From: Pip Cet <pipcet@protonmail.com>
To: "Gerd Möllmann" <gerd.moellmann@gmail.com>
Cc: Helmut Eller <eller.helmut@gmail.com>,
Eli Zaretskii <eliz@gnu.org>, Emacs Devel <emacs-devel@gnu.org>
Subject: Re: MPS: weak hash tables
Date: Wed, 03 Jul 2024 06:33:23 +0000 [thread overview]
Message-ID: <1VNw6cPSIpKfxNRqQBpVleX2BDbQuUqwLQzo-C8N-_PRvNNLG3BnhbcWpUJkiJYnOogBvqRTcLApebjqdZel7CgXVx9T0CnPn6_go_AugDA=@protonmail.com> (raw)
In-Reply-To: <m2ikxn9f8b.fsf@pro2.fritz.box>
On Wednesday, July 3rd, 2024 at 06:11, Gerd Möllmann <gerd.moellmann@gmail.com> wrote:
> Helmut Eller eller.helmut@gmail.com writes:
> > On Tue, Jul 02 2024, Pip Cet wrote:
> >
> > > Also, there's the whole caution thing about weak objects containing
> > > only unaligned words or words pointing directly to a base object,
> > > which is only relevant on Unix/i386, IIRC. (MPS emulates instructions
> > > to simulate fine-grained barriers, which is a really cool idea; I'd
> > > still like an option to turn it off though...). That would mean we
> > > have to replace Lisp_Objects and use the ptr member of our union (and
> > > that's the reason I'm using fixnums rather than plain integers for the
> > > hash).
> >
> > Why do you think that the restriction only applies to 32-bit systems?
> > My interpretation of
> >
> > Section 7.4. Caution
> > ...
> > “Aligned pointer” means a word whose numeric value (that is, its value
> > when treated as an unsigned integer) is a multiple of the size of a
> > pointer. If you’re using a 64-bit architecture, that means that an
> > aligned pointer is a multiple of 8 and its bottom three bits are zero.
> > ...
> >
> > is that it applies to 64-bit machines as well.
>
>
> OTOH, when I see this in a bit broader context, namely
>
> 7.3
> ...
>
> Emulation of accesses to protected objects happens when all of the
> following are true:
>
> The object is a weak object allocated in an AWL pool.
>
> The MPS is running on Linux/IA-32 or Windows/IA-32. Extending this
> list to new (reasonable) operating systems should be tolerable (for
> example, macOS/IA-32). Extending this to new processor architectures
> requires more work.
>
> The processor instruction that is accessing the object is of a
> suitable simple form. The MPS doesn’t contain an emulator for all
> possible instructions that might access memory, so currently it only
> recognizes and emulates a simple MOV from memory to a register or
> vice-versa.
>
> Contact us if you need emulation of access to weak references for new
> operating systems, processor architectures, or memory access
> instructions.
>
> 7.4. Caution
>
> Because of the instruction emulation described in Protection faults
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> above, AWL places the following restriction on the format of objects
> allocated in it:
>
> Each slot in an object must either be a valid word-aligned reference,
> or else the bottom bits of the word must be non-zero so that it does
> not look like an aligned pointer.
>
> “Aligned pointer” means...
>
> I'd bet that these restrictions don't matter when emulation is not done,
> which is the case for 64 bit processors, not IA32 etc. And indeed, on my
> machine with arm64 splatting in the marker vectors works just fine.
>
> I can understand that Ravenbrook documents these restrictions, for
> future (payed) developments and so on, but, you know... implementing
> them gets pretty ugly pretty quickly. (And I wonder if the emulation
> brings enough to warrant the effort.)
Frankly, I wonder how they'd feel about a patch to make emulation optional on all architectures, so the restrictions would be optional as well. I'm playing with a qemu IA32 environment, but haven't been able to trigger the emulation code yet (even on IA32, only some very simple instructions are emulated).
I've run into another issue: finalization. MPS's take on that is rather unusual, in that an object can be "finalized" while weak references to it still exist (and destruction can be vetoed by the finalization code creating a new strong reference to it, IIUC). The upshot of this is that this code:
(setq bignum (1+ most-positive-fixnum))
(setq table (make-hash-table :test 'eq :weakness 'key))
(puthash bignum t table)
table
(setq bignum nil)
(setq values nil)
(igc--collect)
table
produces a table with a nonsensical/random bignum as key, because the memory has been freed and reused for something else.
I have a patch here which "splats" finalized pvecs so they become PVEC_FREE, ignores such objects during iteration, and gets rid of the "count" element, instead counting the elements for (hash-table-count weak-table). I'll install it after some more testing, unless a better solution occurs to someone.
> It would be nice if the ugliness could be encapsulated so that one
> doesn't have to see it all the time, as far as that it possible in C
> :-). Or conditionalized, maybe, because with Helmut's idea (which I find
> the right one), we're using and additional word for weak references.
How hard would it be to "just" add struct igc_headers to the remaining non-headered objects? I don't really want to reopen the "get rid of pure space" discussion again, but that's probably the hard part?
Pip
next prev parent reply other threads:[~2024-07-03 6:33 UTC|newest]
Thread overview: 196+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-07-01 20:47 MPS: weak hash tables Gerd Möllmann
2024-07-01 21:16 ` Pip Cet
2024-07-01 23:10 ` Pip Cet
2024-07-02 4:19 ` Gerd Möllmann
2024-07-02 5:47 ` Gerd Möllmann
2024-07-02 6:23 ` Pip Cet
2024-07-02 6:55 ` Gerd Möllmann
2024-07-02 9:15 ` Pip Cet
2024-07-02 9:37 ` Gerd Möllmann
2024-07-02 10:11 ` Gerd Möllmann
2024-07-02 11:36 ` Gerd Möllmann
2024-07-02 13:15 ` Eli Zaretskii
2024-07-02 13:16 ` Gerd Möllmann
2024-07-02 13:42 ` Eli Zaretskii
2024-07-02 15:03 ` Pip Cet
2024-07-02 15:17 ` Helmut Eller
2024-07-02 15:35 ` Eli Zaretskii
2024-07-02 16:34 ` Pip Cet
2024-07-02 18:20 ` Eli Zaretskii
2024-07-02 20:16 ` Pip Cet
2024-07-03 6:30 ` Gerd Möllmann
2024-07-03 11:23 ` Eli Zaretskii
2024-07-03 11:28 ` Gerd Möllmann
2024-07-02 13:02 ` Eli Zaretskii
2024-07-02 12:45 ` Eli Zaretskii
2024-07-02 11:23 ` Helmut Eller
2024-07-03 6:11 ` Gerd Möllmann
2024-07-03 6:33 ` Pip Cet [this message]
2024-07-03 7:04 ` Gerd Möllmann
2024-07-03 7:24 ` Helmut Eller
2024-07-03 7:25 ` Pip Cet
2024-07-03 7:38 ` Gerd Möllmann
2024-07-03 8:26 ` Gerd Möllmann
2024-07-03 9:31 ` Pip Cet
2024-07-03 10:22 ` Gerd Möllmann
2024-07-03 10:41 ` Pip Cet
2024-07-03 11:17 ` Gerd Möllmann
2024-07-03 20:20 ` Pip Cet
2024-07-04 7:17 ` Gerd Möllmann
2024-07-04 15:24 ` Pip Cet
2024-07-04 16:53 ` Gerd Möllmann
2024-07-04 20:05 ` Pip Cet
2024-07-05 3:50 ` Gerd Möllmann
2024-07-05 12:08 ` Pip Cet
2024-07-05 12:54 ` Gerd Möllmann
2024-07-05 13:27 ` Eli Zaretskii
2024-07-05 20:35 ` Pip Cet
2024-07-06 6:10 ` Eli Zaretskii
2024-07-06 6:31 ` Pip Cet
2024-07-06 7:00 ` Eli Zaretskii
2024-07-06 7:40 ` Gerd Möllmann
2024-07-06 9:13 ` Pip Cet
2024-07-06 10:59 ` Eli Zaretskii
2024-07-05 18:14 ` Helmut Eller
2024-07-05 19:25 ` Pip Cet
2024-07-06 3:39 ` Gerd Möllmann
2024-07-06 5:58 ` Pip Cet
2024-07-06 6:20 ` Gerd Möllmann
2024-07-06 6:29 ` Pip Cet
2024-07-06 6:51 ` Gerd Möllmann
2024-07-06 6:46 ` Eli Zaretskii
2024-07-06 9:23 ` Pip Cet
2024-07-06 11:03 ` Eli Zaretskii
2024-07-06 3:38 ` Gerd Möllmann
2024-07-06 9:47 ` Helmut Eller
2024-07-06 10:38 ` Gerd Möllmann
2024-07-06 11:13 ` Eli Zaretskii
2024-07-06 13:50 ` Helmut Eller
2024-07-06 13:59 ` Eli Zaretskii
2024-07-06 14:38 ` Gerd Möllmann
2024-07-06 16:20 ` Helmut Eller
2024-07-06 16:33 ` Eli Zaretskii
2024-07-06 16:48 ` Helmut Eller
2024-07-06 17:21 ` Eli Zaretskii
2024-07-06 17:59 ` Helmut Eller
2024-07-06 18:14 ` Gerd Möllmann
2024-07-06 18:56 ` Eli Zaretskii
2024-07-06 11:37 ` Pip Cet
2024-07-06 11:40 ` Gerd Möllmann
2024-07-06 11:57 ` Gerd Möllmann
2024-07-06 12:03 ` Eli Zaretskii
2024-07-06 12:16 ` Gerd Möllmann
2024-07-06 12:23 ` Pip Cet
2024-07-06 12:39 ` Gerd Möllmann
2024-07-06 12:30 ` Eli Zaretskii
2024-07-06 12:43 ` Gerd Möllmann
2024-07-06 13:53 ` Eli Zaretskii
2024-07-06 12:36 ` Gerd Möllmann
2024-07-06 14:00 ` Helmut Eller
2024-07-06 14:08 ` Gerd Möllmann
2024-07-06 14:24 ` Gerd Möllmann
2024-07-06 14:44 ` Helmut Eller
2024-07-06 14:52 ` Gerd Möllmann
2024-07-06 15:49 ` Pip Cet
2024-07-06 16:31 ` Gerd Möllmann
2024-07-06 16:56 ` Pip Cet
2024-07-06 17:28 ` Gerd Möllmann
2024-07-06 17:31 ` Gerd Möllmann
2024-07-06 18:30 ` Pip Cet
2024-07-06 20:00 ` Gerd Möllmann
2024-07-06 20:09 ` Ihor Radchenko
2024-07-07 3:55 ` Gerd Möllmann
2024-07-07 4:27 ` Gerd Möllmann
2024-07-07 4:30 ` Gerd Möllmann
2024-07-07 6:38 ` Pip Cet
2024-07-07 7:31 ` Gerd Möllmann
2024-07-07 7:44 ` Helmut Eller
2024-07-07 8:10 ` Gerd Möllmann
2024-07-07 8:24 ` Gerd Möllmann
2024-07-07 8:47 ` Pip Cet
2024-07-07 9:24 ` Gerd Möllmann
2024-07-07 9:26 ` Gerd Möllmann
2024-07-07 10:47 ` Eli Zaretskii
2024-07-07 11:19 ` Gerd Möllmann
2024-07-07 14:09 ` Eli Zaretskii
2024-07-07 14:15 ` Gerd Möllmann
2024-07-07 14:42 ` Eli Zaretskii
2024-07-07 14:52 ` Gerd Möllmann
2024-07-07 15:34 ` Eli Zaretskii
2024-07-07 15:36 ` Gerd Möllmann
2024-07-07 16:00 ` Eli Zaretskii
2024-07-07 17:08 ` Gerd Möllmann
2024-07-07 17:49 ` Eli Zaretskii
2024-07-07 18:15 ` Gerd Möllmann
2024-07-07 18:22 ` Eli Zaretskii
2024-07-07 18:29 ` Gerd Möllmann
2024-07-07 14:16 ` Gerd Möllmann
2024-07-07 14:18 ` Gerd Möllmann
2024-07-07 10:57 ` Pip Cet
2024-07-07 11:35 ` Gerd Möllmann
2024-07-07 11:48 ` Gerd Möllmann
2024-07-07 14:07 ` Gerd Möllmann
2024-07-07 14:21 ` Pip Cet
2024-07-07 14:27 ` Gerd Möllmann
2024-07-07 15:22 ` Helmut Eller
2024-07-07 15:40 ` Gerd Möllmann
2024-07-07 15:52 ` Helmut Eller
2024-07-07 15:56 ` Gerd Möllmann
2024-07-07 15:57 ` Pip Cet
2024-07-07 16:26 ` Helmut Eller
2024-07-07 17:03 ` Gerd Möllmann
2024-07-07 18:40 ` Gerd Möllmann
2024-07-07 18:53 ` Helmut Eller
2024-07-07 19:00 ` Gerd Möllmann
2024-07-07 19:31 ` Pip Cet
2024-07-07 19:36 ` Gerd Möllmann
2024-07-08 9:11 ` MPS: commit limit Gerd Möllmann
2024-07-10 6:46 ` Helmut Eller
2024-07-10 7:08 ` Gerd Möllmann
2024-07-16 15:16 ` Helmut Eller
2024-07-16 15:27 ` Eli Zaretskii
2024-07-16 15:43 ` Helmut Eller
2024-07-16 15:54 ` Eli Zaretskii
2024-07-16 16:29 ` Helmut Eller
2024-07-16 16:39 ` Gerd Möllmann
2024-07-16 16:43 ` Pip Cet
2024-07-16 16:56 ` Gerd Möllmann
2024-07-16 15:32 ` Eli Zaretskii
2024-07-16 16:27 ` Helmut Eller
2024-07-16 18:49 ` Pip Cet
2024-07-17 6:15 ` Helmut Eller
2024-07-16 16:32 ` Pip Cet
2024-07-16 16:48 ` Helmut Eller
2024-07-08 5:11 ` MPS: weak hash tables Pip Cet
2024-07-08 5:17 ` Gerd Möllmann
2024-07-08 5:37 ` Pip Cet
2024-07-08 5:43 ` Gerd Möllmann
2024-07-07 8:49 ` Gerd Möllmann
2024-07-08 9:16 ` Andrea Corallo
2024-07-08 9:24 ` Gerd Möllmann
2024-07-08 9:54 ` Andrea Corallo
2024-07-08 10:10 ` Gerd Möllmann
2024-07-08 11:57 ` MPS: out-of-memory Eli Zaretskii
2024-07-08 13:46 ` Gerd Möllmann
2024-07-08 16:45 ` Pip Cet
2024-07-08 18:26 ` Gerd Möllmann
2024-07-08 19:44 ` Pip Cet
2024-07-09 3:58 ` Gerd Möllmann
2024-07-08 18:27 ` Helmut Eller
2024-07-08 18:31 ` Gerd Möllmann
2024-07-08 19:14 ` Eli Zaretskii
2024-07-08 19:35 ` Gerd Möllmann
2024-07-08 19:08 ` Eli Zaretskii
2024-07-08 19:31 ` Pip Cet
2024-07-04 15:22 ` MPS: weak hash tables Helmut Eller
2024-07-04 15:33 ` Pip Cet
2024-07-04 16:46 ` Gerd Möllmann
2024-07-04 16:43 ` Gerd Möllmann
2024-07-02 13:50 ` Mattias Engdegård
2024-07-02 6:57 ` Gerd Möllmann
2024-07-02 7:15 ` Gerd Möllmann
2024-07-02 8:46 ` Ihor Radchenko
2024-07-02 8:59 ` Gerd Möllmann
2024-07-02 9:33 ` Ihor Radchenko
2024-07-02 9:35 ` Pip Cet
2024-07-02 11:03 ` Ihor Radchenko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='1VNw6cPSIpKfxNRqQBpVleX2BDbQuUqwLQzo-C8N-_PRvNNLG3BnhbcWpUJkiJYnOogBvqRTcLApebjqdZel7CgXVx9T0CnPn6_go_AugDA=@protonmail.com' \
--to=pipcet@protonmail.com \
--cc=eliz@gnu.org \
--cc=eller.helmut@gmail.com \
--cc=emacs-devel@gnu.org \
--cc=gerd.moellmann@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this external index
https://git.savannah.gnu.org/cgit/emacs.git
https://git.savannah.gnu.org/cgit/emacs/org-mode.git
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.