[PATCH net v2] atm: fore200e: fix use-after-free in tasklets during device removal

When the PCA-200E or SBA-200E adapter is being detached, the fore200e
is deallocated. However, the tx_tasklet or rx_tasklet may still be running
or pending, leading to use-after-free bug when the already freed fore200e
is accessed again in fore200e_tx_tasklet() or fore200e_rx_tasklet().

One of the race conditions can occur as follows:

CPU 0 (cleanup)           | CPU 1 (tasklet)
fore200e_pca_remove_one() | fore200e_interrupt()
  fore200e_shutdown()     |   tasklet_schedule()
    kfree(fore200e)       | fore200e_tx_tasklet()
                          |   fore200e-> // UAF

Fix this by ensuring tx_tasklet or rx_tasklet is properly canceled before
the fore200e is released. Add tasklet_kill() in fore200e_shutdown() to
synchronize with any pending or running tasklets. Moreover, since
fore200e_reset() could prevent further interrupts or data transfers,
the tasklet_kill() should be placed after fore200e_reset() to prevent
the tasklet from being rescheduled in fore200e_interrupt().

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Duoming Zhou <redacted>
---
Changes in v2:
  - Move tasklet_kill() after fore200e_reset().

 drivers/atm/fore200e.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/atm/fore200e.c b/drivers/atm/fore200e.c
index f62e3857144..de04c407921 100644
--- a/drivers/atm/fore200e.c
+++ b/drivers/atm/fore200e.c

@@ -362,6 +362,10 @@ fore200e_shutdown(struct fore200e* fore200e)
     if (fore200e->state > FORE200E_STATE_RESET) {
 	/* first, reset the board to prevent further interrupts or data transfers */
 	fore200e_reset(fore200e, 0);
+#ifdef FORE200E_USE_TASKLET
+	tasklet_kill(&fore200e->tx_tasklet);
+	tasklet_kill(&fore200e->rx_tasklet);
+#endif
     }
     
     /* then, release all allocated resources */

-- 
2.34.1

on 2026/2/6 17:48, Duoming Zhou wrote:

I'm sorry if I gave you a confusing comment.

If (fore200e->state <= FORE200E_STATE_RESET), is there no need to do tasklet_kill()?

Tips: If you need to send v3, please don’t repost your patche within one 24h period:
   https://www.kernel.org/doc/html/next/process/maintainer-netdev.html

   Jijie Shao

On Fri, 6 Feb 2026 18:48:38 +0800 Jijie Shao wrote:

quoted

When the PCA-200E or SBA-200E adapter is being detached, the fore200e
is deallocated. However, the tx_tasklet or rx_tasklet may still be running
or pending, leading to use-after-free bug when the already freed fore200e
is accessed again in fore200e_tx_tasklet() or fore200e_rx_tasklet().

One of the race conditions can occur as follows:

CPU 0 (cleanup)           | CPU 1 (tasklet)
fore200e_pca_remove_one() | fore200e_interrupt()
   fore200e_shutdown()     |   tasklet_schedule()
     kfree(fore200e)       | fore200e_tx_tasklet()
                           |   fore200e-> // UAF

Fix this by ensuring tx_tasklet or rx_tasklet is properly canceled before
the fore200e is released. Add tasklet_kill() in fore200e_shutdown() to
synchronize with any pending or running tasklets. Moreover, since
fore200e_reset() could prevent further interrupts or data transfers,
the tasklet_kill() should be placed after fore200e_reset() to prevent
the tasklet from being rescheduled in fore200e_interrupt().

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Duoming Zhou <redacted>
---
Changes in v2:
   - Move tasklet_kill() after fore200e_reset().

  drivers/atm/fore200e.c | 4 ++++
  1 file changed, 4 insertions(+)

diff --git a/drivers/atm/fore200e.c b/drivers/atm/fore200e.c
index f62e3857144..de04c407921 100644
--- a/drivers/atm/fore200e.c
+++ b/drivers/atm/fore200e.c

@@ -362,6 +362,10 @@ fore200e_shutdown(struct fore200e* fore200e)
      if (fore200e->state > FORE200E_STATE_RESET) {
  	/* first, reset the board to prevent further interrupts or data transfers */
  	fore200e_reset(fore200e, 0);
+#ifdef FORE200E_USE_TASKLET
+	tasklet_kill(&fore200e->tx_tasklet);
+	tasklet_kill(&fore200e->rx_tasklet);
+#endif
      }

I'm sorry if I gave you a confusing comment.

If (fore200e->state <= FORE200E_STATE_RESET), is there no need to do tasklet_kill()?

The following four states that are preceding FORE200E_STATE_RESET are only set
during device initialization: FORE200E_STATE_BLANK, FORE200E_STATE_REGISTER, 
FORE200E_STATE_CONFIGURE, and FORE200E_STATE_MAP.

If the device is in any of these states, it means the initialization is
not complete and interrupts have not been registered. Therefore, tasklet
could not be scheduled through fore200e_interrupt().

If the device is in FORE200E_STATE_RESET state, the fore200e_reset() has 
already reset the device, which could prevent further interrupts and 
tasklet scheduling.

So there is no need to do tasklet_kill() if the state of fore200e is
less than or equal to FORE200E_STATE_RESET.

Best regards,
Duoming Zhou

on 2026/2/6 22:56, duoming@zju.edu.cn wrote:

On Fri, 6 Feb 2026 18:48:38 +0800 Jijie Shao wrote:

quoted

quoted

When the PCA-200E or SBA-200E adapter is being detached, the fore200e
is deallocated. However, the tx_tasklet or rx_tasklet may still be running
or pending, leading to use-after-free bug when the already freed fore200e
is accessed again in fore200e_tx_tasklet() or fore200e_rx_tasklet().

One of the race conditions can occur as follows:

CPU 0 (cleanup)           | CPU 1 (tasklet)
fore200e_pca_remove_one() | fore200e_interrupt()
    fore200e_shutdown()     |   tasklet_schedule()
      kfree(fore200e)       | fore200e_tx_tasklet()
                            |   fore200e-> // UAF

Fix this by ensuring tx_tasklet or rx_tasklet is properly canceled before
the fore200e is released. Add tasklet_kill() in fore200e_shutdown() to
synchronize with any pending or running tasklets. Moreover, since
fore200e_reset() could prevent further interrupts or data transfers,
the tasklet_kill() should be placed after fore200e_reset() to prevent
the tasklet from being rescheduled in fore200e_interrupt().

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Duoming Zhou <redacted>
---
Changes in v2:
    - Move tasklet_kill() after fore200e_reset().

   drivers/atm/fore200e.c | 4 ++++
   1 file changed, 4 insertions(+)

diff --git a/drivers/atm/fore200e.c b/drivers/atm/fore200e.c
index f62e3857144..de04c407921 100644
--- a/drivers/atm/fore200e.c
+++ b/drivers/atm/fore200e.c

@@ -362,6 +362,10 @@ fore200e_shutdown(struct fore200e* fore200e)
       if (fore200e->state > FORE200E_STATE_RESET) {
   	/* first, reset the board to prevent further interrupts or data transfers */
   	fore200e_reset(fore200e, 0);
+#ifdef FORE200E_USE_TASKLET
+	tasklet_kill(&fore200e->tx_tasklet);
+	tasklet_kill(&fore200e->rx_tasklet);
+#endif
       }

I'm sorry if I gave you a confusing comment.

If (fore200e->state <= FORE200E_STATE_RESET), is there no need to do tasklet_kill()?

The following four states that are preceding FORE200E_STATE_RESET are only set
during device initialization: FORE200E_STATE_BLANK, FORE200E_STATE_REGISTER,
FORE200E_STATE_CONFIGURE, and FORE200E_STATE_MAP.

If the device is in any of these states, it means the initialization is
not complete and interrupts have not been registered. Therefore, tasklet
could not be scheduled through fore200e_interrupt().

If the device is in FORE200E_STATE_RESET state, the fore200e_reset() has
already reset the device, which could prevent further interrupts and
tasklet scheduling.

So there is no need to do tasklet_kill() if the state of fore200e is
less than or equal to FORE200E_STATE_RESET.

Yeah, I agree.
But I reviewed the fore200e_init() again, and it only needs to do tasklet_kill()
when "fore200e->state >= fFORE200E_STATE_IRQ"(in fore200e_irq_request());
otherwise, "fore200e->tx_tasklet" has not been initialized.

Jijie Shao

On Mon, 9 Feb 2026 14:41:52 +0800 Jijie Shao wrote:

quoted

quoted

quoted

When the PCA-200E or SBA-200E adapter is being detached, the fore200e
is deallocated. However, the tx_tasklet or rx_tasklet may still be running
or pending, leading to use-after-free bug when the already freed fore200e
is accessed again in fore200e_tx_tasklet() or fore200e_rx_tasklet().

One of the race conditions can occur as follows:

CPU 0 (cleanup)           | CPU 1 (tasklet)
fore200e_pca_remove_one() | fore200e_interrupt()
    fore200e_shutdown()     |   tasklet_schedule()
      kfree(fore200e)       | fore200e_tx_tasklet()
                            |   fore200e-> // UAF

Fix this by ensuring tx_tasklet or rx_tasklet is properly canceled before
the fore200e is released. Add tasklet_kill() in fore200e_shutdown() to
synchronize with any pending or running tasklets. Moreover, since
fore200e_reset() could prevent further interrupts or data transfers,
the tasklet_kill() should be placed after fore200e_reset() to prevent
the tasklet from being rescheduled in fore200e_interrupt().

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Duoming Zhou <redacted>
---
Changes in v2:
    - Move tasklet_kill() after fore200e_reset().

   drivers/atm/fore200e.c | 4 ++++
   1 file changed, 4 insertions(+)

diff --git a/drivers/atm/fore200e.c b/drivers/atm/fore200e.c
index f62e3857144..de04c407921 100644
--- a/drivers/atm/fore200e.c
+++ b/drivers/atm/fore200e.c

@@ -362,6 +362,10 @@ fore200e_shutdown(struct fore200e* fore200e)
       if (fore200e->state > FORE200E_STATE_RESET) {
   	/* first, reset the board to prevent further interrupts or data transfers */
   	fore200e_reset(fore200e, 0);
+#ifdef FORE200E_USE_TASKLET
+	tasklet_kill(&fore200e->tx_tasklet);
+	tasklet_kill(&fore200e->rx_tasklet);
+#endif
       }

I'm sorry if I gave you a confusing comment.

If (fore200e->state <= FORE200E_STATE_RESET), is there no need to do tasklet_kill()?

The following four states that are preceding FORE200E_STATE_RESET are only set
during device initialization: FORE200E_STATE_BLANK, FORE200E_STATE_REGISTER,
FORE200E_STATE_CONFIGURE, and FORE200E_STATE_MAP.

If the device is in any of these states, it means the initialization is
not complete and interrupts have not been registered. Therefore, tasklet
could not be scheduled through fore200e_interrupt().

If the device is in FORE200E_STATE_RESET state, the fore200e_reset() has
already reset the device, which could prevent further interrupts and
tasklet scheduling.

So there is no need to do tasklet_kill() if the state of fore200e is
less than or equal to FORE200E_STATE_RESET.

Yeah, I agree.
But I reviewed the fore200e_init() again, and it only needs to do tasklet_kill()
when "fore200e->state >= fFORE200E_STATE_IRQ"(in fore200e_irq_request());
otherwise, "fore200e->tx_tasklet" has not been initialized.

Thank you for your time and suggestions! You are right, I will adopt your 
suggestions in the v3 patch.

Best regards,
Duoming Zhou