[xmlsec] xmlsec1-1.2.16 to xmlsec1-1.2.29 decrypt problems

Floodeenjr, Thomas thomas_floodeenjr at mentor.com
Tue Mar 31 09:27:20 PDT 2020


Thanks!

-----Original Message-----
From: Aleksey Sanin <aleksey at aleksey.com> 
Sent: Tuesday, March 31, 2020 10:26 AM
To: Floodeenjr, Thomas <thomas_floodeenjr at mentor.com>; xmlsec at aleksey.com
Subject: Re: [xmlsec] xmlsec1-1.2.16 to xmlsec1-1.2.29 decrypt problems

Should be fixed on master now.

Aleksey

On 3/30/20 2:29 PM, Floodeenjr, Thomas wrote:
> Sure, thanks for the suggestion.
> 
> https://github.com/lsh123/xmlsec/issues/283
> 
> -Tom
> 
> -----Original Message-----
> From: Aleksey Sanin <aleksey at aleksey.com>
> Sent: Monday, March 30, 2020 2:21 PM
> To: Floodeenjr, Thomas <thomas_floodeenjr at mentor.com>; 
> xmlsec at aleksey.com
> Subject: Re: [xmlsec] xmlsec1-1.2.16 to xmlsec1-1.2.29 decrypt 
> problems
> 
> Could you please file a bug/feature request in github? I don't think it is too difficult to do.
> 
> Aleksey
> 
> On 3/30/20 8:16 AM, Floodeenjr, Thomas wrote:
>> Aleksey,
>>
>> Following the progression of our code, according to this call stack:
>>
>> 	Application.exe!xmlParseCDSect(_xmlParserCtxt * ctxt) Line 9748	C
>>    	Application.exe!xmlParseContent(_xmlParserCtxt * ctxt) Line 9831	C
>>    	Application.exe!xmlParseInNodeContext(_xmlNode * node, const char * data, int datalen, int options, _xmlNode * * lst) Line 13682	C
>>    	Application.exe!xmlSecReplaceNodeBufferAndReturn(_xmlNode * node, const unsigned char * buffer, unsigned int size, _xmlNode * * replaced) Line 632	C
>>    	Application.exe!xmlSecReplaceNodeBuffer(_xmlNode * node, const unsigned char * buffer, unsigned int size) Line 609	C
>>> 	Application.exe!xmlSecEncCtxDecrypt(_xmlSecEncCtx * encCtx, _xmlNode * node) Line 562	C
>>
>> Inside xmlSecReplaceNodeBufferAndReturn(), we call 
>> xmlParseInNodeContext(node->parent, (const char*)buffer, size, 
>> XML_PARSE_NODICT, &results);
>>
>> It looks like the call in xmlSecReplaceNodeBufferAndReturn() hard-codes the option to "XML_PARSE_NODICT", and gives us no option to pass in the XML_PARSE_HUGE parameter we need when we call xmlSecEncCtxDecrypt().
>>
>> Any ideas how to get around this?
>>
>> Thanks,
>> -Tom
>>
>> -----Original Message-----
>> From: Aleksey Sanin <aleksey at aleksey.com>
>> Sent: Friday, March 27, 2020 5:17 PM
>> To: Floodeenjr, Thomas <thomas_floodeenjr at mentor.com>; 
>> xmlsec at aleksey.com
>> Subject: Re: [xmlsec] xmlsec1-1.2.16 to xmlsec1-1.2.29 decrypt 
>> problems
>>
>> Great. I am not aware of any issues with this flag. Since you posted to libxml2 mailing list, I am sure you fill get the answer shortly.
>>
>> Aleksey
>>
>> On 3/27/20 2:55 PM, Floodeenjr, Thomas wrote:
>>> Aleksey,
>>>
>>> It seems we found the problem. Before we decrypt, we call doc = xmlReadFile( filePath, NULL, XML_PARSE_HUGE ); It seems the XML_PARSE_HUGE flag is not honored in libxml2. I am not sure if this is a bug in libxml2, or if it is a new "feature". We are looking for a way to work around this. If we hard-code it in xmlCtxtUseOptionsInternal(), i.e., ctxt->options |= XML_PARSE_HUGE;, then our code decrypts fine.
>>>
>>> Thanks for your replies.
>>>
>>> -Tom
>>>
>>>
>>>
>>> -----Original Message-----
>>> From: Aleksey Sanin <aleksey at aleksey.com>
>>> Sent: Friday, March 27, 2020 10:32 AM
>>> To: Floodeenjr, Thomas <thomas_floodeenjr at mentor.com>; 
>>> xmlsec at aleksey.com
>>> Subject: Re: [xmlsec] xmlsec1-1.2.16 to xmlsec1-1.2.29 decrypt 
>>> problems
>>>
>>> Hi Thomas,
>>>
>>> I am not aware of any limitations on file sizes. I would suggest to either set a breakpoint or dump the decrypted data before the call to xmlParseInNodeContext() to see what's wrong. There are a couple options I can think of:
>>> 1) Decryption is completely incorrect and you will see junk in the buffer.
>>> 2) Decryption is correct but the data is truncated.
>>>
>>> Let me know what do you find.
>>>
>>> Aleksey
>>>
>>> On 3/27/20 6:07 AM, Floodeenjr, Thomas wrote:
>>>> Aleksey,
>>>>
>>>> Is there a limitation with xmlsec1-1.2.29 on the size of the 
>>>> encrypted XML that is being decrypted? The file seems to encrypt 
>>>> OK, but will not decrypt. (290 MB). Smaller files work fine both ways.
>>>> We are using
>>>> libxml2-2.9.9 and openssl-1.1.1d.  (Previously we used 
>>>> libxml2-2.7.8 and openssl-1.0.2g, and it worked fine.)
>>>>
>>>> Thanks,
>>>> -Tom
>>>>
>>>> -----Original Message-----
>>>> From: xmlsec <xmlsec-bounces at aleksey.com> On Behalf Of Floodeenjr, 
>>>> Thomas
>>>> Sent: Thursday, March 26, 2020 12:52 PM
>>>> To: Aleksey Sanin <aleksey at aleksey.com>; xmlsec at aleksey.com
>>>> Subject: Re: [xmlsec] xmlsec1-1.2.16 to xmlsec1-1.2.29 decrypt 
>>>> problems
>>>>
>>>> It returns here with -1
>>>>
>>>>         if(ret != XML_ERR_OK) {
>>>>             xmlSecXmlError("xmlParseInNodeContext", NULL);
>>>>             return(-1);
>>>>         }
>>>>
>>>> -----Original Message-----
>>>> From: Aleksey Sanin <aleksey at aleksey.com>
>>>> Sent: Thursday, March 26, 2020 11:05 AM
>>>> To: Floodeenjr, Thomas <thomas_floodeenjr at mentor.com>; 
>>>> xmlsec at aleksey.com
>>>> Subject: Re: [xmlsec] xmlsec1-1.2.16 to xmlsec1-1.2.29 decrypt 
>>>> problems
>>>>
>>>> What error do you get?
>>>>
>>>> Aleksey
>>>>
>>>> On 3/26/20 8:26 AM, Floodeenjr, Thomas wrote:
>>>>> Hello,
>>>>>
>>>>> We recently upgraded from xmlsec1-1.2.16 to xmlsec1-1.2.29. We 
>>>>> have some data created in 1.2.16 that we can decrypt, but it fails 
>>>>> to decrypt in 1.2.29.
>>>>>
>>>>>                    Application.exe!xmlSecCheckNodeName(_xmlNode * 
>>>>> const cur, const unsigned char * name, const unsigned char * ns) 
>>>>> Line 210 C
>>>>>
>>>>>>                   
>>>>>> Application.exe!xmlSecEncCtxEncDataNodeRead(_xmlSecEncCtx * 
>>>>>> encCtx, _xmlNode * node) Line 696      C
>>>>>
>>>>>                    
>>>>> Application.exe!xmlSecEncCtxDecryptToBuffer(_xmlSecEncCtx
>>>>> * encCtx, _xmlNode * node) Line 597      C
>>>>>
>>>>>                    
>>>>> Application.exe!xmlSecEncCtxDecrypt(_xmlSecEncCtx
>>>>> * encCtx, _xmlNode * node) Line 524      C
>>>>>
>>>>> Other items do decrypt fine with both versions. Any ideas what we 
>>>>> can look into?
>>>>>
>>>>> Thanks,
>>>>>
>>>>> -Tom
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> xmlsec mailing list
>>>>> xmlsec at aleksey.com
>>>>> http://www.aleksey.com/mailman/listinfo/xmlsec
>>>>>
>>>> _______________________________________________
>>>> xmlsec mailing list
>>>> xmlsec at aleksey.com
>>>> http://www.aleksey.com/mailman/listinfo/xmlsec
>>>>
>>> _______________________________________________
>>> xmlsec mailing list
>>> xmlsec at aleksey.com
>>> http://www.aleksey.com/mailman/listinfo/xmlsec
>>>


More information about the xmlsec mailing list