WARNING: at drivers/ata/libata-sff.c:1016

Tejun,

While flailing away the evening on sata_mv hotplug, I managed to get some complaints from libata-sff that do not appear to be due to any bug of my own (ha ha).

Looks like maybe a race or something?

[  353.852823] ata123: mv_err_intr: serr=04010000  <<< hot inserted drive here >>>

[  303.068080] ata123: exception Emask 0x10 SAct 0x0 SErr 0x4010000 action 0xe frozen
[  303.068080] ata123: edma_err 0x00000030, SError=04010000, dev connect, SError
[  303.068080] ata123: SError: { PHYRdyChg DevExch }
[  303.068080] ata123: hard resetting link
[  303.068080] ata123: serr=04010000 serr_mask=019d0000
[  303.068080] ata123: serr=00000000 serr_mask=019c0000
[  303.144771] ata123: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[  303.154762] ------------[ cut here ]------------
[  303.154768] WARNING: at drivers/ata/libata-sff.c:1016 ata_sff_hsm_move+0x26/0x628()
[  303.154770] Modules linked in: sata_mv nfs binfmt_misc rfcomm l2cap bluetooth nfsd lockd nfs_acl auth_rpcgss sunrpc exportfs acpi_cpufreq cpufreq_stats cpufreq_powersave cpufreq_conservative cpufreq_userspace ac container fan battery af_packet firewire_sbp2 fuse mousedev snd_hda_intel snd_pcm_oss snd_pcm snd_mixer_oss snd_seq_dummy snd_seq_oss snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq snd_timer firewire_ohci snd_seq_device firewire_core r8169 sg thermal intel_agp bitrev crc32 snd ehci_hcd uhci_hcd floppy psmouse serio_raw pcspkr sr_mod cdrom crc_itu_t agpgart soundcore snd_page_alloc usbcore button processor unix [last unloaded: sata_mv]
[  303.154839] Pid: 154, comm: ata/1 Not tainted 2.6.25-upstream #12
[  303.154845]  [warn_on_slowpath+65/81] warn_on_slowpath+0x41/0x51
[  303.154861]  [_spin_unlock_irqrestore+54/88] ? _spin_unlock_irqrestore+0x36/0x58
[  303.154874]  [mark_held_locks+65/92] ? mark_held_locks+0x41/0x5c
[  303.154881]  [_spin_unlock_irqrestore+54/88] ? _spin_unlock_irqrestore+0x36/0x58
[  303.154887]  [trace_hardirqs_on+225/258] ? trace_hardirqs_on+0xe1/0x102
[  303.154894]  [_spin_unlock_irqrestore+66/88] ? _spin_unlock_irqrestore+0x42/0x58
[  303.154902]  [try_to_del_timer_sync+72/79] ? try_to_del_timer_sync+0x48/0x4f
[  303.154910]  [ata_sff_hsm_move+38/1576] ata_sff_hsm_move+0x26/0x628
[  303.154917]  [process_timeout+0/10] ? process_timeout+0x0/0xa
[  303.154923]  [schedule_timeout+104/139] ? schedule_timeout+0x68/0x8b
[  303.154929]  [delay_tsc+47/69] ? delay_tsc+0x2f/0x45
[  303.154937]  [ata_pio_task+168/185] ata_pio_task+0xa8/0xb9
[  303.154943]  [run_workqueue+187/390] run_workqueue+0xbb/0x186
[  303.154950]  [run_workqueue+128/390] ? run_workqueue+0x80/0x186
[  303.154956]  [ata_pio_task+0/185] ? ata_pio_task+0x0/0xb9
[  303.154964]  [worker_thread+0/191] ? worker_thread+0x0/0xbf
[  303.154970]  [worker_thread+180/191] worker_thread+0xb4/0xbf
[  303.154976]  [autoremove_wake_function+0/51] ? autoremove_wake_function+0x0/0x33
[  303.154984]  [kthread+59/100] kthread+0x3b/0x64
[  303.154989]  [kthread+0/100] ? kthread+0x0/0x64
[  303.154994]  [kernel_thread_helper+7/16] kernel_thread_helper+0x7/0x10
[  303.155002]  =======================
[  303.155006] ---[ end trace e0a256c019b03b26 ]---
[  303.155289] ata123.00: failed to IDENTIFY (I/O error, err_mask=0x100)
[  303.155293] ata123: failed to recover some devices, retrying in 5 secs
[  303.211417] ata123: hard resetting link
[  303.211424] ata123: serr=00000000 serr_mask=019d0000
[  303.214798] ata123: serr=00000000 serr_mask=019c0000
[  303.261429] ata123: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[  303.271420] ------------[ cut here ]------------
[  303.271425] WARNING: at drivers/ata/libata-sff.c:1016 ata_sff_hsm_move+0x26/0x628()
[  303.271427] Modules linked in: sata_mv nfs binfmt_misc rfcomm l2cap bluetooth nfsd lockd nfs_acl auth_rpcgss sunrpc exportfs acpi_cpufreq cpufreq_stats cpufreq_powersave cpufreq_conservative cpufreq_userspace ac container fan battery af_packet firewire_sbp2 fuse mousedev snd_hda_intel snd_pcm_oss snd_pcm snd_mixer_oss snd_seq_dummy snd_seq_oss snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq snd_timer firewire_ohci snd_seq_device firewire_core r8169 sg thermal intel_agp bitrev crc32 snd ehci_hcd uhci_hcd floppy psmouse serio_raw pcspkr sr_mod cdrom crc_itu_t agpgart soundcore snd_page_alloc usbcore button processor unix [last unloaded: sata_mv]
[  303.271498] Pid: 154, comm: ata/1 Not tainted 2.6.25-upstream #12
[  303.271502]  [warn_on_slowpath+65/81] warn_on_slowpath+0x41/0x51
[  303.271517]  [_spin_unlock_irqrestore+54/88] ? _spin_unlock_irqrestore+0x36/0x58
[  303.271530]  [mark_held_locks+65/92] ? mark_held_locks+0x41/0x5c
[  303.271536]  [_spin_unlock_irqrestore+54/88] ? _spin_unlock_irqrestore+0x36/0x58
[  303.271545]  [trace_hardirqs_on+225/258] ? trace_hardirqs_on+0xe1/0x102
[  303.271551]  [_spin_unlock_irqrestore+66/88] ? _spin_unlock_irqrestore+0x42/0x58
[  303.271559]  [try_to_del_timer_sync+72/79] ? try_to_del_timer_sync+0x48/0x4f
[  303.271566]  [ata_sff_hsm_move+38/1576] ata_sff_hsm_move+0x26/0x628
[  303.271573]  [process_timeout+0/10] ? process_timeout+0x0/0xa
[  303.271580]  [schedule_timeout+104/139] ? schedule_timeout+0x68/0x8b
[  303.271586]  [delay_tsc+47/69] ? delay_tsc+0x2f/0x45
[  303.271593]  [ata_pio_task+168/185] ata_pio_task+0xa8/0xb9
[  303.271600]  [run_workqueue+187/390] run_workqueue+0xbb/0x186
[  303.271605]  [run_workqueue+128/390] ? run_workqueue+0x80/0x186
[  303.271611]  [ata_pio_task+0/185] ? ata_pio_task+0x0/0xb9
[  303.271619]  [worker_thread+0/191] ? worker_thread+0x0/0xbf
[  303.271625]  [worker_thread+180/191] worker_thread+0xb4/0xbf
[  303.271631]  [autoremove_wake_function+0/51] ? autoremove_wake_function+0x0/0x33
[  303.271638]  [kthread+59/100] kthread+0x3b/0x64
[  303.271643]  [kthread+0/100] ? kthread+0x0/0x64
[  303.271649]  [kernel_thread_helper+7/16] kernel_thread_helper+0x7/0x10
[  303.271656]  =======================
[  303.271660] ---[ end trace e0a256c019b03b26 ]---
[  303.271941] ata123.00: failed to IDENTIFY (I/O error, err_mask=0x100)
[  303.271944] ata123: failed to recover some devices, retrying in 5 secs
[  303.331410] ata123: hard resetting link
[  303.331418] ata123: serr=00000000 serr_mask=019d0000
[  303.334790] ata123: serr=00000000 serr_mask=019c0000
[  303.388088] ata123: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[  303.398078] ------------[ cut here ]------------
[  303.398083] WARNING: at drivers/ata/libata-sff.c:1016 ata_sff_hsm_move+0x26/0x628()
[  303.398085] Modules linked in: sata_mv nfs binfmt_misc rfcomm l2cap bluetooth nfsd lockd nfs_acl auth_rpcgss sunrpc exportfs acpi_cpufreq cpufreq_stats cpufreq_powersave cpufreq_conservative cpufreq_userspace ac container fan battery af_packet firewire_sbp2 fuse mousedev snd_hda_intel snd_pcm_oss snd_pcm snd_mixer_oss snd_seq_dummy snd_seq_oss snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq snd_timer firewire_ohci snd_seq_device firewire_core r8169 sg thermal intel_agp bitrev crc32 snd ehci_hcd uhci_hcd floppy psmouse serio_raw pcspkr sr_mod cdrom crc_itu_t agpgart soundcore snd_page_alloc usbcore button processor unix [last unloaded: sata_mv]
[  303.398152] Pid: 154, comm: ata/1 Not tainted 2.6.25-upstream #12
[  303.398156]  [warn_on_slowpath+65/81] warn_on_slowpath+0x41/0x51
[  303.398172]  [_spin_unlock_irqrestore+54/88] ? _spin_unlock_irqrestore+0x36/0x58
[  303.398185]  [mark_held_locks+65/92] ? mark_held_locks+0x41/0x5c
[  303.398192]  [_spin_unlock_irqrestore+54/88] ? _spin_unlock_irqrestore+0x36/0x58
[  303.398198]  [trace_hardirqs_on+225/258] ? trace_hardirqs_on+0xe1/0x102
[  303.398205]  [_spin_unlock_irqrestore+66/88] ? _spin_unlock_irqrestore+0x42/0x58
[  303.398212]  [try_to_del_timer_sync+72/79] ? try_to_del_timer_sync+0x48/0x4f
[  303.398220]  [ata_sff_hsm_move+38/1576] ata_sff_hsm_move+0x26/0x628
[  303.398226]  [process_timeout+0/10] ? process_timeout+0x0/0xa
[  303.398234]  [schedule_timeout+104/139] ? schedule_timeout+0x68/0x8b
[  303.398240]  [delay_tsc+47/69] ? delay_tsc+0x2f/0x45
[  303.398248]  [ata_pio_task+168/185] ata_pio_task+0xa8/0xb9
[  303.398255]  [run_workqueue+187/390] run_workqueue+0xbb/0x186
[  303.398261]  [run_workqueue+128/390] ? run_workqueue+0x80/0x186
[  303.398267]  [ata_pio_task+0/185] ? ata_pio_task+0x0/0xb9
[  303.398274]  [worker_thread+0/191] ? worker_thread+0x0/0xbf
[  303.398281]  [worker_thread+180/191] worker_thread+0xb4/0xbf
[  303.398287]  [autoremove_wake_function+0/51] ? autoremove_wake_function+0x0/0x33
[  303.398295]  [kthread+59/100] kthread+0x3b/0x64
[  303.398299]  [kthread+0/100] ? kthread+0x0/0x64
[  303.398305]  [kernel_thread_helper+7/16] kernel_thread_helper+0x7/0x10
[  303.398312]  =======================
[  303.398314] ---[ end trace e0a256c019b03b26 ]---
[  303.398594] ata123.00: failed to IDENTIFY (I/O error, err_mask=0x100)
[  303.398597] ata123: failed to recover some devices, retrying in 5 secs
[  303.561410] ata123: hard resetting link
[  303.561419] ata123: serr=00000000 serr_mask=019d0000
[  303.564791] ata123: serr=00000000 serr_mask=019c0000
[  303.611408] ata123: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[  303.611429] ata123: exception Emask 0x100 SAct 0x0 SErr 0x0 action 0x1 t4
[  303.611490] ata123: unexpected device interrupt
[  303.611545] ata123: EH complete

..

Mmmm.. got another one, slightly different line perhaps,
though maybe it just moved around (latest libata-dev#upstream this morning).

ata201: SATA link down (SStatus 0 SControl 300)
ata202: SATA link up 3.0 Gbps (SStatus 123 SControl 300)

------------[ cut here ]------------
WARNING: at drivers/ata/libata-sff.c:1211 ata_sff_hsm_move+0x595/0x628()
Modules linked in: sata_mv(+) nfs binfmt_misc rfcomm l2cap bluetooth nfsd lockd nfs_acl auth_rpcgss sunrpc exportfs acpi_cpufreq cpufreq_stats cpufreq_powersave cpufreq_conservative cpufreq_userspace ac container fan battery af_packet firewire_sbp2 fuse mousedev snd_hda_intel snd_pcm_oss snd_pcm snd_mixer_oss snd_seq_dummy snd_seq_oss snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq snd_timer snd_seq_device snd soundcore firewire_ohci ehci_hcd psmouse r8169 firewire_core uhci_hcd sr_mod bitrev sg crc_itu_t intel_agp serio_raw crc32 usbcore floppy pcspkr snd_page_alloc cdrom thermal agpgart button processor unix [last unloaded: sata_mv]

Pid: 150, comm: ata/3 Not tainted 2.6.25-upstream #4 [warn_on_slowpath+65/81] warn_on_slowpath+0x41/0x51 [_spin_unlock_irqrestore+66/88] ? _spin_unlock_irqrestore+0x42/0x58 [rb_insert_color+85/188] ? rb_insert_color+0x55/0xbc [_spin_unlock_irqrestore+54/88] ? _spin_unlock_irqrestore+0x36/0x58 [ata_sff_hsm_move+1429/1576] ata_sff_hsm_move+0x595/0x628 [delay_tsc+47/69] ? delay_tsc+0x2f/0x45 [ata_pio_task+168/185] ata_pio_task+0xa8/0xb9 [run_workqueue+187/390] run_workqueue+0xbb/0x186 [run_workqueue+128/390] ? run_workqueue+0x80/0x186 [ata_pio_task+0/185] ? ata_pio_task+0x0/0xb9 [worker_thread+0/191] ? worker_thread+0x0/0xbf [worker_thread+180/191] worker_thread+0xb4/0xbf [autoremove_wake_function+0/51] ? autoremove_wake_function+0x0/0x33 [kthread+59/100] kthread+0x3b/0x64
[kthread+0/100] ? kthread+0x0/0x64 [kernel_thread_helper+7/16] kernel_thread_helper+0x7/0x10
=======================

---[ end trace b9c8d67c25f0871c ]---

ata202.15: failed to read PMP GSCR[0] (Emask=0x10)
ata202: failed to recover some devices, retrying in 5 secs
ata202: SATA link up 3.0 Gbps (SStatus 123 SControl 300)

..

..

..

That's from here in libata-sff.c:1211:

       case HSM_ST_LAST:
               if (unlikely(!ata_ok(status))) {
                       qc->err_mask |= __ac_err_mask(status);
                       ap->hsm_task_state = HSM_ST_ERR;
                       goto fsm_start;
               }
               /* no more data to transfer */
               DPRINTK("ata%u: dev %u command complete, drv_stat 0x%x\n",
                       ap->print_id, qc->dev->devno, status);

               ap->hsm_task_state = HSM_ST_IDLE;
		...

I thought that maybe this might have been my LLD (sata_mv) noticing that the device (PM in this case) got unplugged mid-stride, and set the qc->err_mask for it. But te LLD doesn't touch polled PIO commands, and there was no log message from the LLD preceding this one, or following it either.

However, there were three of these same WARN_ONs in a row, so perhaps libata-eh was just cancelling the outstanding commands or something, and ???.

No big deal, it didn't crash or anything, and I was abusively unplugging/replugging things rather rapidly at the time. :)

Expect to see the sata_mv IRQ/EH/hotplug patches here in a day or three, after I wind them up and through Marvell's people first.

I'm beginning to *like* this driver, and might actually have to remove the "HIGHLY EXPERIMENTAL" label from it soon-ish. :)