2

Ran into a weird problem trying to lock down our AWS ELBs today, specifically trying to deprecate all support for SHA-1.

Having removed options like AES256-SHA ECDHE-RSA-AES256-SHA (among others) from our external ELB, Chrome and Firefox both refused to connect. Oddly, MS Edge and Safari were still able to connect.

Chrome and FF both reported ERR_SSL_VERSION_OR_CIPHER_MISMATCH.

Based on my best understanding, there's no reason that Chrome or FF shouldn't support larger block sizes for SHA hashing, is there? After all, SHA-1 is deprecated in general. We tried this on several workstations/VMs, each running a different OS (Win10 and OS X El Capitan) and the result was the same for Chrome and FF.

Our certificate was also signed with SHA-256.

Hopefully I'm explaining this reasonably well.

*edit: I found a related list, and it looks like SHA-2 hashing isn't supported by Chrome in conjunction with AES-256.

enter image description here

Improve this question

1 Answer 1

Reset to default
3

After all, SHA-1 is deprecated in general.

No. SHA-1 is deprecated for signatures. It is not deprecated when used as HMAC. See also Obsolete cryptography (SHA1) warning although certificate uses SHA256.

Our certificate was also signed with SHA-256.

This is completely unrelated to the choice of cipher.

...there's no reason that Chrome or FF shouldn't support larger block sizes for SHA hashing, is there?

Apart from performance probably no. But there is also no reason to have a "mine is bigger" approach to security. SHA-1 is save as HMAC and even MD5 is save for this purpose. They both are not safe for use in signatures but this is a different thing.

Improve this answer
2
  • Concur. Also, SHA-256 (input) block size is the same as SHA-1, although state and output are larger.
    – dave_thompson_085
    Nov 4, 2015 at 2:39
  • Thanks guys! Big help. I should I have said - I have a government person insisting that SHA-1 is deprecated entirely, and thus taking issue with it being in use. In trying to explain that it's included as an acceptable in asymmetric key exchanges for hashing in the FIPS 140-2 Annex A, it had little effect, so I was trying to see if there was something we could feasibly do to drop it entirely just to avoid the argument. More of a compliance than a security thing. Appreciate the help.
    – thak
    Nov 4, 2015 at 16:44

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .