NetBSD Problem Report #53296

From www@NetBSD.org  Thu May 17 16:21:12 2018
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [199.233.217.200])
	(using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
	(Client CN "mail.NetBSD.org", Issuer "mail.NetBSD.org CA" (not verified))
	by mollari.NetBSD.org (Postfix) with ESMTPS id 919567A156
	for <gnats-bugs@gnats.NetBSD.org>; Thu, 17 May 2018 16:21:12 +0000 (UTC)
Message-Id: <20180517162111.5B1807A1FB@mollari.NetBSD.org>
Date: Thu, 17 May 2018 16:21:11 +0000 (UTC)
From: tobiasu@tmux.org
Reply-To: tobiasu@tmux.org
To: gnats-bugs@NetBSD.org
Subject: vim is super-slow (hangs) on sparc64 / select issue
X-Send-Pr-Version: www-1.0

>Number:         53296
>Notify-List:    dholland@netbsd.org, martin@netbsd.org, bsiegert@netbsd.org
>Category:       pkg
>Synopsis:       vim is super-slow (hangs) on sparc64 / select issue
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    morr
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu May 17 16:25:00 +0000 2018
>Closed-Date:    Sat Sep 15 12:40:04 +0000 2018
>Last-Modified:  Sat Sep 15 12:50:01 +0000 2018
>Originator:     Tobias Ulmer
>Release:        8.99.17
>Organization:
>Environment:
NetBSD u60 8.99.17 NetBSD 8.99.17 (GENERIC) #5: Wed May 16 13:48:05 UTC 2018  root@u60:/usr/obj/sys/arch/sparc64/compile/GENERIC sparc64
>Description:
HW: Ultra 60 2x450mhz 2G, plenty fast for what I'm asking it to do.

vim /etc/fstab takes 10 or more seconds display the file.

Cursor movement 'hangs' (feels like using vim over a modem line on the other side of the world). Connection is over local ssh and vi works fine.


Had a look with ktrace. Vim is repeatedly calling select() (~1000+ times) with varying timeout less than 1s. select() returns either 0 or EINVAL.

Patched the kernel to see where EINVAL comes from. It's either because the input struct timeval's tv_usec is already negative or the conversion to timespec produces an integer overflow.

Example: sys___select50: tv={0, -18872}

tv_sec is always 0. tv_usec is suspiciously random.


Had a look into vim's os_unix.c. It's calling select() in RealWaitForChar(), passing in a timeout pointer tvp to a local struct timeval tv;

But no matter how hard I try, gdb insists the value of *tvp is  {0, 0}.  -O2, -O0 make no difference.

I've put in code to make it coredump when tv_usec is != 0 to no avail.

os_unix.c:

6229            if (tvp->tv_usec != 0)
6230                    abort();
6231            ret = select(maxfd + 1, &rfds, &wfds, &efds, tvp);
6232            if (ret < 0)
6233                    abort(); // this hits.

Anyone got an idea how to solve this mystery?





>How-To-Repeat:

>Fix:

>Release-Note:

>Audit-Trail:

Responsible-Changed-From-To: port-sparc64-maintainer->martin
Responsible-Changed-By: martin@NetBSD.org
Responsible-Changed-When: Thu, 17 May 2018 16:34:35 +0000
Responsible-Changed-Why:
Take


From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-sparc64/53296: vim is super-slow (hangs) on sparc64 /
 select issue
Date: Fri, 18 May 2018 09:39:18 +0200

 In userland:

 [Switching to LWP 1 of process 842]
 _select (nfds=<optimized out>, readfds=0xffffffffffffd010, 
     writefds=0xffffffffffffd030, exceptfds=0xffffffffffffd050, 
     tim50=0xffffffffffffd000)

 and:

 (gdb) p *tim50
 $4 = {tv_sec = 0, tv_usec = 1078990848}


 If I don't miscount, that is quite a lot of usec (needs to be < 1s).

 Martin

Responsible-Changed-From-To: martin->pkg-manager
Responsible-Changed-By: martin@NetBSD.org
Responsible-Changed-When: Fri, 18 May 2018 11:43:46 +0000
Responsible-Changed-Why:
Not a sparc64 specific issue.


From: Martin Husemann <martin@duskware.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: pkg/53296: vim is super-slow (hangs) on sparc64 / select issue
Date: Fri, 18 May 2018 13:47:40 +0200

 The source plays stupid tricks with the select declaration, see os_unix.c:

 /*
  * Some systems have a prototype for select() that has (int *) instead of
  * (fd_set *), which is wrong. This define removes that prototype. We define
  * our own prototype below.
  * Don't use it for the Mac, it causes a warning for precompiled headers.
  * TODO: use a configure check for precompiled headers?  
  */
 #if !defined(__APPLE__) && !defined(__TANDEM)
 # define select select_declared_wrong
 #endif


 and then it does not include <sys/select.h>, thus missing the function
 renaming on NetBSD. See brute force hack below for a possible fix.

 Martin

 --- src/os_unix.c.orig	2018-04-28 21:56:44.000000000 +0200
 +++ src/os_unix.c	2018-05-18 13:41:00.874162825 +0200
 @@ -25,8 +25,11 @@
   * Don't use it for the Mac, it causes a warning for precompiled headers.
   * TODO: use a configure check for precompiled headers?
   */
 -#if !defined(__APPLE__) && !defined(__TANDEM)
 +#if !defined(__APPLE__) && !defined(__TANDEM) && !defined(__NetBSD__)
  # define select select_declared_wrong
 +#else
 +#include <sys/select.h>
 +#define	HAVE_SYS_SELECT	1
  #endif

  #include "vim.h"
 @@ -77,7 +80,7 @@
  # endif
  #endif

 -#if defined(HAVE_SELECT)
 +#if defined(HAVE_SELECT) && !defined(HAVE_SYS_SELECT)
  extern int   select(int, fd_set *, fd_set *, fd_set *, struct timeval *);
  #endif


From: David Holland <dholland-pbugs@netbsd.org>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: pkg/53296: vim is super-slow (hangs) on sparc64 / select issue
Date: Wed, 6 Jun 2018 18:19:26 +0000

 On Fri, May 18, 2018 at 11:50:01AM +0000, Martin Husemann wrote:
  >  The source plays stupid tricks with the select declaration, see os_unix.c:
  >  [...]

 Please also report this upstream if you haven't as it's unlikely that
 it fails only on netbsd.

 -- 
 David A. Holland
 dholland@netbsd.org

State-Changed-From-To: open->feedback
State-Changed-By: bsiegert@NetBSD.org
State-Changed-When: Sun, 08 Jul 2018 10:34:40 +0000
State-Changed-Why:
Did anyone on the thread actually report this upstream?


Responsible-Changed-From-To: pkg-manager->morr
Responsible-Changed-By: bsiegert@NetBSD.org
Responsible-Changed-When: Sun, 08 Jul 2018 10:35:13 +0000
Responsible-Changed-Why:
Over to vim maintainer.


From: Tobias Ulmer <tobiasu@tmux.org>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-sparc64/53296: vim is super-slow (hangs) on sparc64 /
 select issue
Date: Thu, 13 Sep 2018 14:01:50 +0200

 This issue is fixed in vim 8.1.0371

 https://github.com/vim/vim/commit/643b6140873e0e6f297df0cbca11bc1ea1f21925

State-Changed-From-To: feedback->closed
State-Changed-By: bsiegert@NetBSD.org
State-Changed-When: Sat, 15 Sep 2018 12:40:04 +0000
State-Changed-Why:
Updated vim to a fixed version. Thanks!


From: "Benny Siegert" <bsiegert@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/53296 CVS commit: pkgsrc/editors
Date: Sat, 15 Sep 2018 12:36:41 +0000

 Module Name:	pkgsrc
 Committed By:	bsiegert
 Date:		Sat Sep 15 12:36:41 UTC 2018

 Modified Files:
 	pkgsrc/editors/vim-gtk2: Makefile
 	pkgsrc/editors/vim-gtk3: Makefile
 	pkgsrc/editors/vim-lang: PLIST
 	pkgsrc/editors/vim-share: Makefile PLIST distinfo version.mk

 Log Message:
 Update vim to 8.1.0390.

 Among other minor changes, this correctly detects the argument types for
 select, fixing PR port-sparc64/53296.


 To generate a diff of this commit:
 cvs rdiff -u -r1.77 -r1.78 pkgsrc/editors/vim-gtk2/Makefile
 cvs rdiff -u -r1.9 -r1.10 pkgsrc/editors/vim-gtk3/Makefile
 cvs rdiff -u -r1.23 -r1.24 pkgsrc/editors/vim-lang/PLIST
 cvs rdiff -u -r1.49 -r1.50 pkgsrc/editors/vim-share/Makefile
 cvs rdiff -u -r1.41 -r1.42 pkgsrc/editors/vim-share/PLIST
 cvs rdiff -u -r1.176 -r1.177 pkgsrc/editors/vim-share/distinfo
 cvs rdiff -u -r1.118 -r1.119 pkgsrc/editors/vim-share/version.mk

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

>Unformatted:

Home	PR Database Search