https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112932
Bug ID: 112932
Summary: [14] RISC-V rv64gcv_zvl256b vector: Incorrect behavior
with nested loop array writing
Product: gcc
Version: 14.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: patrick at rivosinc dot com
Target Milestone: ---
Testcase:
int printf(char *, ...);
int a, j, n, b, c, o, d, g, h;
int e[8];
long f[8][6];
void l() {
o = -27;
for (; o; o++) {
*e = 1;
if (a >= n) {
d = 0;
for (; d <= 7; d++)
e[d] = c;
}
}
j = 0;
for (; j < 8; j++) {
g = 0;
for (; g < 2; g++) {
h = 1;
for (; h < 3; h++)
f[j][g * 2 + h] = 1;
}
}
unsigned long *m = &f[1][1];
*m = 0;
}
int main() {
l();
b = f[0][1];
printf("b: %d\n", b);
}
Commands:
rv64gc:
> /scratch/tc-testing/tc-dec-8-trunk/build-rv64gcv/bin/riscv64-unknown-linux-gnu-gcc
> -march=rv64gc -mabi=lp64d -O3 red.c -o rv64gc.out
> QEMU_CPU=rv64,vlen=256,v=true,vext_spec=v1.0
> /scratch/tc-testing/tc-dec-8-trunk/build-rv64gcv/bin/qemu-riscv64 rv64gc.out
f[0][1]: 1
f[0][2]: 1
f[0][3]: 1
f[0][4]: 1
f[1][1]: 2
f[1][2]: 1
f[1][3]: 1
f[1][4]: 1
f[2][1]: 1
f[2][2]: 1
f[2][3]: 1
f[2][4]: 1
f[3][1]: 1
f[3][2]: 1
f[3][3]: 1
f[3][4]: 1
f[4][1]: 1
f[4][2]: 1
f[4][3]: 1
f[4][4]: 1
f[5][1]: 1
f[5][2]: 1
f[5][3]: 1
f[5][4]: 1
f[6][1]: 1
f[6][2]: 1
f[6][3]: 1
f[6][4]: 1
f[7][1]: 1
f[7][2]: 1
f[7][3]: 1
f[7][4]: 1
rv64gcv_zvl256b:
> /scratch/tc-testing/tc-dec-8-trunk/build-rv64gcv/bin/riscv64-unknown-linux-gnu-gcc
> -march=rv64gcv_zvl256b -mabi=lp64d -O3 red.c -o rv64gcv.out
> QEMU_CPU=rv64,vlen=256,v=true,vext_spec=v1.0
> /scratch/tc-testing/tc-dec-8-trunk/build-rv64gcv/bin/qemu-riscv64 rv64gcv.out
f[0][1]: 0
f[0][2]: 0
f[0][3]: 0
f[0][4]: 0
f[1][1]: 2
f[1][2]: 1
f[1][3]: 1
f[1][4]: 1
f[2][1]: 0
f[2][2]: 0
f[2][3]: 0
f[2][4]: 0
f[3][1]: 0
f[3][2]: 0
f[3][3]: 0
f[3][4]: 0
f[4][1]: 0
f[4][2]: 0
f[4][3]: 0
f[4][4]: 0
f[5][1]: 0
f[5][2]: 0
f[5][3]: 0
f[5][4]: 0
f[6][1]: 0
f[6][2]: 0
f[6][3]: 0
f[6][4]: 0
f[7][1]: 0
f[7][2]: 0
f[7][3]: 0
f[7][4]: 0
This issue does not occur when compiled with vlenb=128 (rv64gcv).
Basic analysis:
The print loop is copied from the second set of loops in l().
All of these elements should equal 1 except for f[1][1] since it's set
to 2 via *m.
rv64gcv incorrectly reports that some of these elements are set to 0.
