[v2] arm64/sve: Fix wrong free for task->thread.sve_state
diff mbox series

Message ID 20190927153949.29870-1-msys.mizuma@gmail.com
State New
Headers show
Series
  • [v2] arm64/sve: Fix wrong free for task->thread.sve_state
Related show

Commit Message

Masayoshi Mizuma Sept. 27, 2019, 3:39 p.m. UTC
From: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>

The system which has SVE feature crashed because of
the memory pointed by task->thread.sve_state was destroyed
by someone.

That is because sve_state is freed while the forking the
child process. The child process has the pointer of sve_state
which is same as the parent's because the child's task_struct
is copied from the parent's one. If the copy_process()
fails as an error on somewhere, for example, copy_creds(),
then the sve_state is freed even if the parent is alive.
The flow is as follows.

copy_process
        p = dup_task_struct
            => arch_dup_task_struct
                *dst = *src;  // copy the entire region.
:
        retval = copy_creds
        if (retval < 0)
                goto bad_fork_free;
:
bad_fork_free:
...
        delayed_free_task(p);
          => free_task
             => arch_release_task_struct
                => fpsimd_release_task
                   => __sve_free
                      => kfree(task->thread.sve_state);
                         // free the parent's sve_state

Move child's sve_state = NULL and clearing TIF_SVE flag
to arch_dup_task_struct() so that the child doesn't free the
parent's one.

Cc: stable@vger.kernel.org
Fixes: bc0ee4760364 ("arm64/sve: Core task context handling")
Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
Reported-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Suggested-by: Dave Martin <Dave.Martin@arm.com>
---
 arch/arm64/kernel/process.c | 21 ++++-----------------
 1 file changed, 4 insertions(+), 17 deletions(-)

Comments

Dave Martin Sept. 27, 2019, 4:15 p.m. UTC | #1
On Fri, Sep 27, 2019 at 11:39:49AM -0400, Masayoshi Mizuma wrote:
> From: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> 
> The system which has SVE feature crashed because of
> the memory pointed by task->thread.sve_state was destroyed
> by someone.
> 
> That is because sve_state is freed while the forking the
> child process. The child process has the pointer of sve_state
> which is same as the parent's because the child's task_struct
> is copied from the parent's one. If the copy_process()
> fails as an error on somewhere, for example, copy_creds(),
> then the sve_state is freed even if the parent is alive.
> The flow is as follows.
> 
> copy_process
>         p = dup_task_struct
>             => arch_dup_task_struct
>                 *dst = *src;  // copy the entire region.
> :
>         retval = copy_creds
>         if (retval < 0)
>                 goto bad_fork_free;
> :
> bad_fork_free:
> ...
>         delayed_free_task(p);
>           => free_task
>              => arch_release_task_struct
>                 => fpsimd_release_task
>                    => __sve_free
>                       => kfree(task->thread.sve_state);
>                          // free the parent's sve_state
> 
> Move child's sve_state = NULL and clearing TIF_SVE flag
> to arch_dup_task_struct() so that the child doesn't free the
> parent's one.

You could also add:

--8<--
There is no need to wait until copy_process() to clear TIF_SVE for
dst, becuase the thread flags for dst are initialized already by
copying the src task_struct.

This change simplifies the code, so get rid of comments that are no
longer needed.
-->8--

> 
> Cc: stable@vger.kernel.org

Since SVE only exists from v4.15, it may be helpful to specify that,
i.e., replace that Cc line with:

Cc: <stable@vger.kernel.org> # 4.15.x-


Otherwise, I'm happy to see this applied, but I'd like somebody to
confirm that this change definitely fixes the bug.

Cheers
---Dave

[...]

> Fixes: bc0ee4760364 ("arm64/sve: Core task context handling")
> Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> Reported-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
> Suggested-by: Dave Martin <Dave.Martin@arm.com>
> ---
>  arch/arm64/kernel/process.c | 21 ++++-----------------
>  1 file changed, 4 insertions(+), 17 deletions(-)
> 
> diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
> index f674f28df..6937f5935 100644
> --- a/arch/arm64/kernel/process.c
> +++ b/arch/arm64/kernel/process.c
> @@ -323,22 +323,16 @@ void arch_release_task_struct(struct task_struct *tsk)
>  	fpsimd_release_task(tsk);
>  }
>  
> -/*
> - * src and dst may temporarily have aliased sve_state after task_struct
> - * is copied.  We cannot fix this properly here, because src may have
> - * live SVE state and dst's thread_info may not exist yet, so tweaking
> - * either src's or dst's TIF_SVE is not safe.
> - *
> - * The unaliasing is done in copy_thread() instead.  This works because
> - * dst is not schedulable or traceable until both of these functions
> - * have been called.
> - */
>  int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
>  {
>  	if (current->mm)
>  		fpsimd_preserve_current_state();
>  	*dst = *src;
>  
> +	BUILD_BUG_ON(!IS_ENABLED(CONFIG_THREAD_INFO_IN_TASK));
> +	dst->thread.sve_state = NULL;
> +	clear_tsk_thread_flag(dst, TIF_SVE);
> +
>  	return 0;
>  }
>  
> @@ -351,13 +345,6 @@ int copy_thread(unsigned long clone_flags, unsigned long stack_start,
>  
>  	memset(&p->thread.cpu_context, 0, sizeof(struct cpu_context));
>  
> -	/*
> -	 * Unalias p->thread.sve_state (if any) from the parent task
> -	 * and disable discard SVE state for p:
> -	 */
> -	clear_tsk_thread_flag(p, TIF_SVE);
> -	p->thread.sve_state = NULL;
> -
>  	/*
>  	 * In case p was allocated the same task_struct pointer as some
>  	 * other recently-exited task, make sure p is disassociated from
> -- 
> 2.18.1
> 
> 
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Julien Grall Sept. 27, 2019, 7:56 p.m. UTC | #2
Hi Dave,

On 27/09/2019 17:15, Dave Martin wrote:
> On Fri, Sep 27, 2019 at 11:39:49AM -0400, Masayoshi Mizuma wrote:
>> From: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
>>
>> The system which has SVE feature crashed because of
>> the memory pointed by task->thread.sve_state was destroyed
>> by someone.
>>
>> That is because sve_state is freed while the forking the
>> child process. The child process has the pointer of sve_state
>> which is same as the parent's because the child's task_struct
>> is copied from the parent's one. If the copy_process()
>> fails as an error on somewhere, for example, copy_creds(),
>> then the sve_state is freed even if the parent is alive.
>> The flow is as follows.
>>
>> copy_process
>>          p = dup_task_struct
>>              => arch_dup_task_struct
>>                  *dst = *src;  // copy the entire region.
>> :
>>          retval = copy_creds
>>          if (retval < 0)
>>                  goto bad_fork_free;
>> :
>> bad_fork_free:
>> ...
>>          delayed_free_task(p);
>>            => free_task
>>               => arch_release_task_struct
>>                  => fpsimd_release_task
>>                     => __sve_free
>>                        => kfree(task->thread.sve_state);
>>                           // free the parent's sve_state
>>
>> Move child's sve_state = NULL and clearing TIF_SVE flag
>> to arch_dup_task_struct() so that the child doesn't free the
>> parent's one.
> 
> You could also add:
> 
> --8<--
> There is no need to wait until copy_process() to clear TIF_SVE for
> dst, becuase the thread flags for dst are initialized already by
> copying the src task_struct.
> 
> This change simplifies the code, so get rid of comments that are no
> longer needed.
> -->8--
> 
>>
>> Cc: stable@vger.kernel.org
> 
> Since SVE only exists from v4.15, it may be helpful to specify that,
> i.e., replace that Cc line with:
> 
> Cc: <stable@vger.kernel.org> # 4.15.x-
> 
> 
> Otherwise, I'm happy to see this applied, but I'd like somebody to
> confirm that this change definitely fixes the bug.

I am working on a reproducer for this. So I should be able to test it.

Cheers,
Julien Grall Sept. 30, 2019, 12:23 p.m. UTC | #3
Hi,

On 27/09/2019 16:39, Masayoshi Mizuma wrote:
> From: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> 
> The system which has SVE feature crashed because of
> the memory pointed by task->thread.sve_state was destroyed
> by someone.
> 
> That is because sve_state is freed while the forking the
> child process. The child process has the pointer of sve_state
> which is same as the parent's because the child's task_struct
> is copied from the parent's one. If the copy_process()
> fails as an error on somewhere, for example, copy_creds(),
> then the sve_state is freed even if the parent is alive.
> The flow is as follows.
> 
> copy_process
>          p = dup_task_struct
>              => arch_dup_task_struct
>                  *dst = *src;  // copy the entire region.
> :
>          retval = copy_creds
>          if (retval < 0)
>                  goto bad_fork_free;
> :
> bad_fork_free:
> ...
>          delayed_free_task(p);
>            => free_task
>               => arch_release_task_struct
>                  => fpsimd_release_task
>                     => __sve_free
>                        => kfree(task->thread.sve_state);
>                           // free the parent's sve_state
> 
> Move child's sve_state = NULL and clearing TIF_SVE flag
> to arch_dup_task_struct() so that the child doesn't free the
> parent's one.
> 
> Cc: stable@vger.kernel.org
> Fixes: bc0ee4760364 ("arm64/sve: Core task context handling")

Looking at the log, it looks like THREAD_INFO_IN_TASK was selected before the 
bc0ee4760364. So it should be fine to backport for all the Linux tree contain 
this commit.

> Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> Reported-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
> Suggested-by: Dave Martin <Dave.Martin@arm.com>

I have tested the patch and can confirm that double-free disappeared after the 
patch is applied:

Tested-by: Julien Grall <julien.grall@arm.com>

See below for a few comments.

> ---
>   arch/arm64/kernel/process.c | 21 ++++-----------------
>   1 file changed, 4 insertions(+), 17 deletions(-)
> 
> diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
> index f674f28df..6937f5935 100644
> --- a/arch/arm64/kernel/process.c
> +++ b/arch/arm64/kernel/process.c
> @@ -323,22 +323,16 @@ void arch_release_task_struct(struct task_struct *tsk)
>   	fpsimd_release_task(tsk);
>   }
>   
> -/*
> - * src and dst may temporarily have aliased sve_state after task_struct
> - * is copied.  We cannot fix this properly here, because src may have
> - * live SVE state and dst's thread_info may not exist yet, so tweaking
> - * either src's or dst's TIF_SVE is not safe.
> - *
> - * The unaliasing is done in copy_thread() instead.  This works because
> - * dst is not schedulable or traceable until both of these functions
> - * have been called.
> - */

It would be good to explain in the commit message why tweaking "dst" in 
arch_dup_task_struct() is fine.

 From my understanding, Arm64 used to have thread_info on the stack. So it would 
not be possible to clear TIF_SVE until the stack is initialized.

Now that the thread_info is part of the task, it should be valid to modify the 
flag from arch_dup_task_struct().

Note that technically, TIF_SVE does not need to be cleared from 
arch_dup_task_struct(). It could also be done from copy_thread(). But it is 
easier to keep the both changes together.

>   int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
>   {
>   	if (current->mm)
>   		fpsimd_preserve_current_state();
>   	*dst = *src;
>   
> +	BUILD_BUG_ON(!IS_ENABLED(CONFIG_THREAD_INFO_IN_TASK));

You may want to add a comment on top explaining why TIF_SVE is cleared here.

> +	dst->thread.sve_state = NULL;
> +	clear_tsk_thread_flag(dst, TIF_SVE);
> +
>   	return 0;
>   }
>   
> @@ -351,13 +345,6 @@ int copy_thread(unsigned long clone_flags, unsigned long stack_start,
>   
>   	memset(&p->thread.cpu_context, 0, sizeof(struct cpu_context));
>   
> -	/*
> -	 * Unalias p->thread.sve_state (if any) from the parent task
> -	 * and disable discard SVE state for p:
> -	 */
> -	clear_tsk_thread_flag(p, TIF_SVE);
> -	p->thread.sve_state = NULL;
> -
>   	/*
>   	 * In case p was allocated the same task_struct pointer as some
>   	 * other recently-exited task, make sure p is disassociated from
> 

Cheers,
Dave Martin Sept. 30, 2019, 1:02 p.m. UTC | #4
On Mon, Sep 30, 2019 at 01:23:18PM +0100, Julien Grall wrote:
> Hi,
> 
> On 27/09/2019 16:39, Masayoshi Mizuma wrote:
> >From: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> >
> >The system which has SVE feature crashed because of
> >the memory pointed by task->thread.sve_state was destroyed
> >by someone.
> >
> >That is because sve_state is freed while the forking the
> >child process. The child process has the pointer of sve_state
> >which is same as the parent's because the child's task_struct
> >is copied from the parent's one. If the copy_process()
> >fails as an error on somewhere, for example, copy_creds(),
> >then the sve_state is freed even if the parent is alive.
> >The flow is as follows.
> >
> >copy_process
> >         p = dup_task_struct
> >             => arch_dup_task_struct
> >                 *dst = *src;  // copy the entire region.
> >:
> >         retval = copy_creds
> >         if (retval < 0)
> >                 goto bad_fork_free;
> >:
> >bad_fork_free:
> >...
> >         delayed_free_task(p);
> >           => free_task
> >              => arch_release_task_struct
> >                 => fpsimd_release_task
> >                    => __sve_free
> >                       => kfree(task->thread.sve_state);
> >                          // free the parent's sve_state
> >
> >Move child's sve_state = NULL and clearing TIF_SVE flag
> >to arch_dup_task_struct() so that the child doesn't free the
> >parent's one.
> >
> >Cc: stable@vger.kernel.org
> >Fixes: bc0ee4760364 ("arm64/sve: Core task context handling")
> 
> Looking at the log, it looks like THREAD_INFO_IN_TASK was selected before
> the bc0ee4760364. So it should be fine to backport for all the Linux tree
> contain this commit.
> 
> >Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> >Reported-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
> >Suggested-by: Dave Martin <Dave.Martin@arm.com>
> 
> I have tested the patch and can confirm that double-free disappeared after
> the patch is applied:
> 
> Tested-by: Julien Grall <julien.grall@arm.com>

Good to have that confirmed -- thanks for verifying.

[...]

> >---
> >  arch/arm64/kernel/process.c | 21 ++++-----------------
> >  1 file changed, 4 insertions(+), 17 deletions(-)
> >
> >diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
> >index f674f28df..6937f5935 100644
> >--- a/arch/arm64/kernel/process.c
> >+++ b/arch/arm64/kernel/process.c
> >@@ -323,22 +323,16 @@ void arch_release_task_struct(struct task_struct *tsk)
> >  	fpsimd_release_task(tsk);
> >  }
> >-/*
> >- * src and dst may temporarily have aliased sve_state after task_struct
> >- * is copied.  We cannot fix this properly here, because src may have
> >- * live SVE state and dst's thread_info may not exist yet, so tweaking
> >- * either src's or dst's TIF_SVE is not safe.
> >- *
> >- * The unaliasing is done in copy_thread() instead.  This works because
> >- * dst is not schedulable or traceable until both of these functions
> >- * have been called.
> >- */
> 
> It would be good to explain in the commit message why tweaking "dst" in
> arch_dup_task_struct() is fine.
> 
> From my understanding, Arm64 used to have thread_info on the stack. So it
> would not be possible to clear TIF_SVE until the stack is initialized.
> 
> Now that the thread_info is part of the task, it should be valid to modify
> the flag from arch_dup_task_struct().
> 
> Note that technically, TIF_SVE does not need to be cleared from
> arch_dup_task_struct(). It could also be done from copy_thread(). But it is
> easier to keep the both changes together.
> 
> >  int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
> >  {
> >  	if (current->mm)
> >  		fpsimd_preserve_current_state();
> >  	*dst = *src;

Ack, some more explanation would be a good idea here.

Maybe the following comments are sufficient?

	/* We rely on the above assingment to initialise dst's thread_flags: */

> >+	BUILD_BUG_ON(!IS_ENABLED(CONFIG_THREAD_INFO_IN_TASK));
> 

and

	/*
	 * Detach src's sve_state (if any) from dst so that it does not
	 * get erroneously used or freed prematurely.  dst's sve_state
	 * will be allocated on demand later on if dst uses SVE.
	 * For consistency, also clear TIF_SVE here: this could be done
	 * later in copy_process(), but to avoid tripping up future
	 * maintainers it is best not to leave TIF_SVE and sve_state in
	 * an inconsistent state, even temporarily.
	 */

> >+	dst->thread.sve_state = NULL;
> >+	clear_tsk_thread_flag(dst, TIF_SVE);

(TIF_SVE should not usually be set in the first place of course, since
we are in a fork() or clone() syscall in src.  This may not be true if
a task is created using kernel_thread() while running in the context of
some user task that entered the kernel due to a trap or syscall --
though probably nobody should be doing that.)

[...]

Cheers
---DavE
Masayoshi Mizuma Sept. 30, 2019, 2:29 p.m. UTC | #5
Hi Julien and Dave,

On Mon, Sep 30, 2019 at 02:02:46PM +0100, Dave Martin wrote:
> On Mon, Sep 30, 2019 at 01:23:18PM +0100, Julien Grall wrote:
> > Hi,
> > 
> > On 27/09/2019 16:39, Masayoshi Mizuma wrote:
> > >From: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> > >
> > >The system which has SVE feature crashed because of
> > >the memory pointed by task->thread.sve_state was destroyed
> > >by someone.
> > >
> > >That is because sve_state is freed while the forking the
> > >child process. The child process has the pointer of sve_state
> > >which is same as the parent's because the child's task_struct
> > >is copied from the parent's one. If the copy_process()
> > >fails as an error on somewhere, for example, copy_creds(),
> > >then the sve_state is freed even if the parent is alive.
> > >The flow is as follows.
> > >
> > >copy_process
> > >         p = dup_task_struct
> > >             => arch_dup_task_struct
> > >                 *dst = *src;  // copy the entire region.
> > >:
> > >         retval = copy_creds
> > >         if (retval < 0)
> > >                 goto bad_fork_free;
> > >:
> > >bad_fork_free:
> > >...
> > >         delayed_free_task(p);
> > >           => free_task
> > >              => arch_release_task_struct
> > >                 => fpsimd_release_task
> > >                    => __sve_free
> > >                       => kfree(task->thread.sve_state);
> > >                          // free the parent's sve_state
> > >
> > >Move child's sve_state = NULL and clearing TIF_SVE flag
> > >to arch_dup_task_struct() so that the child doesn't free the
> > >parent's one.
> > >
> > >Cc: stable@vger.kernel.org
> > >Fixes: bc0ee4760364 ("arm64/sve: Core task context handling")
> > 
> > Looking at the log, it looks like THREAD_INFO_IN_TASK was selected before
> > the bc0ee4760364. So it should be fine to backport for all the Linux tree
> > contain this commit.

I think this patch is needed for the kernel has SVE support.
I'll add the Cc tag as Dave said:

Cc: stable@vger.kernel.org # 4.15+

So, I suppose this patch will be backported to stables 5.3.X,
5.2.X and longterm 4.19.X.
Does this make sense?

> > 
> > >Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> > >Reported-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
> > >Suggested-by: Dave Martin <Dave.Martin@arm.com>
> > 
> > I have tested the patch and can confirm that double-free disappeared after
> > the patch is applied:
> > 
> > Tested-by: Julien Grall <julien.grall@arm.com>

Thank you so much!

> 
> Good to have that confirmed -- thanks for verifying.
> 
> [...]
> 
> > >---
> > >  arch/arm64/kernel/process.c | 21 ++++-----------------
> > >  1 file changed, 4 insertions(+), 17 deletions(-)
> > >
> > >diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
> > >index f674f28df..6937f5935 100644
> > >--- a/arch/arm64/kernel/process.c
> > >+++ b/arch/arm64/kernel/process.c
> > >@@ -323,22 +323,16 @@ void arch_release_task_struct(struct task_struct *tsk)
> > >  	fpsimd_release_task(tsk);
> > >  }
> > >-/*
> > >- * src and dst may temporarily have aliased sve_state after task_struct
> > >- * is copied.  We cannot fix this properly here, because src may have
> > >- * live SVE state and dst's thread_info may not exist yet, so tweaking
> > >- * either src's or dst's TIF_SVE is not safe.
> > >- *
> > >- * The unaliasing is done in copy_thread() instead.  This works because
> > >- * dst is not schedulable or traceable until both of these functions
> > >- * have been called.
> > >- */
> > 
> > It would be good to explain in the commit message why tweaking "dst" in
> > arch_dup_task_struct() is fine.
> > 
> > From my understanding, Arm64 used to have thread_info on the stack. So it
> > would not be possible to clear TIF_SVE until the stack is initialized.
> > 
> > Now that the thread_info is part of the task, it should be valid to modify
> > the flag from arch_dup_task_struct().
> > 
> > Note that technically, TIF_SVE does not need to be cleared from
> > arch_dup_task_struct(). It could also be done from copy_thread(). But it is
> > easier to keep the both changes together.

Thanks, let me add some comments to the commit log.

> > 
> > >  int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
> > >  {
> > >  	if (current->mm)
> > >  		fpsimd_preserve_current_state();
> > >  	*dst = *src;
> 
> Ack, some more explanation would be a good idea here.
> 
> Maybe the following comments are sufficient?
> 
> 	/* We rely on the above assingment to initialise dst's thread_flags: */

Thanks, I'll add this comment.

> 
> > >+	BUILD_BUG_ON(!IS_ENABLED(CONFIG_THREAD_INFO_IN_TASK));
> > 
> 
> and
> 
> 	/*
> 	 * Detach src's sve_state (if any) from dst so that it does not
> 	 * get erroneously used or freed prematurely.  dst's sve_state
> 	 * will be allocated on demand later on if dst uses SVE.
> 	 * For consistency, also clear TIF_SVE here: this could be done
> 	 * later in copy_process(), but to avoid tripping up future
> 	 * maintainers it is best not to leave TIF_SVE and sve_state in
> 	 * an inconsistent state, even temporarily.
> 	 */

I'll add this comments.

> 
> > >+	dst->thread.sve_state = NULL;
> > >+	clear_tsk_thread_flag(dst, TIF_SVE);
> 
> (TIF_SVE should not usually be set in the first place of course, since
> we are in a fork() or clone() syscall in src.  This may not be true if
> a task is created using kernel_thread() while running in the context of
> some user task that entered the kernel due to a trap or syscall --
> though probably nobody should be doing that.)

Thanks!
Masa
Dave Martin Sept. 30, 2019, 3:34 p.m. UTC | #6
On Mon, Sep 30, 2019 at 10:29:53AM -0400, Masayoshi Mizuma wrote:
> Hi Julien and Dave,
> 
> On Mon, Sep 30, 2019 at 02:02:46PM +0100, Dave Martin wrote:
> > On Mon, Sep 30, 2019 at 01:23:18PM +0100, Julien Grall wrote:
> > > Hi,
> > > 
> > > On 27/09/2019 16:39, Masayoshi Mizuma wrote:
> > > >From: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> > > >
> > > >The system which has SVE feature crashed because of
> > > >the memory pointed by task->thread.sve_state was destroyed
> > > >by someone.
> > > >
> > > >That is because sve_state is freed while the forking the
> > > >child process. The child process has the pointer of sve_state
> > > >which is same as the parent's because the child's task_struct
> > > >is copied from the parent's one. If the copy_process()
> > > >fails as an error on somewhere, for example, copy_creds(),
> > > >then the sve_state is freed even if the parent is alive.
> > > >The flow is as follows.
> > > >
> > > >copy_process
> > > >         p = dup_task_struct
> > > >             => arch_dup_task_struct
> > > >                 *dst = *src;  // copy the entire region.
> > > >:
> > > >         retval = copy_creds
> > > >         if (retval < 0)
> > > >                 goto bad_fork_free;
> > > >:
> > > >bad_fork_free:
> > > >...
> > > >         delayed_free_task(p);
> > > >           => free_task
> > > >              => arch_release_task_struct
> > > >                 => fpsimd_release_task
> > > >                    => __sve_free
> > > >                       => kfree(task->thread.sve_state);
> > > >                          // free the parent's sve_state
> > > >
> > > >Move child's sve_state = NULL and clearing TIF_SVE flag
> > > >to arch_dup_task_struct() so that the child doesn't free the
> > > >parent's one.
> > > >
> > > >Cc: stable@vger.kernel.org
> > > >Fixes: bc0ee4760364 ("arm64/sve: Core task context handling")
> > > 
> > > Looking at the log, it looks like THREAD_INFO_IN_TASK was selected before
> > > the bc0ee4760364. So it should be fine to backport for all the Linux tree
> > > contain this commit.
> 
> I think this patch is needed for the kernel has SVE support.
> I'll add the Cc tag as Dave said:
> 
> Cc: stable@vger.kernel.org # 4.15+

Based on running git log --grep='stable@vger.*#.*-', the most common
syntax for this situation seems to be

Cc: <stable@vger.kernel.org> # 4.15.x-

> So, I suppose this patch will be backported to stables 5.3.X,
> 5.2.X and longterm 4.19.X.
> Does this make sense?

Yes, I think so.  It's up to the stable maintainers to decide which
trees the patch actually gets applied too.

> > > >Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> > > >Reported-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
> > > >Suggested-by: Dave Martin <Dave.Martin@arm.com>
> > > 
> > > I have tested the patch and can confirm that double-free disappeared after
> > > the patch is applied:
> > > 
> > > Tested-by: Julien Grall <julien.grall@arm.com>
> 
> Thank you so much!
> 
> > 
> > Good to have that confirmed -- thanks for verifying.
> > 
> > [...]
> > 
> > > >---
> > > >  arch/arm64/kernel/process.c | 21 ++++-----------------
> > > >  1 file changed, 4 insertions(+), 17 deletions(-)
> > > >
> > > >diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
> > > >index f674f28df..6937f5935 100644
> > > >--- a/arch/arm64/kernel/process.c
> > > >+++ b/arch/arm64/kernel/process.c
> > > >@@ -323,22 +323,16 @@ void arch_release_task_struct(struct task_struct *tsk)
> > > >  	fpsimd_release_task(tsk);
> > > >  }
> > > >-/*
> > > >- * src and dst may temporarily have aliased sve_state after task_struct
> > > >- * is copied.  We cannot fix this properly here, because src may have
> > > >- * live SVE state and dst's thread_info may not exist yet, so tweaking
> > > >- * either src's or dst's TIF_SVE is not safe.
> > > >- *
> > > >- * The unaliasing is done in copy_thread() instead.  This works because
> > > >- * dst is not schedulable or traceable until both of these functions
> > > >- * have been called.
> > > >- */
> > > 
> > > It would be good to explain in the commit message why tweaking "dst" in
> > > arch_dup_task_struct() is fine.
> > > 
> > > From my understanding, Arm64 used to have thread_info on the stack. So it
> > > would not be possible to clear TIF_SVE until the stack is initialized.
> > > 
> > > Now that the thread_info is part of the task, it should be valid to modify
> > > the flag from arch_dup_task_struct().
> > > 
> > > Note that technically, TIF_SVE does not need to be cleared from
> > > arch_dup_task_struct(). It could also be done from copy_thread(). But it is
> > > easier to keep the both changes together.
> 
> Thanks, let me add some comments to the commit log.
> 
> > > 
> > > >  int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
> > > >  {
> > > >  	if (current->mm)
> > > >  		fpsimd_preserve_current_state();
> > > >  	*dst = *src;
> > 
> > Ack, some more explanation would be a good idea here.
> > 
> > Maybe the following comments are sufficient?
> > 
> > 	/* We rely on the above assingment to initialise dst's thread_flags: */
> 
> Thanks, I'll add this comment.
> 
> > 
> > > >+	BUILD_BUG_ON(!IS_ENABLED(CONFIG_THREAD_INFO_IN_TASK));
> > > 
> > 
> > and
> > 
> > 	/*
> > 	 * Detach src's sve_state (if any) from dst so that it does not
> > 	 * get erroneously used or freed prematurely.  dst's sve_state
> > 	 * will be allocated on demand later on if dst uses SVE.
> > 	 * For consistency, also clear TIF_SVE here: this could be done
> > 	 * later in copy_process(), but to avoid tripping up future
> > 	 * maintainers it is best not to leave TIF_SVE and sve_state in
> > 	 * an inconsistent state, even temporarily.
> > 	 */
> 
> I'll add this comments.

OK, if these make sense to you, feel free to add them.

[...]

Cheers
---Dave

Patch
diff mbox series

diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
index f674f28df..6937f5935 100644
--- a/arch/arm64/kernel/process.c
+++ b/arch/arm64/kernel/process.c
@@ -323,22 +323,16 @@  void arch_release_task_struct(struct task_struct *tsk)
 	fpsimd_release_task(tsk);
 }
 
-/*
- * src and dst may temporarily have aliased sve_state after task_struct
- * is copied.  We cannot fix this properly here, because src may have
- * live SVE state and dst's thread_info may not exist yet, so tweaking
- * either src's or dst's TIF_SVE is not safe.
- *
- * The unaliasing is done in copy_thread() instead.  This works because
- * dst is not schedulable or traceable until both of these functions
- * have been called.
- */
 int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
 {
 	if (current->mm)
 		fpsimd_preserve_current_state();
 	*dst = *src;
 
+	BUILD_BUG_ON(!IS_ENABLED(CONFIG_THREAD_INFO_IN_TASK));
+	dst->thread.sve_state = NULL;
+	clear_tsk_thread_flag(dst, TIF_SVE);
+
 	return 0;
 }
 
@@ -351,13 +345,6 @@  int copy_thread(unsigned long clone_flags, unsigned long stack_start,
 
 	memset(&p->thread.cpu_context, 0, sizeof(struct cpu_context));
 
-	/*
-	 * Unalias p->thread.sve_state (if any) from the parent task
-	 * and disable discard SVE state for p:
-	 */
-	clear_tsk_thread_flag(p, TIF_SVE);
-	p->thread.sve_state = NULL;
-
 	/*
 	 * In case p was allocated the same task_struct pointer as some
 	 * other recently-exited task, make sure p is disassociated from