Or use a PLB

By: dmcq (dmcq.delete@this.fano.co.uk), September 28, 2021 10:59 am
Room: Moderated Discussions
Linus Torvalds (torvalds.delete@this.linux-foundation.org) on September 27, 2021 12:58 pm wrote:
> Linus Torvalds (torvalds.delete@this.linux-foundation.org) on September 27, 2021 10:20 am wrote:
> >
> > Oh, the implementation differences can be huge. Do you "integrate" the pointer
> > with the capability data like Cheri does? Or do you keep them independent so
> > that you can use one "segment" with multiple indexes (like i386 did)?
>
> Thinking about it, the biggest difference is not that the i386 model made the segment
> be very clearly separate from the pointer, but that the i386 segment model itself
> was based on indirection and explicit software management of segments.
>
> IOW, you never loaded "a segment". You loaded a "segment selector", and the hardware then
> looked up the actual segment data from a "segment table" (actually, two different ones,
> but at that point we're going off into "irrelevant implementation details" weeds).
>
> That makes perfect sense in one way, but one of the limitations in that segment model was that it didn't
> allow users of segments to easily "sub-segment" things or do other things on their own. The only thing you
> as a user saw was the selector value, and while you could query certain state (like the segment limit),
> you couldn't really do operations on segments - you had to create a new entry in the segment table.
>
> Contrast this to the Cheri model, which doesn't have that indirection through a segment table,
> and you can create a new segment ("capability pointer") from an existing valid one. So you
> can take a big segment, and create a new smaller subset of it (or a read-only one from a
> read-write one) and just use it without going through that "segment table" thing.
>
> Is that "fundamentally different"? No. The i386 model is one of an explicit software-managed segment
> table, while the Cheri one is comes with hardware support for managing your segments.
>
> Does the Cheri model seem much better? Ahh, there's the nub. It sure as heck looks a lot cleaner
> than i386 segments did (I say "i386", but it was really the 286 that introduced them).
>
> But looks can be deceiving.
>
> The Cheri model requires hidden bits of memory from hardware, for example.
> The Cheri model also makes all pointers really really big.
>
> So it's not some kind of "Cheri is obviously better". Cheri looks much nicer, but
> it has serious downsides, and the big difference is that we know the i386 model
> sucked, we don't have the same kind of lots of experience with the Cheri model.
>
> With Cheri you can do pointer compression (and if Cheri were to become popular, I bet people would
> try to do so), and you'd probably end up making your data structures use a lot of "index off a base
> pointer" instead of using a pointer. Because a 128-bit pointer is just really really big and wastes
> a lot of memory (it also has to be 128-bit aligned, because of how that verification bit works).
>
> That "pointer compression" was actually very very natural in the i386 model. It is not natural in the Cheri
> model. It requires that "one single base pointer". And that's fundamentally what the "segment selector" is.
>
> And yes, we've seen this before: people did some of that kind of work when moving to 64-bit
> architectures, and it sure wasn't all fun and games. On the whole, most projects kept using
> pointers, rather than "index plus base pointer", because managing the base pointer is just
> too painful, and it really doesn't work well when you have multiple allocations.
>
> And that "too painful" becomes even more true when you have
> segments / capability pointers and multiple allocations.
>
> Maybe the i386 segment model with pointer compression using
> explicit near and far pointers wasn't wrong after all?
>
> See? i386 segments really aren't so conceptually different from Cheri capability pointers. Very different
> implementation choices, and very different trade-offs. We know the i386 segment model was horrible
> to use. Does Cheri fix the problems? It's not immediately and trivially obvious. That "use index off
> a base pointer" doesn't look so different from near and far pointers, now does it?
>
> There's a reason Cheri is a "research project".
>
> But there is no real reason to claim that capabilities are fundamentally different from segments.
>
> Linus

THe bits of CHERI that worry me most are anything to do with revoking capabilities and stopping them being saved in a caled domain. THe i386 segments are easier that way but they just don't have the same flexibilty of use. I don't think conflating them achieves any useful purpose. There are other segment based system which are much closer like the Burroughs large systems for instance.
< Previous Post in ThreadNext Post in Thread >
TopicPosted ByDate
POWER10 SAP SD benchmarkanon22021/09/06 03:36 PM
  POWER10 SAP SD benchmarkDaniel B2021/09/07 02:31 AM
    "Cores" (and SPEC)Rayla2021/09/07 07:51 AM
      "Cores" (and SPEC)anon2021/09/07 03:56 PM
  POWER10 SAP SD benchmarkAnon2021/09/07 03:24 PM
    POWER10 SAP SD benchmarkAnon2021/09/07 03:27 PM
  Virtually tagged L1-cachessr2021/09/08 05:49 AM
    Virtually tagged L1-cachesdmcq2021/09/08 08:22 AM
      Virtually tagged L1-cachessr2021/09/08 08:56 AM
      Virtually tagged L1-cachesHugo Décharnes2021/09/08 08:58 AM
        Virtually tagged L1-cachessr2021/09/08 10:09 AM
          Virtually tagged L1-cachesHugo Décharnes2021/09/08 10:46 AM
            Virtually tagged L1-cachessr2021/09/08 11:35 AM
              Virtually tagged L1-cachesHugo Décharnes2021/09/08 12:23 PM
                Virtually tagged L1-cachessr2021/09/08 12:40 PM
                  Virtually tagged L1-cachesanon2021/09/09 03:16 AM
                    Virtually tagged L1-cachesKonrad Schwarz2021/09/10 05:19 AM
                      Virtually tagged L1-cachesHugo Décharnes2021/09/10 06:59 AM
                        Virtually tagged L1-cachesanon2021/09/14 03:17 AM
                          Virtually tagged L1-cachesdmcq2021/09/14 09:34 AM
                            Or use a PLB (NT)Paul A. Clayton2021/09/14 09:45 AM
                              Or use a PLBLinus Torvalds2021/09/14 03:27 PM
                                Or use a PLBanon2021/09/15 12:15 AM
                                  Or use a PLBMichael S2021/09/15 03:21 AM
                                    Or use a PLBdmcq2021/09/15 03:42 PM
                                      Or use a PLBKonrad Schwarz2021/09/16 04:24 AM
                                        Or use a PLBMichael S2021/09/16 10:13 AM
                                          Or use a PLB---2021/09/16 01:02 PM
                                  PLB referencePaul A. Clayton2021/09/18 02:35 PM
                                    PLB referenceMichael S2021/09/18 04:14 PM
                                      Demand paging/translation orthogonalPaul A. Clayton2021/09/19 07:33 AM
                                        Demand paging/translation orthogonalMichael S2021/09/19 09:10 AM
                                      PLB referenceCarson2021/09/20 10:19 PM
                                    PLB referencesr2021/09/20 06:02 AM
                                      PLB referenceMichael S2021/09/20 07:03 AM
                                        PLB referenceLinus Torvalds2021/09/20 12:10 PM
                                  Or use a PLBsr2021/09/20 04:32 AM
                              Or use a PLBsr2021/09/21 09:36 AM
                                Or use a PLBLinus Torvalds2021/09/21 10:04 AM
                                  Or use a PLBsr2021/09/21 10:48 AM
                                    Or use a PLBLinus Torvalds2021/09/21 01:55 PM
                                      Or use a PLBsr2021/09/22 06:55 AM
                                        Or use a PLBrwessel2021/09/22 07:09 AM
                                        Or use a PLBLinus Torvalds2021/09/22 11:50 AM
                                          Or use a PLBsr2021/09/22 01:00 PM
                                            Or use a PLBdmcq2021/09/22 04:07 PM
                                            Or use a PLBEtienne Lorrain2021/09/23 08:50 AM
                                          Or use a PLBanon22021/09/22 04:09 PM
                                            Or use a PLBdmcq2021/09/23 02:35 AM
                                          Or use a PLB2021/09/23 09:37 AM
                                            Or use a PLBLinus Torvalds2021/09/23 12:01 PM
                                              Or use a PLBgpd2021/09/24 03:59 AM
                                                Or use a PLBLinus Torvalds2021/09/24 10:45 AM
                                                  Or use a PLBdmcq2021/09/24 12:43 PM
                                                  Or use a PLBsr2021/09/25 10:19 AM
                                                    Or use a PLBLinus Torvalds2021/09/25 10:44 AM
                                                      Or use a PLBsr2021/09/25 11:11 AM
                                                        Or use a PLBLinus Torvalds2021/09/25 11:31 AM
                                                          Or use a PLBsr2021/09/25 11:52 AM
                                                            Or use a PLBLinus Torvalds2021/09/25 12:05 PM
                                                              Or use a PLBsr2021/09/25 12:23 PM
                                                                Or use a PLBrwessel2021/09/25 03:29 PM
                                                                  Or use a PLBsr2021/10/01 12:22 AM
                                                                    Or use a PLBrwessel2021/10/01 06:19 AM
                                                                      Or use a PLBDavid Hess2021/10/01 10:35 AM
                                                                        Or use a PLBrwessel2021/10/02 04:47 AM
                                                                      Or use a PLBsr2021/10/02 11:16 AM
                                                                        Or use a PLBrwessel2021/10/02 11:53 AM
                                                          Or use a PLBLinus Torvalds2021/09/25 11:57 AM
                                                            Or use a PLBsr2021/09/25 12:07 PM
                                                              Or use a PLBLinus Torvalds2021/09/25 12:21 PM
                                                                Or use a PLBsr2021/09/25 12:40 PM
                                                                  Or use a PLBnksingh2021/09/27 09:07 AM
                                                          Or use a PLB2021/09/27 09:02 AM
                                                            Or use a PLBLinus Torvalds2021/09/27 10:20 AM
                                                              Or use a PLBLinus Torvalds2021/09/27 12:58 PM
                                                                Or use a PLBdmcq2021/09/28 10:59 AM
                                              Or use a PLBsr2021/09/25 10:34 AM
                                                Or use a PLBrwessel2021/09/25 03:44 PM
                                                  Or use a PLBsr2021/10/01 01:04 AM
                                                    Or use a PLBrwessel2021/10/01 06:33 AM
                                                      I386 segmentation highlightssr2021/10/04 07:53 AM
                                                        I386 segmentation highlightsAdrian2021/10/04 09:53 AM
                                                          I386 segmentation highlightssr2021/10/04 10:19 AM
                                                        I386 segmentation highlightsrwessel2021/10/04 04:57 PM
                                                          I386 segmentation highlightssr2021/10/05 11:16 AM
                                                            I386 segmentation highlightsMichael S2021/10/05 12:27 PM
                                                            I386 segmentation highlightsrwessel2021/10/05 04:20 PM
                                                Or use a PLBJohnG2021/09/25 10:18 PM
                                              Or use a PLB2021/09/27 07:37 AM
                                                Or use a PLBHeikki Kultala2021/09/28 03:53 AM
                                                  Or use a PLBrwessel2021/09/28 07:29 AM
                                        Or use a PLBDavid Hess2021/09/23 06:00 PM
                                          Or use a PLBAdrian2021/09/24 01:21 AM
                                            Or use a PLBdmcq2021/09/25 12:41 PM
                                        Or use a PLBblaine2021/09/26 11:19 PM
                                          Or use a PLBDavid Hess2021/09/27 11:35 AM
                                            Or use a PLBblaine2021/09/27 05:19 PM
                                            Or use a PLBAdrian2021/09/27 10:40 PM
                                              Or use a PLBAdrian2021/09/27 10:59 PM
                                                Or use a PLBdmcq2021/09/28 07:45 AM
                                              Or use a PLBrwessel2021/09/28 07:45 AM
                                              Or use a PLBDavid Hess2021/09/28 12:50 PM
                                                Or use a PLBEtienne Lorrain2021/09/30 01:25 AM
                                                  Or use a PLBDavid Hess2021/10/01 10:40 AM
                                  MMU privilegessr2021/09/21 11:07 AM
                                    MMU privilegesLinus Torvalds2021/09/21 01:49 PM
                            Virtually tagged L1-cachesKonrad Schwarz2021/09/16 04:18 AM
                          Virtually tagged L1-cachesCarson2021/09/16 01:12 PM
                            Virtually tagged L1-cachesanon22021/09/16 05:16 PM
                              Virtually tagged L1-cachesrwessel2021/09/16 06:29 PM
                          Virtually tagged L1-cachessr2021/09/20 04:20 AM
              Virtually tagged L1-caches---2021/09/08 02:28 PM
                Virtually tagged L1-cachesanonymou52021/09/08 08:28 PM
                  Virtually tagged L1-cachesanonymou52021/09/08 08:34 PM
                  Virtually tagged L1-caches---2021/09/09 10:14 AM
                    Virtually tagged L1-cachesanonymou52021/09/09 10:44 PM
                Multi-threading?David Kanter2021/09/09 09:32 PM
                  Multi-threading?---2021/09/10 09:19 AM
                Virtually tagged L1-cachessr2021/09/11 01:19 AM
                Virtually tagged L1-cachessr2021/09/11 01:36 AM
                  Virtually tagged L1-caches---2021/09/11 09:53 AM
                    Virtually tagged L1-cachessr2021/09/12 12:43 AM
                      Virtually tagged L1-cachesLinus Torvalds2021/09/12 11:10 AM
                        Virtually tagged L1-cachessr2021/09/12 11:57 AM
                          Virtually tagged L1-cachesdmcq2021/09/13 08:31 AM
                            Virtually tagged L1-cachessr2021/09/20 04:11 AM
            Virtually tagged L1-cachessr2021/09/11 02:49 AM
      Virtually tagged L1-cachesLinus Torvalds2021/09/08 12:34 PM
        Virtually tagged L1-cachesdmcq2021/09/09 02:46 AM
          Virtually tagged L1-cachesdmcq2021/09/09 02:58 AM
          Virtually tagged L1-cachessr2021/09/11 01:29 AM
            Virtually tagged L1-cachesdmcq2021/09/11 08:59 AM
              Virtually tagged L1-cachessr2021/09/12 12:57 AM
                Virtually tagged L1-cachesdmcq2021/09/12 08:44 AM
                  Virtually tagged L1-cachessr2021/09/12 09:48 AM
                    Virtually tagged L1-cachesdmcq2021/09/12 01:22 PM
                      Virtually tagged L1-cachessr2021/09/20 04:40 AM
    Where do you see this information? (NT)anon22021/09/09 02:45 AM
      Where do you see this information?sr2021/09/11 01:40 AM
        Where do you see this information?anon22021/09/11 01:53 AM
          Where do you see this information?sr2021/09/11 02:08 AM
            Thank you (NT)anon22021/09/11 04:31 PM
Reply to this Topic
Name:
Email:
Topic:
Body: No Text
How do you spell avocado?