This error was solved following your method, thank your very much for your kind answers.

Dear developers，

Thank your very much for your valuable time and quick reply.
The first iteration is finished, but the charge density is too big and seems unreasonable.

l_ss=F,l_mperp=F,l_constr=F,l_disp=F,sso_opt=FFT
----> distance of charge densities for spin 1 it= 1: 115.425684 me/bohr**3
----> distance of charge densities for spin 2 it= 1: 95.852274 me/bohr**3
----> distance of charge densities for spin 3 it= 1: 0.828353 me/bohr**3
----> distance of charge densities for it= 1: 205.064444 me/bohr**3
----> distance of spin densities for it= 1: 54.498328 me/bohr**3

Then the error accurred at the second step.

Best wishes
xdzhou

Dear developers,

I checked the internal -> cartesian of real-space (A) and reciprocal space (B). The A in out file should be the transpose of true real space, while the B is the true reciprocal space.
I read the source code of cotra.f. The internal -> cartesian is wrote as :

The real space:
! **********************************************************
SUBROUTINE cotra0(r,x,amat)
! **********************************************************
! ..
! .. Array Arguments ..
REAL, INTENT (IN) :: r(3)
REAL, INTENT (IN) :: amat(3,3)
REAL, INTENT (OUT):: x(3)
!
DO i = 1,3
s = 0.0
DO j = 1,3
s = s + amat(i,j)*r(j)
ENDDO
x(i) = s
ENDDO

RETURN
END SUBROUTINE cotra0

! ***********************************************************


The reciprocal space :
!*********************************************************************
SUBROUTINE cotra3(f,g,bmat)
!*********************************************************************
! ..
! .. Array Arguments ..
REAL, INTENT (IN) :: f(3)
REAL, INTENT (IN) :: bmat(3,3)
REAL, INTENT (OUT):: g(3)
!
DO j = 1,3
g(j) = 0.0
DO i = 1,3
g(j) = g(j) + f(i)*bmat(i,j)
ENDDO
ENDDO

RETURN
END SUBROUTINE cotra3


Therefore, for the vector of spiral, It should be q*B.

For my case above, the internal q should be (1,1,1), but not (0.0000000000 -0.8469595000 1.4264960700) which is a strange vector.

Thank you for your valuable time and giving me useful guidance.

Dear developers,

I want to know, Is the direction of spiral B*q or q*B ？
For fcc, B*q or q*B is the same, while they are different for some lattice.
For example, a rhombohedral primitive cell.

If I input the structure as follows,
##############################################
2.4387895018949948 1.4080357754159085 4.7429872621175369
-2.4387895018949948 1.4080357754159085 4.7429872621175369
0.0000000000000000 -2.8160715508318170 4.7429872621175369
1.889726125
1 1 1
###############################################

The out file will give bravais matrices of real and reciprocal lattices,
4.608644 -4.608644 0.000000 ---- 0.681674 0.393565 0.233673
2.660802 2.660802 -5.321604 ---- -0.681674 0.393565 0.233673
8.962947 8.962947 8.962947 ---- 0.000000 -0.787129 0.233673

Then if spiral direction is along the z direction
For B*q, the q will be (0.0000000000 -0.8469595000 1.4264960700)
That is
( 0.681674 0.393565 0.233673) ---- (0.0000000000) ---- (0)
(-0.681674 0.393565 0.233673) ---- (-0.8469595000) = (0)
(0.000000 -0.787129 0.233673) ---- (1.4264960700) ---- (z)


For q*B, the q will be (1,1,1).
-------------( 0.681674 0.393565 0.233673)
----(1 1 1) (-0.681674 0.393565 0.233673)--------- =(0 0 z)
------------(0.000000 -0.787129 0.233673)