linux-kernel - general protection fault with prefetch

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <333cfb75-1769-c67f-c56f-c9458368751a@molgen.mpg.de>
Date:   Wed, 18 Jul 2018 16:31:00 +0200
From:   Paul Menzel <pmenzel+linux-mm@...gen.mpg.de>
To:     Christoph Lameter <cl@...ux.com>,
        Pekka Enberg <penberg@...nel.org>,
        David Rientjes <rientjes@...gle.com>,
        Joonsoo Kim <iamjoonsoo.kim@....com>,
        Andrew Morton <akpm@...ux-foundation.org>, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org
Subject: general protection fault with prefetch_freepointer

Dear Linux,


Loading the amdgpu module on Ryzen 3 2{2,4}00G (Raven) systems sometimes
causes a general protection fault [1]. At least on my system I am unable
to reliably reproduce the issue.

```
[   35.265941] kfd kfd: kgd2kfd_probe failed
[   35.537445] general protection fault: 0000 [#1] SMP NOPTI
[   35.543209] CPU: 0 PID: 367 Comm: systemd-udevd Not tainted 4.18.0-rc5+ #1
[   35.550371] Hardware name: MSI MS-7A37/B350M MORTAR (MS-7A37), BIOS 1.G1 05/17/2018
[   35.558562] RIP: 0010:prefetch_freepointer+0x10/0x20
[   35.563881] Code: 89 d3 e8 c3 fe 4a 00 85 c0 0f 85 31 75 00 00 48 83 c4 08 5b 5d 41 5c 41 5d c3 0f 1f 44 00 00 48 85 f6 74 13 8b 47 20 48 01 c6 <48> 33 36 48 33 b7 38 01 00 00 0f 18 0e c3 66 90 0f 1f 44 00 00 48 
[   35.584215] RSP: 0018:ffff9c3181f77560 EFLAGS: 00010202
[   35.589849] RAX: 0000000000000000 RBX: 4b2c8be0a60f6ab9 RCX: 0000000000000ccc
[   35.597492] RDX: 0000000000000ccb RSI: 4b2c8be0a60f6ab9 RDI: ffff8d5b1e406e80
[   35.605166] RBP: ffff8d5b1e406e80 R08: ffff8d5b1e824f00 R09: ffffffffc0a45423
[   35.612808] R10: fffff1b60fff64c0 R11: ffff9c3181f77520 R12: 00000000006080c0
[   35.620451] R13: 0000000000000230 R14: ffff8d5b1e406e80 R15: ffff8d5b0c304400
[   35.628116] FS:  00007fb194e4b8c0(0000) GS:ffff8d5b1e800000(0000) knlGS:0000000000000000
[   35.636771] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   35.642931] CR2: 000055c0661d60d0 CR3: 000000040b9ee000 CR4: 00000000003406f0
[   35.650566] Call Trace:
[   35.653198]  kmem_cache_alloc_trace+0xb5/0x1c0
[   35.658077]  ? dal_ddc_service_create+0x38/0x110 [amdgpu]
[   35.663962]  dal_ddc_service_create+0x38/0x110 [amdgpu]
```

Do you have more ideas, how to debug that? The progress in bug report
kind of stalled as people have run out of ideas.

I thought about a hardware/firmware problem, but then searching for the
error, I found the stack trace with Linux 4.17.2 from Fedora [2][3].
This looks a little similar, but can of course be something totally
different.

```
general protection fault: 0000 [#1] SMP PTI
Modules linked in: sctp netlink_diag nfsv3 nfnetlink_queue nfnetlink_log nfnetlink cfg80211 rpcsec_gss_krb5 nfsv4 dns_resolver nfs fscache rfcomm fuse hidp xt_CHECKSUM iptable_mangle ipt_MASQUERADE nf_nat_masquerade_ipv4 xt_conntrack tun devlink ebtable_filter ebtables ip6table_filter ip6_tables iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat nf_conntrack bridge stp llc bnep it87 hwmon_vid btrfs xor zstd_compress raid6_pq libcrc32c zstd_decompress xxhash intel_rapl x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel kvm gpio_ich iTCO_wdt iTCO_vendor_support ppdev irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel intel_cstate intel_uncore intel_rapl_perf btusb btrtl btbcm btintel bluetooth snd_hda_codec_realtek snd_hda_codec_generic snd_hda_codec_hdmi snd_hda_intel
 snd_hda_codec ecdh_generic rfkill snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer i2c_i801 lpc_ich mei_me mei shpchp snd soundcore parport_pc parport nfsd binfmt_misc nfs_acl lockd auth_rpcgss grace sunrpc i915 crc32c_intel i2c_algo_bit drm_kms_helper serio_raw drm r8169 sata_via mii video
CPU: 0 PID: 2690 Comm: nfsd Not tainted 4.17.2-200.fc28.x86_64 #1
Hardware name: Gigabyte Technology Co., Ltd. H81M-S2PV/H81M-S2PV, BIOS F8 06/19/2014
RIP: 0010:prefetch_freepointer+0x10/0x20
RSP: 0018:ffffa33fca4ebc58 EFLAGS: 00010202
RAX: 0000000000000000 RBX: 49f929d076783c7f RCX: 0000000000000f2a
RDX: 0000000000000f29 RSI: 49f929d076783c7f RDI: ffff8da88c9ffe00
RBP: ffff8da88c9ffe00 R08: ffff8da89ee2b100 R09: 0000000000000004
R10: 0000000000000000 R11: 000000000000002c R12: 00000000014080c0
R13: ffffffffc07e0ac1 R14: ffff8da4bc958ae1 R15: ffff8da88c9ffe00
FS:  0000000000000000(0000) GS:ffff8da89ee00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00002001a1783000 CR3: 000000012920a001 CR4: 00000000000606f0
Call Trace:
 kmem_cache_alloc+0xb4/0x1d0
 ? nfsd4_free_file_rcu+0x20/0x20 [nfsd]
```

Help on how to debug this is appreciated.


Kind regards,

Paul


[1]: https://bugs.freedesktop.org/show_bug.cgi?id=105684
[2]: https://bugzilla.redhat.com/show_bug.cgi?id=1600482
[3]: https://retrace.fedoraproject.org/faf/reports/2243945/


Download attachment "smime.p7s" of type "application/pkcs7-signature" (5174 bytes)