On Wed, Aug 15, 2018 at 9:44 PM DJ Lucas <[email protected]> wrote:

> Discussion was requested in bug to see if security issues should be
> fixed before or after release. There are two low severity security
> issues in OpenSSL-1.1.0h currently in rc1. My thoughts, there is a low
> probability of breakage, and since we need to do kernel and headers
> again (again, I assumed this was a necessity instead of discussion), I
> figured OpenSSL would be a shoe-in as a result of that. I don't see a
> problem with it, but understand that there is some risk of breakage. The
> same is true of all the updates, but again, all minimal risk.
>
> I figure at least a few of us haven't updated to the RC yet, so any
> issues would be discovered on those rebuilds (these would only be
> packages between end of LFS and Xorg in BLFS). If nobody else is
> planning on full rebuild for RC, I can commit to rebuilding everything
> that depends on them currently to validate after I complete the few
> items I've assigned to myself.
>
> I've updated all outstanding packages that I've slotted for 8.3,
> including the obligatory glibc rebuild for new kernel headers, and am
> proceeding with those changes in place. Unfortunately, the updates did
> not come in until after I had completed Xorg (2nd milestone on my build
> order). I intend to do a short overnight build of LFS in a different
> prefix.
>
> Thoughts? I guess really, do any editors intend to start from scratch
> for the RC period or is everyone up to date already?
>
> --DJ
>
> --
> http://lists.linuxfromscratch.org/listinfo/lfs-dev
> FAQ: http://www.linuxfromscratch.org/faq/
> Unsubscribe: See the above information page


I'm going to have to start from scratch here at the end of the week to do
validation properly. I think all editors should be running an official
build of the -rc (e.g. not updated), but I'm not 100% sure.

The kernel update contains the L1TF mitigation, which is critical for
anyone who uses qemu or any other virtualization software on an Intel
processor. That in itself is bad enough IMO.

We also need to consider the backlash possible if we do not update to the
latest OpenSSL version prior to release.

In addition, the version of iproute2 in the book currently is designed to
be running against 4.17 headers, not 4.18 headers. As a result, some
features will not be built.

I think a new -rc based off the updates needs to be performed. The kernel
level vulnerability is HIGH (last I checked), and not LOW.

We need to discuss the OpenSSH User Account Disclosure vulnerability that
was just found and shipped straight to oss-security instead of going
through oss-distros first as well, as that was slotted for immediate
disclosure instead of an embargo as is standard [1] - we cannot release
without that fix, unless we plan on making an errata and an announcement
immediately after. This affects every version of OpenSSH from 2.2+ and can
be used to enumerate usernames.

[1] - http://seclists.org/oss-sec/2018/q3/124
-- 
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reply via email to