The program hangs up about pefile

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

The attached file seems to have manipulated or corrupted imports and exports. 
In the export directory, the NumberOfFunctions for the attached file is 
0x10015998 (268523928). So, pefile 
seems to barf at parse_export_directory, trying to enumerate all the ExportData 
and using too much memory. 
Screenshot : http://twitpic.com/1r0xlv

However, the NumberOfNames for the attached file is zero. This could be used as 
an additional check before 
enumerating the exports. These two values (NumberOfNames, NumberOfFunctions) 
are not always equal, but 
the lesser of the two should be used for most purposes. 

temporary fix : NumberOfNames on line 2891 on r73

Original comment by kbandla%[email protected] on 25 May 2010 at 9:31

[GoogleCodeExporter]

Thanks for your answer,

I did as you wrote and it works for the attached file, But I have some more 
files
which still provide a memory leaks. I have attached additionally two files. 

I'm checking with script:

import os
import gc
import pefile

folder = 'folder with tested files'
for f in os.listdir(folder):
  try:
   p = pefile.PE(os.path.join(folder, f), fast_load=False)
   del p
  except KeyboardInterrupt:
   raise KeyboardInterrupt()
  except:
   pass
gc.collect()

Original comment by [email protected] on 26 May 2010 at 6:29

Attachments:

• 0197bdf943b500001493d2dcbabb187b
• 0015a8f056ff3a80b3fb9498a655e8f2

[GoogleCodeExporter]

This seems like expected behavior. You are asking pefile to load all the 
directory info (fast_load=False) on files 
that have a lot of relocation information. (looks like a common pattern for all 
ASPack packed files). 
If you are not using directory information from the pe file, using 
fast_load=True will speed things up for you. 
You can then explicitly call full_load() on the pefile instance to parse all 
directories if you need them. 

Original comment by kbandla%[email protected] on 26 May 2010 at 5:05

[GoogleCodeExporter]

Hello,
I have tried getting info as 
   p = pefile.PE(os.path.join(folder, f), fast_load=True)
   p.parse_data_directories([0,1])
   del p
but the test script still has memory leaks. I have tested my script of a large 
number 
of files and I will try find samples with provide a memory leaks and will send 
you.
Who do I think that it isn't expected behavior?
I delete each PE instance explicitly and call python's GC, but memory still 
continue 
to grow.

Original comment by [email protected] on 28 May 2010 at 12:37

[GoogleCodeExporter]

(Revision 76) Added an upper bound on the number of export entries that will be 
handled. If there are more 
entries than what would fit in what's left until the end of the file we don't 
attempt to process any more
This makes it possible to handle all the files reported although in some larger 
cases it might still take a few 
seconds to load a file

Original comment by [email protected] on 3 Jun 2010 at 12:05

• Changed state: Fixed