Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <1565278859475.1962@mentor.com> <1565358624103.3694@mentor.com> <20190809223831.fk4uyrzscr366syr@master>
In-Reply-To: <20190809223831.fk4uyrzscr366syr@master>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 9 Aug 2019 15:45:59 -0700
Message-ID: <CAHk-=wi_9sdMxurjZ1MbNnxt-pA=dqoyf8Fdn9aYc8xvjjnTBg@mail.gmail.com>
Subject: Re: Resend [PATCH] kernel/resource.c: invalidate parent when freed
 resource has childs
To:     Wei Yang <richard.weiyang@gmail.com>
Cc:     "Schmid, Carsten" <Carsten_Schmid@mentor.com>,
        "bp@suse.de" <bp@suse.de>,
        "dan.j.williams@intel.com" <dan.j.williams@intel.com>,
        "mingo@kernel.org" <mingo@kernel.org>,
        "dave.hansen@linux.intel.com" <dave.hansen@linux.intel.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "bhelgaas@google.com" <bhelgaas@google.com>,
        "osalvador@suse.de" <osalvador@suse.de>,
        "rdunlap@infradead.org" <rdunlap@infradead.org>,
        "richardw.yang@linux.intel.com" <richardw.yang@linux.intel.com>,
        "gregkh@linuxfoundation.org" <gregkh@linuxfoundation.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Aug 9, 2019 at 3:38 PM Wei Yang <richard.weiyang@gmail.com> wrote:
>
> In theory, child may have siblings. Would it be possible to have several
> devices under xhci-hcd?

I'm less interested in the xhci-hcd case - which I certainly *hope* is
fixed already? - than in "if this happens somewhere else".

So if we do want to remove the parent (which may be a good idea with a
warning), and want to make sure that the children are really removed
from the resource hierarchy, we should do somethiing like

  static bool detach_children(struct resource *res)
  {
        res = res->child;
        if (!res)
                return false;
        do {
                res->parent = NULL;
                res = res->sibling;
        } while (res);
        return true;
  }

and then we could write the __release_region() warning as

        /* You should not release a resource that has children */
        WARN_ON_ONCE(detach_children(res));

or something?

NOTE! The above is entirely untested, and written purely in my mail
reader. It might be seriously buggy, including not compiling, or doing
odd things. See it more as a "maybe something like this" code snippet
example than any kind of final form.

               Linus