linux-kernel - Re: [PATCH v7 07/10] tmp, tmp_tis: Implement usage counter for locality

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

Message-ID: <YugCg5s8I5GIRWls@kernel.org>
Date:   Mon, 1 Aug 2022 19:42:43 +0300
From:   Jarkko Sakkinen <jarkko@...nel.org>
To:     Lino Sanfilippo <l.sanfilippo@...bus.com>
Cc:     Lino Sanfilippo <LinoSanfilippo@....de>, peterhuewe@....de,
        jgg@...pe.ca, stefanb@...ux.vnet.ibm.com, linux@...ewoehner.de,
        linux-integrity@...r.kernel.org, linux-kernel@...r.kernel.org,
        lukas@...ner.de, p.rosenberger@...bus.com
Subject: Re: [PATCH v7 07/10] tmp, tmp_tis: Implement usage counter for
 locality

On Thu, Jul 28, 2022 at 07:36:19PM +0200, Lino Sanfilippo wrote:
> 
> 
> On 04.07.22 19:45, Lino Sanfilippo wrote:
> > 
> > 
> > On 01.07.22 01:29, Jarkko Sakkinen wrote:
> > 
> >>
> >> I'm kind of thinking that should tpm_tis_data have a lock for its
> >> contents?
> > 
> > Most of the tpm_tis_data structure elements are set once during init and
> > then never changed but only read. So no need for locking for these. The
> > exceptions I see are
> > 
> > - flags
> > - locality_count
> > - locality
> > 
> > 
> > whereby "flags" is accessed by atomic bit manipulating functions and thus
> > does not need extra locking. "locality_count" is protected by the locality_count_mutex.
> > "locality" is only set in check_locality() which is called from tpm_tis_request_locality_locked()
> > which holds the locality_count_mutex. So check_locality() is also protected by the locality_count_mutex
> > (which for this reason should probably rather be called locality_mutex since it protects both the "locality_count"
> > and the "locality" variable).
> > 
> > There is one other place check_locality() is called from, namely the interrupt handler. This is also the only
> > place in which "locality" could be assigned another value than 0 (aka the default). In this case there
> > is no lock, so this could indeed by racy.
> > 
> > The solution I see for this is:
> > 1. remove the entire loop that checks for the current locality, i.e. this code:
> > 
> > 	if (interrupt & TPM_INTF_LOCALITY_CHANGE_INT)
> > 		for (i = 0; i < 5; i++)
> > 			if (check_locality(chip, i))
> > 				break;
> > 
> > So we avoid "locality" from being changed to something that is not the default.
> > 
> > 
> 
> I wonder if we need tpm_tis_data->locality at all: the claimed locality is already tracked in
> chip->locality and in TPM TIS we never use anything else than locality 0 so it never changes.
> 
> Is there any good reason not to remove it?

I think it would be a great idea to unify them.

BR, Jarkko