netdev - Re: [PATCH net-next 2/2] Crypto/chcr: Checking cra_refcnt before unregistering the algorithms

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20200611064244.GA7402@gondor.apana.org.au>
Date:   Thu, 11 Jun 2020 16:42:44 +1000
From:   Herbert Xu <herbert@...dor.apana.org.au>
To:     Ayush Sawal <ayush.sawal@...lsio.com>
Cc:     ayush.sawal@...cdesigners.com, davem@...emloft.net,
        netdev@...r.kernel.org, manojmalviya@...lsio.com
Subject: Re: [PATCH net-next 2/2] Crypto/chcr: Checking cra_refcnt before
 unregistering the algorithms

On Thu, Jun 11, 2020 at 11:38:39AM +0530, Ayush Sawal wrote:
>
> Sorry for this hack, Our problem was when ipsec is under use and device is
> dettached, then chcr_unregister_alg()
> is called which unregisters the algorithms, but as ipsec is established the
> cra_refcnt is not 1 and it gives a kernel bug.
> So i put a check of cra_refcnt there, taking the reference of a crypto
> driver  "marvell/octeontx/otx_cptvf_algs.c"
> is_any_alg_used(void) function where cra_refcnt is checked before
> unregistering the algorithms.

I understand.  The question is how do you want to deal with the
exception.  IOW do you want to leave the algorithm still registered?
If you can keep the algorithm registered you might as well never
unregister it in the first place.

If it has to go then this code path must wait for the users to
disappear first.

Cheers,
-- 
Email: Herbert Xu <herbert@...dor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt