mm: multi-gen LRU: move lru_gen_add_mm() out of IRQ-off region
authorSebastian Andrzej Siewior <bigeasy@linutronix.de>
Wed, 26 Oct 2022 13:48:30 +0000 (15:48 +0200)
committerAndrew Morton <akpm@linux-foundation.org>
Fri, 28 Oct 2022 20:37:23 +0000 (13:37 -0700)
lru_gen_add_mm() has been added within an IRQ-off region in the commit
mentioned below.  The other invocations of lru_gen_add_mm() are not within
an IRQ-off region.

The invocation within IRQ-off region is problematic on PREEMPT_RT because
the function is using a spin_lock_t which must not be used within
IRQ-disabled regions.

The other invocations of lru_gen_add_mm() occur while
task_struct::alloc_lock is acquired.  Move lru_gen_add_mm() after
interrupts are enabled and before task_unlock().

Link: https://lkml.kernel.org/r/20221026134830.711887-1-bigeasy@linutronix.de
Fixes: bd74fdaea1460 ("mm: multi-gen LRU: support page table walks")
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Yu Zhao <yuzhao@google.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Eric W . Biederman" <ebiederm@xmission.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
fs/exec.c

index 349a5da..7ab1f27 100644 (file)
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -1012,7 +1012,6 @@ static int exec_mmap(struct mm_struct *mm)
        active_mm = tsk->active_mm;
        tsk->active_mm = mm;
        tsk->mm = mm;
-       lru_gen_add_mm(mm);
        /*
         * This prevents preemption while active_mm is being loaded and
         * it and mm are being updated, which could cause problems for
@@ -1025,6 +1024,7 @@ static int exec_mmap(struct mm_struct *mm)
        activate_mm(active_mm, mm);
        if (IS_ENABLED(CONFIG_ARCH_WANT_IRQS_OFF_ACTIVATE_MM))
                local_irq_enable();
+       lru_gen_add_mm(mm);
        task_unlock(tsk);
        lru_gen_use_mm(mm);
        if (old_mm) {