From: Eli Zaretskii <eliz@gnu.org>
To: Daniel Colascione <dancol@dancol.org>
Cc: aurelien.aptel+emacs@gmail.com, p.stephani2@gmail.com,
eggert@cs.ucla.edu, tzz@lifelogs.com, emacs-devel@gnu.org
Subject: Re: Dynamic modules: MODULE_HANDLE_SIGNALS etc.
Date: Wed, 23 Dec 2015 18:07:26 +0200 [thread overview]
Message-ID: <83poxxp2rl.fsf@gnu.org> (raw)
In-Reply-To: <5679BE1D.5070903@dancol.org> (message from Daniel Colascione on Tue, 22 Dec 2015 13:18:21 -0800)
> Cc: eggert@cs.ucla.edu, aurelien.aptel+emacs@gmail.com,
> p.stephani2@gmail.com, tzz@lifelogs.com, emacs-devel@gnu.org
> From: Daniel Colascione <dancol@dancol.org>
> Date: Tue, 22 Dec 2015 13:18:21 -0800
>
> >> Which is why you setjmp in places where you have a significant stack
> >> reserve.
> >
> > There's no way of doing that portably, or even non-portably on many
> > platforms. You simply don't _know_ how much stack is left.
>
> You can probe at program start and pre-allocate as much as is reasonable.
Pre-allocate what? Are you suggesting that Emacs allocates its own
stack, instead of relying on the one provided by the linker and the
OS?
> >> Longjmp, by itself, is simple and clear. What's unreliable is longjmping
> >> to Lisp at completely arbitrary points in the program, even ones marked
> >> "GC can't happen here" and the like.
> >
> > We longjmp to a particular place, not arbitrary place.
>
> But we longjmp _from_ anywhere, and "anywhere" might be in the middle of
> any delicate code sequence, since the compiler can generate code to
> write to new stack slots at any point.
I simply don't see any trouble this could cause, except leaking some
memory. Can you describe in enough detail a single use case where
this could have any other adverse effects that we should care about
when recovering from stack overflow?
> >> You say Emacs shouldn't crash. Fine. We can't make that guarantee
> >> if the crash recovery code breaks program invariants.
> >
> > Crash recovery doesn't need to keep invariants. Or maybe I
> > misunderstand what invariants do you have in mind.
>
> Any stack allocation anywhere in the program can longjmp. It's
> impossible to reason about safety in that situation.
Emacs is not safety-critical software, so there's no requirement to
reason about safety. Since I think the recovery's only role is to
allow the user to exit Emacs in a controlled way without losing work,
I simply don't see any problem that could be caused by longjmping from
an arbitrary stack allocation. After all, stack allocation is just
assignment of value to a register, and sometimes grafting a range of
memory pages into the memory set.
> >> Failing that, we should allocate guard pages, unprotect the guard
> >> pages on overflow
> >
> > Thats what the OS is for. It would be wrong for us to start messing
> > with page protection etc. The exception caused by stack overflow
> > removes protection from the guard page to let you do something simple,
> > like run the exception handler -- are you suggesting we catch the
> > exception and mess with protection bits as well, i.e. replace one of
> > the core functions of a modern OS? All that because what we have now
> > is not elegant enough for us? Doesn't sound right to me.
> We have a program that has its own Lisp runtime, has its own memory
> allocation system, uses its own virtual filesystem access layer, and
> that brings itself back from the dead. We're well past replicating OS
> functionality.
Actually, most of the above is simply untrue: we use system allocators
to allocate memory, use mundane C APIs like 'open' and 'read' to
access files, and if by "bringing itself from the dead" you allude to
unexec, then what it does is a subset of what every linker does,
hardly an OS stuff.
I think we should strive to distance ourselves from the OS business,
not the other way around. There was time when doing complex things
sometimes required messing with low-level functionality like that, but
that time is long passed. Allocating our own stack, setting up and
managing our own guard pages and the related exceptions -- we
shouldn't go back there.
> It's not a matter of elegance: it's a matter of correctness. The current
> scheme is unsafe.
Emacs is not safety-critical software. It doesn't need to be "safe"
by your definition, if I understand it correctly.
> >> and call out_of_memory so that it's obvious Emacs is in a bad
> >> state. This way, we don't have to longjmp out of arbitrary code
> >> sequences.
> >
> > There's no problem longjmping out of arbitrary code sequences. When
> > you debug a program, you do that all the time.
>
> In GDB, interrupting normal control flow is not part of standard
> debugging practice.
??? Every time a debuggee hits a breakpoint, the normal control flow
is interrupted, and you in effect have a huge longjmp -- from the
debuggee to the debugger.
next prev parent reply other threads:[~2015-12-23 16:07 UTC|newest]
Thread overview: 177+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-11-25 18:39 Dynamic modules: MODULE_HANDLE_SIGNALS etc Eli Zaretskii
2015-11-25 18:50 ` Philipp Stephani
2015-11-25 19:24 ` Eli Zaretskii
2015-11-26 21:29 ` Paul Eggert
2015-11-27 7:35 ` Eli Zaretskii
2015-11-27 19:19 ` Philipp Stephani
2015-11-28 10:58 ` Philipp Stephani
2015-11-28 12:10 ` Eli Zaretskii
2015-12-19 21:03 ` Philipp Stephani
2015-12-19 22:57 ` Philipp Stephani
2015-12-20 15:47 ` Eli Zaretskii
2015-12-20 18:34 ` Philipp Stephani
2015-12-20 19:11 ` Eli Zaretskii
2015-12-20 21:40 ` Paul Eggert
2015-12-21 3:33 ` Eli Zaretskii
2015-12-21 11:00 ` Paul Eggert
2015-12-21 11:21 ` Yuri Khan
2015-12-21 11:34 ` Paul Eggert
2015-12-21 15:46 ` Eli Zaretskii
2015-12-21 18:15 ` Paul Eggert
2015-12-21 18:28 ` Daniel Colascione
2015-12-21 19:00 ` Eli Zaretskii
2015-12-21 20:19 ` Philipp Stephani
2015-12-21 19:04 ` Eli Zaretskii
2015-12-22 4:09 ` Paul Eggert
2015-12-22 4:38 ` Daniel Colascione
2015-12-22 4:48 ` Paul Eggert
2015-12-22 4:52 ` Daniel Colascione
2015-12-22 6:09 ` Paul Eggert
2015-12-22 6:14 ` Daniel Colascione
2015-12-22 6:33 ` Paul Eggert
2015-12-22 6:35 ` Daniel Colascione
2015-12-22 6:44 ` Paul Eggert
2015-12-22 6:53 ` Daniel Colascione
2015-12-22 16:13 ` Eli Zaretskii
2015-12-22 16:12 ` Eli Zaretskii
2015-12-22 17:26 ` Philipp Stephani
2015-12-22 17:51 ` Eli Zaretskii
2015-12-22 16:03 ` Eli Zaretskii
2015-12-22 16:39 ` Paul Eggert
2015-12-22 17:46 ` Eli Zaretskii
2015-12-22 23:28 ` Paul Eggert
2015-12-23 16:10 ` Eli Zaretskii
2015-12-23 16:20 ` Philipp Stephani
2015-12-23 16:46 ` Eli Zaretskii
2015-12-23 17:09 ` Paul Eggert
2015-12-23 17:18 ` Daniel Colascione
2015-12-24 2:51 ` Paul Eggert
2015-12-24 3:11 ` Daniel Colascione
2015-12-24 16:10 ` Eli Zaretskii
2015-12-24 17:04 ` Daniel Colascione
2015-12-24 17:17 ` John Wiegley
2016-01-03 14:27 ` Daniel Colascione
2016-01-03 15:46 ` Eli Zaretskii
2016-01-03 15:49 ` Daniel Colascione
2016-01-03 16:40 ` Eli Zaretskii
2016-01-03 16:50 ` Daniel Colascione
2016-01-03 17:20 ` Eli Zaretskii
2016-01-03 16:31 ` Paul Eggert
2016-01-03 16:48 ` Daniel Colascione
2016-01-03 18:07 ` Paul Eggert
2016-01-03 18:22 ` Daniel Colascione
2016-01-03 21:02 ` Paul Eggert
2016-01-03 21:12 ` Daniel Colascione
2016-01-03 23:11 ` Paul Eggert
2016-01-03 23:22 ` Daniel Colascione
2016-01-03 23:29 ` John Wiegley
2016-01-04 1:05 ` Paul Eggert
2016-01-04 1:07 ` Daniel Colascione
2016-01-04 15:38 ` Eli Zaretskii
2016-01-04 15:40 ` Daniel Colascione
2016-01-04 16:07 ` Eli Zaretskii
2016-01-04 20:32 ` John Wiegley
2016-01-04 20:34 ` Daniel Colascione
2016-01-04 20:35 ` Daniel Colascione
2016-01-04 22:06 ` John Wiegley
2016-01-04 15:24 ` Eli Zaretskii
2016-01-04 15:28 ` Daniel Colascione
2016-01-04 16:00 ` Eli Zaretskii
2016-01-03 17:16 ` Eli Zaretskii
2016-01-03 17:22 ` Daniel Colascione
2016-01-03 17:39 ` Eli Zaretskii
2016-01-03 17:49 ` Daniel Colascione
2016-01-03 18:08 ` Eli Zaretskii
2016-01-03 18:24 ` Daniel Colascione
2016-01-03 18:51 ` Eli Zaretskii
2016-01-03 19:04 ` Daniel Colascione
2016-01-03 19:15 ` Eli Zaretskii
2016-01-03 19:26 ` Daniel Colascione
2016-01-03 19:46 ` Eli Zaretskii
2016-01-03 19:47 ` Daniel Colascione
2016-01-03 19:49 ` John Wiegley
2016-01-03 20:14 ` Daniel Colascione
2016-01-04 3:17 ` Richard Stallman
2016-01-03 18:17 ` Paul Eggert
2016-01-03 17:43 ` Eli Zaretskii
2016-01-03 20:25 ` John Wiegley
2016-01-03 20:47 ` Daniel Colascione
2016-01-03 21:07 ` John Wiegley
2016-01-03 21:28 ` Daniel Colascione
2016-01-03 21:31 ` Daniel Colascione
2016-01-04 15:27 ` Eli Zaretskii
2016-01-04 15:29 ` Daniel Colascione
2016-01-04 16:01 ` Eli Zaretskii
2016-01-03 21:45 ` John Wiegley
2016-01-03 22:20 ` Daniel Colascione
2016-01-03 22:43 ` Crash recovery strategies (was: Dynamic modules: MODULE_HANDLE_SIGNALS etc.) John Wiegley
2016-01-03 22:55 ` Crash recovery strategies Daniel Colascione
2016-01-03 22:59 ` John Wiegley
2016-01-03 23:04 ` Daniel Colascione
2016-01-03 23:20 ` John Wiegley
2016-01-03 23:47 ` John Wiegley
2016-01-03 23:51 ` Daniel Colascione
2016-01-04 0:12 ` John Wiegley
2016-01-04 15:40 ` Eli Zaretskii
2016-01-04 15:44 ` Daniel Colascione
2016-01-04 15:33 ` Eli Zaretskii
2016-01-04 15:34 ` Daniel Colascione
2016-01-04 16:02 ` Eli Zaretskii
2016-01-03 23:21 ` Paul Eggert
2016-01-03 23:24 ` Daniel Colascione
2016-01-03 23:28 ` John Wiegley
2016-01-04 0:51 ` Paul Eggert
2016-01-03 23:27 ` John Wiegley
2016-01-03 23:29 ` Daniel Colascione
2016-01-03 23:33 ` Sending automatic crash reports to the FSF (was: Crash recovery strategies) John Wiegley
2016-01-03 23:36 ` Sending automatic crash reports to the FSF Daniel Colascione
2016-01-03 23:39 ` John Wiegley
2016-01-03 23:48 ` Daniel Colascione
2016-01-04 1:34 ` Crash recovery strategies Drew Adams
2016-01-04 15:32 ` Crash recovery strategies (was: Dynamic modules: MODULE_HANDLE_SIGNALS etc.) Eli Zaretskii
2016-01-04 15:35 ` Crash recovery strategies Daniel Colascione
2016-01-04 16:04 ` Eli Zaretskii
2016-01-05 4:48 ` Richard Stallman
2016-01-05 15:52 ` Eli Zaretskii
2016-01-05 16:37 ` Clément Pit--Claudel
2016-01-05 17:08 ` Eli Zaretskii
2016-01-05 17:38 ` Clément Pit--Claudel
2016-01-04 15:31 ` Dynamic modules: MODULE_HANDLE_SIGNALS etc Eli Zaretskii
2016-01-04 15:41 ` Daniel Colascione
2016-01-04 16:13 ` Eli Zaretskii
2016-01-04 15:29 ` Eli Zaretskii
2016-01-04 15:26 ` Eli Zaretskii
2015-12-24 17:36 ` Eli Zaretskii
2015-12-24 18:06 ` Daniel Colascione
2015-12-24 19:15 ` Eli Zaretskii
2015-12-22 16:01 ` Eli Zaretskii
2015-12-22 16:32 ` John Wiegley
2015-12-22 20:31 ` Daniel Colascione
2015-12-22 20:46 ` Eli Zaretskii
2015-12-22 20:52 ` Daniel Colascione
2015-12-22 21:08 ` Eli Zaretskii
2015-12-22 21:18 ` Daniel Colascione
2015-12-23 16:07 ` Eli Zaretskii [this message]
2015-12-23 16:25 ` Crash robustness (Was: Re: Dynamic modules: MODULE_HANDLE_SIGNALS etc.) Daniel Colascione
2015-12-23 17:30 ` Eli Zaretskii
2015-12-23 17:41 ` Daniel Colascione
2015-12-23 17:55 ` Eli Zaretskii
2015-12-23 17:56 ` Daniel Colascione
2015-12-23 18:09 ` Eli Zaretskii
2015-12-23 18:19 ` Daniel Colascione
2015-12-23 18:45 ` Eli Zaretskii
2015-12-24 3:26 ` Daniel Colascione
2015-12-21 18:57 ` Dynamic modules: MODULE_HANDLE_SIGNALS etc Eli Zaretskii
2015-12-21 20:15 ` Philipp Stephani
2015-12-20 15:48 ` Eli Zaretskii
2015-12-20 18:27 ` Philipp Stephani
2015-12-20 19:00 ` Eli Zaretskii
2015-12-20 21:00 ` Philipp Stephani
2017-03-26 20:18 ` Philipp Stephani
2016-02-29 22:48 ` Philipp Stephani
2016-03-01 16:41 ` Paul Eggert
2016-03-01 21:43 ` Philipp Stephani
2016-03-02 18:54 ` Paul Eggert
2016-03-31 18:44 ` Philipp Stephani
2016-04-01 8:29 ` Paul Eggert
2015-11-28 23:20 ` Paul Eggert
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
List information: https://www.gnu.org/software/emacs/
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=83poxxp2rl.fsf@gnu.org \
--to=eliz@gnu.org \
--cc=aurelien.aptel+emacs@gmail.com \
--cc=dancol@dancol.org \
--cc=eggert@cs.ucla.edu \
--cc=emacs-devel@gnu.org \
--cc=p.stephani2@gmail.com \
--cc=tzz@lifelogs.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 public inbox
https://git.savannah.gnu.org/cgit/emacs.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).