Re: Error stack documentation

2010-02-12 Thread Patrick M. Rutkowski
>From the SSL_get_error() man page: The current thread's error queue must be empty before the TLS/SSL I/O operation is attempted, or SSL_get_error() will not work reliably. And don't

Re: Error stack documentation

2010-02-12 Thread John L Veazey
When dealing with the high level SSL_* functions, stick with SSL_get_error() as per s_client.c and s_server.c. If you are doing things on a lower level, you may need to deal with the error stack yourself; but for people new to OpenSSL, just stick with functions starting with SSL_* or BIO_* On Fri

Re: Error stack documentation

2010-02-12 Thread Patrick M. Rutkowski
I should have mentioned ERR_get_error() in my question, that's what's most heavily on my mind. The question is if you have to call clear the error stack with ERR_get_error() after every failed SSL call, even failures that are often "not really failures." For example, my code considered SSL_read() r

Re: Error stack documentation

2010-02-12 Thread John L Veazey
Patrick, I believe the "sane state" implied here is that if you call SSL_get_error() some of the time and not all of the time, you may be getting information about errors that happened any time in the past for that SSL*. It won't necessarily prevent OpenSSL from working correctly, just your perce

Error stack documentation

2010-02-12 Thread Patrick M. Rutkowski
As an OpenSSL newbie, I'm trouble that the man pages for SSL_write() and SSL_read() don't stress that you should clear the error stack upon a failed call. They suggest you check SSL_get_error(), but they make no mention of clearing the error stack afterward, and I hear that clearing the error stack

Re: stunnel crashes with weak certificates... could it be OpenSSL?

2010-02-12 Thread Victor Duchovni
On Fri, Feb 12, 2010 at 01:04:01PM -0700, Thomas J. Hruska wrote: > Roger Cruz wrote: >> I posted the following message in the stunnel group. I'm following that >> posting here because I believe this may be an issue with the underlying >> library which is OpenSSL. Is there a known issue with cer

Re: stunnel crashes with weak certificates... could it be OpenSSL?

2010-02-12 Thread Thomas J. Hruska
Roger Cruz wrote: I posted the following message in the stunnel group. I'm following that posting here because I believe this may be an issue with the underlying library which is OpenSSL. Is there a known issue with certificates for version 0.9.8b that are aware of? What version of OpenSSL con

Re: Subject Alternative Name Help

2010-02-12 Thread Victor Duchovni
On Fri, Feb 12, 2010 at 08:35:09PM +0100, Steffen DETTMER wrote: > (So DER encoding is used, and it is allowing 128 byte long > length fields allowing 2^1024 [a number taking four and a half > line in xterm because 309 decimal digits long] bytes long value > fields sufficient to enumerate

Re: Subject Alternative Name Help

2010-02-12 Thread Steffen DETTMER
* Victor Duchovni wrote on Fri, Feb 12, 2010 at 14:20 -0500: > The limit is not (only?) an X.509 limit, rather the SSL/TLS > record layer cannot carry messages larger than 2^14 bytes (plus > some overhead for compression algorithms which provably need to > be able to make some records larger in ord

Re: Subject Alternative Name Help

2010-02-12 Thread Victor Duchovni
On Fri, Feb 12, 2010 at 12:41:16PM +0100, Steffen DETTMER wrote: > * Victor Duchovni wrote: > > The SSL/TLS record layer has a maximum record size, a > > certificate probably needs to fit into one record, so if your > > 500+ domains generate a certificate that is larger than ~16K > > bytes, you ma

stunnel crashes with weak certificates... could it be OpenSSL?

2010-02-12 Thread Roger Cruz
I posted the following message in the stunnel group. I'm following that posting here because I believe this may be an issue with the underlying library which is OpenSSL. Is there a known issue with certificates for version 0.9.8b that are aware of? What version of OpenSSL contains the fix if the

Re: Subject Alternative Name Help

2010-02-12 Thread rono16
I believe it's an x.509v3 limitation and not specific to openssl. After getting Viktor's hint I did a little research and found that the limitation is set at 2^14. Ron Steffen DETTMER wrote: > > * Victor Duchovni wrote: >> The SSL/TLS record layer has a maximum record size, a >> certificate pr

Re: PKCS#7 extract and verify certificate?

2010-02-12 Thread Patrick Patterson
On 12/02/10 8:51 AM, skillz...@gmail.com wrote: > Is there a way (via the API rather than the tool) to tell OpenSSL that > the sub-CA certificate is trusted and it doesn't need to walk further > up the chain? For my case, I embed the sub-CA certificate in my code > and I'm space constrained so I'd

Re: Subject Alternative Name Help

2010-02-12 Thread Steffen DETTMER
* Victor Duchovni wrote: > The SSL/TLS record layer has a maximum record size, a > certificate probably needs to fit into one record, so if your > 500+ domains generate a certificate that is larger than ~16K > bytes, you may be out of luck. (I just ask for curiosity, not because I have any problem

Re: Subject Alternative Name Help

2010-02-12 Thread rono16
That was it. One @ 13k works and one @18k doesn't. Thanks Victor. Ron Victor Duchovni wrote: > > On Wed, Feb 10, 2010 at 03:23:03PM -0800, rono16 wrote: > >> >> I am using OpenSSL to create a self sign certificate and have a need to >> add >> approximately 4000, yes 4000, DNS entries (don't

Re: PKCS#7 extract and verify certificate?

2010-02-12 Thread skillzero
On Thu, Feb 11, 2010 at 1:31 PM, wrote: > I have a DER-encoded PKCS#7 file that I'd like to extract the > certificate from, verify that certificate against a specific sub-CA > certificate, then use the certificate's public key to verify a > signature. > > I looked at the code for the pkcs7 tool a