By the end of this lesson, you will be able to:
- Explain the importance of "thinking like an attacker"
- Understand basic security practices
- Perform owner checks
- Perform signer checks
- Validate accounts passed into the program
- Perform basic data validation
- Thinking like an attacker means asking "How do I break this?"
- Perform owner checks to ensure that the provided account is owned by the public key you expect, e.g. ensuring that an account you expect to be a PDA is owned by
program_id - Perform signer checks to ensure that any account modification has been signed by the right party or parties
- Account validation entails ensuring that provided accounts are the accounts you expect them to be, e.g. deriving PDAs with the expected seeds to make sure the address matches the provided account
- Data validation entails ensuring that any provided data meets the criteria required by the program
In the last two lessons we worked through building a Movie Review program together. The end result is pretty cool! It's exciting to get something working in a new development environment.
Proper program development, however, doesn't end at "get it working." It's important to think through the possible failure points in your code in order to mitigate them. Failure points are where undesirable behavior in your code could potentially occur. Whether the undesirable behavior happens due to users interacting with your program in unexpected ways or bad actors intentionally trying to exploit your program, anticipating failure points is essential to secure program development.
Remember, you have no control over the transactions that will be sent to your program once it’s deployed. You can only control how your program handles them. While this lesson is far from a comprehensive overview of program security, we'll cover some of the basic pitfalls to look out for.
Neodyme gave a presentation at Breakpoint 2021 entitled "Think Like An Attacker: Bringing Smart Contracts to Their Break(ing) Point." If there's one thing you take away from this lesson, it's that you should think like an attacker.
In this lesson, of course, we cannot cover everything that could possibly go wrong with your programs. Ultimately, every program will have different security risks associated with it. While understanding common pitfalls is essential to engineering good programs, it is insufficient for deploying secure ones. In order to have the broadest security coverage possible, you have to approach your code with the right mindset.
As Neodyme mentioned in their presentation, the right mindset requires moving from the question "Is this broken?" to "How do I break this?" This is the first and most essential step in understanding what your code actually does as opposed to what you wrote it to do.
It's not a question of "if."
Rather, it's a question of "how much effort and dedication would it take."
Our job as developers is to close as many holes as possible and increase the effort and dedication required to break our code. For example, in the Movie Review program we built together over the last two lessons, we wrote code to create new accounts to store movie reviews. If we take a closer look at the code, however, we'll notice how the program also facilitates a lot of unintentional behavior we could easily catch by asking "How do I break this?" We'll dig into some of these problems and how to fix them in this lesson, but remember that memorizing a few pitfalls isn't sufficient. It's up to you to change your mindset toward security.
Before we dive into some of the common security pitfalls and how to avoid them, it's important to know how to use errors in your program. While your code can handle some issues gracefully, other issues will require that your program stop execution and return a program error.
While the solana_program crate provides a ProgramError enum with a list of generic errors we can use, it will often be useful to create your own. Your custom errors will be able to provide more context and detail while you're debugging your code.
We can define our own errors by creating an enum type listing the errors we want to use. For example, the NoteError contains variants Forbidden and InvalidLength. The enum is made into a Rust Error type by using the derive attribute macro to implement the Error trait from the thiserror library. Each error type also has its own #[error("...")] notation. This lets you provide an error message for each particular error type.
use solana_program::{program_error::ProgramError};
use thiserror::Error;
#[derive(Error)]
pub enum NoteError {
#[error("Wrong note owner")]
Forbidden,
#[error("Text is too long")]
InvalidLength,
}The compiler expects errors returned by the program to be of type ProgramError from the solana_program crate. That means we won't be able to return our custom error unless we have a way to convert it into this type. The following implementation handles conversion between our custom error and the ProgramError type.
impl From<NoteError> for ProgramError {
fn from(e: NoteError) -> Self {
ProgramError::Custom(e as u32)
}
}To return the custom error from the program, simply use the into() method to convert the error into an instance of ProgramError.
if pda != *note_pda.key {
return Err(NoteError::Forbidden.into());
}While these won't comprehensively secure your program, there are a few security checks you can keep in mind to fill in some of the larger gaps in your code:
- Ownership checks - used to verify that an account is owned by the program
- Signer checks - used to verify that an account has signed a transaction
- General Account Validation - used to verify that an account is the expected account
- Data Validation - used to verify the inputs provided by a user
An ownership check verifies that an account is owned by the expected public key. Let's use the note-taking app example that we've referenced in previous lessons. In this app, users can create, update, and delete notes that are stored by the program in PDA accounts.
When a user invokes the update instruction, they also provide a pda_account. We presume the provided pda_account is for the particular movie review they want to update, but the user can input any instruction data they want. They could even potentially send data which matches the data format of a note account but was not also created by the note-taking program. This security vulnerability is one potential way to introduce malicious code.
The simplest way to avoid this problem is to always check that the owner of an account is the public key you expect it to be. In this case, we expect the note account to be a PDA account owned by the program itself. When this is not the case, we can report it as an error accordingly.
if note_pda.owner != program_id {
return Err(ProgramError::InvalidNoteAccount);
}As a side note, using PDAs whenever possible is more secure than trusting externally-owned accounts, even if they are owned by the transaction signer. The only accounts that the program has complete control over are PDA accounts, making them the most secure.
A signer check simply verifies that the right parties have signed a transaction. In the note-taking app, for example, we would want to verify that the note creator signed the transaction before we process the update instruction. Otherwise, anyone can update another user's notes by simply passing in the user's public key as the initializer.
if !initializer.is_signer {
msg!("Missing required signature");
return Err(ProgramError::MissingRequiredSignature)
}In addition to checking the signers and owners of accounts, it's important to ensure that the provided accounts are what your code expects them to be. For example, you would want to validate that a provided PDA account's address can be derived with the expected seeds. This ensures that it is the account you expect it to be.
In the note-taking app example, that would mean ensuring that you can derive a matching PDA using the initializer and title as seeds (that's what we're assuming was used when creating the note). That way a user couldn't accidentally pass in a PDA account for the wrong note or, more importantly, that the user isn't passing in a PDA account that represents somebody else's note entirely.
let (pda, bump_seed) = Pubkey::find_program_address(&[initializer.key.as_ref(), title.as_bytes().as_ref(),], program_id);
if pda != *note_pda.key {
msg!("Invalid seeds for PDA");
return Err(ProgramError::InvalidArgument)
}Similar to validating accounts, you should also validate any data provided by the client.
For example, you may have a game program where a user can allocate character attribute points to various categories. You may have a maximum limit in each category of 100, in which case you would want to verify that the existing allocation of points plus the new allocation doesn't exceed the maximum.
if character.agility + new_agility > 100 {
msg!("Attribute points cannot exceed 100");
return Err(AttributeError::TooHigh.into())
}Or, the character may have an allowance of attribute points they can allocate and you want to make sure they don't exceed that allowance.
if attribute_allowance > new_agility {
msg!("Trying to allocate more points than allowed");
return Err(AttributeError::ExceedsAllowance.into())
}Without these checks, program behavior would differ from what you expect. In some cases, however, it's more than just an issue of undefined behavior. Sometimes failure to validate data can result in security loopholes that are financially devastating.
For example, imagine that the character referenced in these examples is an NFT. Further, imagine that the program allows the NFT to be staked to earn token rewards proportional to the NFTs number of attribute points. Failure to implement these data validation checks would allow a bad actor to assign an obscenely high number of attribute points and quickly drain your treasury of all the rewards that were meant to be spread more evenly amongst a larger pool of stakers.
Rust integers have fixed sizes. This means they can only support a specific range of numbers. An arithmetic operation that results in a higher or lower value than what is supported by the range will cause the resulting value to wrap around. For example, a u8 only supports numbers 0-255, so the result of addition that would be 256 would actually be 0, 257 would be 1, etc.
This is always important to keep in mind, but especially so when dealing with any code that represents true value, such as depositing and withdrawing tokens.
To avoid integer overflow and underflow, either:
- Have logic in place that ensures overflow or underflow cannot happen or
- Use checked math like
checked_addinstead of+let first_int: u8 = 5; let second_int: u8 = 255; let sum = first_int.checked_add(second_int);
Let’s practice together with the Movie Review program we've worked on in previous lessons. No worries if you’re just jumping into this lesson without having done the previous lesson - it should be possible to follow along either way.
As a refresher, the Movie Review program lets users store movie reviews in PDA accounts. Last lesson, we finished implementing the basic functionality of adding a movie review. Now, we'll add some security checks to the functionality we've already created and add the ability to update a movie review in a secure manner.
Just as before, we'll be using Solana Playground to write, build, and deploy our code.
To begin, you can find the starter code here. If you've been following along with the Movie Review demos, you'll notice that we've refactored our program.
The refactored starter code is almost the same as what it was before. Since lib.rs was getting rather large and unwieldy, we've separated its code into 3 files: lib.rs, entrypoint.rs, and processor.rs. lib.rs now only registers the code's modules, entrypoint.rs only defines and sets the program's entrypoint, and processor.rs handles the program logic for processing instructions. We've also added an error.rs file where we'll be defining custom errors. The complete file structure is as follows:
- lib.rs - register modules
- entrypoint.rs - entry point to the program
- instruction.rs - serialize and deserialize instruction data
- processor.rs - program logic to process instructions
- state.rs - serialize and deserialize state
- error.rs - custom program errors
In addition to some changes to file structure, we've updated a small amount of code that will let this demo be more focused on security without having you write unnecessary boiler plate.
Since we'll be allowing updates to movie reviews, we also changed account_len in the add_movie_review function (now in processor.rs). Instead of calculating the size of the review and setting the account length to only as large as it needs to be, we're simply going to allocate 1000 bytes to each review account. This way, we don’t have to worry about reallocating size or re-calculating rent when a user updates their movie review.
We went from this:
let account_len: usize = 1 + 1 + (4 + title.len()) + (4 + description.len());To this:
let account_len: usize = 1000;The realloc method was just recently enabled by Solana Labs which allows you to dynamically change the size of your accounts. We will not be using this method for this demo, but it’s something to be aware of.
Finally, we've also implemented some additional functionality for our MovieAccountState struct in state.rs using the impl keyword.
For our movie reviews, we want the ability to check whether an account has already been initialized. To do this, we create an is_initialized function that checks the is_initialized field on the MovieAccountState struct.
Sealed is Solana's version of Rust's Sized trait. This simply specifies that MovieAccountState has a known size and provides for some compiler optimizations.
// inside state.rs
impl Sealed for MovieAccountState {}
impl IsInitialized for MovieAccountState {
fn is_initialized(&self) -> bool {
self.is_initialized
}
}Before moving on, make sure you have a solid grasp on the current state of the program. Look through the code and spend some time thinking through any spots that are confusing to you. It may be helpful to compare the starter code to the solution code from the previous lesson.
Let's begin by writing our custom program errors. We'll need errors that we can use in the following situations:
- The update instruction has been invoked on an account that hasn't been initialized yet
- The provided PDA doesn't match the expected or derived PDA
- The input data is larger than the program allows
- The rating provided does not fall in the 1-5 range
The starter code includes an empty error.rs file. Open that file and add errors for each of the above cases.
// inside error.rs
use solana_program::{program_error::ProgramError};
use thiserror::Error;
#[derive(Debug, Error)]
pub enum ReviewError{
// Error 0
#[error("Account not initialized yet")]
UninitializedAccount,
// Error 1
#[error("PDA derived does not equal PDA passed in")]
InvalidPDA,
// Error 2
#[error("Input data exceeds max length")]
InvalidDataLength,
// Error 3
#[error("Rating greater than 5 or less than 1")]
InvalidRating,
}
impl From<ReviewError> for ProgramError {
fn from(e: ReviewError) -> Self {
ProgramError::Custom(e as u32)
}
}Note that in addition to adding the error cases, we also added the implementation that lets us convert our error into a ProgramError type as needed.
Before moving on, let’s bring ReviewError into scope in the processor.rs. We will be using these errors shortly when we add our security checks.
// inside processor.rs
use crate::error::ReviewError;Now that we have errors to use, let's implement some security checks to our add_movie_review function.
The first thing we should do is ensure that the initializer of a review is also a signer on the transaction. This ensures that you can't submit movie reviews impersonating somebody else. We'll put this check right after iterating through the accounts.
let account_info_iter = &mut accounts.iter();
let initializer = next_account_info(account_info_iter)?;
let pda_account = next_account_info(account_info_iter)?;
let system_program = next_account_info(account_info_iter)?;
if !initializer.is_signer {
msg!("Missing required signature");
return Err(ProgramError::MissingRequiredSignature)
}Next, let's make sure the pda_account passed in by the user is the pda we expect. Recall we derived the pda for a movie review using the initializer and title as seeds. Within our instruction we’ll derive the pda again and then check if it matches the pda_account. If the addresses do not match, we’ll return our custom InvalidPDA error.
// Derive PDA and check that it matches client
let (pda, _bump_seed) = Pubkey::find_program_address(&[initializer.key.as_ref(), account_data.title.as_bytes().as_ref(),], program_id);
if pda != *pda_account.key {
msg!("Invalid seeds for PDA");
return Err(ReviewError::InvalidPDA.into())
}Now let's perform some data validation.
We'll start by making sure rating falls within the 1 to 5 scale. If the rating provided by the user outside of this range, we’ll return our custom InvalidRating error.
if rating > 5 || rating < 1 {
msg!("Rating cannot be higher than 5");
return Err(ReviewError::InvalidRating.into())
}Next, let’s check that the content of the review does not exceed the 1000 bytes we’ve allocated for the account. If the size exceeds 1000 bytes, we’ll return our custom InvalidDataLength error.
let total_len: usize = 1 + 1 + (4 + title.len()) + (4 + description.len())
if total_len > 1000 {
msg!("Data length is larger than 1000 bytes");
return Err(ReviewError::InvalidDataLength.into())
}Lastly, let's checking if the account has already been initialized by calling the is_initialized function we implemented for our MovieAccountState. If the account already exists, then we will return an error.
if account_data.is_initialized() {
msg!("Account already initialized");
return Err(ProgramError::AccountAlreadyInitialized);
}All together, the add_movie_review function should look something like this:
pub fn add_movie_review(
program_id: &Pubkey,
accounts: &[AccountInfo],
title: String,
rating: u8,
description: String
) -> ProgramResult {
msg!("Adding movie review...");
msg!("Title: {}", title);
msg!("Rating: {}", rating);
msg!("Description: {}", description);
let account_info_iter = &mut accounts.iter();
let initializer = next_account_info(account_info_iter)?;
let pda_account = next_account_info(account_info_iter)?;
let system_program = next_account_info(account_info_iter)?;
if !initializer.is_signer {
msg!("Missing required signature");
return Err(ProgramError::MissingRequiredSignature)
}
let (pda, bump_seed) = Pubkey::find_program_address(&[initializer.key.as_ref(), title.as_bytes().as_ref(),], program_id);
if pda != *pda_account.key {
msg!("Invalid seeds for PDA");
return Err(ProgramError::InvalidArgument)
}
if rating > 5 || rating < 1 {
msg!("Rating cannot be higher than 5");
return Err(ReviewError::InvalidRating.into())
}
let total_len: usize = 1 + 1 + (4 + title.len()) + (4 + description.len());
if total_len > 1000 {
msg!("Data length is larger than 1000 bytes");
return Err(ReviewError::InvalidDataLength.into())
}
let account_len: usize = 1000;
let rent = Rent::get()?;
let rent_lamports = rent.minimum_balance(account_len);
invoke_signed(
&system_instruction::create_account(
initializer.key,
pda_account.key,
rent_lamports,
account_len.try_into().unwrap(),
program_id,
),
&[initializer.clone(), pda_account.clone(), system_program.clone()],
&[&[initializer.key.as_ref(), title.as_bytes().as_ref(), &[bump_seed]]],
)?;
msg!("PDA created: {}", pda);
msg!("unpacking state account");
let mut account_data = try_from_slice_unchecked::<MovieAccountState>(&pda_account.data.borrow()).unwrap();
msg!("borrowed account data");
msg!("checking if movie account is already initialized");
if account_data.is_initialized() {
msg!("Account already initialized");
return Err(ProgramError::AccountAlreadyInitialized);
}
account_data.title = title;
account_data.rating = rating;
account_data.description = description;
account_data.is_initialized = true;
msg!("serializing account");
account_data.serialize(&mut &mut pda_account.data.borrow_mut()[..])?;
msg!("state account serialized");
Ok(())
}Now that add_movie_review is more secure, let's turn our attention to supporting the ability to update a movie review.
Let’s begin by updating instruction.rs. We’ll start by adding an UpdateMovieReview variant to MovieInstruction that includes embedded data for the new title, rating, and description.
// inside instruction.rs
pub enum MovieInstruction {
AddMovieReview {
title: String,
rating: u8,
description: String
},
UpdateMovieReview {
title: String,
rating: u8,
description: String
}
}The payload struct can stay the same since aside from the variant type, the instruction data is the same as what we used for AddMovieReview.
Lastly, in the unpack function we need to add UpdateMovieReview to the match statement.
// inside instruction.rs
impl MovieInstruction {
pub fn unpack(input: &[u8]) -> Result<Self, ProgramError> {
let (&variant, rest) = input.split_first().ok_or(ProgramError::InvalidInstructionData)?;
let payload = MovieReviewPayload::try_from_slice(rest).unwrap();
Ok(match variant {
0 => Self::AddMovieReview {
title: payload.title,
rating: payload.rating,
description: payload.description },
1 => Self::UpdateMovieReview {
title: payload.title,
rating: payload.rating,
description: payload.description },
_ => return Err(ProgramError::InvalidInstructionData)
})
}
}Now that we can unpack our instruction_data and determine which instruction of the program to run, we can add UpdateMovieReview to the match statement in the process_instruction function in the processor.rs file.
// inside processor.rs
pub fn process_instruction(
program_id: &Pubkey,
accounts: &[AccountInfo],
instruction_data: &[u8]
) -> ProgramResult {
// unpack instruction data
let instruction = MovieInstruction::unpack(instruction_data)?;
match instruction {
MovieInstruction::AddMovieReview { title, rating, description } => {
add_movie_review(program_id, accounts, title, rating, description)
},
// add UpdateMovieReview to match against our new data structure
MovieInstruction::UpdateMovieReview { title, rating, description } => {
// make call to update function that we'll define next
update_movie_review(program_id, accounts, title, rating, description)
}
}
}Next, we can define the new update_movie_review function. The definition should have the same parameters as the definition of add_movie_review.
pub fn update_movie_review(
program_id: &Pubkey,
accounts: &[AccountInfo],
title: String,
rating: u8,
description: String
) -> ProgramResult {
}All that's left now is to fill in the logic for updating a movie review. Only let's make it secure from the start.
Just like the add_movie_review function, let's start by iterating through the accounts. The only accounts we'll need are the first two: initializer and pda_account.
pub fn update_movie_review(
program_id: &Pubkey,
accounts: &[AccountInfo],
title: String,
rating: u8,
description: String
) -> ProgramResult {
msg!("Updating movie review...");
// Get Account iterator
let account_info_iter = &mut accounts.iter();
// Get accounts
let initializer = next_account_info(account_info_iter)?;
let pda_account = next_account_info(account_info_iter)?;
}Before we continue, let's implement some basic security checks. We'll start with an ownership check on for pda_account to verify that it is owned by our program. If it isn't, we'll return an InvalidOwner error.
if pda_account.owner != program_id {
return Err(ProgramError::InvalidOwner)
}Next, let’s perform a signer check to verify that the initializer of the update instruction has also signed the transaction. Since we are updating the data for a movie review, we want to ensure that the original initializer of the review has approved the changes by signing the transaction. If the initializer did not sign the transaction, we’ll return an error.
if !initializer.is_signer {
msg!("Missing required signature");
return Err(ProgramError::MissingRequiredSignature)
}Next, let’s check that the pda_account passed in by the user is the PDA we expect by deriving the PDA using initializer and title as seeds. If the addresses do not match, we’ll return our custom InvalidPDA error. We'll implement this the same way we did in the add_movie_review function.
// Derive PDA and check that it matches client
let (pda, _bump_seed) = Pubkey::find_program_address(&[initializer.key.as_ref(), account_data.title.as_bytes().as_ref(),], program_id);
if pda != *pda_account.key {
msg!("Invalid seeds for PDA");
return Err(ReviewError::InvalidPDA.into())
}Now that our code ensures we can trust the passed in accounts, let's unpack the pda_account and perform some data validation. We'll start by unpacking pda_account and assigning it to a mutable variable account_data.
msg!("unpacking state account");
let mut account_data = try_from_slice_unchecked::<MovieAccountState>(&pda_account.data.borrow()).unwrap();
msg!("borrowed account data");Now that we have access to the account and its fields, the first thing we need to do is verify that the account has already been initialized. An uninitialized account can't be updated so the program should return our custom UninitializedAccount error.
if !account_data.is_initialized() {
msg!("Account is not initialized");
return Err(ReviewError::UninitializedAccount.into());
}Next, we need to validate the rating, title, and description data just like in the add_movie_review function. We want to limit the rating to a scale of 1 to 5 and limit the overall size of the review to be fewer than 1000 bytes. If the rating provided by the user outside of this range, then we’ll return our custom InvalidRating error. If the review is too long, then we'll return our custom InvalidDataLength error.
if rating > 5 || rating < 1 {
msg!("Rating cannot be higher than 5");
return Err(ReviewError::InvalidRating.into())
}
let total_len: usize = 1 + 1 + (4 + account_data.title.len()) + (4 + description.len());
if total_len > 1000 {
msg!("Data length is larger than 1000 bytes");
return Err(ReviewError::InvalidDataLength.into())
}Now that we've implemented all of the security checks, we can finally update the movie review account by updating account_data and re-serializing it. At that point, we can return Ok from our program.
account_data.rating = rating;
account_data.description = description;
account_data.serialize(&mut &mut pda_account.data.borrow_mut()[..])?;
Ok(())All together, the update_movie_review function should look something like the code snippet below. We've included some additional logging for clarity in debugging.
pub fn update_movie_review(
program_id: &Pubkey,
accounts: &[AccountInfo],
title: String,
rating: u8,
description: String
) -> ProgramResult {
msg!("Updating movie review...");
let account_info_iter = &mut accounts.iter();
let initializer = next_account_info(account_info_iter)?;
let pda_account = next_account_info(account_info_iter)?;
if pda_account.owner != program_id {
return Err(ProgramError::IllegalOwner)
}
if !initializer.is_signer {
msg!("Missing required signature");
return Err(ProgramError::MissingRequiredSignature)
}
let (pda, _bump_seed) = Pubkey::find_program_address(&[initializer.key.as_ref(), account_data.title.as_bytes().as_ref(),], program_id);
if pda != *pda_account.key {
msg!("Invalid seeds for PDA");
return Err(ReviewError::InvalidPDA.into())
}
msg!("unpacking state account");
let mut account_data = try_from_slice_unchecked::<MovieAccountState>(&pda_account.data.borrow()).unwrap();
msg!("review title: {}", account_data.title);
msg!("checking if movie account is initialized");
if !account_data.is_initialized() {
msg!("Account is not initialized");
return Err(ReviewError::UninitializedAccount.into());
}
if rating > 5 || rating < 1 {
msg!("Invalid Rating");
return Err(ReviewError::InvalidRating.into())
}
let update_len: usize = 1 + 1 + (4 + description.len()) + account_data.title.len();
if update_len > 1000 {
msg!("Data length is larger than 1000 bytes");
return Err(ReviewError::InvalidDataLength.into())
}
msg!("Review before update:");
msg!("Title: {}", account_data.title);
msg!("Rating: {}", account_data.rating);
msg!("Description: {}", account_data.description);
account_data.rating = rating;
account_data.description = description;
msg!("Review after update:");
msg!("Title: {}", account_data.title);
msg!("Rating: {}", account_data.rating);
msg!("Description: {}", account_data.description);
msg!("serializing account");
account_data.serialize(&mut &mut pda_account.data.borrow_mut()[..])?;
msg!("state account serialized");
Ok(())
}We're ready to build and upgrade our program! You can test your program by submitting a transaction with the right instruction data. For that, feel free to use this frontend. Remember, to make sure you're testing the right program you'll need to replace MOVIE_REVIEW_PROGRAM_ID with your program ID in Form.tsx and MovieCoordinator.ts.
If you need more time with this project to feel comfortable with these concepts, have a look at the solution code before continuing.
Now it’s your turn to build something independently by building on top of the Student Intro program that you've used in previous lessons. If you haven't been following along or haven't saved your code from before, feel free to use this starter code.
The Student Intro program is a Solana Program that lets students introduce themselves. The program takes a user's name and a short message as the instruction_data and creates an account to store the data on-chain.
Using what you've learned in this lesson, try applying what you've learned to the Student Intro Program. The program should:
- Add an instruction allowing students to update their message
- Implement the basic security checks we've learned in this lesson
Try to do this independently if you can! But if you get stuck, feel free to reference the solution code. Note that your code may look slightly different than the solution code depending on the checks you implement and the errors you write.