linux-kernel - Re: ATA failure regression in kernel 4.2

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CADnq5_Ma0o7iAmXjbYTH_X_sAyH=sHuA0rFR-4-6AbKOX6gZ5Q@mail.gmail.com>
Date:	Tue, 28 Jul 2015 14:19:16 -0400
From:	Alex Deucher <alexdeucher@...il.com>
To:	Jiang Liu <jiang.liu@...ux.intel.com>
Cc:	Tejun Heo <tj@...nel.org>, LKML <linux-kernel@...r.kernel.org>
Subject: Re: ATA failure regression in kernel 4.2

On Mon, Jul 27, 2015 at 12:30 PM, Jiang Liu <jiang.liu@...ux.intel.com> wrote:
> On 2015/7/27 23:21, Alex Deucher wrote:
>> On Sun, Jul 26, 2015 at 11:01 PM, Jiang Liu <jiang.liu@...ux.intel.com> wrote:
>>> On 2015/7/25 1:38, Alex Deucher wrote:
>>>> On Thu, Jul 23, 2015 at 2:44 PM, Alex Deucher <alexdeucher@...il.com> wrote:
>>>>> On Thu, Jul 23, 2015 at 2:35 PM, Tejun Heo <tj@...nel.org> wrote:
>>>>>> Hello,
>>>>>>
>>>>>> On Thu, Jul 23, 2015 at 01:48:24PM -0400, Alex Deucher wrote:
>>>>>>> Something new in kernel 4.2 seems to have broken one of my hard drives
>>>>>>> (ssd) in kernel 4.2.  4.1 and older kernels work fine.  Here are the
>>>>>>> relevant logs.
>>>>>>>
>>>>>> ...
>>>>>>> [    6.547628] ata2.00: qc timeout (cmd 0xec)
>>>>>>> [    6.547721] ata2.00: failed to IDENTIFY (I/O error, err_mask=0x4)
>>>>>>> [    7.007213] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
>>>>>>> [   16.997819] ata2.00: qc timeout (cmd 0xec)
>>>>>>> [   16.997910] ata2.00: failed to IDENTIFY (I/O error, err_mask=0x4)
>>>>>>> [   16.997995] ata2: limiting SATA link speed to 3.0 Gbps
>>>>>>> [   17.457400] ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 320)
>>>>>>> [   47.429257] ata2.00: qc timeout (cmd 0xec)
>>>>>>> [   47.429349] ata2.00: failed to IDENTIFY (I/O error, err_mask=0x4)
>>>>>>> [   47.888822] ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 320)
>>>>>>
>>>>>> Nothing really rings a bell.  Timeouts on IDENTIFY.  Could be IRQ
>>>>>> related.  Which controller is it (lspci -nn)?  Also, can you try to
>>>>>> bisect the issue?
>>>>>
>>>>> 00:11.0 SATA controller [0106]: Advanced Micro Devices, Inc. [AMD] FCH
>>>>> SATA Controller [AHCI mode] [1022:7801] (rev 40)
>>>>> 00:14.1 IDE interface [0101]: Advanced Micro Devices, Inc. [AMD] FCH
>>>>> IDE Controller [1022:780c]
>>>>>
>>>>> I can take a look at bisecting later this week.
>>>>
>>>> You were right about the interrupts.  This is an AMD Kaveri APU system.
>>> Hi Alex,
>>>         Could you please help to provide more information about the
>>> system so we could identify the issue? Dmesg and /proc/interrupts
>>> from good and bad kernels are welcomed.
>>> Thanks!
>>> Gerry
>>
>> See attached.  Thanks!
> Hi Alex,
>         Thanks for the info. Seems something is wrong with multiple-MSI
> support. To narrow down the scope, could you please help to:

I'm also not getting interrupts in my gpu driver.  I haven't bisected
this specifically, but I suspect it is related since it to used to
work in 4.1.  Whether I enable MSIs or not in my driver, I get a huge
numbers of interrupts on all CPUs as soon as the driver is loaded, but
the driver isr never gets called.  E.g.,
  49:  117757835  117763227  117787837  117868913   PCI-MSI
524288-edge      amdgpu
nointremap doesn't seem to help.

Alex

> 1) apply the small patch and retest
> diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
> index 7e62751abfac..35f524cc23b7 100644
> --- a/drivers/ata/ahci.c
> +++ b/drivers/ata/ahci.c
> @@ -1345,6 +1345,7 @@ static int ahci_init_msi(struct pci_dev *pdev,
> unsigned int n_ports,
>         if (nvec < 0)
>                 return nvec;
>
> +       nvec = 1;
>         /*
>          * If number of MSIs is less than number of ports then Sharing Last
>          * Message mode could be enforced. In this case assume that
> advantage
> 2) Disable interrupt remapping by kernel parameter "nointremap"
> and retest.
> Thanks!
> Gerry
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/