arm64/sme: Always exit sme_alloc() early with existing storage

commit dc7eb87 upstream.

When sme_alloc() is called with existing storage and we are not flushing we
will always allocate new storage, both leaking the existing storage and
corrupting the state. Fix this by separating the checks for flushing and
for existing storage as we do for SVE.

Callers that reallocate (eg, due to changing the vector length) should
call sme_free() themselves.

Fixes: 5d0a8d2 ("arm64/ptrace: Ensure that SME is set up for target when writing SSVE state")
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20240115-arm64-sme-flush-v1-1-7472bd3459b7@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Zhaoyang Li <lizy04@hust.edu.cn>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

arch/arm64/kernel/fpsimd.c

@@ -1259,8 +1259,10 @@ void fpsimd_release_task(struct task_struct *dead_task)
  */
 void sme_alloc(struct task_struct *task, bool flush)
 {
-	if (task->thread.za_state && flush) {
-		memset(task->thread.za_state, 0, za_state_size(task));
+	if (task->thread.za_state) {
+		if (flush)
+			memset(task->thread.za_state, 0,
+			       za_state_size(task));
 		return;
 	}