Gert Egede Bøggild Reporter
	Description • 6 years ago

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.110 Safari/537.36

Steps to reproduce:






My problem is that the crypto.subtle.exportKey to public spki pem key in firefox seems to be broken. The key is exported, but the result can't be used! According to https://diafygi.github.io/webcrypto-examples/ it should be posible, but the result from key export don't work in openssl! Chrome and Safari works without problems using p-256, p-384 and p-521.



Actual results:

Must admit i don't know the inner workings of pem keys, but have noticed that the start og the pem key is always identical. this is what i get: MFYwEAYEK4EEcAYIKoZIzj0DAQcDQgAE



Expected results:

This is what i get from Chrome, Safari and openssl: MFkwEwYHKoZIzj0CAQYIKoZIzj0DAQcDQgAE chrome
MFkwEwYHKoZIzj0CAQYIKoZIzj0DAQcDQgAE Safari
MFkwEwYHKoZIzj0CAQYIKoZIzj0DAQcDQgAE openssl

	Christoph Kerschbaumer [:ckerschb]
	Comment 1 • 6 years ago

JC, can you weigh in that webcrypto bug?

	Dana Keeler (she/her) (use needinfo) (:keeler for reviews)
	Comment 2 • 6 years ago

Can you link to the actual code you're running? The strings you posted would be the start of valid subject public key infos for secp256r1, but the actual key data is missing. Maybe that part got truncated accidentally?

	Daniel Veditz [:dveditz]
	Comment 3 • 6 years ago

The steps to reproduce are missing -- maybe add them as an attachment? without that we'll have to close this "incomplete"

	Gert Egede Bøggild Reporter
	Comment 4 • 6 years ago

Tested with openssl:
[root@vm crypto]# openssl version
OpenSSL 1.0.2k-fips  26 Jan 2017

server private ec pem key "privkey.pem":
-----BEGIN EC PRIVATE KEY-----
MIGkAgEBBDAF+58p3iI/fttgLROtsrIglrTTw4Ug5SbOBHY+DhF5rfATdAmLStxG
GOLK5qWpV82gBwYFK4EEACKhZANiAARncHMp7PmQrJCNoxUqpTRrKEoH8mmcUAzn
/J9YAd7W43/UB2qmJLO0+M6odfHnt1XXdxjNSDe/5OsBK9C++B77IHZzMT2QKzOt
do91AoiTxW+BtumF8NqzefoKbhkZFhE=
-----END EC PRIVATE KEY-----

chrome public ec pem key "key.pem.chrome":
-----BEGIN PUBLIC KEY-----
MHYwEAYHKoZIzj0CAQYFK4EEACIDYgAETwEHC0RCseaFHJSXDR1dHcCrkIhwsu8O
2/jrGiS+EF2hcJDYrOCcizYL7aab/TWNIPWokvO8CfXfkxrzO9ODNeUuuZnVLkTr
00HH1GsAD/yC73UdQW3qr+O+PlbO1Phu
-----END PUBLIC KEY-----

firefox public ec pem key "key.pem.firefox":
-----BEGIN PUBLIC KEY-----
MHMwDQYEK4EEcAYFK4EEACIDYgAE2lNfhm/8Zy00uA7HuiZ+mWgP+G3lZ0CV3xK/
IJS/g2T6UP4lCTvrSQkfr8Xu/iMxYomIAENn3PHy2GLEORwmW+SpqAP19QJv8KNV
8D9dtRKT6/B4feeKsfK9fnCOvbYr
-----END PUBLIC KEY-----

[root@vm crypto]# openssl pkeyutl -derive -inkey privkey.pem -peerkey key.pem.chrome -out chrome.shared.key

file "chrome.shared.key":
,ë}fééõuÇÊÑ*–.íZmë‡Ø€i0)!÷½dD´+î,WüÒŽˆ

[root@vm crypto]# openssl pkeyutl -derive -inkey privkey.pem -peerkey key.pem.firefox -out firefox.shared.key
unable to load Peer Key
Error reading peer key key.pem.firefox
139805568268176:error:0609E09C:digital envelope routines:PKEY_SET_TYPE:unsupported algorithm:p_lib.c:239:
139805568268176:error:0B07706F:x509 certificate routines:X509_PUBKEY_get:unsupported algorithm:x_pubkey.c:148:
139805568268176:error:0906700D:PEM routines:PEM_ASN1_read_bio:ASN1 lib:pem_oth.c:83:
Error setting up peer key

openssp exits with error, and file not written. In php 7.3.0 where i use it in openssl_pkey_derive(pubkey,privkey) the key is not derived. Php uses the same openssl version as above. Chrome is version  71.0.3578.98 and Firefox is version 64.0 (64-bit)

	Dana Keeler (she/her) (use needinfo) (:keeler for reviews)
	Comment 5 • 6 years ago

What code did you run to generate the keys?

	Dana Keeler (she/her) (use needinfo) (:keeler for reviews)
	Comment 6 • 6 years ago

Oh this is just bug 1410403.

	Gert Egede Bøggild Reporter
	Comment 7 • 6 years ago

Does it mean that there is a differences in which algorithm to use for ecdh spki in webcrypto. And that the only options to use webcrypto in conjunction with openssl for ecdh key derivation is chromium based browsers and safari? Edge presumably does not support ecdh until they shift javascript engine sometime in the future.

	Dana Keeler (she/her) (use needinfo) (:keeler for reviews)
	Comment 8 • 6 years ago

No it's just a bug in Firefox. The implementation doesn't set the correct OID, so openssl thinks it doesn't know how to handle it.