smartctl on aic94xx causes a kernel warning in libata

smartctl on aic94xx causes a kernel warning in libata

[Grzegorz Nosek]

Hi,

This is a trivially reproducible (in my setup at least) WARN_ON that
triggers whenever I run smartctl on a SATA disk connected via an aic94xx
controller (Adaptec AIC-9410W SAS (Razor ASIC RAID)=20 (rev 09)).

The kernel is (as far as device drivers are concerned) vanilla
2.6.27.12.

Other than the warning and the fact that smartctl requires '-T
permissive' to show the disk status, I can't see any negative effects.

(please keep cc'd)

Best regards,
 Grzegorz Nosek

*** smartctl -a /dev/sdb (without -T permissive) ***

smartctl version 5.37 [i686-pc-linux-gnu] Copyright (C) 2002-6 Bruce Allen
Home page is http://smartmontools.sourceforge.net/

=== START OF INFORMATION SECTION ===
Device Model:     ST3500630NS
Serial Number:    9QG93FV5
Firmware Version: 3.AEK
User Capacity:    500,107,862,016 bytes
Device is:        Not in smartctl database [for details use: -P showall]
ATA Version is:   7
ATA Standard is:  Exact ATA specification draft version not indicated
Local Time is:    Fri Jan 23 20:17:30 2009 CET
SMART support is: Available - device has SMART capability.
SMART support is: Enabled

Error SMART Status command failed
Please get assistance from http://smartmontools.sourceforge.net/
Values from ATA status return descriptor are:
 00     09 0c 00 00 00 00 00 00  00 00 00 00 00 00
A mandatory SMART command failed: exiting. To continue, add one or more '-T permissive' options.

*** the code in question ***

void ata_qc_issue(struct ata_queued_cmd *qc)
{
        struct ata_port *ap = qc->ap;
        struct ata_link *link = qc->dev->link;
        u8 prot = qc->tf.protocol;

        /* Make sure only one non-NCQ command is outstanding.  The
         * check is skipped for old EH because it reuses active qc to
         * request ATAPI sense.
         */
        WARN_ON(ap->ops->error_handler && ata_tag_valid(link->active_tag));

        if (ata_is_ncq(prot)) {
                WARN_ON(link->sactive & (1 << qc->tag));

                if (!link->sactive)
                        ap->nr_active_links++;
                link->sactive |= 1 << qc->tag;
        } else {
                WARN_ON(link->sactive);       /* <-- this is line 4867 */

                ap->nr_active_links++;
                link->active_tag = qc->tag;
        }

*** and the warning itself ***

[251374.059375] ------------[ cut here ]------------
[251374.064287] WARNING: at drivers/ata/libata-core.c:4867 ata_qc_issue+0xe1/0x2ba()
[251374.067805] Modules linked in: drbd i2c_i801 xt_owner ipt_account ipt_MASQUERADE xt_multiport ipt_REJECT xt_comment xt_state xt_hashlimit iptable_mangle iptable_nat nf_nat nf_conntrack_ipv4 nf_conntrack iptable_filter ip_t
ables tg3 libphy psmouse e1000e
[251374.067805] Pid: 9450, comm: smartctl Tainted: G        W 2.6.27.12 #12
[251374.067805]
[251374.067805] Call Trace:
[251374.067805]  [<ffffffff81031103>] warn_on_slowpath+0x58/0x7f
[251374.067805]  [<ffffffff8107c8d0>] ? mempool_alloc_slab+0x11/0x13
[251374.067805]  [<ffffffff8107c9da>] ? mempool_alloc+0x48/0xf9
[251374.067805]  [<ffffffff81247015>] ? scsi_sg_alloc+0x43/0x45
[251374.067805]  [<ffffffff811bb301>] ? __sg_alloc_table+0x78/0xfa
[251374.067805]  [<ffffffff81272adc>] ata_qc_issue+0xe1/0x2ba
[251374.067805]  [<ffffffff812423fd>] ? scsi_done+0x0/0x21
[251374.067805]  [<ffffffff8127a015>] ? ata_scsi_pass_thru+0x0/0x24b
[251374.067805]  [<ffffffff8127756a>] ata_scsi_translate+0x11f/0x155
[251374.067805]  [<ffffffff812423fd>] ? scsi_done+0x0/0x21
[251374.067805]  [<ffffffff812780fc>] ata_sas_queuecmd+0x186/0x1a5
[251374.067805]  [<ffffffff812575f1>] sas_queuecommand+0x82/0x27b
[251374.067805]  [<ffffffff81242b66>] scsi_dispatch_cmd+0x196/0x1ea
[251374.067805]  [<ffffffff812482be>] scsi_request_fn+0x28a/0x35f
[251374.067805]  [<ffffffff811a4776>] elv_insert+0x7f/0x18e
[251374.067805]  [<ffffffff811a4915>] __elv_add_request+0x90/0x98
[251374.067805]  [<ffffffff811a8d0f>] blk_execute_rq_nowait+0x65/0x8e
[251374.067805]  [<ffffffff811a8dd8>] blk_execute_rq+0xa0/0xcc
[251374.067805]  [<ffffffff811a5076>] ? blk_rq_bio_prep+0x3a/0x108
[251374.067805]  [<ffffffff811a8962>] ? blk_rq_append_bio+0x19/0x49
[251374.067805]  [<ffffffff811a8aa6>] ? blk_rq_map_user+0x114/0x1bb
[251374.067805]  [<ffffffff811abbb6>] sg_io+0x263/0x373
[251374.067805]  [<ffffffff811ac326>] scsi_cmd_ioctl+0x1d3/0x3a9
[251374.067805]  [<ffffffff810482b5>] ? sched_clock_cpu+0x134/0x140
[251374.067805]  [<ffffffff81026edb>] ? task_rq_lock+0x40/0x79
[251374.067805]  [<ffffffff8126a5b9>] sd_ioctl+0x7c/0xa1
[251374.067805]  [<ffffffff811a9cbf>] blkdev_driver_ioctl+0x63/0x78
[251374.067805]  [<ffffffff811aa3a5>] blkdev_ioctl+0x6d1/0x6fc
[251374.067805]  [<ffffffff8120062c>] ? n_tty_receive_buf+0xe1a/0xe5a
[251374.067805]  [<ffffffff810821c7>] ? mark_page_accessed+0x20/0x34
[251374.067805]  [<ffffffff8107c696>] ? filemap_fault+0x1c9/0x32b
[251374.067805]  [<ffffffff810c9a08>] block_ioctl+0x1b/0x1f
[251374.067805]  [<ffffffff810af72e>] vfs_ioctl+0x2a/0x77
[251374.067805]  [<ffffffff810af9b0>] do_vfs_ioctl+0x235/0x246
[251374.067805]  [<ffffffff810afa03>] sys_ioctl+0x42/0x67
[251374.067805]  [<ffffffff81026e8e>] ? __wake_up+0x43/0x50
[251374.067805]  [<ffffffff810d4edb>] sg_ioctl_trans+0x1e0/0x248
[251374.067805]  [<ffffffff810d3929>] compat_sys_ioctl+0x2d2/0x316
[251374.067805]  [<ffffffff810a313f>] ? vfs_write+0x122/0x157
[251374.067805]  [<ffffffff81023326>] sysenter_dispatch+0x7/0x27
[251374.067805]
[251374.067805] ---[ end trace 5c5fec5d7ca0c598 ]---

Re: smartctl on aic94xx causes a kernel warning in libata

[Douglas Gilbert]

Grzegorz Nosek wrote:
> Hi,
> 
> This is a trivially reproducible (in my setup at least) WARN_ON that
> triggers whenever I run smartctl on a SATA disk connected via an aic94xx
> controller (Adaptec AIC-9410W SAS (Razor ASIC RAID)=20 (rev 09)).
> 
> The kernel is (as far as device drivers are concerned) vanilla
> 2.6.27.12.
> 
> Other than the warning and the fact that smartctl requires '-T
> permissive' to show the disk status, I can't see any negative effects.
> 
> (please keep cc'd)
> 
> Best regards,
>  Grzegorz Nosek
> 
> *** smartctl -a /dev/sdb (without -T permissive) ***
> 
> smartctl version 5.37 [i686-pc-linux-gnu] Copyright (C) 2002-6 Bruce Allen
> Home page is http://smartmontools.sourceforge.net/
> 
> === START OF INFORMATION SECTION ===
> Device Model:     ST3500630NS
> Serial Number:    9QG93FV5
> Firmware Version: 3.AEK
> User Capacity:    500,107,862,016 bytes
> Device is:        Not in smartctl database [for details use: -P showall]
> ATA Version is:   7
> ATA Standard is:  Exact ATA specification draft version not indicated
> Local Time is:    Fri Jan 23 20:17:30 2009 CET
> SMART support is: Available - device has SMART capability.
> SMART support is: Enabled
> 
> Error SMART Status command failed
> Please get assistance from http://smartmontools.sourceforge.net/
> Values from ATA status return descriptor are:
>  00     09 0c 00 00 00 00 00 00  00 00 00 00 00 00
> A mandatory SMART command failed: exiting. To continue, add one or more '-T permissive' options.

Putting my smartmontools hat on that SAT ATA status
return descriptor for an ATA SMART RETURN STATUS
command looks broken, most likely not correctly "wired"
as the payload is zeros. Using a LSI SAS controller
(which implements SAT in firmware) here is what the
response should look like:

Values from ATA Return Descriptor are:
  00     09 0c 00 00 00 00 00 00  00 4f 00 c2 00 50

assuming the disk is healthy. [And if it was unhealthy
then 4f -> f4 and c2 -> 2c (t13 dyslexic humour ?).]

To my knowledge that usage of the ATA SMART RETURN
status command is the only occasion when smartctl
needs to read the ATA registers _after_ command completion.

Doug Gilbert