IRQ #14 nobody cared?

[Fai Wong]

Hi all,

I just updated the kernel on my 2f box to 2.6.33 from rt4ls git source.
It seems to have VERY slow hard disk speed if I enable libata and
PATA_AMD configuration, and it will show something like this during
bootup.

irq 14: nobody cared (try booting with the "irqpoll" option)
Call Trace:
[<ffffffff80214ae8>] dump_stack+0x8/0x4c
[<ffffffff802b24a0>] __report_bad_irq+0x40/0x100
[<ffffffff802b2758>] note_interrupt+0x1f8/0x270
[<ffffffff802b38ac>] handle_level_irq+0x18c/0x200
[<ffffffff8021d504>] do_IRQ+0x44/0x58
[<ffffffff80200400>] ret_from_irq+0x0/0x4
[<ffffffff80255580>] __do_softirq+0x70/0x1e8
[<ffffffff802561d0>] irq_exit+0x108/0x128
[<ffffffff80200400>] ret_from_irq+0x0/0x4
[<ffffffff8027c520>] srcu_notifier_call_chain+0x60/0x108
[<ffffffff805a54ec>] cpufreq_notify_transition+0x1ac/0x200
[<ffffffff8022f24c>] loongson2_cpufreq_target+0x164/0x170
[<ffffffff805a7c94>] do_dbs_timer+0x314/0x320
[<ffffffff8026ba80>] worker_thread+0x1d0/0x320
[<ffffffff80272a78>] kthread+0xa0/0xa8
[<ffffffff8021db00>] kernel_thread_helper+0x28/0x30

handlers:
[<ffffffff8054b3f0>] (ata_sff_interrupt+0x0/0x1d8)
Disabling IRQ #14

Is it a known issue of libata or it is loongson specific? It is OK if
I disable libata and use old IDE driver in kernel (ie. /dev/hdaX
instead of /dev/sdaX )

Cheers,
Fai

[Wu Zhangjin]

Hi,

Thanks very much for your report.

You need to do some modification in drivers/ide/ide-iops.c
b/drivers/ide/ide-iops.c to enable the quirk for fujtsu disk:

void ide_check_nien_quirk_list(ide_drive_t *drive)
{
const char **list, *m = (char *)&drive->id[ATA_ID_PROD];

- if (mips_machtype != MACH_LEMOTE_YL2F89)
- return;

for (list = nien_quirk_list; *list != NULL; list++)
if (strstr(m, *list) != NULL) {
drive->dev_flags |= IDE_DFLAG_NIEN_QUIRK;
return;
}
}

The cause here is, the fujitsu disk need the quirk:

diff --git a/drivers/ide/ide-iops.c b/drivers/ide/ide-iops.c
index b14fa9a..222c1ef 100644
--- a/drivers/ide/ide-iops.c
+++ b/drivers/ide/ide-iops.c
@@ -292,6 +292,7 @@ static const char *nien_quirk_list[] = {
"QUANTUM FIREBALLP KX27.3",
"QUANTUM FIREBALLP LM20.4",
"QUANTUM FIREBALLP LM20.5",
+ "FUJITSU MHZ2160BH G2",
NULL
};

But in some fuloong-2f series, they used a different switch interace,
with the quirk, the kernel will stop on booting, so, compromise the
problem, we only enable the quirk for YeeLoong netbook for there is no
switch interface used in it.

Therefore, for some old fuloong-2f machines, please "disable libata and
use old IDE driver in kernel" or enable the quirk for fujitsu disk
yourselves.(perhaps I need to add a new kernel option to do that job,
but whatever, there is no suitable solution possible to upstream.).

Regards,
Wu Zhangjin

[Fai Wong]

看来要用 libata 的话，需要修改一下 drivers/ata/libata-core.c 中的
ata_device_blacklist，做类似 ide_check_nien_quirk_list 的检查了。

2010/3/11 Wu Zhangjin <wuzha...@gmail.com>:

> --
> You received this message because you are subscribed to the Google Groups "loongson-dev" group.
> To post to this group, send email to loongs...@googlegroups.com.
> To unsubscribe from this group, send email to loongson-dev...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/loongson-dev?hl=en.
>
>

[Mads]

Hi!

I too get this error, but on linux-loongson-community kernel from
yesterday..

Running this on a Fuloong 2F.

irq 14: nobody cared (try booting with the "irqpoll" option)
Call Trace:

[<ffffffff802143d8>] dump_stack+0x8/0x40
[<ffffffff80279dbc>] __report_bad_irq+0x58/0xe4
[<ffffffff80279fc4>] note_interrupt+0x17c/0x23c
[<ffffffff8027af88>] handle_level_irq+0x10c/0x1b8
[<ffffffff8021ac58>] do_IRQ+0x38/0x48
[<ffffffff80200400>] ret_from_irq+0x0/0x4
[<ffffffff8027d358>] __rcu_process_callbacks+0x90/0x598
[<ffffffff8027d87c>] rcu_process_callbacks+0x1c/0x48
[<ffffffff80245f0c>] __do_softirq+0xd0/0x188
[<ffffffff80246014>] do_softirq+0x50/0x84
[<ffffffff80246210>] irq_exit+0x48/0xb0
[<ffffffff80200400>] ret_from_irq+0x0/0x4
[<ffffffff8025fc54>] __srcu_read_lock+0x0/0x74
[<ffffffff8026037c>] __srcu_notifier_call_chain+0x34/0x88
[<ffffffff8048c024>] cpufreq_notify_transition+0xec/0x12c
[<ffffffff80229518>] loongson2_cpufreq_target+0x104/0x130
[<ffffffff8048e80c>] do_dbs_timer+0x29c/0x324
[<ffffffff80255f90>] worker_thread+0x1f4/0x2f8
[<ffffffff8025a59c>] kthread+0x88/0x90
[<ffffffff8021b520>] kernel_thread_helper+0x1c/0x24

handlers:
[<ffffffff80454fa0>] (ata_bmdma_interrupt+0x0/0x2bc)
Disabling IRQ #14

The "nien_quirk_list" FUJITSU line was already added in git, but I removed
those two lines of "if (mips_machtype != MACH_LEMOTE_YL2F89) return;" which
you suggested, but the problem persists.

Appreciate your efforts!

Best Regards,

Mads

On Thu, 11 Mar 2010 09:38:48 +0800, Wu Zhangjin <wuzha...@gmail.com>
wrote:

[Wu Zhangjin]

Hi,

Which FuLoong box and kernel version are you using?

Regards,
Wu Zhangjin

[Mads]

Hi!

My kernel is from the master branch of
git://dev.lemote.com/linux-loongson-community.git

Running the version with the latest git commit:
commit 76083a9e776d913b93fac79f80e61fe46d9c5ba2
Merge: 59530f2 386f40c
Author: Wu Zhangjin <wuzha...@gmail.com>
Date: Mon Jun 7 21:34:13 2010 +0800

Lemote linux # uname -a
Linux Lemote 2.6.33 #3 Sun Apr 11 19:40:12 CEST 2010 mips64 ICT Loongson-2
V0.3 FPU V0.1 lemote-fuloong-2f-box GNU/Linux

Lemote linux # cat /proc/cpuinfo
system type : lemote-fuloong-2f-box
processor : 0
cpu model : ICT Loongson-2 V0.3 FPU V0.1
BogoMIPS : 528.38
wait instruction : yes
microsecond timers : yes
tlb_entries : 64
extra interrupt vector : no
hardware watchpoint : yes, count: 0, address/irw mask: []
ASEs implemented :
shadow register sets : 1
core : 0
VCED exceptions : not available
VCEI exceptions : not available

It's a Lemote Fuloong 2F, I think it's the FL6005

And, I applied the patch you mentioned, see my first mail:

> The "nien_quirk_list" FUJITSU line was already added in git, but I
removed
> those two lines of "if (mips_machtype != MACH_LEMOTE_YL2F89) return;"
> which you suggested, but the problem persists.

- Mads

On Sat, 12 Jun 2010 19:25:29 +0800, Wu Zhangjin <wuzha...@gmail.com>

[Wu Zhangjin]

Hi,

On Sat, 2010-06-12 at 17:36 +0200, Mads wrote:
> Hi!
>
> My kernel is from the master branch of
> git://dev.lemote.com/linux-loongson-community.git
>
> Running the version with the latest git commit:
> commit 76083a9e776d913b93fac79f80e61fe46d9c5ba2
> Merge: 59530f2 386f40c
> Author: Wu Zhangjin <wuzha...@gmail.com>
> Date: Mon Jun 7 21:34:13 2010 +0800
>
> Lemote linux # uname -a
> Linux Lemote 2.6.33 #3 Sun Apr 11 19:40:12 CEST 2010 mips64 ICT Loongson-2

2.6.33 and the master branch info is what I need, thanks!

> V0.3 FPU V0.1 lemote-fuloong-2f-box GNU/Linux
>
> Lemote linux # cat /proc/cpuinfo
> system type : lemote-fuloong-2f-box

FuLoong 6003? 6015? 6014? which one is yours?

Best Regards,
Wu Zhangjin

[Mads]

Ehm...

*checks receipt*

It says "fuloong 6005 mini pc"

Hope this helps :)

- Mads

On Sat, 12 Jun 2010 23:46:23 +0800, Wu Zhangjin <wuzha...@gmail.com>

[wu zhangjin]

Hi,

>> Lemote linux # uname -a
>> Linux Lemote 2.6.33 #3 Sun Apr 11 19:40:12 CEST 2010 mips64 ICT Loongson-2
>
> 2.6.33 and the master branch info is what I need, thanks!

The current master branch is 2.6.25-rc3? could you please pull it into
your local source code, then we can try to find out the exact problem.

I'm not sure how did you get the source code, if you have cloned it,
then just do "git pull" to update your local source code:

$ git pull git://dev.lemote.com/code/linux-loongson-community.git master

If you have downloaded the source code, then init your local source
code and pull it.

$ git init && git add . && git commit
$ git pull git://dev.lemote.com/code/linux-loongson-community.git master

Regards,
Wu Zhangjin

[wu zhangjin]

Hi,

On Sun, Jun 13, 2010 at 12:20 AM, Mads <ma...@ab3.no> wrote:
> Ehm...
>
> *checks receipt*
>
> It says "fuloong 6005 mini pc"
>
> Hope this helps :)
>

Thanks!

BTW:

Now, It's time to fix this problem utterly, at first, we must ensure
which boxes have such problems, which disk and interface card are they
using. I don't have any FuLoong box currently, so, I hope you people
can provide such informations, thanks!

If we want to share the same kernel image among all of the Lemote 2F
machines, the potential fix is checking which interface have such
problem and add related workaround for them, but it is a little hard
for we can not get the interfaces info before booting the kernel.
fortunately, some versions of PMON have passed the PMON_VER info to
the kernel and this info includes the box revision. so if we know
which boxes have such problem, we can add the fixup to these machines.

In the future, I think we'd better to port the FDT to Loongson
bootloader to let it pass the necessary board devices information to
the kernel, of course, we also need to add related support to the
kernel. some kernel related patches have been pushed into the
linux-mips mailing list, perhaps we can refer to it:
http://patchwork.linux-mips.org/project/linux-mips/list/?state=*&q=The+Device+Tree+Patch+for+MIPS+

Regards,

--
Studying engineer. Wu Zhangjin
Lanzhou University http://www.lzu.edu.cn
Distributed & Embedded System Lab http://dslab.lzu.edu.cn
School of Information Science and Engeneering http://xxxy.lzu.edu.cn
wuzha...@gmail.com http://falcon.oss.lzu.edu.cn
Address:Tianshui South Road 222,Lanzhou,P.R.China Zip Code:730000
Tel:+86-931-8912025

[项宇]

Can you please give the info of the /proc/cmdline?
It can give us the version of bios.

Mads 写道:

> Ehm...
>
> *checks receipt*
>
> It says "fuloong 6005 mini pc"
>
> Hope this helps :)
>
>
> - Mads
>
> On Sat, 12 Jun 2010 23:46:23 +0800, Wu Zhangjin <wuzha...@gmail.com>
> wrote:
>

--
Xiang Yu (Software Department)

Tel: 0512-52308631
Jiangsu Lemote Technology Co., Ltd
Menglan, Yushan, Changshu, Jiangsu, China

[Mads]

I guess I'm not so fluent in git...

The only thing I did was write "git clone
git://dev.lemote.com/code/linux-loongson-community.git", is that not the
master branch?

Lemote linux # git branch
* master

Lemote linux # cat include/config/kernel.release
2.6.35-rc2+

- Mads

On Sun, 13 Jun 2010 10:08:26 +0800, wu zhangjin <wuzha...@gmail.com>
wrote:

[Mads]

Lemote linux # cat /proc/cmdline
console=tty root=/dev/sda5 PMON_VER=LM6004-1.3.2 EC_VER=undefined
machtype=lemote-fuloong-2f-box ide_core.ignore_cable=0

- Mads

[Wu Zhangjin]

On Sun, 2010-06-13 at 14:39 +0200, Mads wrote:
> I guess I'm not so fluent in git...
>
> The only thing I did was write "git clone
> git://dev.lemote.com/code/linux-loongson-community.git", is that not the
> master branch?
>
> Lemote linux # git branch
> * master
>
> Lemote linux # cat include/config/kernel.release
> 2.6.35-rc2+
>

That is it, but what you have shown is 2.6.33 ;)

Thanks & Regards,

[Mads]

Whops, I'm sorry! It was my rt4ls-kernel which that uname came from...

Lemote ~ # uname -a
Linux Lemote 2.6.35-rc2+ #8 PREEMPT Sun Jun 13 15:14:29 CEST 2010 mips64
ICT Loongson-2 V0.3 FPU V0.1 lemote-fuloong-2f-box GNU/Linux

There!

- Mads

On Sun, 13 Jun 2010 22:20:00 +0800, Wu Zhangjin <wuzha...@gmail.com>
wrote:

[Wu Zhangjin]

On Sun, 2010-06-13 at 16:27 +0200, Mads wrote:
> Whops, I'm sorry! It was my rt4ls-kernel which that uname came from...
>

no problem, so, your problem is with 2.6.35-rc2?

Regards,

[Mads]

Yep, I booted the old kernel to check the hw info, the problem appeared
with linux-loongson-community kernel. There was no such trouble with the
rt4ls-kernel, I'm using /dev/sdX-devices there too...

- Mads

On Sun, 13 Jun 2010 22:30:26 +0800, Wu Zhangjin <wuzha...@gmail.com>