ahci error during install of 6.4

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

ahci error during install of 6.4

Paul Swanson
Hi,

I'm currently trying to install 6.4 on a Dell Latitude E7470 laptop (Intel Skylake).

During the whole disk (G) partitioning process, setup fails with the following messages:

newfs: wtfs: write error on block 8352576: Input / output error
ahci0: attempting to idle devices
atascsi_disk_sync_done: error
ahci0: NCQ errored slot 14 is idle (20000000 active)

Assuming that perhaps there might be a bad block on the drive (nvme ssd) I've run read / write bad block tests on the whole drive, but nothing showed.

The drive has had a working install of Ubuntu up till now, and I've subsequently installed Xubuntu on it successfully.

As it stands I can't proceed with the install; very sad.

Any help would be appreciated.

Regards,

Paul S.

Sent from ProtonMail mobile
Reply | Threaded
Open this post in threaded view
|

Re: ahci error during install of 6.4

Juan Francisco Cantero Hurtado
On Fri, Dec 28, 2018 at 08:18:38AM +0000, Paul Swanson wrote:

> Hi,
>
> I'm currently trying to install 6.4 on a Dell Latitude E7470 laptop (Intel Skylake).
>
> During the whole disk (G) partitioning process, setup fails with the following messages:
>
> newfs: wtfs: write error on block 8352576: Input / output error
> ahci0: attempting to idle devices
> atascsi_disk_sync_done: error
> ahci0: NCQ errored slot 14 is idle (20000000 active)
>
> Assuming that perhaps there might be a bad block on the drive (nvme ssd) I've run read / write bad block tests on the whole drive, but nothing showed.
>
> The drive has had a working install of Ubuntu up till now, and I've subsequently installed Xubuntu on it successfully.
>
> As it stands I can't proceed with the install; very sad.
>
> Any help would be appreciated.

Install OpenBSD on a usb stick, run OpenBSD from there and use dd to
write zeroes to the disk. If the disk has bad blocks you will see
similar errors in the dmesg. You can do the same with linux.

Sometimes bad units pass the checks of badblocks programs because these
run read-only tests by default and the flash controller lies. You only
see the bad sectors when you try to write to the disk.


--
Juan Francisco Cantero Hurtado http://juanfra.info

Reply | Threaded
Open this post in threaded view
|

Re: ahci error during install of 6.4

Nick Holland
On 12/28/18 5:37 PM, Juan Francisco Cantero Hurtado wrote:

> On Fri, Dec 28, 2018 at 08:18:38AM +0000, Paul Swanson wrote:
>> Hi,
>>
>> I'm currently trying to install 6.4 on a Dell Latitude E7470 laptop (Intel Skylake).
>>
>> During the whole disk (G) partitioning process, setup fails with the following messages:
>>
>> newfs: wtfs: write error on block 8352576: Input / output error
>> ahci0: attempting to idle devices
>> atascsi_disk_sync_done: error
>> ahci0: NCQ errored slot 14 is idle (20000000 active)
>>
>> Assuming that perhaps there might be a bad block on the drive (nvme ssd) I've run read / write bad block tests on the whole drive, but nothing showed.
>>
>> The drive has had a working install of Ubuntu up till now, and I've subsequently installed Xubuntu on it successfully.
>>
>> As it stands I can't proceed with the install; very sad.
>>
>> Any help would be appreciated.
>
> Install OpenBSD on a usb stick, run OpenBSD from there and use dd to
> write zeroes to the disk. If the disk has bad blocks you will see
> similar errors in the dmesg. You can do the same with linux.
>
> Sometimes bad units pass the checks of badblocks programs because these
> run read-only tests by default and the flash controller lies. You only
> see the bad sectors when you try to write to the disk.

Actually...you won't see most SSD style write errors --they will be
silently remapped.

After writing zeros with dd, do it again with 0xff (377 octal) --

tr '\0' '\377' < /dev/zero | dd bs=1m if=- of=/dev/rsdXc

That will run a lot slower than the zeros, but now you have tested every
bit of the disk for one and zero storage and remapped them.

Did this recently with some annoying SSDs that have been bugging me for
years, and the results have been ... promising (NO problems since).

Nick.

Reply | Threaded
Open this post in threaded view
|

Re: ahci error during install of 6.4

Paul Swanson
In reply to this post by Juan Francisco Cantero Hurtado
Cheers Juan and everyone else!

Turns out my SSD had an intermittent controller fault that not even a firmware update would resolve.

Had it replaced under warranty for a different model, OpenBSD installs and works great.

Very thankful, if OpenBSD hadn't failed so hard upon detecting the controller error, I'd never have noticed the issue. Dell tech reckons heaps of those early Intel SATA M.2 SSDs have failed completely.

It's quite likely that OpenBSD saved me a heck of a lot of data loss; needless to say I'm very impressed, this is my first time on this platform.

Thanks all!

Sent from ProtonMail mobile

-------- Original Message --------
On Dec 29, 2018, 08:37, Juan Francisco Cantero Hurtado wrote:

> On Fri, Dec 28, 2018 at 08:18:38AM +0000, Paul Swanson wrote:
>> Hi,
>>
>> I'm currently trying to install 6.4 on a Dell Latitude E7470 laptop (Intel Skylake).
>>
>> During the whole disk (G) partitioning process, setup fails with the following messages:
>>
>> newfs: wtfs: write error on block [8352576](tel:8352576): Input / output error
>> ahci0: attempting to idle devices
>> atascsi_disk_sync_done: error
>> ahci0: NCQ errored slot 14 is idle (20000000 active)
>>
>> Assuming that perhaps there might be a bad block on the drive (nvme ssd) I've run read / write bad block tests on the whole drive, but nothing showed.
>>
>> The drive has had a working install of Ubuntu up till now, and I've subsequently installed Xubuntu on it successfully.
>>
>> As it stands I can't proceed with the install; very sad.
>>
>> Any help would be appreciated.
>
> Install OpenBSD on a usb stick, run OpenBSD from there and use dd to
> write zeroes to the disk. If the disk has bad blocks you will see
> similar errors in the dmesg. You can do the same with linux.
>
> Sometimes bad units pass the checks of badblocks programs because these
> run read-only tests by default and the flash controller lies. You only
> see the bad sectors when you try to write to the disk.
>
> --
> Juan Francisco Cantero Hurtado http://juanfra.info