[BUG] cfi_cmdset_0001.c: possible recursive locking detected

[BUG] cfi_cmdset_0001.c: possible recursive locking detected

[Antonio Ospite]

[-- Attachment #1: Type: text/plain, Size: 2629 bytes --]

Hi,

I am attaching a trace of a possible recursive locking in
cfi_cmdset_0001.c

The problem shows up at shutdown, with
CONFIG_MTD_CFI_INTELEXT=y
CONFIG_MTD_PXA2XX=y

Platform is armv5te (Motorola A780 smartphone), but I guess this doesn't
matter. If you think it does, let me know and I'll provide more
information about board code and config.

Thanks,
   Antonio Ospite

=============================================
[ INFO: possible recursive locking detected ]
2.6.32-rc2-ezxdev #39
---------------------------------------------
halt/785 is trying to acquire lock:
 (&chip->_spinlock){+.+...}, at: [<c01801e0>] get_chip+0xe4/0x1e0

but task is already holding lock:
 (&chip->_spinlock){+.+...}, at: [<c01801a4>] get_chip+0xa8/0x1e0

other info that might help us debug this:
2 locks held by halt/785:
 #0:  ((reboot_notifier_list).rwsem){.+.+..}, at: [<c005d2d0>] __blocking_notifier_call_chain+0x30/0x64
 #1:  (&chip->_spinlock){+.+...}, at: [<c01801a4>] get_chip+0xa8/0x1e0

stack backtrace:
[<c002fd44>] (unwind_backtrace+0x0/0xe0) from [<c00683a8>] (validate_chain+0x5b0/0xdd8)
[<c00683a8>] (validate_chain+0x5b0/0xdd8) from [<c0069410>] (__lock_acquire+0x840/0x910)
[<c0069410>] (__lock_acquire+0x840/0x910) from [<c006a3b4>] (lock_acquire+0x60/0x74)
[<c006a3b4>] (lock_acquire+0x60/0x74) from [<c02a2cb4>] (_spin_lock+0x40/0x50)
[<c02a2cb4>] (_spin_lock+0x40/0x50) from [<c01801e0>] (get_chip+0xe4/0x1e0)
[<c01801e0>] (get_chip+0xe4/0x1e0) from [<c0180314>] (cfi_intelext_reset+0x38/0x90)
[<c0180314>] (cfi_intelext_reset+0x38/0x90) from [<c0180400>] (cfi_intelext_reboot+0x10/0x1c)
[<c0180400>] (cfi_intelext_reboot+0x10/0x1c) from [<c005d0ac>] (notifier_call_chain+0x2c/0x70)
[<c005d0ac>] (notifier_call_chain+0x2c/0x70) from [<c005d2ec>] (__blocking_notifier_call_chain+0x4c/0x64)
[<c005d2ec>] (__blocking_notifier_call_chain+0x4c/0x64) from [<c005d31c>] (blocking_notifier_call_chain+0x18/0x20)
[<c005d31c>] (blocking_notifier_call_chain+0x18/0x20) from [<c00525a4>] (kernel_shutdown_prepare+0x20/0x38)
[<c00525a4>] (kernel_shutdown_prepare+0x20/0x38) from [<c00525cc>] (kernel_power_off+0x10/0x44)
[<c00525cc>] (kernel_power_off+0x10/0x44) from [<c0053844>] (sys_reboot+0x118/0x1d8)
[<c0053844>] (sys_reboot+0x118/0x1d8) from [<c0029f20>] (ret_fast_syscall+0x0/0x34)

-- 
Antonio Ospite
http://ao2.it

PGP public key ID: 0x4553B001

A: Because it messes up the order in which people normally read text.
   See http://en.wikipedia.org/wiki/Posting_style
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

[PATCH] MTD: add missing put_chip() in cfi_intelext_reset()

[Nicolas Pitre]

Signed-off-by: Nicolas Pitre <nico@fluxnic.net>

---

On Fri, 23 Oct 2009, Antonio Ospite wrote:

> Hi,
> 
> I am attaching a trace of a possible recursive locking in
> cfi_cmdset_0001.c

This patch should fix it.

diff --git a/drivers/mtd/chips/cfi_cmdset_0001.c b/drivers/mtd/chips/cfi_cmdset_0001.c
index e7563a9..d923a2d 100644
--- a/drivers/mtd/chips/cfi_cmdset_0001.c
+++ b/drivers/mtd/chips/cfi_cmdset_0001.c
@@ -2564,6 +2564,7 @@ static int cfi_intelext_reset(struct mtd_info *mtd)
 		if (!ret) {
 			map_write(map, CMD(0xff), chip->start);
 			chip->state = FL_SHUTDOWN;
+			put_chip(map, chip, chip->start);
 		}
 		spin_unlock(chip->mutex);
 	}

Re: [PATCH] MTD: add missing put_chip() in cfi_intelext_reset()

[Antonio Ospite]

[-- Attachment #1: Type: text/plain, Size: 1212 bytes --]

On Fri, 23 Oct 2009 16:02:42 -0400 (EDT)
Nicolas Pitre <nico@fluxnic.net> wrote:

> 
> Signed-off-by: Nicolas Pitre <nico@fluxnic.net>
> 
> ---
> 
> On Fri, 23 Oct 2009, Antonio Ospite wrote:
> 
> > Hi,
> > 
> > I am attaching a trace of a possible recursive locking in
> > cfi_cmdset_0001.c
> 
> This patch should fix it.
> 
> diff --git a/drivers/mtd/chips/cfi_cmdset_0001.c b/drivers/mtd/chips/cfi_cmdset_0001.c
> index e7563a9..d923a2d 100644
> --- a/drivers/mtd/chips/cfi_cmdset_0001.c
> +++ b/drivers/mtd/chips/cfi_cmdset_0001.c
> @@ -2564,6 +2564,7 @@ static int cfi_intelext_reset(struct mtd_info *mtd)
>  		if (!ret) {
>  			map_write(map, CMD(0xff), chip->start);
>  			chip->state = FL_SHUTDOWN;
> +			put_chip(map, chip, chip->start);
>  		}
>  		spin_unlock(chip->mutex);
>  	}

I've just tested this and it seems to fix the issue, indeed.

Many thanks,
   Antonio

-- 
Antonio Ospite
http://ao2.it

PGP public key ID: 0x4553B001

A: Because it messes up the order in which people normally read text.
   See http://en.wikipedia.org/wiki/Posting_style
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [PATCH] MTD: add missing put_chip() in cfi_intelext_reset()

[Antonio Ospite]

[-- Attachment #1: Type: text/plain, Size: 1217 bytes --]

On Fri, 23 Oct 2009 16:02:42 -0400 (EDT)
Nicolas Pitre <nico@fluxnic.net> wrote:

> 
> Signed-off-by: Nicolas Pitre <nico@fluxnic.net>
>

Hi, has this fix been submitted yet?
I can't see it in 2.6.32-rc8.

Thanks,
   Antonio

> ---
> 
> On Fri, 23 Oct 2009, Antonio Ospite wrote:
> 
> > Hi,
> > 
> > I am attaching a trace of a possible recursive locking in
> > cfi_cmdset_0001.c
> 
> This patch should fix it.
> 
> diff --git a/drivers/mtd/chips/cfi_cmdset_0001.c b/drivers/mtd/chips/cfi_cmdset_0001.c
> index e7563a9..d923a2d 100644
> --- a/drivers/mtd/chips/cfi_cmdset_0001.c
> +++ b/drivers/mtd/chips/cfi_cmdset_0001.c
> @@ -2564,6 +2564,7 @@ static int cfi_intelext_reset(struct mtd_info *mtd)
>  		if (!ret) {
>  			map_write(map, CMD(0xff), chip->start);
>  			chip->state = FL_SHUTDOWN;
> +			put_chip(map, chip, chip->start);
>  		}
>  		spin_unlock(chip->mutex);
>  	}


-- 
Antonio Ospite
http://ao2.it

PGP public key ID: 0x4553B001

A: Because it messes up the order in which people normally read text.
   See http://en.wikipedia.org/wiki/Posting_style
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [PATCH] MTD: add missing put_chip() in cfi_intelext_reset()

[Artem Bityutskiy]

On Fri, 2009-11-20 at 11:14 +0100, Antonio Ospite wrote:
> On Fri, 23 Oct 2009 16:02:42 -0400 (EDT)
> Nicolas Pitre <nico@fluxnic.net> wrote:
> 
> > 
> > Signed-off-by: Nicolas Pitre <nico@fluxnic.net>
> >
> 
> Hi, has this fix been submitted yet?
> I can't see it in 2.6.32-rc8.

The current situation indicates it is not going to be there.

This patch is in my l2-mtd tree, though.

-- 
Best Regards,
Artem Bityutskiy (Артём Битюцкий)

Re: [PATCH] MTD: add missing put_chip() in cfi_intelext_reset()

[Antonio Ospite]

[-- Attachment #1: Type: text/plain, Size: 1029 bytes --]

On Thu, 26 Nov 2009 09:01:18 +0200
Artem Bityutskiy <dedekind1@gmail.com> wrote:

> On Fri, 2009-11-20 at 11:14 +0100, Antonio Ospite wrote:
> > On Fri, 23 Oct 2009 16:02:42 -0400 (EDT)
> > Nicolas Pitre <nico@fluxnic.net> wrote:
> > 
> > > 
> > > Signed-off-by: Nicolas Pitre <nico@fluxnic.net>
> > >
> > 
> > Hi, has this fix been submitted yet?
> > I can't see it in 2.6.32-rc8.
> 
> The current situation indicates it is not going to be there.
> 
> This patch is in my l2-mtd tree, though.
>

Ok, thanks, I just want to be sure this will find its way to
mainline, I don't care which path it takes :)

> -- 
> Best Regards,
> Artem Bityutskiy (Артём Битюцкий)
> 

Regards,
   Antonio

-- 
Antonio Ospite
http://ao2.it

PGP public key ID: 0x4553B001

A: Because it messes up the order in which people normally read text.
   See http://en.wikipedia.org/wiki/Posting_style
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]