Re: rm fails but returns success

[Peter Rosin <peda at lysator dot liu dot se>]

Corinna Vinschen skrev 2012-02-03 17:37:
> On Feb  3 17:27, Peter Rosin wrote:
>> Corinna Vinschen skrev 2012-02-03 17:17:
>>> On Feb  3 17:02, Peter Rosin wrote:
>>>> Hi!
>>>>
>>>> I have this annoying leftover file from a automake testsuite run.
>>>> I don't know if it was created by an MSYS process or a Cygwin
>>>> process, but I can't get rid of it.  I can't take ownership of
>>>> it either, not even as admin.  I haven't tried stopping all
>>>> MSYS/Cygwin processes yet, nor rebooting, but I'd rather not.
>>>>
>>>> Any help with that is appreciated.  No, not rebooting :-)
>>>>
>>>> However, that is not really why I'm writing, I'm writing to
>>>> report the following bug related to the above file.
>>>>
>>>> $ uname -a
>>>> CYGWIN_NT-6.1-WOW64 peda-pc 1.7.10s(0.259/5/3) 20120123 00:15:09 i686 Cygwin
>>>> $ ls -l aclibobj.log-t
>>>> -rw-r----- 1 ???????? ???????? 2113 Jan 31 16:09 aclibobj.log-t
>>>> $ rm aclibobj.log-t; echo $?
>>>> rm: remove write-protected regular file `aclibobj.log-t'? yes
>>>> 0
>>>
>>> Send an strace of this, please.  One reason that rm (better: unlink(2))
>>> reports success is if the file is still in use by another process but
>>> it's already marked as "delete pending" in the OS.  This should only
>>> occur if a non-Cygwin process is still holding a handle to the file.
>>
>> Here's the strace.
>> [...]
>>    17   11885 [main] rm 16748 unlink_nt: Trying to delete \??\C:\cygwin\home\peda\automake\tests\aclibobj.log-t, isdir = 0
>>    26   11911 [main] rm 16748 unlink_nt: Delete \??\C:\cygwin\home\peda\automake\tests\aclibobj.log-t already pending
>>    16   11927 [main] rm 16748 unlink_nt: \??\C:\cygwin\home\peda\automake\tests\aclibobj.log-t, return status = 0x0
>>    15   11942 [main] rm 16748 unlink: 0 = unlink(/home/peda/automake/tests/aclibobj.log-t)
> 
> Yup, as I supected.  The delete is pending already.  Therefore the file
> is logically already successfully deleted.  Only the fact that some
> importunate non-Cygwin process is still holding an open handle to the
> file forestalls its disappearance.
> 
> There's nothing at all you can do about that, other than to identify the
> process holding the open handle and kill it.

I killed the suspended MSYS expr process that I mentioned in the other
mail (I also confirmed that it indeed had the file open) and the file
went away.

Thanks for your help!

Cheers,
Peter

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple