Am 28.06.2014 04:42, schrieb Viktor Dukhovni:
On Sat, Jun 28, 2014 at 12:28:45AM +0200, Stephan M?hlstrasser wrote:
Maybe b is uninitialized?
I would expect function arguments to be initialized by virtual of
being passed to the function, and the use of uninitialized values
to be attributed to the caller. However, if valgrind does not
consider pushing uninitialized data onto the stack as "use", then
perhaps...
Valgrind reports the problem of an uninitialized memory access at the
moment when a calculation or a jump depends on it. As long as
uninitialized data is only copied around this does not trigger a message.
Note, ECDSA signatures do involve a random nonce, and if that uses
some uninitialized data, that could well be intentional.
Maybe it is possible to suppress the random nonce with the -DPURIFY flag?
This seems unlikely to be a real problem, but if you do file a
ticket, make sure to provide all relevant detail:
* commit id of build
* build platform and Configure target
* how to reproduce the problem
If ECDSA signatures are actually valid, this is most likely a false
positive.
All the signatures I checked were actually valid.
--
Stephan
______________________________________________________________________
OpenSSL Project http://www.openssl.org
User Support Mailing List [email protected]
Automated List Manager [email protected]
