Looks neat.
I'd rather have ipv6 first, but +1 for this
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
__
OpenSSL Project
Please, not mdoc. It doesn't offer any particular feature it's just different.
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
I've converted all the divisibility rules for all the primes less than 25
into binary. All the sums can be calculated at once.
Nice work!
/r$
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
The core team should come up with a list and announce the decision. SOON. Be
firm. Say something like in xxx months, support for these platforms will be
dropped and we will start to remove that code. Encourage folks interested in
supporting those platforms to maintain a fork. I don't care
Thanks. In particular, since SSL_OP_ALL is a compile-time constant,
applications compiled with older releases will not send the extension by
default. Only applications compiled against 1.0.1g or later that use
SSL_OP_ALL, or specifically enable this work-around, will send the extension.
especially Stephen Henson, who has kept it together in much the same way as
Keith Richards did the Stones.
With no disrespect intended to either man, I have to say that this is an
analogy that never would have occurred to me in a million years.
/r$
--
Principal Security Engineer
Is there somebody working on it to get Chacha/Poly cipher suites production
ready?
It's expected that the way the ciphers are used will change as it goes through
the IETF TLS group. Therefore, Google has not been encouraging folks to pick up
and use these patches other than an on your own
Perhaps Configure should have a -f nnn flag, that lets folks add their own
local table without having to patch the script
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
-Original Message-
From: owner-openssl-...@openssl.org
I think this misses the point, one can already just pass a table entry on the
command-line as a colon-separated target name.
Yes, you're right, I was mis-using the thread.
But putting a config spec on the command line is, shall we say, awkward. And
adding the flag would help with code
Is it possible to adapt the Configure tool in order to
- first execute the preprocessing stage (macro expand and source code
generation) like gcc -E
- execute some custom source code manipulation (free/malloc enhance, array
bound checks, etc) of my own
One way to do this would be to use
A colleague here noticed that the pthreads-based locking loses the distinction
between read and write locks. We've collected mutex contention data, and found
that the CRYPTO_ERR lock, used while getting error info, is one of the biggest
offenders.
It turns out that pthreads_locking_callback
And I want to reduce the number of exposed APIs.
Except that as we (hopefully) move to making struct's opaque, then we'll need
add lots of accessors. I assume you know that, but just want to make sure
folks realize it.
In the medium term, I'd like to see things like this BN foo; break at
http://opensslrampage.org/post/88383880093
The rampager is wrong; see Adam Langley's comments on twitter;
https://twitter.com/agl__/status/476420434095648768
/r$
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
What kinds of operations are protected by read locks?
Looking at almost any of the global data structures, such as error tables, OID
tables, and so on.
Often, RW locks aren't a win because maintaining just the read locks (without
any writers) introduces contention at the hardware level, and
Ø Preload them all at startup with a global lock held, delete them at shutdown
with a global lock held. If all the other access is 'read' the structures don't
need a lock between times.
Ø Might be something to consider putting on the to do list. I can understand
things being done like that
Hey, that's very neat.
The REPORT part looks automated; are the REMARKS your commentary or does the
tool do that too?
/r$
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
For what it's worth, the policy at IBM (where I used to work, and where they
know quite a few things about software intellectual property), is that you only
update the copyright on an individual file *when you modify it.*
/r$
--
Principal Security Engineer
Akamai Technologies,
Does openssl handle a clientHello (or any handshake message) that splits across
records?
I can't quite tell ... :)
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.memailto:rs...@jabber.me; Twitter: RichSalz
Ø Mostly yes (I know because I made the changes to allow this a long time ago).
That’s what it seemed to me (because read_message will fill a buffer as
needed), but the intern here was pretty sure of himself.
Now, maybe not so sure ☺
--
Principal Security Engineer
Akamai Technologies,
It is perhaps appropriate that my comment had a typo. We can't change it.
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
From: owner-openssl-...@openssl.org [mailto:owner-openssl-...@openssl.org] On
Behalf Of Erik Forsberg
What would be the best equivalent yo pthread_once on Windows ?
I was once looking for one, and back then, years ago, I didnt like the
choices.
Perhaps
We need to support embedded clients that only speak SSL2 :(
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
__
OpenSSL Project
I have no problem disabling it by default and think that should have been done
awhile ago, actually.
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.memailto:rs...@jabber.me; Twitter: RichSalz
The website is written using a tool called wml.
It would be great if someone wanted to make it more modern and properly use
things like CSS. Then it might make sense to put it into a github repository.
Want to volunteer?
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM:
(Sorry if this is a duplicate message.)
Right now the website is written in WML (http://thewml.org) so it's not clear
how useful it would be to put the pages up on github.
I think it would be great if they were converted to HTML+CSS. Then it would
make more sense.
/r$
--
Principal
I fixed that one on master :)
commit 327f3c040ed7451e6f7fb461e13044884607273c
Author: Rich Salz rs...@akamai.com
Date: Sun Jun 29 11:40:05 2014 -0400
Fix typo in message (RT 3107)
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
Platform in the h/w and s/w sense, not just hardware.
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
-Original Message-
From: owner-openssl-...@openssl.org [mailto:owner-openssl-
d...@openssl.org] On Behalf Of Philip A.
Feel free to re-open :)
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
-Original Message-
From: owner-openssl-...@openssl.org [mailto:owner-openssl-
d...@openssl.org] On Behalf Of Kurt Roeckx via RT
Sent: Monday, June 30,
There are several tickets about mingw and djgpp builds breaking, or building
software that crashes, and so on.
If you can help me understand the current state of things with those
toolchains, please drop me a line.
Thanks.
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM:
I thought until now, that as long there are developers who are willing to
develop for a certain platform and there is some community interest in using
that - the platform will be supported as odd might it be in the Windows and
Linux dominated World.
With the releases now in github, one
Hope you are right...but not sure..
Neither are we. That is why the current roadmap says that we're working on it.
It's important to realize that supporting a platform incurs a cost, and we need
to have some way of making the appropriate trade-offs. Clearly, we don't want
to end up where
So now gcc/clang is required to build OpenSSL?
No, nobody's said that. The phrase was perhaps And if openssl ships with a
default set of dependencies, which it does, there's no issue about which
compiler you use at all. Once we fix the make depend requirement.
--
Principal Security
I was wondering why 'make depend' output was saved in the Makefiles.
Because way back when (think like early X and xmkmf) that's the way things were
done.
So I guess adding the .d files to the repository and using include statements
in the Makefiles is a reasonable possibility? (That's the
Really? Its much more efficient to update the .d files when you compile the
(changed) source - which more-or-less implies one per source file.
Not necessarily. One process scanning all the sources, and one file open/parse
in make is often more efficient.
I read this on the internet
However, I feel that the developer group is a bit closed to outsiders.
More communication and transparency is coming, as we have a bigger and more
invigorated developer team. It will take time. But not everything will always
be discussed in public mailing lists right away, parciularly
-openssl-
d...@openssl.org] On Behalf Of Loganaden Velvindron
Sent: Wednesday, July 02, 2014 2:24 PM
To: openssl-dev@openssl.org
Subject: Re: OpenSSL roadmap
On Wed, Jul 2, 2014 at 9:48 PM, Salz, Rich rs...@akamai.com wrote:
However, I feel that the developer group is a bit closed to outsiders
Looks to me like you've only fixed this (and many others) in master - surely
should also go to 1.0.2 at least (and probably older branches, too)?
Okay, tell me which branches.
Also, we generally rebase rather than merge...
I don't know the difference. But okay, if that's the practice, I'll
No, I don't mean to imply that you are one of the bad guys. It's just that we
have only one real way of knowing who the good guys are, and that is being part
of the development team. Yes, that can be very inconvenient. Trust me, I
know, it took more than 10 years for the team to open up and
Of Ben Laurie
Sent: Thursday, July 03, 2014 7:15 AM
To: OpenSSL development
Cc: Jeffrey Walton
Subject: Re: [openssl.org #3277] OpenSSL s_client doc missing option
On 3 July 2014 12:04, Salz, Rich rs...@akamai.com wrote:
Looks to me like you've only fixed this (and many others) in master
Closed, thanks.
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.memailto:rs...@jabber.me; Twitter: RichSalz
release processes at various distributions. (Given that Microsoft has weekly
patch Tuesdays, if even slow moving *Microsoft* can turn around a
security update in a week, what's your excuse? :-)
They have a regular release train, but it doesn't mean that everything gets
fixed in one week.
Why not just have bn_expand_internal call memset?
; git diff bn_lib.c
diff --git a/crypto/bn/bn_lib.c b/crypto/bn/bn_lib.c
index b1e224b..86d1d37 100644
--- a/crypto/bn/bn_lib.c
+++ b/crypto/bn/bn_lib.c
@@ -324,6 +324,9 @@ static BN_ULONG *bn_expand_internal(const BIGNUM *b, int
words)
There's a bunch of hacks in apps/openssl.c to work around some old VMS
releases; the coment is dated 2011-03-22.
I am going to delete it.
Speak up now if you can justify keeping it.
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.memailto:rs...@jabber.me;
Those who forget history are doomed to re-implement it, wrongly.
SO_REUSEADDR was implemented in 4.2BSD so that a server could restart without
waiting for the various FIN_WAIT timeouts to happen.
:)
/r$
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM:
...but can not let the less popular platforms decline, therefore I decided to
set up Jenkins builds on polarhome.com's 30+ rare operating systems and
Wow, that is really great. Thank you!
As Ben said, we haven't decided on *anything* yet.
/r$
--
Principal Security Engineer
Akamai
Steve,
Thanks for the explanation. I'll refactor it a bit, and keep it.
/r$
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
__
OpenSSL Project
The Globus syntax is strange. :)
We should support the ISO date/time standard, and use that throughout and not
invent yet another syntax, or yet another flag. It's fairly simple to parse,
and handles timezones, relative times, date/time mixing, and so on. The XML
XSD spec, for example, has a
But then it has to be supported for, like ever. :)
If the right thing to do is the ISO format, and I strongly believe it is, then
we should just work toward that and not add variants to solve a short-term need
that will require long-term care and confusion.
/r$
--
Principal
Would it work to *always* copy argv on VMS?
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
__
OpenSSL Project http://www.openssl.org
The amount of time it took me to read the comment and figure out what is going
on, and your time to write email explaining it, and Ted's time to chime in
about the necessity of doing all this far outweighs the new code which is
#ifdef VMS'd Because now main() looks really simple and
do you realistically think we'll ever drop support for the -days argument
though? Dropping -days would break a million scripts.
No, we'll never drop support for -days. But whether the code is atoi() or
atof() is a big difference and might cause important silent failures for new
scripts
date '+%Y%m%d%H%M%SZ' -d '1 month 12 hours'
Wow. Old code never dies; that's my get_date code from August 1990 :)
/r$
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
I agree with that as well. I did not look at the actual code in openssl so I
did
not know that the fractional argument with the current version does not
error out.
I have a branch that adds pretty comprehensive option-checking to all the
openssl commands:
; ./openssl x509 --CA
You've declared -days to take only positive numbers, it should allow
negative numbers.
Pushed, thanks.
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.me; Twitter: RichSalz
__
OpenSSL
keyform, OPT_KEYFORM, 'f', Private key file format (PEM or ENGINE)
while the valid choices seem to be PEM or DER, not PEM or ENGINE:
No, it depends on the command. Some, for example, expect keys to be stored in
the ENGINE (presumably an HSM).
The docs are often outdated. But pem/der is
The right thing to do is change opt_format to be generic, and specify exactly
which types of formats are supported.
Done and pushed. Some of the bit-settings are probably more loose than I'd
like, but it works.
/r$
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
Yes, you’re totally right about the root cause being poor abstractions.
We will probably remove all mention of MSDOS, which should be another way to
fix the problem, right?
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.memailto:rs...@jabber.me; Twitter:
You are preaching to the choir. Look at the rsalz-monolith branch in
akamai/openssl on github.
If you have a patch to go into that, I'll take it right away.
We’re going to address the larger issues, in time. For now: does removing
MSDOS fix the problem?
--
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: rs...@jabber.memailto:rs...@jabber.me; Twitter: RichSalz
Where did you get your SSL package? Did it come with the OS? If so, then ask
them for an update.
If not, then where did you get it? Contact them.
If you built it internally, you'll have to learn or buy expertise.
--
Principal Security Engineer, Akamai Technologies, Cambridge, MA
IM:
If you're doing multi-threaded builds with GCC, the following performance hack
can help a great deal.
#ifdef _GNU_SOURCE
int gnu_builtin_sync(int *pointer, int amount, int type, const char *file, int
line)
{
int ret;
if (amount 0)
ret = __sync_add_and_fetch(pointer, amount);
else
Can you take a look at http://rt.openssl.org/Ticket/Display.html?id=549
And let us know what you think?
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.memailto:rs...@jabber.me Twitter: RichSalz
Thanks for your kind words. We do post a notice that we're putting out a
security update. Not sure how you missed it...
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.me Twitter: RichSalz
What's the programming model for using session cache with a multi-threaded
server?
When a client connections, a refcount on the object is incremented. But then
fields can be changed (such as ecpointformat). Does it make more sense for
session to deep-copy the session from the cache?
--
We're using the standard internal session (maintained per SSL_CTX object); not
tickets.
We're seeing that the sessions are shared, a refcount is maintained, but that
SSL does modified fields within a session while it's being used. Most notably
an address sanitizer build found the EC point
please dont do that! I maintained it in the past (and try to do in future as
my
time permits), and currently it still builds (except for asm support were ich
Okay. Thanks for your efforts.
The NETWARE port is really messy, with about 130 #ifdef flags in 70 files. It
would be great if we
Thanks for the info!
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.me Twitter: RichSalz
__
OpenSSL Project http://www.openssl.org
Development Mailing List
Just a comment. the OpenSSL build already depends on Perl and Perl already
has a Make of it's own .
Ooh, that could be interesting. What's the perl make thing called? A web
search for perl make was too voluminous...
/r$
--
Principal Security Engineer
Akamai Technologies,
Problem solved by me three years ago. Still using old platform. Works fine.
Glad it works! Anything worth sharing or was it very specific?
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.me Twitter: RichSalz
Ugh, you're right. Re-opening this. BIO sockets are a tangle that will take
some time to figure out.
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.me Twitter: RichSalz
Does anyone want to speak up for the requirement that we continue to support
BEOS (apparently B/1 and R5?), OS/2, or pre-Windows MSDOS?
Unless there is strong interest and commitment, we will drop these after 1.0.2
/r$
--
Principal Security Engineer
Akamai Technologies,
Minor clarification is appropriate. MSDOS is supported in single stance,
namely DJGPP, which is 32-bit environment.
Good point.
So the idea is that MSDOS gets turned into DJGPP. BEOS and OS/2 are removed in
HEAD (i.e., after 1.0.2), and Microsoft means WINDOWS of various flavors.
If this is
I'm not sure what WINDOWS means. And I'm not sure MSFT knows either :)
Less flippantly, the goal is that OPENSSL_SYS_WINDOWS means any Windows
platform, and then there are subtypes within that. We'll figure it out as we
go along. It's gonna take a while to clean up the #ifdef world without
Thanks for the feedback!
There are 70 files that have OS2 in them, for a total of 130 instances. That's
rather a lot for a platform that hasn't had an update in five years.
This is my personal opinion, as a team member. We will release 1.0.2 this
year. At that time we will announce end of
So I would not understand that we go in a hurry to remove WCE compatibility
I do not think we are in a hurry to do that.
Your patch looks nice. I am CC'ing rt, so that this thread becomes an issue
and we'll see the link to your mail.
--
Principal Security Engineer
Akamai Technologies,
Did I miss something, or did you happen to count the includes of e_os2.h
which is not OS/2 specific at all? Or both?
No, I made the stupid mistake.
The current version of eComStation, 2.1, was released only a year
after version 2.0, in May 2011.
We were not aware of eComStation. Thanks.
Just generated a pull request for this; let me know if it's what you actually
had in mind:
https://github.com/openssl/openssl/pull/161
I already had the fix in-hand :) See attached.
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.me Twitter: RichSalz
It'd be good to fix this.
Opening an RT (email to r...@openssl.org) is the simplest way. Thanks!
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.me Twitter: RichSalz
Find a C version (which I have written) of the utility at:
http://git.alpinelinux.org/cgit/aports/plain/main/openssl/c_rehash.c
That's pretty cool. We'd need to modify it to not use the XXXat functions or
fnmatch, but definitely something we should consider for a future release.
--
I changed the bug title, since the test directory isn't ever removed. But yes,
something strange is going on.
:��IϮ��r�m
(Z+�7�zZ)���1���x��hW^��^��%�� ��jם.+-1�ځ��j:+v���h�
Don't rush. It'll be a while until (or if) we switch over. Neat job tho.
Perhaps it should be merged into the openssl command?
(see https://github.com/akamai/openssl/tree/rsalz-monolith)
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.me Twitter: RichSalz
BTW, as you work on this, also take a look at RT items 2272 and 2973 :)
__
OpenSSL Project http://www.openssl.org
Development Mailing List openssl-dev@openssl.org
Automated
Think of this as pre-release software. The changes are too large to disrupt
the 1.0.2 release, which is already in beta.
We haven’t yet figured out how to make early-access to branches available, so
for now I just did it via Akamai. At some point, I’d expect that branch to
“move” over to
FWIW, most of us picking up 1.0.2 will be in it for the long haul, I wouldn't
expect many to shift from 1.0.2 again to 1.0.3 over the course of a year or
several. It might be worth rethinking the 1.0.2 release plan to pick
I understand the concern. But we have already declared that 1.0.2
Would it be an idea to create branches in the official repo for (certain
classes of) bugfixes, which can be merged onto the respective branches at set
times ? For instance one for documentation fixes ? You could
Yes. But we (the dev team) haven't figured out all of the details of our
i don't think that's really true. else, why is autoconf friends relying on
a
shell and not perl ? those see way more distribution than openssl.
Last I looked, autoconf doesn't use anything that really wasn't in Version 7
Bourne shell. In my comment, I deliberately used the term posix
These all first appeared in ksh: functions, local, return, $((math))
But to my mind, the question is moot, since post-1.0.2 we'll almost
definitely have c_rehash builtin to the openssl command.
that would also work
:)
It will also be much much much faster, since it doesn't have to call
Not according to the PKIX RFC 5280
CAs conforming to this profile MUST always encode certificate
validity dates through the year 2049 as UTCTime; certificate validity
dates in 2050 or later MUST be encoded as GeneralizedTime.
Conforming applications MUST be able to process validity
What about usoing stunnel?
You can't use partial writes.
The size of your UDP packet depends on the MTU supported by everyone along the
path. (BTW, that's what heartbeat was created.)
I suggest you get your program working properly for your definition of what
properly means, without DTLS. Then add DTLS.
And have you
My point is that since stunnel has a different goal of wrapping almost any
protocol, that might be a better place for it, rather than going down the
slippery slope of putting a binary hack into s_client which wouldn't let you
actually USE the protocol.
If Frank doesn't want SSLv2 then he needs to disable it in the SSL_CTX first,
no?
The mechanism to say what ciphers you want is orthogonal to the mechanism to
say what protcols you want. That's unfortunate and a source of confusion, but
is unlikely to change any time soon.
--
Principal
Of no less importance is to emphasise that it adds additional keyform
parameter to functions defined in ts.c and utilized by -reply function, that
will *break* compatibility with any previously existing code.
How does it break? We don't care about source-level compatibility within the
apps
RFC 5280 requires that serial numbers MUST be positive, negative serial
numbers do not conform with RFC (see 4.1.2.2).
Yes, thanks for the clarification.
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.me Twitter: RichSalz
Partial writes do not work over UDP; by design.
As to whether or not you can use a packet as big as 16K, in depends on the
path MTU -- what's the maximum transmission size between you and the
destination, along the communication path. You'll have to make your packets
smaller then that. This
You are right - it should not break anything as the patch only affects the ts
app.
I put this on my dev branch for post-1.0.2 release:
https://github.com/akamai/openssl/tree/rsalz-monolith
--
Principal Security Engineer
Akamai Technologies, Cambridge MA
IM: rs...@jabber.me Twitter:
I think there's interest for 1.0.1 and beyond.
But I thought we already had a similar alias mechanism?
These configuration options do not build. I started to try and fix them, but
after fixing the first few problems, things got really sticky.
We hear that OpenSSL on embedded devices is important. Is anyone using this,
willing to share their fixes, and help maintain it? If not, it will be
Anyone?
This mail was sent one minute after your previous mail. A little patience
perhaps? :)
:��IϮ��r�m
(Z+�7�zZ)���1���x��hW^��^��%�� ��jם.+-1�ځ��j:+v���h�
It is from real world application. In some case the X509_find_by_subject
(called from ocsp_req_find_signer) returned NULL, and the whole
application halted.
Ah, I misunderstood the ticket. Add if (!signer) return 0; after the call to
X509_find_by_subject.
I'll submit that shortly. Thanks!
101 - 200 of 1073 matches
Mail list logo