OASIS Static Analysis Results Interchange Format (SARIF) TC

If length is negative, then it is downwards growing. If it is positive, then it is upwards growing. If it is null then it is unknown as it is just an address not a region. Jim On 4/29/19 4:46 PM, Larry Golding (Myriad Consulting Inc) wrote: > Oh good heavens, but that means that a negative offsetFromParent might not signal a downward-growing structure, so I can't write (as I already have) that "a negative value SHALL mean that the data structure grows from higher addresses towards lower addresses". > > So what constraints _can_ I put on length/offsetFromParent/relative address, and what _is_ the reliable signal that I have a downward growing structure? > > Larry > >

Got it
Original Message----- From: James Kupsch <kupsch@cs.wisc.edu> Sent: Monday, April 29, 2019 2:50 PM To: Larry Golding (Myriad Consulting Inc) <v-lgold@microsoft.com>; Michael Fanning <Michael.Fanning@microsoft.com>; OASIS SARIF TC Discussion List <sarif@lists.oasis-open.org> Subject: Re: [sarif] #401: Improved design of address object design If length is negative, then it is downwards growing. If it is positive, then it is upwards growing. If it is null then it is unknown as it is just an address not a region. Jim On 4/29/19 4:46 PM, Larry Golding (Myriad Consulting Inc) wrote: > Oh good heavens, but that means that a negative offsetFromParent might not signal a downward-growing structure, so I can't write (as I already have) that "a negative value SHALL mean that the data structure grows from higher addresses towards lower addresses". > > So what constraints _can_ I put on length/offsetFromParent/relative address, and what _is_ the reliable signal that I have a downward growing structure? > > Larry > >

Larry: I'm on travel, but I asked Tom Johnson here to confirm that the recent changes in this area will be good for our needs. He replied yes, but spotted a typo. See the attached. Best, -Paul On 4/29/2019 5:51 PM, Larry Golding (Myriad Consulting Inc) wrote: Got it
Original Message----- From: James Kupsch <kupsch@cs.wisc.edu> Sent: Monday, April 29, 2019 2:50 PM To: Larry Golding (Myriad Consulting Inc) <v-lgold@microsoft.com>; Michael Fanning <Michael.Fanning@microsoft.com>; OASIS SARIF TC Discussion List <sarif@lists.oasis-open.org> Subject: Re: [sarif] #401: Improved design of address object design If length is negative, then it is downwards growing. If it is positive, then it is upwards growing. If it is null then it is unknown as it is just an address not a region. Jim On 4/29/19 4:46 PM, Larry Golding (Myriad Consulting Inc) wrote: Oh good heavens, but that means that a negative offsetFromParent might not signal a downward-growing structure, so I can't write (as I already have) that "a negative value SHALL mean that the data structure grows from higher addresses towards lower addresses". So what constraints _can_ I put on length/offsetFromParent/relative address, and what _is_ the reliable signal that I have a downward growing structure? Larry
Original Message----- From: James Kupsch <kupsch@cs.wisc.edu> Sent: Monday, April 29, 2019 2:44 PM To: Larry Golding (Myriad Consulting Inc) <v-lgold@microsoft.com>; Michael Fanning <Michael.Fanning@microsoft.com>; OASIS SARIF TC Discussion List <sarif@lists.oasis-open.org> Subject: Re: [sarif] #401: Improved design of address object design They might not all have the same sign. Generally they should, but if you are trying to report a memory access that is outside of a container the sign might be the opposite, so probably a SHOULD instead of a SHALL. Jim On 4/29/19 4:36 PM, Larry Golding (Myriad Consulting Inc) wrote: Let's talk about length. I argued to Michael that it should just be a positive number of bytes. The only time you would need length to be negative is if you had a downward-growing structure, and your address object had _neither_ offsetFromParent _nor_ relativeAddress to signal the negative value for you. Right? I'm willing to allow length to be negative for that situation (and therefore, Nullable to signal absence). It seems a pity that we have this Threefold Path for determining whether the data structure is downward growing. And I suppose I have to say that of those that are present, all of them have to have the same sign! Now once again I wish for Boolean growsTowardLowAddresses, but ok, we agree that it's too error prone. So Michael are you ok with length/offsetFromParent/relativeAddress all being present, all allowed to be negative, and all required to have the same sign? Larry