[dev] [commits] Horde branch master updated. 900790ecbb506e850e5e6f03823ad293dc2d370d
Jan Schneider
jan at horde.org
Fri Apr 28 14:10:35 UTC 2017
Zitat von Michael J Rubinsky <mrubinsk at horde.org>:
> Quoting Michael J Rubinsky <mrubinsk at horde.org>:
>
>> Quoting Jan Schneider <jan at horde.org>:
>>
>>> Zitat von Michael J. Rubinsky <mrubinsk at horde.org>:
>>>
>>>> The branch "master" has been updated.
>>>> The following is a summary of the commits.
>>>>
>>>> from: 076b00cfe4db6a0fa44b8b1b7d07e96996ab1b33
>>>>
>>>> 76c0d9d Set the micalg header parameter correctly when using SHA-256.
>>>> fa2dca3 Pass the openssl path too.
>>>> 900790e BFN
>>>>
>>>> Summary:
>>>> http://github.com/horde/horde/compare/076b00cfe4db6a0fa44b8b1b7d07e96996ab1b33...900790ecbb506e850e5e6f03823ad293dc2d370d
>>>>
>>>> -----------------------------------------------------------------------
>>>>
>>>> commit 76c0d9d3f2af854e8b34e3536af68100332a203c
>>>> Author: Michael J Rubinsky <mrubinsk at horde.org>
>>>> Date: Thu Apr 27 00:42:20 2017 -0400
>>>>
>>>> Set the micalg header parameter correctly when using SHA-256.
>>>>
>>>> This took me the better part of a day to research and figure out,
>>>> so would appreciate a second pair of eyes and/or some testing.
>>>>
>>>> New-ish versions of openssl use SHA-256 as the message digest alg.
>>>> when smime signing while other versions use SHA-1. This causes some
>>>> clients to reject the signature, which I believe to be the correct
>>>> behavior.
>>>>
>>>> PHP's openssl_pkcs7_* methods don't allow setting or reading the md
>>>> method so we are left with either parsing the entire DER binary
>>>> stream using something like phpseclib and pulling out the digest
>>>> method, forgoing the openssl_* methods and call the openssl executable's
>>>> smime tool to directly to do the signature, or we can use
>>>> openssl executable's
>>>> asn1parse command and search for a known string indicating
>>>> SHA-256 is being used.
>>>>
>>>> The first option is overkill, the second option would defeat the
>>>> purpose of having the more efficient openssl_* methods and would
>>>> require writing out a copy of the private key to temporary storage, so
>>>> I went with the third option.
>>>>
>>>> framework/Crypt/lib/Horde/Crypt/Smime.php | 18 +++++++++++++++++-
>>>> 1 file changed, 17 insertions(+), 1 deletion(-)
>>>>
>>>> http://github.com/horde/horde/commit/76c0d9d3f2af854e8b34e3536af68100332a203c
>>>>
>>>> -----------------------------------------------------------------------
>>>
>>> According to the OpenSSL documentation at the latest stable
>>> version
>>> (https://www.openssl.org/docs/man1.1.0/crypto/PKCS7_sign.html) but
>>> also at the current master version
>>> (https://www.openssl.org/docs/manmaster/man3/PKCS7_sign.html) is
>>> says:
>>>
>>> "If a signer is specified it will use the default digest for the
>>> signing algorithm. This is SHA1 for both RSA and DSA keys."
>>>
>>> The source tells a different story though:
>>> https://github.com/openssl/openssl/blob/6f0ac0e2f27d9240516edb9a23b7863e7ad02898/crypto/dsa/dsa_ameth.c#L499
>>>
>>> And this is the commit that changed the behavior:
>>> https://github.com/openssl/openssl/commit/44e0c2bae4bfd87d770480902618dbccde84fd81
>>>
>>> If we could find out what the default digest method for the used
>>> public keys is, we can use that. I would prefer a solution without
>>> using the openssl binary, because we are actually trying to get
>>> rid of it: https://github.com/horde/horde/pull/218
>>
>>
>> Agreed. I'll take another look at this during my next time block.
>
> Actually, isn't this the actual issue - knowing what openssl's
> default digest method is for any particular signing algorithm? The
> default digest method for these signing algorithms changed from
> SHA-1 to SHA-256 in openssl, none of php's openssl_* methods expose
> this value, we can't explicitly specify one, and we can't parse the
> ASN.1 s/mime signature data using PHP's methods.
>
> Since it looks like it changed in openssl between version 1.0.1 and
> 1.0.2 (thank you for finding that file - I gave up looking). Maybe
> we could version sniff, but i was hesitant to do that since it looks
> like there are upstream patches, at least in Debian, that made this
> change in earlier versions.
>
> Another option would be to use a third party library like phpseclib
> to do the ASN.1 parsing instead of calling out to the openssl binary.
Turned out that all required parameters had already been set by
openssl and are available in $mime_message.
--
Jan Schneider
The Horde Project
https://www.horde.org/
More information about the dev
mailing list