Re: 2.6.27-rc1 (2fca5c): libata: kernel cant boot

[Petr Vandrovec <vandrove@vc.cvut.cz>]

Jens Axboe wrote:
> On Sat, Oct 25 2008, Jens Axboe wrote:
>> On Sat, Oct 25 2008, Petr Vandrovec wrote:
>>> Jens Axboe wrote:
>>>> On Fri, Oct 24 2008, Elias Oltmanns wrote:
>>>>> Jens Axboe <jens.axboe@oracle.com> wrote:
>>>>> >From e598055dde1951c47c8b3522616f6ebff0ed9847 Mon Sep 17 00:00:00 2001
>>>>>> From: Jens Axboe <jens.axboe@oracle.com>
>>>>>> Date: Fri, 24 Oct 2008 09:22:42 +0200
>>>>>> Subject: [PATCH] libata: fix bug with non-ncq devices
>>>>>>
>>>>>> The recent commit 201f1b98822078c808b5e2d379a6ddbfc0a06ee1 to enable
>>>>> Wouldn't that be commit 2fca5ccf97d2c28bcfce44f5b07d85e74e3cd18e?
>>>> Yes that is correct, the other commit is actually a private one in my
>>>> tree for other libata changes. Updated patch below, thanks for checking!
>>>>
>>> >From e598055dde1951c47c8b3522616f6ebff0ed9847 Mon Sep 17 00:00:00 2001
>>>> From: Jens Axboe <jens.axboe@oracle.com>
>>>> Date: Fri, 24 Oct 2008 09:22:42 +0200
>>>> Subject: [PATCH] libata: fix bug with non-ncq devices
>>> Hello,
>>>   this fixes my DVD, but unfortunately NCQ devices connected to PMP are 
>>> still dead - apparently as soon as mount() tries to do serious I/O on 
>>> the drive.  Backing out both post-2.6.28-rc1 fix as well as your 
>>> original change brings storage back.  I suspect that problem is that 
>>> with PMP same tag cannot be (should not be?  must not be?) used on 
>>> multiple devices behind PMP - and before your change tags were allocated 
>>> per-port, while now they are allocated per-device.
>> That would indeed break, this requires allocating the tag map in the
> 
> Totally untested, does this work?
> 
> diff --git a/drivers/ata/libata-scsi.c b/drivers/ata/libata-scsi.c
> index 4b95c43..0785c46 100644
> --- a/drivers/ata/libata-scsi.c
> +++ b/drivers/ata/libata-scsi.c
> @@ -1107,6 +1107,10 @@ static int ata_scsi_dev_config(struct scsi_device *sdev,
>  
>  		depth = min(sdev->host->can_queue, ata_id_queue_depth(dev->id));
>  		depth = min(ATA_MAX_QUEUE - 1, depth);
> +
> +		if (dev->link->ap->pmp_link)
> +			scsi_init_shared_tag_map(sdev->host, ATA_MAX_QUEUE - 1);
> +
>  		scsi_set_tag_type(sdev, MSG_SIMPLE_TAG);
>  		scsi_activate_tcq(sdev, depth);
>  	}

No.  It went through same story as without patch - first it declared 
drive #2 hung, after port reset drives #0,1,2 were declared hung, after 
second port reset drive #2 was declared dead, after third port reset 
drive #3 was hung, after fourth reset it said that /dev/sde changed 
capacity from 0 to 1TB, and at that point I decided that it is time to 
hit alt-sysrq-b to prevent damage...

Also I think that this change leaks memory a bit...
						Petr
						Petr