[RFC][PATCH v4 0/3] make efivars/efi_pstore interrupt-safe

[RFC][PATCH v4 0/3] make efivars/efi_pstore interrupt-safe

[Seiji Aguchi]

Changelog
v3 -> v4
  - Patch 2/3
    Move cancel_work_sync() above an efi_enabled test in efivars_exit().

v2 -> v3
  - Patch 1/3
    Replace spin_lock_irqsave/spin_unlock_irqrestore with spin_lock_irq/spin_unlock_irq in efivars_unregister(),
    efivar_create(), efivar_store_raw() and efivar_delete() which are called in a process context. 

 - Patch 2/3
    Change a name of delete_sysfs_entry() to delete_all_stale_sysfs_entries().
    Also, don't release an efivar->lock while searching efivar->list in delete_all_stale_sysfs_entries().

 - Patch 3/3
    Remove a logic in efi_pstore_erase() which freshly created in patch v2.

v1 -> v2
 - Patch 1/3
    Add spin_lock_irq/spin_unlock_irq to open/close callbacks of efi_pstore 
    instead of moving spin_locks to a read callback.    

 - Patch 2/3
    Replace a periodical timer with schedule_work().

 - Patch 3/3
    freshly create to kick a workqueue in oops case only.

[Problem]
 There are following problems related to an interrupt context in efivar/efi_pstore.

 Currently, efivars enables interrupt while taking efivars->lock.
 So, there is a risk to be deadlocking in a write callback of efi_pstore if kernel panics 
 in interrupt context while taking efi_lock.

 Also, efi_pstore creates sysfs entries ,which enable users to access to NVRAM, in a write callback.
 If a kernel panic happens in interrupt contexts, pstore may fail because it could sleep due to dynamic 
 memory allocations during creating sysfs entries.

To resolve the problems above, a goal of this patchset is making efivars/efi_pstore interrupt-safe.

[Patch Description]
 Patch 1/3 efivars: Disable external interrupt while holding efivars->lock
   This patch replaces spin_lock/spin_unlock with spin_lock_irqsave/spin_lock_irqrestore to make efivars interrupt safe

 Patch 2/3 efi_pstore: Introducing workqueue updating sysfs entries
   This patch removes sysfs operations from write callback by introducing a workqueue updating sysfs entries

 Patch 3/3 efi_pstore: Skiping scheduling a workqueue in cases other than oops
   This patch restricts a schedule of a workqueue in case where users erase entries or oops happen which is truly needed for users.

 drivers/firmware/efivars.c |  167 +++++++++++++++++++++++++++++++++++--------
 include/linux/efi.h        |    3 +-
 2 files changed, 138 insertions(+), 32 deletions(-)

RE: [RFC][PATCH v4 0/3] make efivars/efi_pstore interrupt-safe

[Seiji Aguchi]

Tony,

Cloud you please review this patchset?

Seiji

> -----Original Message-----
> From: Seiji Aguchi
> Sent: Tuesday, August 21, 2012 4:35 PM
> To: 'Mike Waychison'; 'linux-kernel@vger.kernel.org'; 'Luck, Tony (tony.luck@intel.com)'; 'Matthew Garrett (mjg@redhat.com)';
> 'dzickus@redhat.com'
> Cc: 'dle-develop@lists.sourceforge.net'; Satoru Moriya
> Subject: [RFC][PATCH v4 0/3] make efivars/efi_pstore interrupt-safe
> 
> Changelog
> v3 -> v4
>   - Patch 2/3
>     Move cancel_work_sync() above an efi_enabled test in efivars_exit().
> 
> v2 -> v3
>   - Patch 1/3
>     Replace spin_lock_irqsave/spin_unlock_irqrestore with spin_lock_irq/spin_unlock_irq in efivars_unregister(),
>     efivar_create(), efivar_store_raw() and efivar_delete() which are called in a process context.
> 
>  - Patch 2/3
>     Change a name of delete_sysfs_entry() to delete_all_stale_sysfs_entries().
>     Also, don't release an efivar->lock while searching efivar->list in delete_all_stale_sysfs_entries().
> 
>  - Patch 3/3
>     Remove a logic in efi_pstore_erase() which freshly created in patch v2.
> 
> v1 -> v2
>  - Patch 1/3
>     Add spin_lock_irq/spin_unlock_irq to open/close callbacks of efi_pstore
>     instead of moving spin_locks to a read callback.
> 
>  - Patch 2/3
>     Replace a periodical timer with schedule_work().
> 
>  - Patch 3/3
>     freshly create to kick a workqueue in oops case only.
> 
> [Problem]
>  There are following problems related to an interrupt context in efivar/efi_pstore.
> 
>  Currently, efivars enables interrupt while taking efivars->lock.
>  So, there is a risk to be deadlocking in a write callback of efi_pstore if kernel panics  in interrupt context while taking efi_lock.
> 
>  Also, efi_pstore creates sysfs entries ,which enable users to access to NVRAM, in a write callback.
>  If a kernel panic happens in interrupt contexts, pstore may fail because it could sleep due to dynamic  memory allocations during
> creating sysfs entries.
> 
> To resolve the problems above, a goal of this patchset is making efivars/efi_pstore interrupt-safe.
> 
> [Patch Description]
>  Patch 1/3 efivars: Disable external interrupt while holding efivars->lock
>    This patch replaces spin_lock/spin_unlock with spin_lock_irqsave/spin_lock_irqrestore to make efivars interrupt safe
> 
>  Patch 2/3 efi_pstore: Introducing workqueue updating sysfs entries
>    This patch removes sysfs operations from write callback by introducing a workqueue updating sysfs entries
> 
>  Patch 3/3 efi_pstore: Skiping scheduling a workqueue in cases other than oops
>    This patch restricts a schedule of a workqueue in case where users erase entries or oops happen which is truly needed for users.
> 
>  drivers/firmware/efivars.c |  167 +++++++++++++++++++++++++++++++++++--------
>  include/linux/efi.h        |    3 +-
>  2 files changed, 138 insertions(+), 32 deletions(-)
> 
> 
>