[PATCH] ocxl: Fix page fault handler in case of fault on dying process

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

[PATCH] ocxl: Fix page fault handler in case of fault on dying process

Frederic Barrat-3
If a process exits without doing proper cleanup, there's a window
where an opencapi device can try to access the memory of the dying
process and may trigger a page fault. That's an expected scenario and
the ocxl driver holds a reference on the mm_struct of the process
until the opencapi device is notified of the process exiting.
However, if mm_users is already at 0, i.e. the address space of the
process has already been destroyed, the driver shouldn't try resolving
the page fault, as it will fail, but it can also try accessing already
freed data.

It is fixed by only calling the bottom half of the page fault handler
if mm_users is greater than 0 and get a reference on mm_users instead
of mm_count. Otherwise, we can safely return a translation fault to
the device, as its associated memory context is being removed. The
opencapi device will be properly cleaned up shortly after when closing
the file descriptors.

Fixes: 5ef3166e8a32 ("ocxl: Driver code for 'generic' opencapi devices")
Cc: [hidden email] # v4.16+
Signed-off-by: Frederic Barrat <[hidden email]>
---
 drivers/misc/ocxl/link.c | 24 +++++++++++++++---------
 1 file changed, 15 insertions(+), 9 deletions(-)

diff --git a/drivers/misc/ocxl/link.c b/drivers/misc/ocxl/link.c
index f30790582dc0..2aeaf34e7eda 100644
--- a/drivers/misc/ocxl/link.c
+++ b/drivers/misc/ocxl/link.c
@@ -136,7 +136,7 @@ static void xsl_fault_handler_bh(struct work_struct *fault_work)
  int rc;
 
  /*
- * We need to release a reference on the mm whenever exiting this
+ * We must release a reference on mm_users whenever exiting this
  * function (taken in the memory fault interrupt handler)
  */
  rc = copro_handle_mm_fault(fault->pe_data.mm, fault->dar, fault->dsisr,
@@ -172,7 +172,7 @@ static void xsl_fault_handler_bh(struct work_struct *fault_work)
  }
  r = RESTART;
 ack:
- mmdrop(fault->pe_data.mm);
+ mmput(fault->pe_data.mm);
  ack_irq(spa, r);
 }
 
@@ -184,6 +184,7 @@ static irqreturn_t xsl_fault_handler(int irq, void *data)
  struct pe_data *pe_data;
  struct ocxl_process_element *pe;
  int lpid, pid, tid;
+ bool schedule = false;
 
  read_irq(spa, &dsisr, &dar, &pe_handle);
  trace_ocxl_fault(spa->spa_mem, pe_handle, dsisr, dar, -1);
@@ -226,14 +227,19 @@ static irqreturn_t xsl_fault_handler(int irq, void *data)
  }
  WARN_ON(pe_data->mm->context.id != pid);
 
- spa->xsl_fault.pe = pe_handle;
- spa->xsl_fault.dar = dar;
- spa->xsl_fault.dsisr = dsisr;
- spa->xsl_fault.pe_data = *pe_data;
- mmgrab(pe_data->mm); /* mm count is released by bottom half */
-
+ if (mmget_not_zero(pe_data->mm)) {
+ spa->xsl_fault.pe = pe_handle;
+ spa->xsl_fault.dar = dar;
+ spa->xsl_fault.dsisr = dsisr;
+ spa->xsl_fault.pe_data = *pe_data;
+ schedule = true;
+ /* mm_users count released by bottom half */
+ }
  rcu_read_unlock();
- schedule_work(&spa->xsl_fault.fault_work);
+ if (schedule)
+ schedule_work(&spa->xsl_fault.fault_work);
+ else
+ ack_irq(spa, ADDRESS_ERROR);
  return IRQ_HANDLED;
 }
 
--
2.17.1

Reply | Threaded
Open this post in threaded view
|

Re: [PATCH] ocxl: Fix page fault handler in case of fault on dying process

Alastair D'Silva
On Mon, 2018-06-18 at 14:14 +0200, Frederic Barrat wrote:

> If a process exits without doing proper cleanup, there's a window
> where an opencapi device can try to access the memory of the dying
> process and may trigger a page fault. That's an expected scenario and
> the ocxl driver holds a reference on the mm_struct of the process
> until the opencapi device is notified of the process exiting.
> However, if mm_users is already at 0, i.e. the address space of the
> process has already been destroyed, the driver shouldn't try
> resolving
> the page fault, as it will fail, but it can also try accessing
> already
> freed data.
>
> It is fixed by only calling the bottom half of the page fault handler
> if mm_users is greater than 0 and get a reference on mm_users instead
> of mm_count. Otherwise, we can safely return a translation fault to
> the device, as its associated memory context is being removed. The
> opencapi device will be properly cleaned up shortly after when
> closing
> the file descriptors.
>
> Fixes: 5ef3166e8a32 ("ocxl: Driver code for 'generic' opencapi
> devices")
> Cc: [hidden email] # v4.16+
> Signed-off-by: Frederic Barrat <[hidden email]>
> ---
>  drivers/misc/ocxl/link.c | 24 +++++++++++++++---------
>  1 file changed, 15 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/misc/ocxl/link.c b/drivers/misc/ocxl/link.c
> index f30790582dc0..2aeaf34e7eda 100644
> --- a/drivers/misc/ocxl/link.c
> +++ b/drivers/misc/ocxl/link.c
> @@ -136,7 +136,7 @@ static void xsl_fault_handler_bh(struct
> work_struct *fault_work)
>   int rc;
>  
>   /*
> - * We need to release a reference on the mm whenever exiting
> this
> + * We must release a reference on mm_users whenever exiting
> this
>   * function (taken in the memory fault interrupt handler)
>   */
>   rc = copro_handle_mm_fault(fault->pe_data.mm, fault->dar,
> fault->dsisr,
> @@ -172,7 +172,7 @@ static void xsl_fault_handler_bh(struct
> work_struct *fault_work)
>   }
>   r = RESTART;
>  ack:
> - mmdrop(fault->pe_data.mm);
> + mmput(fault->pe_data.mm);
>   ack_irq(spa, r);
>  }
>  
> @@ -184,6 +184,7 @@ static irqreturn_t xsl_fault_handler(int irq,
> void *data)
>   struct pe_data *pe_data;
>   struct ocxl_process_element *pe;
>   int lpid, pid, tid;
> + bool schedule = false;
>  
>   read_irq(spa, &dsisr, &dar, &pe_handle);
>   trace_ocxl_fault(spa->spa_mem, pe_handle, dsisr, dar, -1);
> @@ -226,14 +227,19 @@ static irqreturn_t xsl_fault_handler(int irq,
> void *data)
>   }
>   WARN_ON(pe_data->mm->context.id != pid);
>  
> - spa->xsl_fault.pe = pe_handle;
> - spa->xsl_fault.dar = dar;
> - spa->xsl_fault.dsisr = dsisr;
> - spa->xsl_fault.pe_data = *pe_data;
> - mmgrab(pe_data->mm); /* mm count is released by bottom half
> */
> -
> + if (mmget_not_zero(pe_data->mm)) {
> + spa->xsl_fault.pe = pe_handle;
> + spa->xsl_fault.dar = dar;
> + spa->xsl_fault.dsisr = dsisr;
> + spa->xsl_fault.pe_data = *pe_data;
> + schedule = true;
> + /* mm_users count released by bottom half */
> + }
>   rcu_read_unlock();
> - schedule_work(&spa->xsl_fault.fault_work);
> + if (schedule)
> + schedule_work(&spa->xsl_fault.fault_work);
> + else
> + ack_irq(spa, ADDRESS_ERROR);
>   return IRQ_HANDLED;
>  }
>  

Reviewed-By: Alastair D'Silva <[hidden email]>

--
Alastair D'Silva
Open Source Developer
Linux Technology Centre, IBM Australia
mob: 0423 762 819

Reply | Threaded
Open this post in threaded view
|

Re: [PATCH] ocxl: Fix page fault handler in case of fault on dying process

Andrew Donnellan
In reply to this post by Frederic Barrat-3
On 18/06/18 22:14, Frederic Barrat wrote:

> If a process exits without doing proper cleanup, there's a window
> where an opencapi device can try to access the memory of the dying
> process and may trigger a page fault. That's an expected scenario and
> the ocxl driver holds a reference on the mm_struct of the process
> until the opencapi device is notified of the process exiting.
> However, if mm_users is already at 0, i.e. the address space of the
> process has already been destroyed, the driver shouldn't try resolving
> the page fault, as it will fail, but it can also try accessing already
> freed data.
>
> It is fixed by only calling the bottom half of the page fault handler
> if mm_users is greater than 0 and get a reference on mm_users instead
> of mm_count. Otherwise, we can safely return a translation fault to
> the device, as its associated memory context is being removed. The
> opencapi device will be properly cleaned up shortly after when closing
> the file descriptors.
>
> Fixes: 5ef3166e8a32 ("ocxl: Driver code for 'generic' opencapi devices")
> Cc: [hidden email] # v4.16+
> Signed-off-by: Frederic Barrat <[hidden email]>

I think I understand what's going on here and it looks right.

Acked-by: Andrew Donnellan <[hidden email]>

--
Andrew Donnellan              OzLabs, ADL Canberra
[hidden email]  IBM Australia Limited

Reply | Threaded
Open this post in threaded view
|

Re: ocxl: Fix page fault handler in case of fault on dying process

Michael Ellerman-3
In reply to this post by Frederic Barrat-3
On Mon, 2018-06-18 at 12:14:36 UTC, Frederic Barrat wrote:

> If a process exits without doing proper cleanup, there's a window
> where an opencapi device can try to access the memory of the dying
> process and may trigger a page fault. That's an expected scenario and
> the ocxl driver holds a reference on the mm_struct of the process
> until the opencapi device is notified of the process exiting.
> However, if mm_users is already at 0, i.e. the address space of the
> process has already been destroyed, the driver shouldn't try resolving
> the page fault, as it will fail, but it can also try accessing already
> freed data.
>
> It is fixed by only calling the bottom half of the page fault handler
> if mm_users is greater than 0 and get a reference on mm_users instead
> of mm_count. Otherwise, we can safely return a translation fault to
> the device, as its associated memory context is being removed. The
> opencapi device will be properly cleaned up shortly after when closing
> the file descriptors.
>
> Fixes: 5ef3166e8a32 ("ocxl: Driver code for 'generic' opencapi devices")
> Cc: [hidden email] # v4.16+
> Signed-off-by: Frederic Barrat <[hidden email]>
> Reviewed-By: Alastair D'Silva <[hidden email]>
> Acked-by: Andrew Donnellan <[hidden email]>

Applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/d497ebf5fb3a026c0817f8c96cde57

cheers