Instruction Combining did not reach a fixpoint

[KatCe]

Hello,
I get the error below with the following synthesis commands and program below:

commands:

set-src ../myfunc.c
set-clock-period 23
compile   --buffer-algorithm fpl22
write-hdl --hdl verilog

Dynamatic version used: 8074f73

error:

LLVM ERROR: Instruction Combining did not reach a fixpoint after 1 iterations
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace.
Stack dump:
0.	Program arguments: /data//workspace/dynamatic/bin/opt -S -passes=inline,mem2reg,consthoist,instcombine,function(loop-mssa(licm<no-allowspeculation>)),function(loop(loop-idiom,indvars,loop-deletion)),simplifycfg,loop-rotate,simplifycfg,sink,lowerswitch,simplifycfg,dce /home//myfunc/synth/../out/comp/clang.ll

Program (compiles with gcc for x86):

// var_0_secret

// Modified from WolfSSL

typedef unsigned char byte;
typedef unsigned short word16;

#define F25519_SIZE 32

// void fe_normalize(byte *x)
int myfunc (byte var_0_secret)
{
    byte minusp[F25519_SIZE];
    word16 c;
    int i;

    // The original argument, filling it with some random secret-dependent data
    unsigned char x[32];
    for (i = 0; i < 32; i++) {
        x[i] = (unsigned char)(i + var_0_secret);
    }

    // The original function

    /* Reduce using 2^255 = 19 mod p */
    c = (x[31] >> 7) * 19;
    x[31] &= 127;

    for (i = 0; i < F25519_SIZE; i++) {
        c += x[i];
        x[i] = (byte)c;
        c >>= 8;
    }

    /* The number is now less than 2^255 + 18, and therefore less than
     * 2p. Try subtracting p, and conditionally load the subtracted
     * value if underflow did not occur.
     */
    c = 19;

    for (i = 0; i + 1 < F25519_SIZE; i++) {
        c += x[i];
        minusp[i] = (byte)c;
        c >>= 8;
    }

    c += ((word16)x[i]) - 128;
    minusp[31] = (byte)c;

    /* Load x-p if no underflow */
    // fe_select(x, minusp, x, (c >> 15) & 1);
    // inlining it:
    const byte mask = -(c >> 15) & 1;

    for (i = 0; i < F25519_SIZE; i++)
        x[i] = minusp[i] ^ (mask & (x[i] ^ minusp[i]));

    // Some dummy code to make sure the result is not optimized away
    int result = 0;
    for (i = 0; i < 32; i++)
        result += x[i];
    return result;
}


#include <stdlib.h>
#include "dynamatic/Integration.h"
int main(void) {
    unsigned char fn_arg_0_secret = rand() % 8;

    CALL_KERNEL(myfunc,fn_arg_0_secret);
    return 0;
}