Using IQueryable.Contains() in an expression causes a memory leak

[jessebarocio]

If you use IQueryable<T>.Contains() in an expression it will cause a memory leak. The leak appears to be in the CompiledQueryCache as memory snapshots show a MemoryCache growing like crazy.

The obvious workaround is to not use an IQueryable in the expression. I never intended to and I've already fixed it. But it was difficult to track down the leak so I still think it should be fixed if possible.

Steps to reproduce

Here's a complete sample to reproduce the issue.

class Program
{
    private static void Main(string[] args)
    {
        using (var context = new Context())
        {
            context.Database.EnsureCreated();
            for (int i = 0; i < 10000; i++)
            {
                var genders = (new[] { Gender.MALE, Gender.FEMALE }).AsQueryable();
                context.Persons.Where(p => genders.Contains(p.Gender)).ToList();  
            }
        }
        
        Console.WriteLine("Done.");
        Console.ReadKey();
    }
}

class Context : DbContext
{
    public DbSet<Person> Persons { get; set; }

    protected override void OnConfiguring(DbContextOptionsBuilder optionsBuilder)
        => optionsBuilder.UseSqlite("Data Source=memleak.db");
}

public class Person
{
    public int Id { get; set; }
    public string FirstName { get; set; }
    public string LastName { get; set; }
    public Gender Gender { get; set; }
}

public enum Gender { UNKNOWN, MALE, FEMALE };

Further technical details

EF Core version: 2.0.1
Database Provider: Tested with SqlServer & Sqlite providers
Operating system: Windows 10
IDE: Visual Studio 2017 15.5.2

[smitpatel]

Related #9301

[smitpatel]

@jessebarocio - We get separate cache entry for each query since, EnumerableQueries are opaque to us. We would not know genders would enumerate too. Hence they don't match existing cache entry. If you move genders declaration outside then we do reference equality on value causing cached entry to be used.

If we parametrize the Queryable then we can use same query cache while still enumerating to different SQL. Current pipeline already supports that with Contains. We will be doing that as a part of fixing this issue.