 0.245 , 0.270 , 0.295 , 0.325 , &
0.355 , 0.375 , 0.410 , 0.450 , 0.480 , 0.510 , 0.550 , 0.620 , 0.700
, 0.800 , 0.910/
DATA d0 , d1 , d2 , d3 , d4 , d5/ 501.911804844 , 643.013652461 ,
254.053681062 , 53.4172368776 , &
5.47296691662 , .227574393371/
DATA e0 , e1 , e2 , e3 , e4 , e5 , e6/ 983.676282822 , 2649.64781997 ,
1365.75884763 , 373.419845898 , &
181216.034989 , 23651.1207875 , 1029.9067531/
END
$ gcc -O3 -march=znver2 doduc.f90 -c
doduc.f90:14:0:
14 | END
|
Error: could not split insn
(insn:TI 17 37 19 2 (set (reg:V4DF 20 xmm0 [orig:92 vect__6.9 ] [92])
(vec_select:V4DF (mem/c:V4DF (symbol_ref:DI ("__BLNK__") [flags 0x2]
<var_decl 0x7ffff7fc7ab0 __BLNK__>) [1 MEM <vector(4) real(kind=8)>
[(real(kind=8) *)&__BLNK__]+0 S32 A256])
(parallel [
(const_int 0 [0]) repeated x4
]))) "doduc.f90":1:0 4926 {*avx_vperm_broadcast_v4df}
(expr_list:REG_EQUIV (mem/c:V4DF (const:DI (plus:DI (symbol_ref:DI
("__BLNK__") [flags 0x2] <var_decl 0x7ffff7fc7ab0 __BLNK__>)
(const_int 40 [0x28]))) [1 MEM <vector(4) real(kind=8)>
[(real(kind=8) *)&__BLNK__ + 40B]+0 S32 A64])
(nil)))
during RTL pass: final
doduc.f90:14:0: internal compiler error: in final_scan_insn_1, at final.c:3073
0x6d6597 _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
/home/marxin/Programming/gcc/gcc/rtl-error.c:108
0x659456 final_scan_insn_1
/home/marxin/Programming/gcc/gcc/final.c:3073
0xb3f1db final_scan_insn(rtx_insn*, _IO_FILE*, int, int, int*)
/home/marxin/Programming/gcc/gcc/final.c:3152
0xb3f4a9 final_1
/home/marxin/Programming/gcc/gcc/final.c:2020
0xb3fe04 rest_of_handle_final
/home/marxin/Programming/gcc/gcc/final.c:4658
0xb3fe04 execute
/home/marxin/Programming/gcc/gcc/final.c:4736
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.
Started with r10-6215-g5d782a8d909c5cc472c911c0ab4de0b890aad868
