csaf — archive

[Vincent Danen]

On 04/26/2017, at 10:53 AM, Feng Cao wrote: Hi Stefan, I went through the same idea of making it generic, but found out there were some obstacles which need to be removed by CVSS working group. That is the reason the current 1.2 proposal makes a small change instead of making it generic. Here are the obstacles. First, we cannot control what will be changed in the future CVSS versions, for example, the BaseScore range might be changed, and TemporalScore range might be changed too. In addition, we cannot rely on v3 schema all the times. The safe way so far is to make sure the version of CVSS can be consistent with the exact schema for that version referred inside CVRF. I did talk to my colleague in CVSS working group before, asking if their WG can always commit to [0, 10] and generate a generic data schema which can be independent of the versions. But he cannot make such a promise for such a ongoing effort in CVSS. So these obstacles cannot be removed for CVRF. So I think this idea can be put into CVRF 2.0. We need to keep CVRF 1.2 simple and get 1.2 out ASAP Apologies for the slow response, but I agree with Feng here. I don't think CVRF 1.2 is meant to be a future-looking spec, but a bridge to make CVSSv3 (which has been out for many months now) viable to use with CVRF and then make CSAF 2.0 the more future-looking spec. I think keeping the changes in CVRF 1.2 as simple as possible so that we (Red Hat, at least, is _only_ using CVSSv3 now so we are very invested in seeing 1.2 with this support come out ASAP) can use it is to everyone's benefit. CSAF 2.0 should be where all the future-proofing happens. On 4/25/2017 11:25 PM, Mr. Stefan Hagen wrote: Dear members, to facilitate the business of our TC, I did many simple transformations on the contributed content from v1.1 and the XSD kindly provided by Feng Cao which seeded the v1.2 drafts until now. This night I did a first real modelling that would IMO better represent the transitive task of embedding and relating CVSS information of various versions. I hereby suggest to replace in vuln.xsd the hard wired V2/V3 mimic with the following: A) Type for the vector (cvssVector): Only one type (which constrains the length to the maximum needed for v2 and v3) and hopefully the future versions. String representing the components needed to compute the various scores for CVSS versions 2 and 3. Note, that v3 scores can be longer than v2 scores (up to 138 characters). It is assumed, that future versions of CVSS maintain the structure of the information break down and the required vector length. B) Only a ScoreSet *but* with a required Version attribute (xs:positiveInteger) This will hold the CVSS version that the content conforms to. Versions 2 and 3 fit the structure. It is assumed, that future versions 4, 5, ... will also fit. C) The Container content of every [0, nifty] ScoreSet to be a sequence: BaseScore (cvssv3:zeroToTenDecimalType) [1, 1] TemporalScore (cvssv3:zeroToTenDecimalType) [0, 1] EnvironmentalScore (cvssv3:zeroToTenDecimalType) [0, 1] Vector (vuln:cvssVector) [0, 1] vuln:ProductID [0, infty] I suggest we use the cvssv3:zeroToTenDecimalType or to be more concise declare our own finite field for these 0.0, 0.1, ... 10.0 allowed values. D) Change the path value of the selector in the uniqueness constraint UniqueScoreSetProductID into: .//vuln:CVSSScoreSets/vuln:ScoreSet/vuln:ProductID All in all this embraces the future in a reasonable way, and should allow for easy migration of older documents. Please note, that currently the per CVSSv3 required fixed prefix for vectors is not always used "in the wild" (and understandably so, as our cvrf v1.1 length might forbid legitimate values in CVSS v3. So a Version attribute on a single ScoreSet element instance might be an enhancement for existing documents when being transformed (see the Oracle example, where the CVSS3.0 prefix would now be signalled via the Version attribute value "3". Any feedback greatly appreciated esp. practical problems not known to me in my tower of thought ;-) Details: A) <xs:simpleType name="cvssVector"> <xs:annotation> <xs:documentation xml:lang="en">String representing the components needed to compute the various scores for CVSS versions 2 and 3. Note, that v3 scores can be longer than v2 scores (up to 138 characters). It is assumed, that future versions of CVSS maintain the structure of the information break down and the required vector length.</xs:documentation> </xs:annotation> <xs:restriction base="xs:token"> <xs:maxLength value="140"/> </xs:restriction> </xs:simpleType> B) and C) <xs:element name="ScoreSet" minOccurs="0" maxOccurs="unbounded"> <xs:annotation> <xs:documentation xml:lang="en">CVSS scores for a given product ID. If the ProductID attribute is omitted, the score applies to all vulnerable products.</xs:documentation> </xs:annotation> <xs:complexType> <xs:sequence> <xs:element name="BaseScore" type="cvssv3:zeroToTenDecimalType" minOccurs="1" maxOccurs="1"> <xs:annotation> <xs:documentation xml:lang="en">The CVSS Base Score is the numeric value of the computed CVSS Base Score which should be a float from 0 ? 10.0.</xs:documentation> </xs:annotation> </xs:element> <xs:element name="TemporalScore" type="cvssv3:zeroToTenDecimalType" minOccurs="0" maxOccurs="1"> <xs:annotation> <xs:documentation xml:lang="en">The CVSS Base Score is the numeric value of the computed CVSS Temporal Score which should be a float from 0 ? 10.0.</xs:documentation> </xs:annotation> </xs:element> <xs:element name="EnvironmentalScore" type="cvssv3:zeroToTenDecimalType" minOccurs="0" maxOccurs="1"> <xs:annotation> <xs:documentation xml:lang="en">The CVSS Base Score is the numeric value of the computed CVSS Environmental Score which should be a float from 0 ? 10.0.</xs:documentation> </xs:annotation> </xs:element> <xs:element name="Vector" type="vuln:cvssVector" minOccurs="0" maxOccurs="1"> <xs:annotation> <xs:documentation xml:lang="en">The CVSS Vector string is the official notation that contains all of the values used to compute the Base, Temporal, and Environmental scores.</xs:documentation> </xs:annotation> </xs:element> <xs:element ref="vuln:ProductID" minOccurs="0" maxOccurs="unbounded"/> </xs:sequence> <xs:attribute name="Version" type="xs:positiveInteger" use="required"> <xs:annotation> <xs:documentation xml:lang="en">The CVSS version that the content conforms to. Versions 2 and 3 fit the structure. It is assumed, that future versions 4, 5, ... will also fit.</xs:documentation> </xs:annotation> </xs:attribute> </xs:complexType> D) <xs:unique name="UniqueScoreSetProductID"> <xs:annotation> <xs:documentation xml:lang="en">This is to ensure that each CVSS score set mentions a given ProductID only one.</xs:documentation> </xs:annotation> <xs:selector xpath=".//vuln:CVSSScoreSets/vuln:ScoreSet/vuln:ProductID"/> <xs:field xpath="."/> </xs:unique> All the best, Stefan. --------------------------------------------------------------------- To unsubscribe from this mail list, you must leave the OASIS TC that generates this mail. Follow this link to all your TCs in OASIS at: https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php -- Vincent Danen / Red Hat Product Security