NetBSD Problem Report #55760

From gson@gson.org  Wed Oct 28 07:11:45 2020
Return-Path: <gson@gson.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 D950A1A9239
	for <gnats-bugs@gnats.NetBSD.org>; Wed, 28 Oct 2020 07:11:45 +0000 (UTC)
Message-Id: <20201028071140.0F725253EDE@guava.gson.org>
Date: Wed, 28 Oct 2020 09:11:40 +0200 (EET)
From: gson@gson.org (Andreas Gustafsson)
Reply-To: gson@gson.org (Andreas Gustafsson)
To: gnats-bugs@NetBSD.org
Subject: Multiple t_tsan_*:*profile test cases fail
X-Send-Pr-Version: 3.95

>Number:         55760
>Category:       toolchain
>Synopsis:       Multiple t_tsan_*:*profile test cases fail
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    toolchain-manager
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Oct 28 07:15:00 +0000 2020
>Last-Modified:  Tue Nov 17 08:30:02 +0000 2020
>Originator:     Andreas Gustafsson
>Release:        NetBSD-current
>Organization:
>Environment:
System: NetBSD
Architecture: x86_64
Machine: amd64
>Description:

The following test cases have been consistently failing on real amd64
hardware ever since they were added on September 14:

  usr.bin/c++/t_tsan_data_race:data_race_profile
  usr.bin/c++/t_tsan_heap_use_after_free:heap_use_after_free_profile
  usr.bin/c++/t_tsan_lock_order_inversion:lock_order_inversion_profile
  usr.bin/c++/t_tsan_locked_mutex_destroy:locked_mutex_destroy_profile
  usr.bin/c++/t_tsan_signal_errno:signal_errno_profile
  usr.bin/c++/t_tsan_thread_leak:thread_leak_profile
  usr.bin/c++/t_tsan_vptr_race:vptr_race_profile
  usr.bin/cc/t_tsan_data_race:data_race_profile
  usr.bin/cc/t_tsan_heap_use_after_free:heap_use_after_free_profile
  usr.bin/cc/t_tsan_lock_order_inversion:lock_order_inversion_profile
  usr.bin/cc/t_tsan_locked_mutex_destroy:locked_mutex_destroy_profile
  usr.bin/cc/t_tsan_signal_errno:signal_errno_profile
  usr.bin/cc/t_tsan_thread_leak:thread_leak_profile

Log output from a recent failure on real hardware:

  http://www.gson.org/netbsd/bugs/build/amd64-baremetal/2020/2020.10.26.21.34.10/test.html#usr.bin_cc_t_tsan_data_race_data_race_profile

They are also failing on the qemu-based testbeds:

  http://releng.netbsd.org/b5reports/amd64/2020/2020.10.27.00.07.18/test.html#usr.bin_cc_t_tsan_data_race_data_race_profile

>How-To-Repeat:

>Fix:

>Audit-Trail:
From: Kamil Rytarowski <n54@gmx.com>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: toolchain/55760: Multiple t_tsan_*:*profile test cases fail
Date: Sat, 14 Nov 2020 17:16:23 +0100

 On 28.10.2020 08:15, Andreas Gustafsson wrote:
 > The following test cases have been consistently failing on real amd64
 > hardware ever since they were added on September 14:
 >

 The tests were added in 2018 and merged in 2019. They were developed,
 tested on real amd64 hardware with Clang/LLVM and unless something
 broke, they should still work.

 GCC uses still very old sanitizer snapshot predating finishing of my
 work and the time of developing the tests (in 2018!). At this point I
 recommend to mark these tests as expected-failure and wait for GCC10 in
 base. Meanwhile someone can debug it independently.

From: Andreas Gustafsson <gson@gson.org>
To: Kamil Rytarowski <n54@gmx.com>
Cc: gnats-bugs@NetBSD.org
Subject: Re: toolchain/55760: Multiple t_tsan_*:*profile test cases fail
Date: Sat, 14 Nov 2020 20:47:00 +0200

 Kamil Rytarowski wrote:
 >  > The following test cases have been consistently failing on real amd64
 >  > hardware ever since they were added on September 14:
 >
 >  The tests were added in 2018 and merged in 2019.

 OK, so I should have said "... since they were enabled on 64-bit
 systems using gcc on September 14".

 >  They were developed, tested on real amd64 hardware with Clang/LLVM
 >  and unless something broke, they should still work.

 Maybe they do with Clang/LLVM, but have they ever worked using gcc
 (on NetBSD)?

 >  GCC uses still very old sanitizer snapshot predating finishing of my
 >  work and the time of developing the tests (in 2018!). At this point I
 >  recommend to mark these tests as expected-failure and wait for GCC10 in
 >  base. Meanwhile someone can debug it independently.

 Sure, I can mark them as expected failures.
 -- 
 Andreas Gustafsson, gson@gson.org

From: "Andreas Gustafsson" <gson@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/55760 CVS commit: src/tests/usr.bin
Date: Tue, 17 Nov 2020 08:25:57 +0000

 Module Name:	src
 Committed By:	gson
 Date:		Tue Nov 17 08:25:57 UTC 2020

 Modified Files:
 	src/tests/usr.bin/c++: t_tsan_data_race.sh
 	    t_tsan_heap_use_after_free.sh t_tsan_lock_order_inversion.sh
 	    t_tsan_locked_mutex_destroy.sh t_tsan_signal_errno.sh
 	    t_tsan_thread_leak.sh t_tsan_vptr_race.sh
 	src/tests/usr.bin/cc: t_tsan_data_race.sh t_tsan_heap_use_after_free.sh
 	    t_tsan_lock_order_inversion.sh t_tsan_locked_mutex_destroy.sh
 	    t_tsan_signal_errno.sh t_tsan_thread_leak.sh

 Log Message:
 Mark t_tsan_*:*profile test cases as expected failures as discussed in
 PR toolchain/55760.


 To generate a diff of this commit:
 cvs rdiff -u -r1.4 -r1.5 src/tests/usr.bin/c++/t_tsan_data_race.sh \
     src/tests/usr.bin/c++/t_tsan_heap_use_after_free.sh \
     src/tests/usr.bin/c++/t_tsan_lock_order_inversion.sh \
     src/tests/usr.bin/c++/t_tsan_locked_mutex_destroy.sh \
     src/tests/usr.bin/c++/t_tsan_signal_errno.sh
 cvs rdiff -u -r1.3 -r1.4 src/tests/usr.bin/c++/t_tsan_thread_leak.sh \
     src/tests/usr.bin/c++/t_tsan_vptr_race.sh
 cvs rdiff -u -r1.4 -r1.5 src/tests/usr.bin/cc/t_tsan_data_race.sh \
     src/tests/usr.bin/cc/t_tsan_heap_use_after_free.sh \
     src/tests/usr.bin/cc/t_tsan_lock_order_inversion.sh \
     src/tests/usr.bin/cc/t_tsan_locked_mutex_destroy.sh \
     src/tests/usr.bin/cc/t_tsan_signal_errno.sh \
     src/tests/usr.bin/cc/t_tsan_thread_leak.sh

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

Home	PR Database Search