[Oisf-devel] libhtp 0.5.5 test suite fails on some architectures

[Ivan Ristić]

On 08/10/2013 12:55, Pierre Chifflier wrote:
> On Mon, Oct 07, 2013 at 06:39:32PM +0100, Ivan Ristić wrote:
>> On 07/10/2013 17:24, Victor Julien wrote:
>>> On 10/07/2013 05:35 PM, Pierre Chifflier wrote:
>>>> I am slowly catching up after vacation, so I have updated versions of
>>>> suricata in unstable, and prepared the 2.0-beta1 in experimental.
>>>>
>>>> The build daemons seems to have caught something, since the test suite
>>>> fails on several tests on some architectures:
>>>> https://buildd.debian.org/status/package.php?p=libhtp&suite=experimental
>>>>
>>>> https://buildd.debian.org/status/fetch.php?pkg=libhtp&arch=powerpc&ver=0.5.5-1&stamp=1377013001
>>>> https://buildd.debian.org/status/fetch.php?pkg=libhtp&arch=s390&ver=0.5.5-1&stamp=1377013856
>>>> https://buildd.debian.org/status/fetch.php?pkg=libhtp&arch=armel&ver=0.5.5-1&stamp=1377019860
>>>>
>>>> For example:
>>>> [  FAILED  ] 6 tests, listed below:
>>>> [  FAILED  ] HybridParsing.CompressedResponse
>>>> [  FAILED  ] HybridParsing.CompressedResponseNoDecompression
>>>> [  FAILED  ] HybridParsing.ForcedDecompression
>>>> [  FAILED  ] HybridParsing.DisableDecompression
>>>> [  FAILED  ] Base64.Single
>>>> [  FAILED  ] Base64.Decode
>>>
>>> Ivan has released 0.5.7, so it may be worth looking at that instead.
>>
>> I don't think that anything I did in 0.5.7 is going to make a
>> difference. I was not aware of those tests failing.
>>
>> Do you know if there are some VM images I could use to deploy on those
>> architectures?
>>
> 
> Hi Ivan and Victor,
> 
> Looking at the changelog it seems that 0.5.7 will not change anything, I
> will try later though.

Some base64 code (from http://sourceforge.net/projects/libb64/) had
assumptions that chars are always signed, and that failed on PowerPC.
After I fixed that all the tests worked.

If you can, please test the latest from the 0.5.x branch:

    https://github.com/ironbee/libhtp/tree/0.5.x

Thanks.

> I'm afraid there is no VM available. Debian developers have access to
> some porter boxes, so I will try to build the package directly on one of
> the failing architectures to see if I can get some more details.
> 
> Note that the build logs give some more details on the tests output, for
> ex in
> https://buildd.debian.org/status/fetch.php?pkg=libhtp&arch=armhf&ver=0.5.5-1&stamp=1378587543
> 
> [ RUN      ] Base64.Single
> test_utils.cpp:59: Failure
> Value of: htp_base64_decode_single(',')
>   Actual: 255
> Expected: -1
> [  FAILED  ] Base64.Single (1 ms)
> [ RUN      ] Base64.Decode
> test_utils.cpp:69: Failure
> Value of: bstr_cmp_c(out, "this is a test..")
>   Actual: 1
> Expected: 0
> [  FAILED  ] Base64.Decode (0 ms)
> 
> and
> 
> [ RUN      ] HybridParsing.CompressedResponse
> test_hybrid.cpp:512: Failure
> Value of: tx->response_message_len
>   Actual: 189
> Expected: 187
> 
> Pierre
> 
> 


-- 
Ivan