svn commit: r270249 - head/sys/cam/ata
Warner Losh
imp at bsdimp.com
Fri Aug 22 05:34:55 UTC 2014
On Aug 21, 2014, at 10:31 PM, Neel Natu <neelnatu at gmail.com> wrote:
> Hi Warner,
>
> On Wed, Aug 20, 2014 at 3:58 PM, Warner Losh <imp at freebsd.org> wrote:
>> Author: imp
>> Date: Wed Aug 20 22:58:12 2014
>> New Revision: 270249
>> URL: http://svnweb.freebsd.org/changeset/base/270249
>>
>> Log:
>> Turns out that IDENTIFY DEVICE and IDENTIFY PACKET DEVICE return data
>> that's only mostly similar. Specifically word 78 bits are defined for
>> IDENTIFY DEVICE as
>> 5 Supports Hardware Feature Control
>> while a IDENTIFY PACKET DEVICE defines them as
>> 5 Asynchronous notification supported
>> Therefore, only pay attention to bit 5 when we're talking to ATAPI
>> devices (we don't use the hardware feature control at this time).
>> Ignore it for ATA devices. Remove kludge that papered over this issue
>> for Samsung SATA SSDs, since Micron drives also have the bit set and
>> the error was caused by this bad interpretation of the spec (which is
>> quite easy to do, since bits aren't normally overlapping like this).
>>
>> Modified:
>> head/sys/cam/ata/ata_xpt.c
>>
>> Modified: head/sys/cam/ata/ata_xpt.c
>> ==============================================================================
>> --- head/sys/cam/ata/ata_xpt.c Wed Aug 20 22:39:26 2014 (r270248)
>> +++ head/sys/cam/ata/ata_xpt.c Wed Aug 20 22:58:12 2014 (r270249)
>> @@ -458,12 +458,18 @@ negotiate:
>> 0, 0x02);
>> break;
>> case PROBE_SETAN:
>> - /* Remember what transport thinks about AEN. */
>> - if (softc->caps & CTS_SATA_CAPS_H_AN)
>> + /*
>> + * Only ATAPI defines this bit to mean AEN, but remember
>> + * what transport thinks about AEN.
>> + */
>> + if ((softc->caps & CTS_SATA_CAPS_H_AN) &&
>> + periph->path->device->protocol == PROTO_ATAPI)
>> path->device->inq_flags |= SID_AEN;
>> else
>> path->device->inq_flags &= ~SID_AEN;
>> xpt_async(AC_GETDEV_CHANGED, path, NULL);
>> + if (periph->path->device->protocol != PROTO_ATAPI)
>> + break;
>> cam_fill_ataio(ataio,
>> 1,
>> probedone,
>> @@ -750,14 +756,6 @@ out:
>> goto noerror;
>>
>> /*
>> - * Some Samsung SSDs report supported Asynchronous Notification,
>> - * but return ABORT on attempt to enable it.
>> - */
>> - } else if (softc->action == PROBE_SETAN &&
>> - status == CAM_ATA_STATUS_ERROR) {
>> - goto noerror;
>> -
>> - /*
>> * SES and SAF-TE SEPs have different IDENTIFY commands,
>> * but SATA specification doesn't tell how to identify them.
>> * Until better way found, just try another if first fail.
>>
>
> This change causes a panic for me on boot. Here is the boot log:
>
> ahci0: <Intel Patsburg AHCI SATA controller> port
> 0xf050-0xf057,0xf040-0xf043,0xf030-0xf037,0xf020-0xf023,0xf000-0xf01f
> mem 0xfbb21000-0xfbb217ff irq 18 at device 31.2 on pci0
> ahci0: AHCI v1.30 with 6 6Gbps ports, Port Multiplier not supported
> ahcich0: <AHCI channel> at channel 0 on ahci0
> ahcich1: <AHCI channel> at channel 1 on ahci0
> ahcich2: <AHCI channel> at channel 2 on ahci0
> ahcich3: <AHCI channel> at channel 3 on ahci0
> ahcich4: <AHCI channel> at channel 4 on ahci0
> ahcich5: <AHCI channel> at channel 5 on ahci0
> ahciem0: <AHCI enclosure management bridge> on ahci0
> ...
> xpt_action_default: CCB type 0xdeadc0de not supported
> ...
> run_interrupt_driven_hooks: still waiting after 60 seconds for xpt_config
> run_interrupt_driven_hooks: still waiting after 120 seconds for xpt_config
> run_interrupt_driven_hooks: still waiting after 180 seconds for xpt_config
> run_interrupt_driven_hooks: still waiting after 240 seconds for xpt_config
> run_interrupt_driven_hooks: still waiting after 300 seconds for xpt_config
> panic: run_interrupt_driven_config_hooks: waited too long
> cpuid = 0
> KDB: stack backtrace:
> db_trace_self_wrapper() at db_trace_self_wrapper+0x2b/frame 0xffffffff81d92920
> kdb_backtrace() at kdb_backtrace+0x39/frame 0xffffffff81d929d0
> vpanic() at vpanic+0x189/frame 0xffffffff81d92a50
> kassert_panic() at kassert_panic+0x139/frame 0xffffffff81d92ac0
> boot_run_interrupt_driven_config_hooks() at
> boot_run_interrupt_driven_config_hooks+0x111/frame 0xffffffff81d92b50
> mi_startup()fffff81d92b70
> btext() at btext+0x2c
> KDB: enter: panic
> [ thread pid 0 tid 100000 ]
> Stopped at kdb_enter+0x3e: movq $0,kdb_why
> db>
>
> The peripheral in question is a SATA attached CDROM:
>
> % camcontrol devlist
> <INTEL SSDSC2CW240A3 400i> at scbus0 target 0 lun 0 (pass0,ada0)
> <ATAPI iHAS524 C LL23> at scbus2 target 0 lun 0 (cd0,pass1)
> <WDC WD1000CHTZ-04JCPV0 04.06A00> at scbus3 target 0 lun 0 (pass2,ada1)
> <Corsair Neutron GTX SSD M306> at scbus4 target 0 lun 0 (pass3,ada2)
> <AHCI SGPIO Enclosure 1.00 0001> at scbus6 target 0 lun 0 (ses0,pass4)
>
> pass1 at ahcich2 bus 0 scbus2 target 0 lun 0
> pass1: <ATAPI iHAS524 C LL23> Removable CD-ROM SCSI-0 device
> pass1: Serial Number 3524472 2N8225501140
> pass1: 150.000MB/s transfers (SATA 1.x, UDMA5, ATAPI 12bytes, PIO 8192bytes)
>
> The following patch fixes the panic.
>
> Index: sys/cam/ata/ata_xpt.c
> ===================================================================
> --- sys/cam/ata/ata_xpt.c (revision 270249)
> +++ sys/cam/ata/ata_xpt.c (working copy)
> @@ -468,7 +468,8 @@
> else
> path->device->inq_flags &= ~SID_AEN;
> xpt_async(AC_GETDEV_CHANGED, path, NULL);
> - if (periph->path->device->protocol != PROTO_ATAPI)
> + if (periph->path->device->protocol != PROTO_ATAPI &&
> + periph->path->device->protocol != PROTO_SCSI)
> break;
> cam_fill_ataio(ataio,
> 1,
I think the more proper test is == PROTO_ATA elsewhere, since that’s what
distinguishes the ATA_IDENTIFY from the ATAPI_IDENTIFY.
> However, there seem to be a couple of issues with the original patch:
>
> 1. The 'periph->path->device->protocol' is not initialized to
> PROTO_ATAPI anywhere in the tree so the not-equal-to test is a no-op.
We test here to determine which identify command to send:
if (periph->path->device->protocol == PROTO_ATA)
ata_28bit_cmd(ataio, ATA_ATA_IDENTIFY, 0, 0, 0);
else
ata_28bit_cmd(ataio, ATA_ATAPI_IDENTIFY, 0, 0, 0);
and that is working to send the right command.
> 2. It seems not right to break out of switch in 'probestart()' without
> providing a way for 'probedone()' to be called. I believe that this
> stops the state machine from making forward progress and results in
> 'xpt_config()' not completing.
That’s a problem, you’re right. Let me rework.
> If you need more information to debug this some more or test a proper
> fix then I am happy to help.
Please try the one included here. I think it will address things. I’ve tried it on one system, and am trying it on others in parallel to sending this.
Warner
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ata-xpt-patch
Type: application/octet-stream
Size: 1437 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/svn-src-all/attachments/20140821/5f4ef757/attachment.obj>
-------------- next part --------------
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 842 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.freebsd.org/pipermail/svn-src-all/attachments/20140821/5f4ef757/attachment.sig>
More information about the svn-src-all
mailing list