help-gnutls
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Help-gnutls] gnutls_record_send() problem


From: Laurent Birtz
Subject: Re: [Help-gnutls] gnutls_record_send() problem
Date: Tue, 22 Jan 2008 18:44:29 -0500
User-agent: Mozilla-Thunderbird 2.0.0.4 (X11/20070622)


Nikos Mavrogiannopoulos wrote:
On Monday 21 January 2008, Laurent Birtz wrote:
I've encountered a weird problem with the stock code examples (anonymous
server and client). By itself the example works fine, but if I replace
the string sent ("GET / HTTP/1.0\r\n\r\n") by a zeroed buffer, then the
client hangs. This does not occur if I send a buffer containing bytes
with the value 1.

Where does the client hang? gnutls does not parse or use the data to be sent in any way, so it's quite unlikely that sending nulls hangs gnutls.
OK, I found the problem. The server calls strlen() on the buffer,
which result in 0 bytes being sent to the client, c.f.
gnutls_record_send (session, buffer, strlen (buffer)) instead of
gnutls_record_send (session, buffer, ret).

gnutls does not parse or use the data to be sent in any way

In version 2.0.4-1, valgrind was passing me this warning when I run it on
the example. I assumed some kind of compression bug was going on in when I
used a zeroed buffer.

==1785== Conditional jump or move depends on uninitialised value(s)
==1785==    at 0x4235432: (within /usr/lib/libz.so.1.2.3.3)
==1785==    by 0x4235DCF: deflate (in /usr/lib/libz.so.1.2.3.3)
==1785==    by 0x406128B: _gnutls_compress (in /usr/lib/libgnutls.so.13.9.1)
==1785==    by 0x40541FD: _gnutls_m_plaintext2compressed (in 
/usr/lib/libgnutls.so.13.9.1)
==1785==    by 0x4054A13: _gnutls_encrypt (in /usr/lib/libgnutls.so.13.9.1)
==1785==    by 0x4052C5D: _gnutls_send_int (in /usr/lib/libgnutls.so.13.9.1)
==1785==    by 0x4052F0A: gnutls_record_send (in /usr/lib/libgnutls.so.13.9.1)
==1785==    by 0x8048D6B: main (ex-client1.c:73)


Thanks,
Laurent Birtz




reply via email to

[Prev in Thread] Current Thread [Next in Thread]