Hi,
I previously reported this issue with the subject "ssl3_get_server_hello
not threadsafe(?)" to -dev. Rephrased a bit and reposted to -bugs. This
is definitely relevant for 0.9.8i, but seems to be so for current
snapshots too.
I've a simple program (used as a stress tool) that creates a numb
Wow. This may explain a once-in-a-million anomaly going on here.
Hm... Checked this and yes, same story in 0.9.9 HEAD.
Note that this is a particular wicked insect, because, as it is the
stack/stack.c internal_find() function who exhibits the _side effect_
of sorting a yet unsorted stack, _any
Whoops, missed a spot:
When you've processed the stuff reported in this thread before now,
recall that line about the lock only having to surround the sort() to
make the stuff threadsafe... ASSUMING a repeatable sort. (As in
parallel access fringe cases the 'sorted' flag can be read and found
to b
Ger Hobbelt wrote:
lock(l);
if ( ! is_locked())
{
sort();
sorted=1;
}
unlock(l);
I suspect is_locked() needs to be is_sorted()?
I suggest this instead:
--
if ( ! is_sorted()) {
lock(l);
if ( ! is_sorted())
{
sort();
sorted=1;
}
Please merge.
On 11/23/08, Alon Bar-Lev <[EMAIL PROTECTED]> wrote:
> > I finish the tests with native build (MSYS environment):
> >
> > The source tree contain recent version (2008-11-23) plus patches:
> >
> > from ticket 1747:
> > - openssl-cvs-mingw-capi.patch.gz
> > last from this t
is_sorted() indeed. Mis-wrote that.
Sequencing multiple is_sorted() calls like you suggest may be done;
the race condition recurs as you allow access to the 'sorted' flag
outside the critical section, yet this time the inner is_sorted()
check prevents repeating the sort in case of race. Nice idea
Hi!
from enc.c:
/* Note that str is NULL if a key was passed on the command
* line, so we get no salt in that case. Is this a bug?
*/
I say yes and have a fix.
Cheers!
Kosta
salt.patch
Description: Binary data
I use openssl 0.9.8 on SPARC Solaris 10 and try to send a text message to a
server running on Windows XP Professional SP3 with activated firewall using a
non-blocking socket with BIO_puts().
The write to the socket returns with ENOTCONN which is a non-fatal error
because BIO_should_retry() retu
[ Not subscribed to -dev, so replying out-of-band from reading web
archives - apologies for formatting foul-ups]
There are two ways about this:
a) either forego on the side effect and resort to a linear (slow)
search/scan when the stack is unsorted (which can happen, for
instance, after an ele
Hi for all
I wrote simple example from book "Network Security with OpenSSL"
when I want compile, I take error: undefined reference to `BIO_read`
and not only BIO_read, all other functions which of located in header files.
my environment is: CentOS 5
of course I know for some body this question
10 matches
Mail list logo
