NetBSD Problem Report #55632

From martin@aprisoft.de  Mon Aug 31 13:55:55 2020
Return-Path: <martin@aprisoft.de>
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 8155B1A923A
	for <gnats-bugs@gnats.NetBSD.org>; Mon, 31 Aug 2020 13:55:55 +0000 (UTC)
Message-Id: <20200831135546.888DE5CC7A5@emmas.aprisoft.de>
Date: Mon, 31 Aug 2020 15:55:46 +0200 (CEST)
From: martin@NetBSD.org
Reply-To: martin@NetBSD.org
To: gnats-bugs@NetBSD.org
Subject: tests/net/ipsec/t_ipsec_misc: timeout tests fail randomly
X-Send-Pr-Version: 3.95

>Number:         55632
>Category:       kern
>Synopsis:       tests/net/ipsec/t_ipsec_misc: timeout tests fail randomly
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Aug 31 14:00:00 +0000 2020
>Last-Modified:  Wed Apr 05 00:56:40 +0000 2023
>Originator:     Martin Husemann
>Release:        NetBSD 9.99.72
>Organization:
The NetBSD Foundation, Inc.
>Environment:
System: NetBSD h-pulse.aprisoft.de 9.99.72 NetBSD 9.99.72 (GENERIC64) #155: Sat Aug 29 13:31:41 CEST 2020 martin@seven-days-to-the-wolves.aprisoft.de:/work/src/sys/arch/evbarm/compile/GENERIC64 evbarm
Architecture: aarch64
Machine: evbarm
>Description:

The t_ipsec_misc "timeout" tests randomly fail, and if they fail leave
busy rump_server processes around. This eventually breaks later tests.

>How-To-Repeat:

cd /usr/tests/net/ipsec && atf-run t_ipsec_misc | atf-report

>Fix:
n/a

>Release-Note:

>Audit-Trail:
From: "Martin Husemann" <martin@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/55632 CVS commit: src/tests/net/ipsec
Date: Mon, 31 Aug 2020 14:03:56 +0000

 Module Name:	src
 Committed By:	martin
 Date:		Mon Aug 31 14:03:56 UTC 2020

 Modified Files:
 	src/tests/net/ipsec: t_ipsec_misc.sh

 Log Message:
 Skip timeout tests, pointing to PR 55632.


 To generate a diff of this commit:
 cvs rdiff -u -r1.23 -r1.24 src/tests/net/ipsec/t_ipsec_misc.sh

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

From: matthew green <mrg@eterna.com.au>
To: gnats-bugs@NetBSD.org, martin@NetBSD.org, netbsd-bugs@netbsd.org,
    kern-bug-people@netbsd.org, gnats-admin@netbsd.org
Cc: tech-net@netbsd.org
Subject: Re: kern/55632: tests/net/ipsec/t_ipsec_misc: timeout tests fail randomly
Date: Sun, 06 Dec 2020 19:40:35 +1100

 these tests are generally extremely slow.  the tests/net/ipsec
 group itself takes about 9 minutes on a ryzen 3600 system (for
 reference, the tests/net/npf group takes 2.3 seconds.)

 on a sparc station 20 with 90mhz CPU, tests/net/ipsec takes
 over half a day to complete.

 there's something really awful here because the tests martin
 disabled take a long time to be skipped:

     ipsec_spi_ah_hmacsha512_preferred_new_timeout: [47.253920s] Skipped: PR 55632: test fails randomly, leaving spurious rump_server around

 there's nothing about "skipped" it self being slow eg, this is
 earlier in the log:

     scalbnl_inf_neg: [0.393539s] Skipped: Requires long double support

 while fixing this is less useful than fixing the global
 slowness, it would go a short distance to helping..

 this table shows the per-group and totals for the ipsec groups:

                                                  test    total
 t_ipsec_ah_keys (520/895): 20 test cases         624s     624s
 t_ipsec_esp_keys (521/895): 22 test cases        852s    1476s
 t_ipsec_gif (522/895): 16 test cases            4420s    5896s
 t_ipsec_l2tp (523/895): 16 test cases           4727s   10623s
 t_ipsec_misc (524/895): 44 test cases           5462s   16085s
 t_ipsec_natt (525/895): 4 test cases             555s   16640s
 t_ipsec_sockopt (526/895): 10 test cases        1328s   17968s
 t_ipsec_sysctl (527/895): 3 test cases            52s   18020s
 t_ipsec_tcp (528/895): 15 test cases            1665s   19685s
 t_ipsec_transport (529/895): 44 test cases      5646s   25331s
 t_ipsec_tunnel (530/895): 42 test cases         9739s   36759s
 t_ipsec_tunnel_ipcomp (531/895): 8 test cases   1689s   35070s
 t_ipsec_tunnel_odd (532/895): 42 test cases     9952s   46711s


 are these tests really worth this much CPU?  can we reduce
 this to at least 10% of current run time?  please?


 .mrg.

>Unformatted:

Home	PR Database Search