Message-ID: <003501c2517e$2e4fdc30$0100a8c0@p4>
From: "Andrew Cottrell" <acottrel AT ihug DOT com DOT au>
To: "Charles Sandmann" <sandmann AT clio DOT rice DOT edu>
Cc: "Richard Dawe" <rich AT phekda DOT freeserve DOT co DOT uk>, <djgpp-workers AT delorie DOT com>
References: <10208312054 DOT AA20561 AT clio DOT rice DOT edu>
Subject: Re: Two rm.exe issues on XP
Date: Sun, 1 Sep 2002 16:09:18 +1000
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Reply-To: djgpp-workers AT delorie DOT com

Charles & Richard,

> > PROBLEM 1 - RM.EXE SIGABRT
>
> > Test directory setup:
> >     -Created a test directory C:\dj204\test
> >     -Created a sub directory C:\dj204\test\test
> >     -Created one file in the subdirectory C:\dj204\test\test\test.c
>
> Using the rm.exe from simtel's fil41b.zip; and creating the exact
> directories/names as you did above, this works fine on my Win2K system.
I didn't try this yesterday, but I did once I read the email and the rm.exe
from simtel's fil41b.zip works fine with the example. I now expect that the
problem is in a 2.04 CVS LIBC change or series of changes.

> A quick inspection of the remove.c makes it unclear to me how have_device
> could not be set at this point; this code should be harmless but is
> probably unneeded on DOS/Windows systems (it's a security fix).

I have occasionally seen the inode error display on me for no apparent
reason. The error is :-
        "ERROR: the directory %s initially had device/inode\n\
        numbers %lu/%lu, but now (after a chdir into it), the numbers for
`.'\n\
        are %lu/%lu.  That means that while rm was running, the directory\n\
        was replaced with either another directory or a link to another
directory."),

I have added allot of printf's and debug to the remove.c and I am still
tracing it out, but so far no joy. I suspect something in one of the
functions within the remove_cwd_entries() as have_device goes to 0 after the
fspec_init_dp (&fs, dp) call. The following is what I have got so far. I
have not included the remove.c as it would be too big. I need to do some
more debugging to find out where the problem is.

line 336 (340) => fspec_init_common (struct File_spec *fs)
line 353 (355) => fspec_init_dp (struct File_spec *fs, struct dirent *dp)
line 564 (562)=>   fspec_init_dp (&fs, dp);
The numbers in the brackets correspond to the numbers below.

DJGPP_204 C:\dj204\gnu\filutil4.1\src>rm.exe -rf test
347     ++++ 340        872     886     422     374     377     381     FAIL
426
        897     943     957     985     738 test

filename >test< have_filetype_mode 1    have_full_mode 1        have_device
1
mode 31ED       st_ino 10000001 st_dev 2
514     559, .  514     559, .. 514     559, test       562     355     ++++
340

remove.c 567

filename >test< have_filetype_mode 1    have_full_mode 0        have_device
0
mode 3000       st_ino 1        st_dev 1CFF70
872     886     422     FAIL 426        897     943     957     985     738
test
/test

filename >test< have_filetype_mode 1    have_full_mode 0        have_device
0
mode 3000       st_ino 1        st_dev 1CFF70
Assertion failed at remove.c line 775: fs->have_device
Exiting due to signal SIGABRT
.... SNIP....

Call frame traceback EIPs:
  0x0000d706 ___djgpp_traceback_exit+54
  0x0000d7f6 _raise+86
  0x00009a5e ___dj_assert+46
  0x00003386 _fspec_init_file+4406
  0x0000357a _rm+426
  0x00002636 _fspec_init_file+998
  0x000030cb _fspec_init_file+3707
  0x0000357a _rm+426
  0x00001e88 _main+424
  0x00009217 ___crt1_startup+199