[Bug 1084812] New: [aarch64] IPv4 DNS leading to segfaults
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812 Bug ID: 1084812 Summary: [aarch64] IPv4 DNS leading to segfaults Classification: openSUSE Product: openSUSE Tumbleweed Version: Current Hardware: aarch64 OS: openSUSE Factory Status: NEW Severity: Major Priority: P5 - None Component: Network Assignee: bnc-team-screening@forge.provo.novell.com Reporter: afaerber@suse.com QA Contact: qa-bugs@suse.de CC: agraf@suse.com, dimstar@opensuse.org, dmueller@suse.com, malcolmlewis@cableone.net, mbrugger@suse.com, schwab@suse.com, yousaf.kaukab@suse.com Found By: --- Blocker: --- After a zypper dup to latest 20180305 on aarch64 based Raspberry Pi 3, commands such as zypper or ping segfault. Observed both on 4.15.4 and 4.16-rc4. Same observation on other aarch64 devices (e.g., Odroid-C2). # ping google.com Segmentation fault (core dumped) # ping6 google.com PING google.com(fra15s28-in-x0e.1e100.net (2a00:1450:4001:80b::200e)) 56 data bytes 64 bytes from fra15s28-in-x0e.1e100.net (2a00:1450:4001:80b::200e): icmp_seq=1 ttl=57 time=14.0 ms ... This issue is prohibiting further zypper dup's. On armv7hl, Tumbleweed is already at 20180309 and no such symptoms are observed. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c1
--- Comment #1 from Andreas Färber
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c2
Stefan Brüns
... This issue is prohibiting further zypper dup's.
Workaround: add "195.135.221.134 download.opensuse.org" to /etc/hosts -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c3
--- Comment #3 from Stefan Brüns
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c4
--- Comment #4 from Stefan Brüns
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c5
--- Comment #5 from Malcolm Lewis
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
Matthias Brugger
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
Jeroen Pluimers
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
matthias sweertvaegher
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
Remco Beekmans
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
Cor Slot
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c11
--- Comment #11 from Freek de Kruijf
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c16
--- Comment #16 from Jeroen Pluimers
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
Alexander Graf
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c37
--- Comment #37 from Jeroen Pluimers
--fix-cortex-a53-843419 isn't even documented.
I found the 2015 threads https://gcc.gnu.org/ml/gcc-patches/2015-05/msg00017.html and https://sourceware.org/ml/binutils-cvs/2015-04/msg00012.html having a bit more documentation. But I'm confused: these are 2015 indicating "Some early revisions of the Cortex-A53 have an erratum (843419)" which would suggest that more recent revisions are not affected, and having fairly recent Raspberry Pi 3 systems, I kind of expected them to have more recent revisions. Next from the assumptions above, I tried finding out what sets these revisions apart, but my Google foo failed me here. Anyone who can shed some more light on this? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c43
--- Comment #43 from Jeroen Pluimers
... ARM says r0p0, r0p1, r0p2, r0p3, r0p4. ...
Found it at http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0500g/index.h... (In reply to Michael Matz from comment #41)
... I'm guessing it has something to do with bang-for-buck and margins :)
You are right. Seems 2016 only had a second release of r0p4, so no actual change. Bummer. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
Tulio Magno Quites Machado Filho
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c54
--- Comment #54 from Jeroen Pluimers
(In reply to Andreas Färber from comment #0)
... This issue is prohibiting further zypper dup's.
Workaround: add "195.135.221.134 download.opensuse.org" to /etc/hosts
Somehow that does not always work and makes zypper still segfailt (I think because of the MirrorBrain based mirror system in combination with no zypper --verbose level being able to show which mirror it uses before doing the IPv4 lookup over DNS). So I've put your builds in a repository on gitlab including instructions on how to update: https://gitlab.com/wiert.me/public/linux/opensuse/tumbleweed/aarch64/1084182... That should allow many more people to get their system in a kind of working state again. I've tested this on a few systems, one I tried to later do a `zypper dist-upgrade` on, but that failed (not yet found the cause for that, to figure that one out I need to install a few more 20180202 based systems from the JeOS at http://download.opensuse.org/ports/aarch64/tumbleweed/images/) Thanks for those builds. It helped me getting a system to work again, and made me learn a lot of new things in the mean time. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c56
Guillaume GARDET
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c57
--- Comment #57 from matthias sweertvaegher
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c58
--- Comment #58 from Guillaume GARDET
@guillaume , are you suggesting Build20180403 has been compiled with the workarounds enabled?
No, it is affected. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c59
--- Comment #59 from Andreas Färber
Firefox is also affected. You can check openSUSE Tumbleweed AArch64 - Build20180403. Or try to run Firefox on AArch64 hardware.
Please be more specific: Are you saying it needs a rebuild due to being affected by the missing compiler workaround? Can you clarify how to reproduce a crash? Or are you saying Firefox' JIT needs to enable the workaround for dynamic code? In that case please open a new bug depending on this one. Note there's more JITs that may need to be checked for whether they're affected - OpenJDK comes to mind, Mono, ... Our kernel itself has the erratum enabled. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
Andreas Färber
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c60
Guillaume GARDET
Please be more specific: Are you saying it needs a rebuild due to being affected by the missing compiler workaround? Can you clarify how to reproduce a crash?
Just try to start firefox on AArch64 and it will segfault in libnss_dns.so.2 in gaih_getanswer_slice.isra function. If the compiler is fixed, then yes, I guess it would need a rebuild. If I need to open a new bug, please tell me, but it really seems to be the same problem as zypper. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812
http://bugzilla.opensuse.org/show_bug.cgi?id=1084812#c65
Guillaume GARDET
participants (1)
-
bugzilla_noreply@novell.com