Debugging .Net String value in windbg

问题:

I have a .Net application dump which captured an exception, I'm analysing using windbg and interested in the value of a String parameter on one of the methods. I've isolated the String object. My windbg working is:

0:000> .loadby sos mscorwks
0:000> !dso
OS Thread Id: 0x16f0 (0)
RSP/REG          Object           Name
00000000001fe908 000000000f011440 System.AppDomainSetup
00000000001fe918 000000000f0335f8 System.ArgumentException
00000000001fe920 000000000f011b60 System.String

0:000> !do 000000000f011b60

Name: System.String
MethodTable: 000007feef477a80
EEClass: 000007feef07e530
Size: 538(0x21a) bytes
 (C:\Windows\assembly\GAC_64\mscorlib\2.0.0.0__b77a5c561934e089\mscorlib.dll)
String: C:\Windows\Installer\MSI2D87.tmp
Fields:
              MT    Field   Offset                 Type VT     Attr            Value Name
000007feef47ecf0  4000096        8         System.Int32  1 instance              257 m_arrayLength
000007feef47ecf0  4000097        c         System.Int32  1 instance              179 m_stringLength
000007feef4794c8  4000098       10          System.Char  1 instance               43 m_firstChar
000007feef477a80  4000099       20        System.String  0   shared           static Empty
                                 >> Domain:Value  00000000029d02d0:000000000f011308 <<
000007feef479378  400009a       28        System.Char[]  0   shared           static WhitespaceChars
                                 >> Domain:Value  00000000029d02d0:000000000f0121f8 <<

The m_stringLength member variable indicates that the string is 179 characters long, however inspecting the string it seems to be only 32 characters long. Looking at the memory for this string is appears to be NULL terminated. There are more characters after the NULL terminating character. This may be reused memory or a string corruption however the path looks correct as displayed. The exception being thrown is 'Illegal characters in path' but there are no illegal characters in this path. So the call stack for this exception is:

0:000> !CLRStack
OS Thread Id: 0xbac (0)
Child-SP         RetAddr          Call Site
000000000021e9a0 000007feeea64dec System.IO.Path.CheckInvalidPathChars(System.String)
000000000021e9e0 000007feee9c0e66 System.IO.Path.NormalizePathFast(System.String, Boolean)
000000000021eaa0 000007feee9badf8 System.AppDomainSetup.NormalizePath(System.String, Boolean)
000000000021eb10 000007feeea630ad System.AppDomainSetup.SetupDefaultApplicationBase(System.String)
000000000021eb70 000007feee9bb27b System.AppDomain.SetupFusionStore(System.AppDomainSetup)
000000000021ebc0 000007feef87d4a2 System.AppDomain.SetupDomain(Boolean, System.String, System.String)

Does the System.IO.Path.CheckInvalidPathChars method process the string using the length found in m_stringLength, or does it take into account the NULL termination in the string itself? I'm also open to the fact that there is something else wrong if you can spot something I haven't.

回答1:

Here's what System.IO.Path.CheckInvalidPathChars is doing (at least in .NET 2.0):

for (int i = 0; i < path.Length; i++)
{
    int num2 = path[i];
    if (((num2 == 0x22) || (num2 == 60)) || (((num2 == 0x3e) || (num2 == 0x7c)) || (num2 < 0x20)))
    {
        throw new ArgumentException(Environment.GetResourceString("Argument_InvalidPathChars"));
    }
}

Note that the interesting part of string.Length, that is, how string actually exposes its length property, is missing:

public int Length { [MethodImpl(MethodImplOptions.InternalCall)] get; }

I would try to simulate the issue if possible, by retrieving the exact string (up to to length stored in m_stringLength) and trying to reproduce the issue.

回答2:

I would dump the actual string from the memory in to file to check the contents instead of just looking at the output within windbg.

Here is a Windbg script I wrote a while back to dump the strings to a file.

$$ Dumps the managed strings to a file
$$ Platform x86
$$ Usage $$>a<"c:\temp\dumpstringtofolder.txt" 6544f9ac 5000 c:\temp\stringtest
$$ First argument is the string method table pointer
$$ Second argument is the Min size of the string that needs to be used filter the strings
$$ Third is the path of the file
.foreach ($string {!dumpheap -short -mt ${$arg1}  -min ${$arg2}})
{ 

  $$ MT        Field      Offset               Type  VT     Attr    Value Name
  $$ 65452978  40000ed        4         System.Int32  1 instance    71117 m_stringLength
  $$ 65451dc8  40000ee        8          System.Char  1 instance       3c m_firstChar
  $$ 6544f9ac  40000ef        8        System.String  0   shared   static Empty

  $$ start of string is stored in the 8th offset, which can be inferred from above
  $$ Size of the string which is stored in the 4th offset
  r@$t0=  poi(${$string}+4)*2
  .writemem ${$arg3}${$string}.txt ${$string}+8 ${$string}+8+@$t0
}

The whole idea is to use the .writemem command to dump the contents to a file.

The dumped contents would be in Unicode format and to view its contents use something like this

Console.WriteLine(ASCIIEncoding.Unicode.GetString(File.ReadAllBytes(@"c:\temp\stringtest03575270.txt")));

HTH