Index Out of Range Exception in Thread Pooling in C# -Task.Run() - multithreading

Question

When I remove Thread pooling or uncomment Console.WriteLine() from code then code works fine, but to improve performance I want to process each DataTable column on separate Task. It throws index out of range exception.
DataTable dt = new DataTable();
private async void btnExcel_Click(object sender, RoutedEventArgs e)
{
dt = new DataTable("worksheet");
dt.Columns.Add("Id");
dt.Columns.Add("MobileNo");
dt.Columns.Add("Name");
dt.Columns.Add("Name1");
dt.Columns.Add("Name2");
dt.Columns.Add("Name3");
dt.Columns.Add("Name4");
for (int i = 0; i < 100; i++)
dt.Rows.Add(i, "99999", "ABC" + i, "n1", "n2", "n3", "n4");
//var tasksInFlight = new Task[dt.Columns.Count];
var tasksInFlight = new List<Task>();
for (int index = 0; index < dt.Columns.Count; index++)
{
tasksInFlight.Add(updateDt(index, "col " + index));
}
//await Task.Factory.ContinueWhenAll(tasksInFlight, cT => { string a = "abc"; });
await Task.WhenAll(tasksInFlight);
}
public async Task updateDt(int colNum, string data)
{
try
{
Task t = Task.Run(() =>
{
for (int i = 0; i < dt.Rows.Count; i++)
{
// Console.WriteLine("Col Num : " + colNum + " i = " + i);
dt.Rows[i][colNum] = data;
}
});
await t;
}
catch (Exception ex)
{
}
}

Answer 1

When you call tasksInFlight.Add(updateDt(index, "col " + index)); the value of index is not read; you merely store a task that will be executed later - when you call Task.WhenAll. It is when the tasks are executed when the value of index is evaluated. Which happens after the loop is over and the value of index is now equal to dt.Columns.Count, which is out of the range of the array.
Read about C# closures here.
To fix it, you can do like this:
for (int index = 0; index < dt.Columns.Count; index++)
{
int tmpIndex = index;
tasksInFlight.Add(updateDt(tmpIndex, "col " + tmpIndex));
}
EDIT: After further investigation it turns out that DataTable is not thread-safe.
In addition to the fix above, DataTable should be accessed in a lock:
lock (dt)
{
dt.Rows[i][colNum] = data;
}
However, unless retrieving the actual data to put in the DataTable CPU-intensive, the lock eliminates all benefits of concurrency in this case.