unofficial mirror of bug-gnu-emacs@gnu.org 
 help / color / mirror / code / Atom feed
From: Adam Van Ymeren <adam@vany.ca>
To: Dmitry Gutov <dgutov@yandex.ru>, 40907@debbugs.gnu.org
Subject: bug#40907: Font-lock-mode hanging in Java buffers
Date: Mon, 27 Apr 2020 18:21:09 +0000	[thread overview]
Message-ID: <ecbcc2f2-cb31-86e8-9273-06647f48153a@vany.ca> (raw)
In-Reply-To: <215d8224-219b-ad06-ba4a-f99250e0a56d@yandex.ru>


On 2020-04-27 5:38 p.m., Dmitry Gutov wrote:
> On 27.04.2020 20:30, Adam Van Ymeren wrote:
>> I've been doing a lot of Java work lately, and I'm having issues with 
>> emacs hanging and spinning at 100% CPU.  I believe the culprit to be 
>> font-lock-mode.  C-g doesn't save me either, I usually have to kill 
>> emacs.  Any advice on how to debug this if C-g doesn't work?
>
> You can send the USR2 signal to the Emacs process to try to interrupt 
> what it's doing and enter the debugger:
>
> kill -SIGUSR2 <pid>
>
> To fix this, we'll most likely need the contents of the buffer that 
> has the problem. If it's proprietary, you could try recreating the 
> problem from scratch, or replacing all identifiers in the code with 
> something else, etc.
>
> You could also try a build of Emacs 27 first before doing all that.


Wow SIGUSR2 is magical, thanks for that tip.  I reproduced it once 
visiting a particular file but it's not happening reliably for me now.  
Below is the stacktrace and the contents of the associated buffer, but 
it's not reproducing reliably.  Now that I know about SIGUSR2 I'll 
report back with a reliable repro once I figure one out.


Here's the emacs stacktrace:

Debugger entered--entering a function:

* #f(compiled-function () #<bytecode 0xe5f179>)()

   c-parse-state()

   c-syntactic-skip-backward("^;{}" 5344 t)

   c-fl-decl-start(5927)

   c-context-expand-fl-region(5927 6427)

   #f(compiled-function (fn) #<bytecode 
0xec98d5>)(c-context-expand-fl-region)

   mapc(#f(compiled-function (fn) #<bytecode 0xec98d5>) 
(c-context-expand-fl-region))

   c-before-context-fl-expand-region(5927 6427)

   c-font-lock-fontify-region(5927 6427 nil)

   font-lock-fontify-region(5927 6427)

   #f(compiled-function (fun) #<bytecode 
0x14a42f5>)(font-lock-fontify-region)

   run-hook-wrapped(#f(compiled-function (fun) #<bytecode 0x14a42f5>) 
font-lock-fontify-region)

   jit-lock--run-functions(5927 6427)

   jit-lock-fontify-now(5927 6427)

   jit-lock-function(5927)

   redisplay_internal\ \(C\ function\)()


And the buffer contents are given after ====

===============

/**

  * @license

  * Copyright 2017 The FOAM Authors. All Rights Reserved.

  * http://www.apache.org/licenses/LICENSE-2.0

  */

package foam.dao;

import foam.core.*;

import foam.dao.index.*;

import foam.mlang.order.Comparator;

import foam.mlang.predicate.Or;

import foam.mlang.predicate.Predicate;

import foam.mlang.sink.GroupBy;

import foam.nanos.logger.Logger;

import foam.nanos.pm.PM;

import java.util.ArrayList;

import java.util.List;

import java.util.HashSet;

import java.util.Set;

/**

  The MDAO class for an ordering, fast lookup, single value,

  index multiplexer, or any other MDAO select() assistance class.

  The assitance class TreeIndex implements the

  data nodes that hold the indexed items and plan and execute

  queries. For any particular operational Index, there may be

  many IndexNode instances:

  <pre>

  1---------> TreeIndex(id)

  MDAO: AltIndex 2---------> TreeIndex(propA) ---> TreeIndex(id) -------------> ValueIndex

  | 1x AltIndexNode    | 1x TreeIndexNode    | 14x TreeIndexNodes         | (DAO size)x ValueIndexNodes

  (2 alt subindexes)     (14 nodes)             (each has 0-5 nodes)

  </pre>

  The base AltIndex has two complete subindexes (each holds the entire DAO).

  The TreeIndex on property A has created one TreeIndexNode, holding one tree of 14 nodes.

  Each tree node contains a tail instance of the next level down, thus

  the TreeIndex on id has created 14 TreeIndexNodes. Each of those contains some number

  of tree nodes, each holding one tail instance of the ValueIndex at the end of the chain.

  */

  // TODO: clone and freeze objects stored in memory

public class MDAO

   extends AbstractDAO

{

   protected AltIndex index_;

   protected Object   state_ = null;

   protected Object   writeLock_ = new Object();

   protected Set      unindexed_ = new HashSet();

   public MDAO(ClassInfo of) {

     setOf(of);

     index_ = new AltIndex(new TreeIndex((PropertyInfo) this.of_.getAxiomByName("id")));

   }

   public void addIndex(Index index) {

     synchronized ( writeLock_ ) {

       state_ = index_.addIndex(state_, index);

     }

   }

   /** Add an Index which is for a unique value. Use addIndex() if the index is not unique. **/

   public void addUniqueIndex(PropertyInfo... props) {

     Index i = ValueIndex.instance();

     for ( PropertyInfo prop : props ) i = new TreeIndex(prop, i);

     addIndex(i);

   }

   /** Add an Index which is for a non-unique value. The 'id' property is

    * appended to property list to make it unique.

    **/

   public void addIndex(PropertyInfo... props) {

     Index i = new TreeIndex((PropertyInfo) this.of_.getAxiomByName("id"));

     for ( PropertyInfo prop : props ) i = new TreeIndex(prop, i);

     addIndex(i);

   }

   synchronized Object getState() {

     return state_;

   }

   synchronized void setState(Object state) {

     state_ = state;

   }

   public FObject objIn(FObject obj) {

     return obj.fclone().freeze();

   }

   public FObject objOut(FObject obj) {

     return obj;

   }

   public FObject put_(X x, FObject obj) {

     // Clone and freeze outside of lock to minimize time spent under lock

     obj = objIn(obj);

     synchronized ( writeLock_ ) {

       FObject oldValue = find_(x, obj);

       Object  state    = getState();

       if ( oldValue != null ) {

         state = index_.remove(state, oldValue);

       }

       setState(index_.put(state, obj));

     }

     onPut(obj);

     return obj;

   }

   public FObject remove_(X x, FObject obj) {

     if ( obj == null ) return null;

     FObject found;

     synchronized ( writeLock_ ) {

       found = find_(x, obj);

       if ( found != null ) {

         setState(index_.remove(getState(), found));

       }

     }

     if ( found != null ) {

       onRemove(found);

     }

     return found;

   }

   public FObject find_(X x, Object o) {

     Object state;

     state = getState();

     if ( o == null ) return null;

     // TODO: PM unindexed plans

     return objOut(

       getOf().isInstance(o)

         ? (FObject) index_.planFind(state, getPrimaryKey().get(o)).find(state, getPrimaryKey().get(o))

         : (FObject) index_.planFind(state, o).find(state, o)

     );

   }

   public Sink select_(X x, Sink sink, long skip, long limit, Comparator order, Predicate predicate) {

     Logger     logger = (Logger) x.get("logger");

     SelectPlan plan;

     Predicate  simplePredicate = null;

     PM         pm = null;

     // use partialEval to wipe out such useless predicate such as: And(EQ()) ==> EQ(), And(And(EQ()),GT()) ==> And(EQ(),GT())

     if ( predicate != null ) simplePredicate = predicate.partialEval();

     Object state = getState();

     // We handle OR logic by seperate request from MDAO. We return different plan for each parameter of OR logic.

     if ( simplePredicate instanceof Or ) {

       Sink dependSink = new ArraySink();

       // When we have groupBy, order, skip, limit such requirement, we can't do it separately so I replace a array sink to temporarily holde the whole data

       //Then after the plan wa slelect we change it to the origin sink

       int length = ((Or) simplePredicate).getArgs().length;

       List<Plan> planList = new ArrayList<>();

       for ( int i = 0 ; i < length ; i++ ) {

         Predicate arg = ((Or) simplePredicate).getArgs()[i];

         planList.add(index_.planSelect(state, dependSink, 0, AbstractDAO.MAX_SAFE_INTEGER, null, arg));

       }

       plan = new OrPlan(simplePredicate, planList);

     } else {

       plan = index_.planSelect(state, sink, skip, limit, order, simplePredicate);

     }

     if ( state != null && predicate != null && plan.cost() > 10 && plan.cost() >= index_.size(state) ) {

       pm = new PM(this.getClass(), "MDAO:UnindexedSelect:" + getOf().getId());

       if ( ! unindexed_.contains(getOf().getId())) {

         if ( ! predicate.equals(simplePredicate) &&

              logger != null ) {

             logger.debug(String.format("The original predicate was %s but it was simplified to %s.", predicate.toString(), simplePredicate.toString()));

         }

         unindexed_.add(getOf().getId());

         if ( logger != null ) {

           logger.warning("Unindexed search on MDAO", getOf().getId(), simplePredicate.toString());

         }

       }

     }

     plan.select(state, sink, skip, limit, order, simplePredicate);

     if ( pm != null ) pm.log(x);

     sink.eof();

     return sink;

   }

   public void removeAll_(X x, long skip, long limit, Comparator order, Predicate predicate) {

     if ( predicate == null && skip == 0 && limit == MAX_SAFE_INTEGER ) {

       synchronized ( writeLock_ ) {

         setState(null);

       }

     } else {

       super.removeAll_(x, skip, limit, order, predicate);

     }

   }

}







  reply	other threads:[~2020-04-27 18:21 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-04-27 17:30 bug#40907: Font-lock-mode hanging in Java buffers Adam Van Ymeren
2020-04-27 17:38 ` Dmitry Gutov
2020-04-27 18:21   ` Adam Van Ymeren [this message]
2020-04-29 18:36     ` Alan Mackenzie
2021-08-10 15:59       ` Lars Ingebrigtsen
2021-08-10 16:58         ` Adam Van Ymeren

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=ecbcc2f2-cb31-86e8-9273-06647f48153a@vany.ca \
    --to=adam@vany.ca \
    --cc=40907@debbugs.gnu.org \
    --cc=dgutov@yandex.ru \
    /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).