Dr S N Henson <[EMAIL PROTECTED]> writes:
> Michael Shanzer wrote:
> >
> > Is there a select equivlant or is my work around good
> > enough (if SSL_read returns -1 check the return value
> > of SSL_get_error, and if it returns 2, try again...).
> >
>
> If you mean retry SSL_read then that will work (eventually...) but it
> will be inefficient. What you should do is to redo the select, using the
> return value of SSL_get_error to decide whether to select on read or
> write. When select indicated data is available then you can retry
> SSL_read, but it could of course return -1 and want more data.
Quite so. I misspoke previously. The correct idiom looks like
this.
/* Set up the select masks here */
r=select(width,&readfds,&writefds,0,0);
if(r==0)
continue;
/* Now check if there's data to read */
if(FD_ISSET(sock,&readfds)){
do {
r=SSL_read(ssl,s2c,BUFSIZZ);
switch(SSL_get_error(ssl,r)){
case SSL_ERROR_NONE:
/* Do whatever you'd do with the data */
break;
case SSL_ERROR_ZERO_RETURN:
/* End of data */
if(!shutdown_wait)
SSL_shutdown(ssl);
goto end;
break;
case SSL_ERROR_WANT_READ:
break;
default:
berr_exit("SSL read problem");
}
} while (SSL_pending(ssl));
}
-Ekr
______________________________________________________________________
OpenSSL Project http://www.openssl.org
User Support Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
