[OpenWrt-Devel] [PATCH 2/3] Fix SSL negotiation being interrupted by .notify_write from BIO method.

[Felix Fietkau]

On 2014-12-12 05:16, Yousong Zhou wrote:
> On 12 December 2014 at 00:42, Felix Fietkau <nbd at openwrt.org> wrote:
>> On 2014-11-11 11:34, Yousong Zhou wrote:
>>> ustream_ssl_check_conn() may be called by .notify_write while a previous
>>> SSL_connect() is still in process.  This can happen because the
>>> .notify_write callback will may be triggered by writes in the BIO
>>> methods.
>>>
>>> Signed-off-by: Yousong Zhou <yszhou4tech at gmail.com>
>>> ---
>>>  ustream-ssl.c |   19 +++++++++++++++----
>>>  ustream-ssl.h |    1 +
>>>  2 files changed, 16 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/ustream-ssl.c b/ustream-ssl.c
>>> index dd0faf9..84104b0 100644
>>> --- a/ustream-ssl.c
>>> +++ b/ustream-ssl.c
>>> @@ -34,12 +34,22 @@ static void ustream_ssl_error_cb(struct uloop_timeout *t)
>>>               us->notify_error(us, error, __ustream_ssl_strerror(us->error, buffer, sizeof(buffer)));
>>>  }
>>>
>>> +static enum ssl_conn_status ustream_ssl_do_connect(struct ustream_ssl *us)
>>> +{
>>> +     enum ssl_conn_status status;
>>> +
>>> +     us->connecting = true;
>>> +     status = __ustream_ssl_connect(us);
>>> +     us->connecting = false;
>>> +     return status;
>>> +}
>>> +
>> I think this can be fixed in a much simpler way. Simply prevent
>> re-entrant calls to __ustream_ssl_connect through a static variable.
> 
> Guarding it with a single static variable do not work well with
> multiple instances of ustream_ssl.
How so? Even with multiple instances a call stack from one instance that
calls do_connect should not end up with re-entrant calls for a different
instance.

- Felix
_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel