GNU bug report logs - #64613
Signalling byte ops should show up in backtraces

Previous Next

To reply to this bug, email your comments to 64613 AT debbugs.gnu.org.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Mattias Engdegård <mattias.engdegard <at> gmail.com>
To: Emacs Bug Report <bug-gnu-emacs <at> gnu.org>
Cc: Stefan Monnier <monnier <at> iro.umontreal.ca>
Subject: Signalling byte ops should show up in backtraces
Date: Fri, 14 Jul 2023 16:20:37 +0200

Many byte ops don't appear in the backtrace if they signal an error. This makes it unnecessarily difficult to find the precise location of the error.

Most operations that replace a function call, such as `cdr`, `+`, `memq` and so on, have this problem.

The tricky part is to include the function in the backtrace without slowing down normal execution.

Message #8 received at 64613 <at> debbugs.gnu.org (full text, mbox):

From: Mattias Engdegård <mattiase <at> acm.org>
To: 64613 <at> debbugs.gnu.org
Cc: Stefan Monnier <monnier <at> iro.umontreal.ca>
Subject: bug#64613: Signalling byte ops should show up in backtraces
Date: Fri, 14 Jul 2023 19:51:51 +0200

A small improvement, only covering car, cdr, setcar, sector, and (partially) nth and elt, was pushed in 8acd52bba4.
This is low-hanging fruit and worth doing for the benefit it brings even if we end up using a more general solution later on.

Message #11 received at 64613 <at> debbugs.gnu.org (full text, mbox):

From: Mattias Engdegård <mattias.engdegard <at> gmail.com>
To: 64613 <at> debbugs.gnu.org
Cc: Alan Mackenzie <acm <at> muc.de>, Stefan Monnier <monnier <at> iro.umontreal.ca>
Subject: Re: bug#64613: Signalling byte ops should show up in backtraces
Date: Wed, 26 Jul 2023 19:07:27 +0200

Now (82f5f3b8a2) the `aref` and `aset` byte-ops provide backtrace records for errors that are discovered in the fast path:

- index argument not a fixnum
- for vector and record arguments, index out of range

For other types there is no such help yet since we call out to Faref and Faset.

We need a general strategy for dealing with such call-outs from the bytecode engine. So far, the backtrace records have all been to cold paths with no new hot-path code.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.