OpenSSL on C writing garbage values after EOF - c

Question

So I am writing an archiver program to also encrypt-decrypt files sing block cipher. When decrpting from a archive file, and I have checked, in situations where the remaining btes of the fle before EOF is smaller than the read buffer some garbage values are written to the file. I am using OpenSSL's EVP API with C language. Any ideas what is wrong? You can find the code for AES 256 bit decryption below.
int TARIM_decrypt_aes256(FILE* infile, FILE* outfile, unsigned char* key, unsigned char* iv, unsigned long long int fileLoc, unsigned long int fileSize)
{
if (infile == NULL || outfile == NULL || key == NULL || iv == NULL)
{
printf("(ERROR) decrypt_aes256: One/more parameters are NULL pointers\n");
return 1;
}
EVP_CIPHER_CTX* ctx = EVP_CIPHER_CTX_new();
if (ctx == NULL)
{
printf("(ERROR) decrypt_aes256: Failed to generate cipher context. OpenSSL: %s\n", ERR_error_string(ERR_get_error(), NULL));
return 1;
}
EVP_CIPHER* cipher = EVP_CIPHER_fetch(NULL, "AES-256-CBC", NULL);
if (cipher == NULL)
{
printf("(ERROR) decrypt_aes256: Failed to fetch cipher. OpenSSL: %s\n", ERR_error_string(ERR_get_error(), NULL));
EVP_CIPHER_CTX_free(ctx);
return 1;
}
// Explicitly Enable Padding
EVP_CIPHER_CTX_set_padding(ctx, 1);
int num_read = 0;
int block_size = EVP_CIPHER_block_size(cipher);
int inLen = 1024;
int outLen = inLen + block_size;
unsigned char inbuffer[inLen], outbuffer[outLen]; // Allow space for additional block in outbuffer
if (!EVP_CipherInit_ex2(ctx, cipher, NULL, NULL, 0, NULL))
{
printf("(ERROR) decrypt_aes256: Failed to Initiate Cipher. OpenSSL: %s\n", ERR_error_string(ERR_get_error(), NULL));
EVP_CIPHER_free(cipher);
EVP_CIPHER_CTX_free(ctx);
return 1;
}
// Assert size of key and iv
OPENSSL_assert(EVP_CIPHER_CTX_key_length(ctx) == 32);
OPENSSL_assert(EVP_CIPHER_CTX_iv_length(ctx) == 16);
// Set key and iv
if (!EVP_CipherInit_ex2(ctx, NULL, key, iv, 0, NULL))
{
printf("(ERROR) decrypt_aes256: Failed to set key and iv. OpenSSL: %s\n", ERR_error_string(ERR_get_error(), NULL));
EVP_CIPHER_free(cipher);
EVP_CIPHER_CTX_free(ctx);
return 1;
}
// Seek to File Location
fseek(infile, fileLoc, SEEK_SET);
// Loop until bytes read
unsigned long long int b_count = 0;
while (1)
{
num_read = fread(inbuffer, sizeof(unsigned char), inLen, infile);
b_count += num_read;
if (!EVP_CipherUpdate(ctx, outbuffer, &outLen, inbuffer, num_read))
{
printf("(ERROR) decrypt_aes256: Failed to pass bytes to the cipher. OpenSSL: %s\n", ERR_error_string(ERR_get_error(), NULL));
EVP_CIPHER_free(cipher);
EVP_CIPHER_CTX_free(ctx);
return 1;
}
fwrite(outbuffer, sizeof(unsigned char), outLen, outfile);
// EOF
if (b_count >= fileSize || num_read < inLen)
{ break; }
}
// Cipher Final block with padding
if (!EVP_CipherFinal_ex(ctx, outbuffer, &outLen))
{
printf("(ERROR) decrypt_aes256: Failed to pass bytes from final block to cipher. OpenSSL: %s\n", ERR_error_string(ERR_get_error(), NULL));
EVP_CIPHER_free(cipher);
EVP_CIPHER_CTX_free(ctx);
return 1;
}
fwrite(outbuffer, sizeof(unsigned char), outLen, outfile);
// Clean up
EVP_CIPHER_free(cipher);
EVP_CIPHER_CTX_free(ctx);
return 0;
}
I tried changing the inLen variable to block_size but that gives me OpenSSL's bad decrypt error.
int inLen = block_size;
The Error comes out like this dring decryption:
(ERROR) decrypt_aes256: Failed to pass bytes from final block to cipher. OpenSSL: error:1C800064:Provider routines::bad decrypt
[Edit]
I did find a weird behavior of my program. So the program can encrypt files together into a single file. The files are not padded to differentiate to save space, instead I keep the record of the file size and figure out the start byte position in the code using their sizes and extract them. The only files that give me trouble when using inLen = block_size were the files who were a multiple for 'block_size' and only when they were clustered in a single file. Here is the code that figures out the start byte of a file:
// Get Relative Location
unsigned long long int relativeLoc = 0;
for (long long int itr = 0; itr < option_num; itr++) {
if (block_size == 0 || fArray[itr].fsize == 0) {
relativeLoc += fArray[itr].fsize;
} else {
// Adjust for padding
relativeLoc += ((fArray[itr].fsize / block_size) + 1) * block_size;
}
}