https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126234
Bug ID: 126234
Summary: [17 Regression] gfortran reject SPEC CPU wrf and wrf_r
since
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Keywords: rejects-valid
Severity: normal
Priority: P3
Component: fortran
Assignee: unassigned at gcc dot gnu.org
Reporter: pheeck at gcc dot gnu.org
CC: jvdelisle at gcc dot gnu.org
Blocks: 26163
Target Milestone: ---
Host: x86_64-pc-linux-gnu
Target: x86_64-pc-linux-gnu
GCC rejects both the SPEC CPU 2006 benchmark wrf and the SPEC CPU 2017
benchmark wrf_r. For example wrf_r:
gfortran -c -I. -I./netcdf/include -I./inc -O0 -std=legacy
module_pbl_driver.fppized.f90
module_pbl_driver.fppized.f90.orig.f90:107:8:
107 | USE module_bl_qnsepbl09
| 1
Error: Name ‘a1’ at (1) is an ambiguous reference to ‘a1’ from module
‘module_bl_myjpbl’
module_pbl_driver.fppized.f90.orig.f90:107:8:
107 | USE module_bl_qnsepbl09
| 1
Error: Name ‘a2s’ at (1) is an ambiguous reference to ‘a2s’ from module
‘module_bl_myjpbl’
module_pbl_driver.fppized.f90.orig.f90:107:8:
107 | USE module_bl_qnsepbl09
| 1
Error: Name ‘a2x’ at (1) is an ambiguous reference to ‘a2x’ from module
‘module_bl_myjpbl’
module_pbl_driver.fppized.f90.orig.f90:107:8:
107 | USE module_bl_qnsepbl09
| 1
Error: Name ‘a3s’ at (1) is an ambiguous reference to ‘a3s’ from module
‘module_bl_myjpbl’
module_pbl_driver.fppized.f90.orig.f90:107:8:
107 | USE module_bl_qnsepbl09
| 1
Error: Name ‘a4s’ at (1) is an ambiguous reference to ‘a4s’ from module
‘module_bl_myjpbl’
module_pbl_driver.fppized.f90.orig.f90:107:8:
107 | USE module_bl_qnsepbl09
| 1
Error: Name ‘adnh’ at (1) is an ambiguous reference to ‘adnh’ from module
‘module_bl_myjpbl’
module_pbl_driver.fppized.f90.orig.f90:107:8:
107 | USE module_bl_qnsepbl09
| 1
and so on...
Bisected to
commit f935a7a9013d0748ba1f42dc532df3d278707bf6
Author: Jerry DeLisle <[email protected]>
AuthorDate: Thu Jul 9 19:24:49 2026 -0700
Commit: Jerry DeLisle <[email protected]>
CommitDate: Fri Jul 10 11:06:39 2026 -0700
fortran: [PR126170] Fix LTO ICE in copy_function_or_variable
I've noticed
commit 69a2daacba6353d1aab3118415b89afd479c4c76
Author: Jerry DeLisle <[email protected]>
AuthorDate: Sat Jul 11 10:35:29 2026 -0700
Commit: Jerry DeLisle <[email protected]>
CommitDate: Sat Jul 11 10:39:34 2026 -0700
fortran: [PR126210] Restrict host-assoc symbol skip to derived types
but this fix doesn't seem to help here.
I reduced module_pbl_driver.fppized.f90, but my experience with Fortran is very
limited so I wasn't able to reduce the other files required to reproduce this.
---- reduced module_pbl_driver.fppized.f90 ----
CONTAINS
SUBROUTINE pbl_driver(
ims ,scalar,tracer
)
USE module_state_description
USE module_bl_qnsepbl09
USE module_bl_acm
USE module_bl_myjurb
REAL, DIMENSION( ims) :: rho, dz8w
REAL, DIMENSION( ims) :: EXCH_H,EXCH_M
REAL ,DIMENSION(:) :: tracer
REAL ,DIMENSION(:) :: scalar
REAL ,DIMENSION(ims) :: scalar_tend
LOGICAL run_param
IF (run_param) THEN
DO ij = 1 , ENDDO
IF (bl_pbl_physics .EQ. QNSEPBLSCHEME ) THEN
CALL diff4d (DTdtbl,dz8w, scalar, .true.
,rho,EXCH_H,EXCH_M,SCALAR_TEND,num_scalar, param_first_scalar
,IDS,ide,jds,jde,kds,KDE,IMS,ime,jms,jme,kms,KME,its,ite,jts,jte,kts,KTE)
CALL diff4d (DTdtbl,dz8w, tracer, .false. ,rho,EXCH_H,exch_m
,SCALAR_TEND,num_tracer,
PARAM_FIRST_SCALAR,ids,IDE,jds,jde,kds,KDE,ims,ime,jms,jme,kms,KME,its,ite,jts,jte,kts,kte)
ENDIF
ENDDO
ENDIF
END
SUBROUTINE diff4d(DT,DZ,SCALAR,IS_SCALAR,RHO
,EXCH_H,EXCH_M ,SCALAR_TEND ,NUM_SCALAR, PARAM_FIRST_SCALAR
,IDS,IDE,JDS,JDE,KDS,KDE ,IMS,IME,JMS,JME,KMS,KME
,ITS,ITE,JTS,JTE,KTS,KTE )
INTEGER PARAM_FIRST_SCALAR
LOGICAL IS_SCALAR
REAL,DIMENSION(IMS) :: DZ
REAL,DIMENSION(KMS) :: SCALAR
REAL,DIMENSION(:) :: RHO
REAL,DIMENSION(IMS) :: EXCH_H
REAL,DIMENSION(IMS) :: EXCH_M
REAL,DIMENSION(JMS) :: SCALAR_TEND
END
END
---- ----
Referenced Bugs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26163
[Bug 26163] [meta-bug] missed optimization in SPEC (2026, 2k17, 2k and 2k6 and
95)
