Segfaults during compilation on AMD Ryzen.

Message

mblnx · Post by **mblnx** » Mon May 15, 2017 12:09 am

I have seem at least one other segfault on my system after disabling opcode cache option, but it happened on a program that I killed, everything else looks stable.

My system has also completely froze at least two times since I built it (once with opcode and once without it). I've got absolutely no messages about the freeze at all, very hard to debug or understand what is going on.

boudin · Post by **boudin** » Mon May 15, 2017 7:32 pm

Hi everybody.
I'm facing the same issue with this config :
* Ryzen R5 1600
* Corsair Vengeance 2666 (2x8 GB)
* Gygabyte B350 gaming 3 (bios F6, AGESA 1004a)
* Kernel 4.9.16 non hardened
* GCC 5.4.0

When I have a segfault during compilation, my whole system becomes unstable. I need to reboot afterward.

I can't disable OP cache in my bios (or I missed the option).

I noticed that the frequency of the error seems mainly to be related to my memory settings.
With default bios settings (extreme profile disabled and lower timings), it was impossible to build anything. With higher timings I was able to build my base system until it showed up
By enabling the extreme profile it is highly unstable again.

Now, by enabling extreme profile but with a voltage of 1.35 instead of 1.2 it's more stable. Still the segfault issue though, but not as much and my system still run, no need to reboot.
I'm gonna continue to tweak my memory settings, but it really seems that this issue is more related to my ram than anything else...

dpaddy · Post by **dpaddy** » Tue May 16, 2017 6:38 pm

Gigabyte AB350M-Gaming 3, BIOS F4 with default settings, four sticks of KHX2133C14/16G.

Code: Select all

tux ~ # uname -a
Linux tux 4.10.8-gentoo #8 SMP Mon May 1 15:03:51 EDT 2017 x86_64 AMD Ryzen 7 1700 Eight-Core Processor AuthenticAMD GNU/Linux

Code: Select all

# gcc-config -l
 [1] x86_64-pc-linux-gnu-4.9.4
 [2] x86_64-pc-linux-gnu-5.4.0
 [3] x86_64-pc-linux-gnu-6.3.0 *

CPU_FLAGS_X86="aes avx avx2 fma3 mmx mmxext popcnt sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3"
CFLAGS="-march=native -O2 -pipe"
MAKEOPTS="-j16"

Had stability issues -- like as described above -- but that was before I reseated stock cooler using arctic silver... now stable.

boudin · Post by **boudin** » Tue May 16, 2017 9:22 pm

I've disabled PowerNow and Cool and Quiet in the bios and switch to the ACPI driver for freq scaling.
It seems to be stable now, I've been building things for 45 minutes without issue.

liewyec · Post by **liewyec** » Thu May 18, 2017 4:35 pm

So on my system the segfaults almost never happen now. I was using grub-mkconfig to generate grub configuration and the older kernel was always first on the list so i was booting old versions of kernel. I noticed it after I compiled stable 4.11. Since then it seems ok i had one segfault in 2 days of compilation. Now i am having different problem. The system started to crash during compilation without any error in logs. I refleshed the latest bios and it seems ok and i am testing stability but that was realy strange.

alfonsor · Post by **alfonsor** » Fri May 19, 2017 9:27 am

hello all

gigabyte gaming k7, ryzen 1600, corsair 3200

this problem is making me crazy, compilation fails if -j12 is in use, it fails less when -j6 is in use

and I can't understand if it is the memory, the mb, the cpu, in my opionion there must be a hardware problem somewhere, but where?

do you people have any news about this damned bug?

Tony0945 · Post by **Tony0945** » Fri May 19, 2017 12:11 pm

alfonsor wrote:gigabyte gaming k7, ryzen 1600, corsair 3200

A lot of people are having trouble with 3200 memory. Make sure you have the latest bios and/or run the memory at stock clock (no tweaking).
If Gigabyte advertises 3200MHZ, complain to Gigabyte support. (That's not blowing you off. They owe you.)

Naib · Post by **Naib** » Fri May 19, 2017 12:36 pm

alfonsor wrote:hello all

gigabyte gaming k7, ryzen 1600, corsair 3200

this problem is making me crazy, compilation fails if -j12 is in use, it fails less when -j6 is in use

and I can't understand if it is the memory, the mb, the cpu, in my opionion there must be a hardware problem somewhere, but where?

do you people have any news about this damned bug?

what bug? you have said nothing about your system just the mobo and the cpu. I can do that as well

dmesg | grep -i carbon; echo; uname -a; echo; /lib/libc.so.6
[ 0.000000] DMI: Micro-Star International Co., Ltd MS-7A32/X370 GAMING PRO CARBON (MS-7A32), BIOS 1.50 04/27/2017

Linux fluidmotion 4.11.0-gentoo #4 SMP PREEMPT Sat May 13 22:22:26 BST 2017 x86_64 AMD Ryzen 5 1600 Six-Core Processor AuthenticAMD GNU/Linux

GNU C Library (Gentoo 2.24-r1 p6) stable release version 2.24, by Roland McGrath et al.
Copyright © 2016 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.
There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
PARTICULAR PURPOSE.
Compiled by GNU CC version 7.1.0.
Available extensions:
C stubs add-on version 2.1.2
crypt add-on version 2.1 by Michael Glad and others
GNU Libidn by Simon Josefsson
Native POSIX Threads Library by Ulrich Drepper et al
BIND-8.2.3-T5B
libc ABIs: UNIQUE IFUNC
For bug reporting instructions, please see:
<http://bugs.gentoo.org/>.

my system is working fine

If you are using a kernel < 4.10 you will have potential issues
If you are using gcc < 5.x you will have potential issues (6.3 starts being good, 7.1 is nice)
If you are not using the latest BIOS you will have potential issues
If your BIOS has set the wrong RAM info (voltage & timing) you will have issues

alfonsor · Post by **alfonsor** » Fri May 19, 2017 1:29 pm

naib, of course I am talking about the topic of the thread; during compilation I have various segfault, expecially when compiling gcc, mesa and gtk2, but not also, f.e. a kernel compilation also failed twice

it seems there is nothing I can do about it, any memory timing, any memory downclocking seems to have no effect at all, so I am not sure it is a memory problem

the system is stable, gentoo run fines, only compilation triggers the bug

Naib · Post by **Naib** » Fri May 19, 2017 1:34 pm

again...
what bug? point me to a ticket.
again...
you have said nothing about your system.

right now you are soapboxing and if you cannot be bothered to help yourself, how do you expect others to help you

alfonsor · Post by **alfonsor** » Fri May 19, 2017 1:54 pm

you are an esquisite person, naib

thanks for you help

Tony0945 · Post by **Tony0945** » Fri May 19, 2017 2:05 pm

alfonsor, you should open a new thread, as your problem may be different from Kresp's and
c1pherx's. Perhaps there should be three threads. I'll leave that to the moderator.

I'm afraid responders will get confused as to what has been tried and what has not. I know I am!

Post by **NeddySeagoon** » Fri May 19, 2017 4:06 pm

alfonsor,

Naib made some good points. You will have problems if you use the stable kernel and gcc.
Ryzen is new, the OSS community is still getting to grips with it.

Post your

Code: Select all

emerge --info

so that we can see what you are running on your Ryzen.

drizzt · Post by **drizzt** » Fri May 19, 2017 7:12 pm

Naib wrote:If you are using a kernel < 5.x you will have potential issues (6.3 starts being good, 7.1 is nice)

Just to avoid confusion:
"gcc" is meant here

Is gcc-7.1 usable without major issues ?

Naib · Post by **Naib** » Fri May 19, 2017 8:16 pm

drizzt wrote:
Naib wrote:If you are using a kernel < 5.x you will have potential issues (6.3 starts being good, 7.1 is nice)
Just to avoid confusion:
"gcc" is meant here

Is gcc-7.1 usable without major issues ?

corrected thanks.

I am using gcc-7.1 and a few others are.
6 to 7 is relatively smooth. There are a few troublesome packages and either so.e CFLAG tweaks or a patch is required.
I ran into issues with inkscape and upstream have fixed for their next reeaae

drizzt · Post by **drizzt** » Sat May 20, 2017 9:18 am

Naib wrote:I am using gcc-7.1 and a few others are.
6 to 7 is relatively smooth. There are a few troublesome packages and either so.e CFLAG tweaks or a patch is required.
I ran into issues with inkscape and upstream have fixed for their next reeaae

Are gcc-6 and gcc-7 binary compatible so I can switch back if needed ?

Post by **NeddySeagoon** » Sat May 20, 2017 10:00 am

drizzt,

Yes, but keep an eye on USE=pie. There has been talk of switching the default setting between gcc-6 and gcc-7.
Do not permit that and switching between gcc-6 and gcc-7 is gcc-config.

drizzt · Post by **drizzt** » Sat May 20, 2017 11:03 am

NeddySeagoon,
thanks for the info. I've set the useflag "-pie" to ensure no pie is used and I am now emerging gcc-7.1.0

alfonsor · Post by **alfonsor** » Sat May 20, 2017 11:15 am

NeddySeagoon wrote:alfonsor,

Naib made some good points. You will have problems if you use the stable kernel and gcc.
Ryzen is new, the OSS community is still getting to grips with it.

Post your
Code: Select all
emerge --info
so that we can see what you are running on your Ryzen.

it seems that this problem, I mean segfault during compilation, happens to many users of linux around the world; it happens to me with a funtoo installation, two gentoo installations and a very frash gentoo installation during emerge @world in a stage 3

to replicate the problem, I just
- open a shell and do a "emerge gcc"
- open another shell and do a "emerge -1 mesa; emerge -1 mesa; emerge -1 mesa"

sooner or later the compilation of mesa (but also of gcc from time to time) stops with a segfault, dmesg shows error 4 or 6 (so it is probably a memory problem)

In gentoo, I re emerged @system with no optimizations (I mean CFLAGS="-O2 -pipe"), I tried gcc 4, 5 and 6, but the results are always the same.

Note that gentoo runs fine, I have no problem under any kind of stress, only compilation triggers the problem. Also, if I use -j6 the problem disappear.

I tried three different ram kits with any frequencies and timings. I can run prime95 for hours in windows with no problem, stability or temperatures related.

So I came here to ask if there were any news about it. F.e. is it a cpu problem? I am starting to think some ryzen cpu come with a broken cache?

Of course this problem is a deal breaker. If I can't fix it I will move to another hw platform asap.

My emerge --info

Code: Select all

Portage 2.3.3 (python 3.4.5-final-0, default/linux/amd64/13.0/desktop/plasma, gcc-5.4.0, glibc-2.23-r3, 4.11.1 x86_64)
=================================================================
System uname: Linux-4.11.1-x86_64-AMD_Ryzen_5_1600_Six-Core_Processor-with-gentoo-2.3
KiB Mem:    16477356 total,   4391984 free
KiB Swap:    8389628 total,   8389628 free
Timestamp of repository gentoo: Wed, 26 Apr 2017 15:00:01 +0000
sh bash 4.3_p48-r1
ld GNU ld (Gentoo 2.26.1 p1.0) 2.26.1
app-shells/bash:          4.3_p48-r1::gentoo
dev-lang/perl:            5.24.1-r1::gentoo
dev-lang/python:          2.7.12::gentoo, 3.4.5::gentoo
dev-util/cmake:           3.7.2::gentoo
dev-util/pkgconfig:       0.28-r2::gentoo
sys-apps/baselayout:      2.3::gentoo
sys-apps/openrc:          0.24.2::gentoo
sys-apps/sandbox:         2.10-r3::gentoo
sys-devel/autoconf:       2.13::gentoo, 2.69::gentoo
sys-devel/automake:       1.11.6-r1::gentoo, 1.15::gentoo
sys-devel/binutils:       2.26.1::gentoo
sys-devel/gcc:            5.4.0-r3::gentoo
sys-devel/gcc-config:     1.7.3::gentoo
sys-devel/libtool:        2.4.6-r3::gentoo
sys-devel/make:           4.2.1::gentoo
sys-kernel/linux-headers: 4.4::gentoo (virtual/os-headers)
sys-libs/glibc:           2.23-r3::gentoo
Repositories:

gentoo
    location: /usr/portage
    sync-type: rsync
    sync-uri: rsync://rsync.gentoo.org/gentoo-portage
    priority: -1000

local
    location: /usr/local/portage
    masters: gentoo

ACCEPT_KEYWORDS="amd64"
ACCEPT_LICENSE="* -@EULA"
CBUILD="x86_64-pc-linux-gnu"
CFLAGS="-O2 -pipe -mno-fma4 -mno-fma -fno-delete-null-pointer-checks"
CHOST="x86_64-pc-linux-gnu"
CONFIG_PROTECT="/etc /usr/share/config /usr/share/gnupg/qualified.txt /usr/share/themes/oxygen-gtk/gtk-2.0"
CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/dconf /etc/env.d /etc/fonts/fonts.conf /etc/gconf /etc/gentoo-release /etc/revdep-rebuild /etc/sandbox.d /etc/terminfo"
CXXFLAGS="-O2 -pipe -mno-fma4 -mno-fma -fno-delete-null-pointer-checks"
DISTDIR="/usr/portage/distfiles"
FCFLAGS="-O2 -pipe"
FEATURES="assume-digests binpkg-logs config-protect-if-modified distlocks ebuild-locks fixlafiles merge-sync news parallel-fetch preserve-libs protect-owned sandbox sfperms strict unknown-features-warn unmerge-logs unmerge-orphans userfetch userpriv usersandbox usersync xattr"
FFLAGS="-O2 -pipe"
GENTOO_MIRRORS="http://distfiles.gentoo.org"
LANG="it_IT.utf8"
LDFLAGS="-Wl,-O1 -Wl,--as-needed"
MAKEOPTS="-j22"
PKGDIR="/usr/portage/packages"
PORTAGE_CONFIGROOT="/"
PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --omit-dir-times --compress --force --whole-file --delete --stats --human-readable --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages --exclude=/.git"
PORTAGE_TMPDIR="/var/tmp"
USE="X a52 aac acl acpi alsa amd64 berkdb bluetooth branding bzip2 cairo cdda cddb cdr cli consolekit cracklib crypt cups cxx dbus declarative dri dts dvd dvdr emboss encode exif fam firefox flac fortran gdbm gif glamor gpm gtk iconv infinality ipv6 jack jpeg kde kipi kwallet lame lcms ldap libnotify lm_sensors mad mng modules mp3 mp4 mpeg multilib ncurses nls nptl nvidia ogg opengl openmp pam pango pcre pdf phonon plasma png policykit ppds qml qt3support qt4 qt5 readline samba sdl seccomp semantic-desktop session spell ssl startup-notification svg tcpd tiff truetype udev udisks unicode upower usb vorbis widgets wxwidgets x264 xattr xcb xcomposite xinerama xml xscreensaver xv xvid zlib" ABI_X86="64" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" APACHE2_MODULES="authn_core authz_core socache_shmcb unixd actions alias auth_basic authn_alias authn_anon authn_dbm authn_default authn_file authz_dbm authz_default authz_groupfile authz_host authz_owner authz_user autoindex cache cgi cgid dav dav_fs dav_lock deflate dir disk_cache env expires ext_filter file_cache filter headers include info log_config logio mem_cache mime mime_magic negotiation rewrite setenvif speling status unique_id userdir usertrack vhost_alias" CALLIGRA_FEATURES="kexi words flow plan sheets stage tables krita karbon braindump author" COLLECTD_PLUGINS="df interface irq load memory rrdtool swap syslog" CPU_FLAGS_X86="aes avx avx2 fma3 mmx mmxext popcnt sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3" ELIBC="glibc" GPSD_PROTOCOLS="ashtech aivdm earthmate evermore fv18 garmin garmintxt gpsclock isync itrax mtk3301 nmea ntrip navcom oceanserver oldstyle oncore rtcm104v2 rtcm104v3 sirf skytraq superstar2 timing tsip tripmate tnt ublox ubx" INPUT_DEVICES="libinput keyboard mouse" KERNEL="linux" L10N="it" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LIBREOFFICE_EXTENSIONS="presenter-console presenter-minimizer" LINGUAS="it en" OFFICE_IMPLEMENTATION="libreoffice" PHP_TARGETS="php5-6" PYTHON_SINGLE_TARGET="python3_4" PYTHON_TARGETS="python2_7 python3_4" RUBY_TARGETS="ruby21 ruby22" USERLAND="GNU" VIDEO_CARDS="nvidia" XTABLES_ADDONS="quota2 psd pknock lscan length2 ipv4options ipset ipp2p iface geoip fuzzy condition tee tarpit sysrq steal rawnat logmark ipmark dhcpmac delude chaos account"
Unset:  CC, CPPFLAGS, CTARGET, CXX, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LC_ALL, PORTAGE_BUNZIP2_COMMAND, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS, USE_PYTHON

Naib · Post by **Naib** » Sat May 20, 2017 11:28 am

You don't have a -march set in your CFLAGS. What are binaries being built for? Without a march does gcc default to native ? If so that's an issue as you are using gcc-5.x

-mtune=generic -march=x86-64". Is the default of not set

alfonsor · Post by **alfonsor** » Sat May 20, 2017 11:35 am

Naib wrote:You don't have a -march set in your CFLAGS. What are binaries being built for? Without a march does gcc default to native ? If so that's an issue as you are using gcc-5.x

-mtune=generic -march=x86-64". Is the default of not set

jane / # gcc -Q --help=target | grep march
-march= x86-64

Doesn't matter · Post by **Doesn't matter** » Fri May 26, 2017 9:00 am

X1600 on a Gigabyte AB350-Gaming, Bios F4. I ran a (not very extensive) test and it seems that disabling 'Core Performance Boost' in the Bios works for me. All other options are enabled or on auto.
Disabling 'Cool n Quiet' reduced the segfaults but didn't eliminate them completely. Disabling 'Global C-State control' didn't change anything.

trippels · Post by **trippels** » Fri May 26, 2017 10:27 am

It is obviously a nasty CPU bug. Until AMD fixes it with a microcode update there is no workaround.

(I've hit the issue myself, because we now have two Ryzen machines in the gcc compile farm.
For me the bug (random non reproducible gcc segfault) only happens very rarely: about once every three days.)

Naib · Post by **Naib** » Fri May 26, 2017 12:26 pm

bold statement. Lets try to remove some of the heysay...
the more people that can fill this out, the greater the chance of finding a common issue

Gentoo & Ryzen config & stability questionair (public)

drizzt · Post by **drizzt** » Fri May 26, 2017 1:48 pm

Naib wrote:bold statement. Lets try to remove some of the heysay...
the more people that can fill this out, the greater the chance of finding a common issue

Gentoo & Ryzen config & stability questionair (public)

Send you my informations. There's a typo within my answer containing the "Mortar Arctic" but I hope this isn't going to be a problem.

My opinion:
I suspect there is a general problem since I tested 2 CPU Versions (1700 & 1600), 2 Motherboards (ASUS PRIME B350M-A and MSI Mortar Arctic), 2 Sets of RAM (both on the OVL) and I always get the same behaviour of rare segfaults during compiling regardless of compiler version (5, 6 & 7) and kernel version(4.10 & 4.11 & 4.12-rc).
But since these segfaults are rare and until now fixable by retrying multiple times I can live with that and wait for a fix.

Edit added RAM