On Thu, Apr 11, 2002 at 12:39:59PM +0200, Pavel Tsekov wrote:
> Hello, there! :)
>
> My colleague Nedelcho Stanev and myself have identified what we would
> think to be a bug (or a flaw) in the EVP_DecodeUpdate() routine. We
> were trying to read base64 encoded data with the base64 BIO which has
> the 'next' member pointed to membuf BIO in which we write the encoded
> data. So far so good ... Eventually it turned up that some of the data
> we pass to the BIO chain is properly decoded while other not - or more
> properly would be to say - partially decoded. We tried to debug the
> problem and here is what we have found:
>
> The EVP_DecodeUpdate() routine errnously returns 0 (EOF) if the buffer
> passed it, ends with CRLF and EVP_DecodeBlock() has just processed
> the data upto the ending CRLF.
>
> Attached is a patch and a testcase. I tried to follow the coding style
> and to make the change not intrusive :) though its a very small change
> :)
>
> Here is how to reproduce:
>
> 1. Base64 encode a file lets say > 200kb
> 2. unix2dos the output
> 3. Use the attached testcase to decode the file
I could easily reproduce your problem and solution.
I am currently checking in your patch.
Thanks,
Lutz
--
Lutz Jaenicke [EMAIL PROTECTED]
http://www.aet.TU-Cottbus.DE/personen/jaenicke/
BTU Cottbus, Allgemeine Elektrotechnik
Universitaetsplatz 3-4, D-03044 Cottbus
______________________________________________________________________
OpenSSL Project http://www.openssl.org
User Support Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
