http://bugzilla.opensuse.org/show_bug.cgi?id=1165535 http://bugzilla.opensuse.org/show_bug.cgi?id=1165535#c2 Ulrich Windl changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(Ulrich.Windl@rz.u | |ni-regensburg.de) | --- Comment #2 from Ulrich Windl --- (In reply to Hannes Reinecke from comment #1)

Can you give some more details on how you configured 'encrypted swap' ? dm-crypt? Hardware encryption?

It's a LUKS volume via /etc/crypttab: cr_swap /dev/disk/by-id/ata-Samsung_SSD_860_EVO_500GB_S4YAMG8N878778W-part2 none discard,swap

Would it be possible use 'normal' swap and check if the error persists?

I think it's hard enough to reproduce the memory pressure (it that was the trigger at all). Maybe even the encrypted swap does not matter; maybe the discard option matters; I don't really know. Also I had been using this setup for a long time. Only a few weeks ago I replaced a dying harddisk with the SSD, adjusting the setup. So far it was the first time that this problem had occurred. (the advantage of the old and noisy Seagate disk was that you could clearly hear when the system started paging, explaining the slowness of the system ;-) With SSD you hardly experience a major slowdown when the system begins paging) Maybe "ata2: log page 10h reported inactive tag 5" indicates that something's wrong in the kernel. That was the message after which everything began. I can exclude memory errors, as the system uses an old AMD Phenom II CPU with ECC RAM that is absolutely "rock solid stable". -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1165535 http://bugzilla.opensuse.org/show_bug.cgi?id=1165535#c9 --- Comment #9 from Ulrich Windl --- (In reply to Sebastian Chlad from comment #8)

@Dear Ulrich: could you please see your system, specifically what controller there is? Perhaps we are hitting the same issue here?

(from lspci) 00:11.0 SATA controller: Advanced Micro Devices, Inc. [AMD/ATI] SB7x0/SB8x0/SB9x0 SATA Controller [IDE mode] 00:14.1 IDE interface: Advanced Micro Devices, Inc. [AMD/ATI] SB7x0/SB8x0/SB9x0 IDE Controller # lspci -nn -vv -s 0:14.1 00:14.1 IDE interface [0101]: Advanced Micro Devices, Inc. [AMD/ATI] SB7x0/SB8x0/SB9x0 IDE Controller [1002:439c] (prog-if 8a [ISA Compatibility mode controller, supports both channels switched to PCI native mode, supports bus mastering]) Subsystem: Gigabyte Technology Co., Ltd Device [1458:5002] Control: I/O+ Mem- BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx- Status: Cap+ 66MHz+ UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- TAbort-

Since the last comment, have you observed the same problem again?

No. Maybe it was fixed in the kernel. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1165535 http://bugzilla.opensuse.org/show_bug.cgi?id=1165535#c10 --- Comment #10 from Ulrich Windl --- When not expecting it, the unexpected happens: Today a few seconds after resuming from an automatic suspend to disk (due to inactivity), first Firefox crashed, the I realized that any access to /home resulted in an I/O error. This may sound different from the original issue, but /home is an encrypted logical volume (LV), and the error message in syslog reads quite similar: Jul 27 08:02:52 pc kernel: ata1.00: exception Emask 0x0 SAct 0x8000000 SErr 0xc0000 action 0x0 Jul 27 08:02:52 pc kernel: ata1.00: irq_stat 0x40000008 Jul 27 08:02:52 pc kernel: ata1: SError: { CommWake 10B8B } Jul 27 08:02:52 pc kernel: ata1.00: failed command: READ FPDMA QUEUED Jul 27 08:02:52 pc kernel: ata1.00: cmd 60/20:d8:e0:72:19/00:00:22:00:00/40 tag 27 ncq dma 16384 in res 41/40:00:f1:72:19/00:00:22:00:00/40 Emask 0x409 (media error) <F> Jul 27 08:02:52 pc kernel: ata1.00: status: { DRDY ERR } Jul 27 08:02:52 pc kernel: ata1.00: error: { UNC } Jul 27 08:02:52 pc kernel: ata1: hard resetting link Jul 27 08:02:52 pc kernel: ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300) Jul 27 08:02:52 pc kernel: ata1.00: configured for UDMA/133 Jul 27 08:02:52 pc kernel: sd 0:0:0:0: [sda] tag#27 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE Jul 27 08:02:52 pc kernel: sd 0:0:0:0: [sda] tag#27 Sense Key : Medium Error [current] Jul 27 08:02:52 pc kernel: sd 0:0:0:0: [sda] tag#27 Add. Sense: Unrecovered read error - auto reallocate failed Jul 27 08:02:52 pc kernel: sd 0:0:0:0: [sda] tag#27 CDB: Read(10) 28 00 22 19 72 e0 00 00 20 00 Jul 27 08:02:52 pc kernel: print_req_error: I/O error, dev sda, sector 572093169 Jul 27 08:02:52 pc kernel: ata1: EH complete Jul 27 08:02:52 pc kernel: XFS (dm-8): metadata I/O error: block 0x7e312e0 ("xfs_trans_read_buf_map") error 5 numblks 32 Jul 27 08:02:52 pc kernel: XFS (dm-8): xfs_imap_to_bp: xfs_trans_read_buf() returned error -5. Jul 27 08:02:53 pc kernel: XFS (dm-8): xfs_do_force_shutdown(0x8) called from line 3496 of file ../fs/xfs/xfs_inode.c. Return address = 0xffffffffa12b5c2a Jul 27 08:02:53 pc kernel: XFS (dm-8): Corruption of in-memory data detected. Shutting down filesystem Jul 27 08:02:53 pc kernel: XFS (dm-8): Please umount the filesystem and rectify the problem(s) ... Jul 27 12:27:15 pc kernel: ata2.00: exception Emask 0x50 SAct 0x780 SErr 0x800 action 0x6 frozen Jul 27 12:27:15 pc kernel: ata2.00: irq_stat 0x08000000, interface fatal error Jul 27 12:27:15 pc kernel: ata2: SError: { HostInt } Jul 27 12:27:15 pc kernel: ata2.00: failed command: WRITE FPDMA QUEUED Jul 27 12:27:15 pc kernel: ata2.00: cmd 61/00:38:00:c0:09/08:00:00:00:00/40 tag 7 ncq dma 1048576 ou res 40/00:38:00:c0:09/00:00:00:00:00/40 Emask 0x50 (ATA bus error) Jul 27 12:27:15 pc kernel: ata2.00: status: { DRDY } Jul 27 12:27:15 pc kernel: ata2.00: failed command: WRITE FPDMA QUEUED Jul 27 12:27:15 pc kernel: ata2.00: cmd 61/00:40:00:c8:09/08:00:00:00:00/40 tag 8 ncq dma 1048576 ou res 40/00:38:00:c0:09/00:00:00:00:00/40 Emask 0x50 (ATA bus error) Jul 27 12:27:15 pc kernel: ata2.00: status: { DRDY } Jul 27 12:27:15 pc kernel: ata2.00: failed command: WRITE FPDMA QUEUED Jul 27 12:27:15 pc kernel: ata2.00: cmd 61/00:48:00:d0:09/08:00:00:00:00/40 tag 9 ncq dma 1048576 ou res 40/00:38:00:c0:09/00:00:00:00:00/40 Emask 0x50 (ATA bus error) Jul 27 12:27:15 pc kernel: ata2.00: status: { DRDY } Jul 27 12:27:15 pc kernel: ata2.00: failed command: WRITE FPDMA QUEUED Jul 27 12:27:15 pc kernel: ata2.00: cmd 61/58:50:00:d8:09/08:00:00:00:00/40 tag 10 ncq dma 1093632 ou res 40/00:38:00:c0:09/00:00:00:00:00/40 Emask 0x50 (ATA bus error) Jul 27 12:27:15 pc kernel: ata2.00: status: { DRDY } Jul 27 12:27:15 pc kernel: ata2: hard resetting link Jul 27 12:27:15 pc kernel: ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 300) Jul 27 12:27:15 pc kernel: ata2.00: supports DRM functions and may not be fully accessible Jul 27 12:27:15 pc kernel: ata2.00: supports DRM functions and may not be fully accessible Jul 27 12:27:15 pc kernel: ata2.00: configured for UDMA/133 Jul 27 12:27:15 pc kernel: ata2: EH complete Jul 27 12:27:15 pc kernel: ata2.00: Enabling discard_zeroes_data Note: The system has ECC RAM, so an im-memory data corruption by bit-rot is rather unlikely. Kernel in use was 4.12.14-lp151.28.52-default -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1165535 http://bugzilla.opensuse.org/show_bug.cgi?id=1165535#c13 --- Comment #13 from Ulrich Windl --- Created attachment 842196 --> http://bugzilla.opensuse.org/attachment.cgi?id=842196&action=edit Screenshot with top running when freeze occurred (In reply to Miroslav Beneš from comment #11)

Frankly, I am not sure what to do here. It would be too easy to blame it on HW incompatibility, which Sebastian mentioned.

Counter-argument: On a completely different and new hardware (Machine with four Xeon Gold (72 cores, 144 threads, > 700GB RAM) I had a kernel "freeze" while performing some memory load test that just uses a few processes that shuffle process-local memory around. The machine in question only had SSDs, configured as two RAIDs via PERC: RAID1 for base OS, and RAID6 for additional data. When performing the test without having any swap configured, the kernel started to use the OOM killer to terminate a few processes. After having configured 5GB swap @ prio=4 and >700GB swap @prio=2 the kernel froze shortly after having used swap. The special thing about the new setup was that the bug swap was configured via LVM and a thin volume. The kernel in question was 5.3.18-24.15-default (actually from SLES15 SP2), and there was plenty of swap space available when this had happened. Interestingly the kernel still responded to PING, but that was all: No VT-switching, not console getty, no ssh connection, no reaction on existing SSH connections. However I had a "top" with sleep time 1s running at the moment when the kernel froze. I gave it more than 10 minutes to recover, but nothing had happened and there was no message in /var/log/messages after the date top last had refreshed. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=1165535 http://bugzilla.opensuse.org/show_bug.cgi?id=1165535#c15 Nikolai Nikolaevskii changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |kaykaykay123@gmail.com --- Comment #15 from Nikolai Nikolaevskii --- ILL problem with Samsung SATA SSDs. Problem was solved by recent kernel update. More info: https://forums.opensuse.org/showthread.php/559734-Avoid-using-Samsung-SATA-S... -- You are receiving this mail because: You are on the CC list for the bug.