linux-kernel - Re: Proposal: rename tpm1_eventlog.c and tpm2

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: <6d3c18e4-6033-be6f-5783-fa36e7bbbd86@linux.vnet.ibm.com>
Date:   Tue, 31 Oct 2017 00:04:59 +0530
From:   Nayna Jain <nayna@...ux.vnet.ibm.com>
To:     Jarkko Sakkinen <jarkko.sakkinen@...ux.intel.com>,
        linux-integrity@...r.kernel.org
Cc:     linux-kernel@...r.kernel.org, linux-security-module@...r.kernel.org
Subject: Re: Proposal: rename tpm1_eventlog.c and tpm2_eventlog.c



On 10/25/2017 03:51 AM, Jarkko Sakkinen wrote:
> I noticed when making slides for KS that the naming for event log stuff
> that the naming is so broken that it is hard to understand the code.
> Here it really would make sense to have a patch set just to clean up the
> cruft.
>
> Random examples of more senseful naming:
>
> * tpm2_bios_measurements_start() should be rather something like
>    tpm_eventlog_seq_agile_start().
> * tpm_bios_measurements_start() should be rather something like
>    tpm_eventlog_seq_sha1_start().
BIOS eventlog being exposed as securityfs files are named as
ascii_bios_measurements and binary_bios_measurements.
I thought that the function names were originally written corresponding
to the files they write into. In that context, I didn't find them 
misleading.

Still if we want to rename, I think having tpm_eventlog_seq_sha1_start() for
TPM1.2 might be confusing as even TPM2.0 supports SHA1. And there might
be systems which use only SHA1 even in the case of TPM2.0. I was 
thinking if
we can just call them as tpm2_eventlog_seq_start() and 
tpm1_eventlog_seq_start().
> Corresponding structs would be tpm_eventlog_agile_seq_ops and
> tpm_eventlog_sha1_seq_ops.
>
> Finally, I would place the file operations, being so complicated, in
> separate files:
>
> * tpm_eventlog_seq_sha1.c
> * tpm_eventlog_seq_eventlog.c
>
> And move all the management code that is right now illogically located
> in tpm1_eventlog.c to tpm_eventlog.c that would be the entry point for
> the event log.
By management code, do you mean the functions common to both TPM1.2 and 
TPM2.0 as:

tpm_bios_measurements_open()
tpm_read_log()
tpm_bios_log_setup()
tpm_bios_log_teardown()

So, do you mean to move these to tpm_eventlog.c and keep only specific 
functions in
tpm1_eventlog.c and tpm2_eventlog.c ? If so, then yeah I also agree with 
this.

Thanks & Regards,
    - Nayna
> The code is laid out so badly right now that I have really hard time
> understanding it if I haven't looked at it within last couple of weeks.
> It's really a trainwreck at the moment. We must clean up it up fast.
>
> Getting this done will help me to review patches to this area faster
> so it would be a benefit for everyone. The current structure makes every
> event log patch a pain to review.
>
> /Jarkko
>