多线程问题，可能是使用Foreach的死锁
本文关键字：Foreach 死锁 问题 多线程 | 更新日期: 2023-09-27 18:30:03
Parallel.ForEach继续运行，我的程序没有结束。我无法追踪它在第一次迭代后的去向。我的猜测是，这会导致死锁，并继续进行上下文切换。
private void ReadInputFile()
{
var collection = new ConcurrentBag<PropertyRecord>();
var lines = System.IO.File.ReadLines(InputFileName);
int i = 0;
int RecordsCount = lines.Count();
Parallel.ForEach(lines, line =>
{
if (string.IsNullOrWhiteSpace(line))
{
return;
}
var tokens = line.Split(',');
var postalCode = tokens[0];
var country = tokens.Length > 1 ? tokens[1] : "england";
SetLabelNotifyTwoText(
string.Format(
"Reading PostCode {0} out of {1}"
i,
lines.Length));
var tempRecord = GetAllAddesses(postalCode, country);
if (tempRecord != null)
{
foreach (PropertyRecord r in tempRecord)
{
collection.Add(r);
}
}
});
}
private List<PropertyRecord> GetAllAddesses(
string postalCode,
string country = "england")
{
SetLabelNotifyText("");
progressBar1.Value = 0;
progressBar1.Update();
var records = new List<PropertyRecord>();
using (WebClient w = new WebClient())
{
var url = CreateUrl(postalCode, country);
var document = w.DownloadString(url);
var pagesCount = GetPagesCount(document);
if (pagesCount == null)
{
return null;
}
for (int i = 0; i < pagesCount; i++)
{
SetLabelNotifyText(
string.Format(
"Reading Page {0} out of {1}",
i,
pagesCount - 1));
url = CreateUrl(postalcode,country, i);
document = w.DownloadString(url);
var collection = Regex.Matches(
document,
"<div class='"soldDetails'">(.|''n|''r)*?class=" +
"'"soldAddress'".*?>(?<address>.*?)(</a>|</div>)" +
"(.|''n|''r)*?class='''"noBed'''">(?<noBed>.*?)" +
"</td>|</tbody>");
foreach (var match in collection)
{
var r = new PropertyRecord();
var bedroomCount = match.Groups["noBed"].Value;
if(!string.IsNullOrEmpty(bedroomCount))
{
r.BedroomCount = bedroomCount;
}
else
{
r.BedroomCount = "-1";
}
r.address = match.Groups["address"].Value;
var line = string.Format(
"'"{0}'",{1}",
r.address
r.BedroomCount);
OutputLines.Add(line);
Records.Add(r);
}
}
}
return Records;
}
它在没有Parallel.ForEach的情况下运行良好，但需要使用Parallel.ForEach。
我已经调试过了，第一次从GetAllAdresses-方法返回后，Step Next按钮停止，它只是在后台继续调试。我放的任何书签上都找不到它。
正如您在评论中所说，您的SetLabelNotifyText和SetLabelNotifyTwoText方法调用Control.Invoke。
为了使Control.Invoke工作，主线程必须是空闲的，但在您的情况下，您似乎通过调用其中的Parallel.ForEach来阻塞主线程
这是一个最小复制：
private void button1_Click(object sender, EventArgs e)
{
Parallel.ForEach(Enumerable.Range(1, 100), (i) =>
{
Thread.Sleep(10);//Simulate some work
this.Invoke(new Action(() => SetText(i)));
});
}
private void SetText(int i)
{
textBox1.Text = i.ToString();
}
主线程等待Parallel.ForEach，工作线程等待主线程，从而导致死锁。
如何修复：不要使用Invoke，只需使用BeginInvoke或不阻塞MainThread。
如果sscce后不是这样，这将对我们有帮助
这样更改代码，使用async和await。这是使用BeginInvoke和其他异步代码模型的现代替代方案。
private async Task ReadInputFile()
{
var collection = new ConcurrentBag<PropertyRecord>();
var lines = System.IO.File.ReadLines(InputFileName);
int i = 0;
int RecordsCount = lines.Count();
Parallel.ForEach(lines, line =>
{
if (string.IsNullOrWhiteSpace(line))
{
return;
}
var tokens = line.Split(',');
var postalCode = tokens[0];
var country = tokens.Length > 1 ? tokens[1] : "england";
SetLabelNotifyTwoText(
string.Format(
"Reading PostCode {0} out of {1}"
i,
lines.Length));
var tempRecord = await GetAllAddesses(postalCode, country);
if (tempRecord != null)
{
foreach (PropertyRecord r in tempRecord)
{
collection.Add(r);
}
}
});
}
private async Task<List<PropertyRecord>> GetAllAddesses(
string postalCode,
string country = "england")
{
SetLabelNotifyText("");
progressBar1.Value = 0;
progressBar1.Update();
var records = new List<PropertyRecord>();
using (WebClient w = new WebClient())
{
var url = CreateUrl(postalCode, country);
var document = await w.DownloadStringTaskAsync(url);
var pagesCount = GetPagesCount(document);
if (pagesCount == null)
{
return null;
}
for (int i = 0; i < pagesCount; i++)
{
SetLabelNotifyText(
string.Format(
"Reading Page {0} out of {1}",
i,
pagesCount - 1));
url = CreateUrl(postalcode,country, i);
document = await w.DownloadStringTaskAsync(url);
var collection = Regex.Matches(
document,
"<div class='"soldDetails'">(.|''n|''r)*?class=" +
"'"soldAddress'".*?>(?<address>.*?)(</a>|</div>)" +
"(.|''n|''r)*?class='''"noBed'''">(?<noBed>.*?)" +
"</td>|</tbody>");
foreach (var match in collection)
{
var r = new PropertyRecord();
var bedroomCount = match.Groups["noBed"].Value;
if(!string.IsNullOrEmpty(bedroomCount))
{
r.BedroomCount = bedroomCount;
}
else
{
r.BedroomCount = "-1";
}
r.address = match.Groups["address"].Value;
var line = string.Format(
"'"{0}'",{1}",
r.address
r.BedroomCount);
OutputLines.Add(line);
Records.Add(r);
}
}
}
return Records;
}
那就这样叫吧
ReadInputFile.Wait();
或者，更好的是，呼叫者是async，
await ReadInputFile();