List Archive

Thread

Thread Index

Message

From: Remi Collet <remi%fedoraproject.org@localhost>
To: libzip-discuss%nih.at@localhost
Subject: Re: Version 1.3.0 - Test failure on 32-bit arch
Date: Fri, 15 Sep 2017 10:49:35 +0200

Hi,

Sorry for late answer...

Le 07/09/2017 à 15:54, Thomas Klausner a écrit :
> On Mon, Sep 04, 2017 at 02:15:27PM +0200, Remi Collet wrote:
>> On 64-bit arch, test suite is OK.
>>
>> On 32-bit (e.g. i686 or arm)
>>
>> FAIL: encryption-nonrandom-aes128.test
>> FAIL: encryption-nonrandom-aes192.test
>> FAIL: encryption-nonrandom-aes256.test
>>
>> Any idea ?
> 
> You can run the tests manually with
> 
> cd regress
> ./runtest -v encryption-nonrandom-aes256.test
> 
> Runtest has a couple of options, including leaving over the sandbox.
> Can you please check if the zipfiles in the sandboxes, when run
> multiple times, are the same?

Yes

$ ./runtest -v --keep-broken --no-cleanup  encryption-nonrandom-aes256.test
Binary files ../encrypt-aes256-noentropy.zip and encrypt.zzip differ
encryption-nonrandom-aes256 -- FAIL: files

$ sha256sum ./encrypt-aes256-noentropy.zip
./sandbox-encryption-nonrandom-aes256.d41/encrypt.zzip
2e9488731a5c578b0d30de4e329807d94aec19cb7501daeb9a5e10dc63c136f3
./encrypt-aes256-noentropy.zip
fb754068aa24c5fe8c48a84512698f609807eaaf9602eab27bb447966654216f
./sandbox-encryption-nonrandom-aes256.d41/encrypt.zzip

Various runs always give the same sha256


Remi

> 
> If yes, then there's probably a 32-bit bug somewhere.
> 
> If not, then the random-ness disabling doesn't work as intended (we
> have another report of that, but I hadn't had time to look at that
> yet).
> 
>  Thomas
> 


Attachment: signature.asc
Description: OpenPGP digital signature

Made by MHonArc.