[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <506021.1614264211@warthog.procyon.org.uk>
Date: Thu, 25 Feb 2021 14:43:31 +0000
From: David Howells <dhowells@...hat.com>
To: Eric Snowberg <eric.snowberg@...cle.com>
Cc: dhowells@...hat.com, Jarkko Sakkinen <jarkko@...nel.org>,
=?utf-8?Q?Micka=C3=ABl_Sala=C3=BCn?= <mic@...ux.microsoft.com>,
David Woodhouse <dwmw2@...radead.org>,
keyrings@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] cert: Add kconfig dependency for validate_trust
Eric Snowberg <eric.snowberg@...cle.com> wrote:
> > I've added an extra config option to turn on SYSTEM_REVOCATION_LIST support.
>
> I believe this is ok. However currently, whenever the kernel finds either a
> EFI_CERT_SHA256_GUID or EFI_CERT_X509_SHA256_GUID entry in the dbx, it loads
> it into the blacklist keyring. Then anytime signature validation takes
> place, these entries are referenced. If there is a match, the signature
> check fails. Now with the inclusion of EFI_CERT_X509_GUID, I question why
> we want to enable it thru a Kconfig option, when we don’t for the other two
> types.
But we do.
mark_hash_blacklisted(), is_hash_blacklisted() and is_binary_blacklisted() do
nothing if CONFIG_SYSTEM_BLACKLIST_KEYRING=n.
David
Powered by blists - more mailing lists