There's now support for OpenSSL 1.1.0-pre4 support in v3.1.x. This was actually a fair amount of work. OpenSSL has switched to private structs, meaning we have to allocate things like HMAC_CTX on the heap now instead of using the stack. It also means we no longer have direct access to fields within SSL session contexts and have to use OpenSSL accessor functions in many places. Alan D completed the first part of the work a few months ago, and I finished off the changes yesterday. Our basic EAP test suite passes, but it would be useful if those who rely heavily on TLS could test this out in their lab environment. For large deployments of EAP-TLS/EAP-TTLS/EAP-PEAP there may be a noticeable performance improvement, as OpenSSL has (finally) removed their global object/type-specific mutexes. -Arran Arran Cudbard-Bell <a.cudbardb@freeradius.org> FreeRADIUS Development Team FD31 3077 42EC 7FCD 32FE 5EE2 56CF 27F9 30A8 CAA2
Hi,
Our basic EAP test suite passes, but it would be useful if those who rely heavily on TLS could test this out in their lab environment.
would need to have an OpenSSL 1.1.0 system - will look into it
For large deployments of EAP-TLS/EAP-TTLS/EAP-PEAP there may be a noticeable performance improvement, as OpenSSL has (finally) removed their global object/type-specific mutexes.
....especially for that reason... many thanks! alan
On Fri, Apr 01, 2016 at 10:34:51AM -0600, Arran Cudbard-Bell wrote:
There's now support for OpenSSL 1.1.0-pre4 support in v3.1.x.
Nice.
Our basic EAP test suite passes, but it would be useful if those who rely heavily on TLS could test this out in their lab environment.
I'll try and check it out here in the next couple of weeks if I get a spare 10 minutes. Cheers, Matthew -- Matthew Newton, Ph.D. <mcn4@le.ac.uk> Systems Specialist, Infrastructure Services, I.T. Services, University of Leicester, Leicester LE1 7RH, United Kingdom For IT help contact helpdesk extn. 2253, <ithelp@le.ac.uk>
On 1 Apr 2016, at 14:01, Matthew Newton <mcn4@leicester.ac.uk> wrote:
On Fri, Apr 01, 2016 at 10:34:51AM -0600, Arran Cudbard-Bell wrote:
There's now support for OpenSSL 1.1.0-pre4 in v3.1.x.
Nice.
Our basic EAP test suite passes, but it would be useful if those who rely heavily on TLS could test this out in their lab environment.
I'll try and check it out here in the next couple of weeks if I get a spare 10 minutes.
Thanks Alan B/Matthew! -Arran Arran Cudbard-Bell <a.cudbardb@freeradius.org> FreeRADIUS Development Team FD31 3077 42EC 7FCD 32FE 5EE2 56CF 27F9 30A8 CAA2
On 2 Apr 2016, at 21:39, Arran Cudbard-Bell <a.cudbardb@freeradius.org> wrote:
On 1 Apr 2016, at 14:01, Matthew Newton <mcn4@leicester.ac.uk> wrote:
On Fri, Apr 01, 2016 at 10:34:51AM -0600, Arran Cudbard-Bell wrote:
There's now support for OpenSSL 1.1.0-pre4 in v3.1.x.
Nice.
Our basic EAP test suite passes, but it would be useful if those who rely heavily on TLS could test this out in their lab environment.
I'll try and check it out here in the next couple of weeks if I get a spare 10 minutes.
Thanks Alan B/Matthew!
Whilst I was digging through the 1.1.0 I found some undocumented callbacks added for EAP-FAST that allow you to construct custom session tickets. This may allow us to serialize &session-state:[*] in the session ticket, and have the supplicant hand back any authorizational info required when they resume their session :) I'm sure there's a reason why this is a terrible idea from a security (or other) perspective, but i've not figured it out yet. If anyone else has any views on it, i'd appreciate the feedback. Not sure of what the limit on serialised data would be, extension length is 2^24, guessing the record layer can fragment extensions, else that size wouldn't make sense. The real limit would probably be the number of roundtrips :) Although Session-Tickets are 'obsoleted' in TLS 1.3 (or at least the latest draft), a very similar ticket based mechanism is provided. 0RTT and 1RTT modes in TLS 1.3 look very nice! Should be possible to get EAP-TLS/EAP-TTLS resumed in two rounds! -Arran Arran Cudbard-Bell <a.cudbardb@freeradius.org> FreeRADIUS Development Team FD31 3077 42EC 7FCD 32FE 5EE2 56CF 27F9 30A8 CAA2
On 3 Apr 2016, at 15:10, Arran Cudbard-Bell <a.cudbardb@freeradius.org> wrote:
On 2 Apr 2016, at 21:39, Arran Cudbard-Bell <a.cudbardb@freeradius.org> wrote:
On 1 Apr 2016, at 14:01, Matthew Newton <mcn4@leicester.ac.uk> wrote:
On Fri, Apr 01, 2016 at 10:34:51AM -0600, Arran Cudbard-Bell wrote:
There's now support for OpenSSL 1.1.0-pre4 in v3.1.x.
Nice.
Our basic EAP test suite passes, but it would be useful if those who rely heavily on TLS could test this out in their lab environment.
I'll try and check it out here in the next couple of weeks if I get a spare 10 minutes.
Thanks Alan B/Matthew!
Whilst I was digging through the 1.1.0 I found some undocumented callbacks added for EAP-FAST that allow you to construct custom session tickets.
This may allow us to serialize &session-state:[*] in the session ticket, and have the supplicant hand back any authorizational info required when they resume their session :)
I'm sure there's a reason why this is a terrible idea from a security (or other) perspective, but i've not figured it out yet. If anyone else has any views on it, i'd appreciate the feedback.
Not sure of what the limit on serialised data would be, extension length is 2^24, guessing the record layer can fragment extensions, else that size wouldn't make sense. The real limit would probably be the number of roundtrips :)
Ah in both TLS 1.3 and RFC 5077 struct { uint32 ticket_lifetime_hint; opaque ticket<0..2^16-1>; } NewSessionTicket; 16K is still useful! :) Arran Cudbard-Bell <a.cudbardb@freeradius.org> FreeRADIUS Development Team FD31 3077 42EC 7FCD 32FE 5EE2 56CF 27F9 30A8 CAA2
On Apr 3, 2016, at 3:10 PM, Arran Cudbard-Bell <a.cudbardb@freeradius.org> wrote:
This may allow us to serialize &session-state:[*] in the session ticket, and have the supplicant hand back any authorizational info required when they resume their session :)
I'm sure there's a reason why this is a terrible idea from a security (or other) perspective, but i've not figured it out yet. If anyone else has any views on it, i'd appreciate the feedback.
The ticket is opaque to everyone but the server. So it's safe from that point of view. The real problem is size. Large session tickets mean that session resumption now takes many round trips.
Not sure of what the limit on serialised data would be, extension length is 2^24, guessing the record layer can fragment extensions, else that size wouldn't make sense. The real limit would probably be the number of roundtrips :)
Yes. I would suggest limiting data in the session ticket to less than 1K. Anything more will cause too many round trips. Alan DeKok.
On Sat, Apr 02, 2016 at 09:39:20PM -0400, Arran Cudbard-Bell wrote:
On 1 Apr 2016, at 14:01, Matthew Newton <mcn4@leicester.ac.uk> wrote:
On Fri, Apr 01, 2016 at 10:34:51AM -0600, Arran Cudbard-Bell wrote:
There's now support for OpenSSL 1.1.0-pre4 in v3.1.x.
Nice.
Our basic EAP test suite passes, but it would be useful if those who rely heavily on TLS could test this out in their lab environment.
I'll try and check it out here in the next couple of weeks if I get a spare 10 minutes.
Thanks Alan B/Matthew!
Took a while, sorry. Recommendation is to not get ill... :( openssl 1.1.0-pre5-dev (34da11b39d24) freeradius v3.1.x (dc43c5913064) The clients are Windows 7, and use PEAP/EAP-TLS. (Which I guess is fairly unusual in itself.) PEAP seems to come up fine, and SoH data is retrieved. Then inner-tunnel does EAP-TLS which looks like it's all there, but the client then goes silent on an Access-Challenge. Haven't yet tested with same FreeRADIUS and openssl 1.0.x, so I don't know if this is due to the new openssl code or if it's the previous EAP fragment issue. Assuming it's something to do with openssl code, to debug much further I'll need to see if I can get hold of a laptop to test with, though they're like hen's teeth around here. Cheers, Matthew -- Matthew Newton, Ph.D. <mcn4@le.ac.uk> Systems Specialist, Infrastructure Services, I.T. Services, University of Leicester, Leicester LE1 7RH, United Kingdom For IT help contact helpdesk extn. 2253, <ithelp@le.ac.uk>
On 19 Apr 2016, at 08:29, Matthew Newton <mcn4@leicester.ac.uk> wrote:
On Sat, Apr 02, 2016 at 09:39:20PM -0400, Arran Cudbard-Bell wrote:
On 1 Apr 2016, at 14:01, Matthew Newton <mcn4@leicester.ac.uk> wrote:
On Fri, Apr 01, 2016 at 10:34:51AM -0600, Arran Cudbard-Bell wrote:
There's now support for OpenSSL 1.1.0-pre4 in v3.1.x.
Nice.
Our basic EAP test suite passes, but it would be useful if those who rely heavily on TLS could test this out in their lab environment.
I'll try and check it out here in the next couple of weeks if I get a spare 10 minutes.
Thanks Alan B/Matthew!
Took a while, sorry. Recommendation is to not get ill... :(
openssl 1.1.0-pre5-dev (34da11b39d24) freeradius v3.1.x (dc43c5913064)
The clients are Windows 7, and use PEAP/EAP-TLS. (Which I guess is fairly unusual in itself.) PEAP seems to come up fine, and SoH data is retrieved. Then inner-tunnel does EAP-TLS which looks like it's all there, but the client then goes silent on an Access-Challenge.
Yeah that's pre-existing. I think Win 7 might actually produce useful output though... If you can get a trace that'd help. I found the same issue with v3.0.x with OpenSSL 1.0.1 on Win 10. Assumed it was Win 10 though... Seems something more fundamental is broken. -Arran Arran Cudbard-Bell <a.cudbardb@freeradius.org> FreeRADIUS Development Team FD31 3077 42EC 7FCD 32FE 5EE2 56CF 27F9 30A8 CAA2
Hi,
The clients are Windows 7, and use PEAP/EAP-TLS. (Which I guess is fairly unusual in itself.) PEAP seems to come up fine, and SoH data is retrieved. Then inner-tunnel does EAP-TLS which looks like it's all there, but the client then goes silent on an Access-Challenge.
this doesnt work with 1.0.x OpenSSL either - known issue....and in fact, PEAP only just started working too alan
On Tue, Apr 19, 2016 at 05:57:00PM +0000, A.L.M.Buxey@lboro.ac.uk wrote:
The clients are Windows 7, and use PEAP/EAP-TLS. (Which I guess is fairly unusual in itself.) PEAP seems to come up fine, and SoH data is retrieved. Then inner-tunnel does EAP-TLS which looks like it's all there, but the client then goes silent on an Access-Challenge.
this doesnt work with 1.0.x OpenSSL either - known issue....and in fact, PEAP only just started working too
Yeah, I couldn't remember if that had been fixed yet or not. Spent a while trying to revert the patch that broke it so I could test the openssl 1.1 stuff, but it's too ingrained now to do easily. So looks like it's back to trying to fix the fragmentation problem. I really need a Windows 7 laptop... :( I really don't believe I just wrote that... :( :( Matthew -- Matthew Newton, Ph.D. <mcn4@le.ac.uk> Systems Specialist, Infrastructure Services, I.T. Services, University of Leicester, Leicester LE1 7RH, United Kingdom For IT help contact helpdesk extn. 2253, <ithelp@le.ac.uk>
Hi,
I really need a Windows 7 laptop... :(
? yuck. I believe Windows 8.1 and 10 exhibit the same issue - so at least you can be up to date
I really don't believe I just wrote that... :( :(
its late.....you're tired and still recovering from some illness.....some illness that makes you want to run Windows...serious stuff. see the Doctor! ;-) alan
On Tue, Apr 19, 2016 at 08:15:34PM +0000, A.L.M.Buxey@lboro.ac.uk wrote:
I really need a Windows 7 laptop... :(
? yuck. I believe Windows 8.1 and 10 exhibit the same issue - so at least you can be up to date
They're worse :( Win7 is probably easiest to get my mitts on. But around here you get a desktop or a laptop; both is considered excessive. And I like my three screens. (If my main machine was a laptop I'd wipe any official image and install Debian anyway.)
I really don't believe I just wrote that... :( :(
its late.....you're tired and still recovering from some illness.....some illness that makes you want to run Windows...serious stuff. see the Doctor! ;-)
Waste of time that was. Apparently "everyone's got it". Not sure if she was talking about the nasty virus or Windows. Maybe both. Have been wondering if I can persuade hostapd to work on a "wired" interface to a VM. Would make testing Windows 802.1X much easier. Matthew -- Matthew Newton, Ph.D. <mcn4@le.ac.uk> Systems Specialist, Infrastructure Services, I.T. Services, University of Leicester, Leicester LE1 7RH, United Kingdom For IT help contact helpdesk extn. 2253, <ithelp@le.ac.uk>
participants (4)
-
A.L.M.Buxey＠lboro.ac.uk -
Alan DeKok -
Arran Cudbard-Bell -
Matthew Newton