[PATCH v3] arm64/sve: Fix wrong free for task->thread.sve_state

From: Masayoshi Mizuma
Date: Mon Sep 30 2019 - 16:56:16 EST


From: Masayoshi Mizuma <m.mizuma@xxxxxxxxxxxxxx>

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.
There is no need to wait until copy_process() to clear TIF_SVE for
dst, because 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.

As a note, arm64 used to have thread_info on the stack. So it
would not be possible to clear TIF_SVE until the stack is initialized.
>From commit c02433dd6de3 ("arm64: split thread_info from task stack"),
the thread_info is part of the task, so it should be valid to modify
the flag from arch_dup_task_struct().

Cc: stable@xxxxxxxxxxxxxxx # 4.15.x-
Fixes: bc0ee4760364 ("arm64/sve: Core task context handling")
Signed-off-by: Masayoshi Mizuma <m.mizuma@xxxxxxxxxxxxxx>
Reported-by: Hidetoshi Seto <seto.hidetoshi@xxxxxxxxxxxxxx>
Suggested-by: Dave Martin <Dave.Martin@xxxxxxx>
Tested-by: Julien Grall <julien.grall@xxxxxxx>
---
arch/arm64/kernel/process.c | 32 +++++++++++++++-----------------
1 file changed, 15 insertions(+), 17 deletions(-)

diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
index a47462def04b..ef7aa909bfda 100644
--- a/arch/arm64/kernel/process.c
+++ b/arch/arm64/kernel/process.c
@@ -332,22 +332,27 @@ 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;

+ /* We rely on the above assignment to initialize dst's thread_flags: */
+ BUILD_BUG_ON(!IS_ENABLED(CONFIG_THREAD_INFO_IN_TASK));
+
+ /*
+ * 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);
+
return 0;
}

@@ -360,13 +365,6 @@ 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