Hey there,
On Sun, 26 Nov 2000, Dr S N Henson wrote:
> You normally call the RSA_get_ex_new_index() at startup and then store
> the index somewhere static. Then use that for later access. It then ends
> up calling the various ex callbacks whenever RSA_new() and RSA_free()
> are called, the dup() callback never gets called for most things.
Yeah I follow, it just doesn't really seem to fit for "method-specific"
things.
> In this case though if you want different things on a per method basis I
> agree that init() is the place to do it and ex_data should be
> initialised earlier on.
Cool ... I feel secure enough then that I'm either OK, or at least
breaking something in cooperation with others rather than on my own. :-)
> Then of course you'd better look at what happens in RSA_free(). The
> ex_data structure is destroyed *before* the finish() method is called.
Oh *god* ... *sigh*. I hadn't looked that far ... well, I can't imagine
how anyone would be writing a finish() handler that depends on the ex_data
having been cleaned *before* it's called - so the corresponding reversal
there shouldn't be a problem.
I'll give this some more thought and if no objections are raised in the
interim, I'll probably commit a change to this tomorrow.
Cheers,
Geoff
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
