Mailing List vkapps-user@2rosenthals.com Archived Message #7

From: Lewis <lgrosenthal@2rosenthals.com> Full Headers
Undecoded message
Subject: Re: [vkapps-user] CRC error during PF unpack
Date: Sun, 12 Aug 2018 15:14:46 -0400
To: VKApps Users Mailing List <vkapps-user@2rosenthals.com>

On 08/11/18 06:11 pm, Steven Levine wrote:
In <list-480648@2rosenthals.com>, on 08/11/18
    at 04:51 PM, Lewis <lgrosenthal@2rosenthals.com> said:

Hi,

Okay. that would imply that even if one or more of the files inside the
PF  was broken, the CRC should still match.
Depending on your definition of broken, the answer probaby yes.

If I extract the PF to a temporary directory:
[d:\] e_pf.exe AOSBOOT.PF test\ <Enter>
That's the only test mode that's available.  Well that's not quite true.
There's also

   e_pf.exe AOSBOOT.PF test\ /B

which requests bootdisk unpacker mode, which I have never used.  I'm
pretty sure I've never had the need to build e_pf either.

Can I assume you copied AOSBOOT.PF from the stick?

all files extract cleanly. If I then compare those files to the files
used  to create the PF, they all compare without errors.
Can I assume you rebuilt the failing .pf file and the USB stick and that
the failure persists?

Did you happen to be able to see the name of the failing file?  I suspect
not.

A count of the dots might help too.  I guess we are configured for dots
rather than the default progress bar character.

If the problem persists, please open a ticket for this in the mantis
bugtracker, since it is related to the ArcaOS installer, and attach the
failing .pf file.

The unpacking code in e_pf and in memdisk are entirely different.  e_pf
does not appear to check the CRCs.  I guess Veit only implemented what he
needed at the time.


This may have been a red herring, or at least, real corruption introduced to the FAT filesystem on the flash drive. While I have confirmed - in some instances - that the md5sums match between the PF on the stick and the PF in the work directory, at other times, the stick has been dirty, with lost clusters. Still investigating. Right now, though, it seems as though the CRC error is indeed a symptom of another problem, and is correct, so this may all be WAD.

--
Lewis


Subscribe: Feed, Digest, Index.
Unsubscribe
Mail to ListMaster