On 08/15/2018 08:49 PM, Yang Shi wrote:
> We need check if mm or vma has uprobes in the following patch to check
> if a vma could be unmapped with holding read mmap_sem. The checks and
> pre-conditions used by uprobe_munmap() look just suitable for this
> purpose.
>
> Extracting those checks into a helper function, has_uprobes().
>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> Cc: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
> Cc: Alexander Shishkin <alexander.shishkin@xxxxxxxxxxxxxxx>
> Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
> Cc: Namhyung Kim <namhyung@xxxxxxxxxx>
> Cc: Vlastimil Babka <vbabka@xxxxxxx>
> Signed-off-by: Yang Shi <yang.shi@xxxxxxxxxxxxxxxxx>
> ---
> include/linux/uprobes.h | 7 +++++++
> kernel/events/uprobes.c | 23 ++++++++++++++++-------
> 2 files changed, 23 insertions(+), 7 deletions(-)
>
> diff --git a/include/linux/uprobes.h b/include/linux/uprobes.h
> index 0a294e9..418764e 100644
> --- a/include/linux/uprobes.h
> +++ b/include/linux/uprobes.h
> @@ -149,6 +149,8 @@ struct uprobes_state {
> extern bool arch_uprobe_ignore(struct arch_uprobe *aup, struct pt_regs *regs);
> extern void arch_uprobe_copy_ixol(struct page *page, unsigned long vaddr,
> void *src, unsigned long len);
> +extern bool has_uprobes(struct vm_area_struct *vma, unsigned long start,
> + unsigned long end);
> #else /* !CONFIG_UPROBES */
> struct uprobes_state {
> };
> @@ -203,5 +205,10 @@ static inline void uprobe_copy_process(struct task_struct *t, unsigned long flag
> static inline void uprobe_clear_state(struct mm_struct *mm)
> {
> }
> +static inline bool has_uprobes(struct vm_area_struct *vma, unsigned long start,
> + unsgined long end)
> +{
> + return false;
> +}
> #endif /* !CONFIG_UPROBES */
> #endif /* _LINUX_UPROBES_H */
> diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
> index aed1ba5..568481c 100644
> --- a/kernel/events/uprobes.c
> +++ b/kernel/events/uprobes.c
> @@ -1114,22 +1114,31 @@ int uprobe_mmap(struct vm_area_struct *vma)
> return !!n;
> }
>
> -/*
> - * Called in context of a munmap of a vma.
> - */
> -void uprobe_munmap(struct vm_area_struct *vma, unsigned long start, unsigned long end)
> +bool
> +has_uprobes(struct vm_area_struct *vma, unsigned long start, unsigned long end)
The name is not really great...
> {
> if (no_uprobe_events() || !valid_vma(vma, false))
> - return;
> + return false;
>
> if (!atomic_read(&vma->vm_mm->mm_users)) /* called by mmput() ? */
> - return;
> + return false;
>
> if (!test_bit(MMF_HAS_UPROBES, &vma->vm_mm->flags) ||
> test_bit(MMF_RECALC_UPROBES, &vma->vm_mm->flags))
This means that vma might have uprobes, but since RECALC is already set,
we don't need to set it again. That's different from "has uprobes".
Perhaps something like vma_needs_recalc_uprobes() ?
But I also worry there might be a race where we initially return false
because of MMF_RECALC_UPROBES, then the flag is cleared while vma's
still have uprobes, then we downgrade mmap_sem and skip uprobe_munmap().
Should be checked if e.g. mmap_sem and vma visibility changes protects
this case from happening.
> - return;
> + return false;
>
> if (vma_has_uprobes(vma, start, end))
> + return true;
> +
> + return false;
Simpler:
return vma_has_uprobes(vma, start, end);
> +}
> +
> +/*
> + * Called in context of a munmap of a vma.
> + */
> +void uprobe_munmap(struct vm_area_struct *vma, unsigned long start, unsigned long end)
> +{
> + if (has_uprobes(vma, start, end))
> set_bit(MMF_RECALC_UPROBES, &vma->vm_mm->flags);
> }
>
>
