Hello CIP core group members,
I'd formally like to propose to add security packages as revision 3. I've already the description sheet before, but I'll share all files for this review again in this mail.
Contents: proposal_SecurityWG_rev03.yml: the full flat proposal file including all source and binary sets with reason, security tag information and so on Requirements_for_proposal_SecurityWG_rev03.xlsx: the same file which I've already sent before to explain the requirement in the standard 2_src-bin_sort_SecurityWG.txt: the 95 proposed package lists simplified with source and binary names 2_src-bin_sort_all.txt: the 179 package lists consisted of the 95 lists for this proposal and the 84 lists already approved by CIP core as a minimal base shown in brackets
I'd like to set the due date for reviewing this proposal by February 21, Friday. It would be very helpful if I can get your feedback, concerning or question in this week due to resolve by the due date.
@kazuhiro3.hayashi@..., Could you proceed this proposal? Thank you for many cooperation.
Thank you all for considering my request, Kent
toggle quoted messageShow quoted text
-----Original Message----- From: cip-dev <cip-dev-bounces@...> On Behalf Of Kento Yoshida Sent: Friday, February 7, 2020 5:58 PM To: Punit Agrawal <punit1.agrawal@...>; kazuhiro3.hayashi@... Cc: cip-security@...; cip-dev@... Subject: Re: [cip-dev] RESUME REQUEST: [cip-core] Package Proposal #1 (Security packages)
Hello reviews,
Thank you for your supporting against our proposal. I'd like to share you the description sheet for our proposal of security packages. Please consider my attachment and the following note.
Note: 1. Added "fail2ban" as the alternative "pam-shield" because "pam-shield" is not well-maintained and replace with "fail2ban" 2. There are 3 packages in bottom that are under discussion to add. They are out of scope for this review but I'd like to explain them, so let me know your ideas if you have. 3. The requirements for hardware functions are out of scope for this review, but tpm2 is concrete example mentioned in the standard, so I'd like to add some packages related tpm2. However, they are options for only using tpm2, so let me know your comments against adding the packages for a specific use case.
BTW,
@kazuhiro3.hayashi@...,
I'd like to create new proposal to add "fail2ban", but the script for generating proposal shows the following error, and I could not generate it.
------------------------------- Source package name: Binary packages:
any> Traceback (most recent call last): File "./generate-proposal.py", line 218, in <module> generate_proposal(common.PDPProposal.ProposalInfo()) File "./generate-proposal.py", line 176, in generate_proposal deb_src_pkg_info = prepare_src_pkg_info(apt, cve, dep_src_pkg, dep_pkg_info.keys()) File "./generate-proposal.py", line 51, in prepare_src_pkg_info dp_list_final = gpd.get_pkg_depends(pkg, apt) File "/home/yoshidak/cip-pkglist/get_pkg_depends.py", line 102, in get_pkg_depends dp_list, dp_vir_pkg_dict = apt.apt_cache_get_depends_list(pkg_name) File "/home/yoshidak/cip-pkglist/common.py", line 222, in apt_cache_get_depends_list dp_info=c[pkg_name] File "/usr/lib/python2.7/dist-packages/apt/cache.py", line 238, in __getitem__ raise KeyError('The cache has no package named %r' % key) KeyError: "The cache has no package named 'any>'" -------------------------------
I think that the reason is below. When I enter the information of "fail2ban", the script get the dependency for it as <python3:any>.
------------------------------- Enter the source package name: fail2ban Choose the required binary packages: 1: fail2ban Input the numbers in comma separated (eg: 1,3,4): 1
fail2ban Choose one of the virtual package provider: <python3:any> 1: python3 Input the number: 1 -<python3:any>:python3 -lsb-base Are any of the binary packages used in target rootfs? 1: True 2: False -------------------------------
Would you confirm this issue?
Best regards, Kent
-----Original Message----- From: Punit Agrawal <punit1.agrawal@...> Sent: Thursday, January 23, 2020 4:35 PM To: Kento Yoshida <kento.yoshida.wz@...> Cc: cip-dev@...; cip-security@... Subject: Re: [cip-dev] RESUME REQUEST: [cip-core] Package Proposal #1 (Security packages)
Thank you for your comments, Yoshida-san. Follow up comments inline.
Kento Yoshida <kento.yoshida.wz@...> writes:
Hello and thank you for your comment, Punit,
-----Original Message----- From: Punit Agrawal <punit1.agrawal@...> Sent: Monday, January 20, 2020 7:29 PM To: Kento Yoshida <kento.yoshida.wz@...> Cc: cip-dev@...; kazuhiro3.hayashi@...; cip-security@... Subject: Re: [cip-dev] RESUME REQUEST: [cip-core] Package Proposal #1 (Security packages)
Hello Yoshida-san,
Kento Yoshida <kento.yoshida.wz@...> writes:
Hello,
I would like to resume our proposal from security working group. As you know, Kazu has modified the script to generate a proposal and posted the minimum base system proposal, and then I created the new proposal.
The difference from the original (rev01) proposal is below: 1. We remove 'duplicity', 'google-authenticator', 'pam-shield' and 'suricata' in the new proposal because they have an issue such as non-well maintained, python version, too much dependencies and so on. We'll separately propose them after solved these issues.
2. The new proposal shows all source package as flat. Thanks to the new script, Kazu!
3. Actually several packages overlap with the proposed packages for minimum base system in Debian, so I added comment them like that.
@kazuhiro3.hayashi@..., Would you check this proposal and set the due date to review it?
Please reply if you have any comments or questions. I have a comment about packages in the proposal that depend on hardware / system features -
* Some packages in the proposal depend on special purpose hardware to provide their functionality. e.g., TPM.
In systems, where TPM is not present (or similar functionality is provided by alternate mechanisms), the TPM related packages will not be useful. e.g., the non-x86 platforms in the CIP reference hardware list.
* Similarly, some packages require the system to be connected to the network.
In both of these situations, I am wondering what is the impact on compliance? Is there a need to also define minimal set of hardware features expected from reference hardware to be able to meet compliance requirements?
How each reference hardware satisfies the requirements should be considered by each reference hardware provider. Agreed.
But without an explicit statement of the requirement, how can a hardware vendor wanting to develop system for CIP users know what features to enable in their system?
If we provide hardware mechanisms similar with TPM to protect credentials and authentications, we can meet compliance requirements. TPM2 specification is more than 2000 pages long with many features and functions. I believe the IEC standard requires a subset of this functionality. The Security WG maybe intimately familiar with the required features but for the reviewers on this list, there isn't any criteria to use for evaluation.
Stating these functional requirements explicitly will serve the dual purpose of -
* Provide an objective criteria for evaluating the package proposal (and discuss alternatives)
* Give hardware / system vendors the features / functions needed by CIP users.
What do you think?
TPM related packages are options in only systems where TPM is implemented as you said. If supporting these packages require too much costs, the necessity of them will diminish. Actually the standard lists TPM as a typical example, so we thought it will be useful to maintain TPM related packages for many users, but their necessities depend on supporting cost. I see - thanks for the background of the TPM-related packages in the proposal.
To help review the package list (and also discuss alternatives), it would help to define the underlying functionality that is required in more detail, e.g., secure key storage, verified boot, etc. It'll make it possible review the proposal more concretely. [...]
|