Subject: Re: [PATCH] PCI: hv: Fix a bug on removing child devices on the bus

On Thu, Aug 26, 2021 at 08:09:19PM +0000, Long Li wrote:

quoted

quoted

Subject: Re: [PATCH] PCI: hv: Fix a bug on removing child devices on
the bus

On Thu, Aug 26, 2021 at 04:50:28PM +0000, Michael Kelley wrote:

quoted

From: Long Li <longli@microsoft.com> Sent: Wednesday, August 25,
2021
1:25 PM

quoted

quoted

I thought list_for_each_entry_safe() is for use when list
manipulation is *not* protected by a lock and you want to
safely walk the list even if an entry gets removed.  If the
list is protected by a lock or not subject to contention (as
is the case here), then
list_for_each_entry() is the simpler implementation.  The
original implementation didn't need to use the _safe version
because of the spin

lock.

quoted

quoted

quoted

Or do I have it backwards?

Michael

I think we need list_for_each_entry_safe() because we delete the
list

elements while going through them:

quoted

quoted

Here is the comment on list_for_each_entry_safe():
/**
 * Loop through the list, keeping a backup pointer to the element.
This
 * macro allows for the deletion of a list element while looping
through the
 * list.
 *
 * See list_for_each_entry for more details.
 */

Got it.  Thanks (and to Rob Herring).   I read that comment but
with the wrong assumptions and didn't understand it correctly.

Interestingly, pci-hyperv.c has another case of looping through
this list and removing items where the _safe version is not used.
See pci_devices_present_work() where the missing children are
moved to a list on the stack.

That can be converted too, I think.

The original code is not wrong per-se. It is just not as concise as
using list_for_each_entry_safe.

Wei.

I assume we are talking about the following code in

pci_devices_present_work():

quoted

                list_for_each_entry(hpdev, &hbus->children, list_entry) {
                        if (hpdev->reported_missing) {
                                found = true;
                                put_pcichild(hpdev);
                                list_move_tail(&hpdev->list_entry, &removed);
                                break;
                        }
                }

This code is correct as there is a "break" after a list entry is
removed from the list. So there is no need to use the _safe version.
This code can be converted to use the _safe version.

After this block there is another block like

  while (!list_empty(removed)) {
	...
  	list_del(...)

  }

I assumed Michael was referring to that block. :-)

Wei.

This block is also correct. We don't have a bug here but there is a better way to code it.

Long