Bug #6939
closed
Uninformative exception in FIPS mode
Added by vo.x (Vit Ondruch) over 11 years ago.
Updated over 11 years ago.
Description
cat /proc/sys/crypto/fips_enabled¶
1
]# irb
irb(main):001:0> require 'openssl'
=> true
irb(main):002:0> OpenSSL::PKey::DH.new(1024)
=> -----BEGIN DH PARAMETERS-----
MIGHAoGBAMjWrD9U8wfqxMEMPBaBnihhTJb6CGgy7Auy1Aark27nFER3RuYY4ZXC
2lZ11/mDhyymW/LPNr8cupYgs5AsZttguT/zhpr6j2sobnjkcvj8T6FkQ42TC4Dw
PS+O+Mdvz1BP8ZUWXV8QBxyxCKCanPVWvPGI8tC5amj9QM66VyUTAgEC
-----END DH PARAMETERS-----
irb(main):003:0> OpenSSL::PKey::DH.new(128)
OpenSSL::PKey::DHError: BN lib
from (irb):3:in initialize' from (irb):3:in
new'
from (irb):3
from /bin/irb:12:in `'
irb(main):004:0>
Could you please provide better exception message? While it is fine that DH.new fails with short key, it is not obvious from the message what is the reason. Thank you.
- Assignee changed from duerst (Martin Dürst) to MartinBosslet (Martin Bosslet)
I'm not sure whether this is possible at all - the message being generated is what OpenSSL itself generates at this point. I'd have to check if there is a reliable way to detect whether we are in FIPS mode or not. Still, I'd prefer if OpenSSL itself provided a better exception message.
- Status changed from Open to Assigned
- Status changed from Assigned to Third Party's Issue
This is indeed a third party issue. The exception message (or better, the lack thereof) is generated by OpenSSL. They're fine for RSA, DSA and EC ("key too short"), but miserable for DH, agreed.
I really wouldn't want to start to improve OpenSSL error messages in the Ruby extension, that's not where this belongs IMHO. We could try to open an issue on the OpenSSL tracker, though.
Closing as TPI for now, please feel free to reopen if your opinions differ from mine.
MartinBosslet (Martin Bosslet) wrote:
We could try to open an issue on the OpenSSL tracker, though.
Would be cool if you can do this.
Also available in: Atom
PDF
Like0
Like0Like0Like0Like0Like0