powerpc/eeh: Clean up EEH PEs after recovery finishes

When the last device in an eeh_pe is removed the eeh_pe structure itself
(and any empty parents) are freed since they are no longer needed. This
results in a crash when a hotplug driver is involved since the following
may occur:

1. Device is suprise removed.
2. Driver performs an MMIO, which fails and queues and eeh_event.
3. Hotplug driver receives a hotplug interrupt and removes any
   pci_devs that were under the slot.
4. pci_dev is torn down and the eeh_pe is freed.
5. The EEH event handler thread processes the eeh_event and crashes
   since the eeh_pe pointer in the eeh_event structure is no
   longer valid.

Crashing is generally considered poor form. Instead of doing that use
the fact PEs are marked as EEH_PE_INVALID to keep them around until the
end of the recovery cycle, at which point we can safely prune any empty
PEs.

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20190903101605.2890-2-oohall@gmail.com

1 files changed, 34 insertions, 2 deletions

diff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c
index a31cd32c4ce9..75266156943f 100644
--- a/arch/powerpc/kernel/eeh_driver.c
+++ b/arch/powerpc/kernel/eeh_driver.c
@@ -734,6 +734,33 @@ static int eeh_reset_device(struct eeh_pe *pe, struct pci_bus *bus,
  */
 #define MAX_WAIT_FOR_RECOVERY 300
 
+
+/* Walks the PE tree after processing an event to remove any stale PEs.
+ *
+ * NB: This needs to be recursive to ensure the leaf PEs get removed
+ * before their parents do. Although this is possible to do recursively
+ * we don't since this is easier to read and we need to garantee
+ * the leaf nodes will be handled first.
+ */
+static void eeh_pe_cleanup(struct eeh_pe *pe)
+{
+	struct eeh_pe *child_pe, *tmp;
+
+	list_for_each_entry_safe(child_pe, tmp, &pe->child_list, child)
+		eeh_pe_cleanup(child_pe);
+
+	if (pe->state & EEH_PE_KEEP)
+		return;
+
+	if (!(pe->state & EEH_PE_INVALID))
+		return;
+
+	if (list_empty(&pe->edevs) && list_empty(&pe->child_list)) {
+		list_del(&pe->child);
+		kfree(pe);
+	}
+}
+
 /**
  * eeh_handle_normal_event - Handle EEH events on a specific PE
  * @pe: EEH PE - which should not be used after we return, as it may
@@ -772,8 +799,6 @@ void eeh_handle_normal_event(struct eeh_pe *pe)
 		return;
 	}
 
-	eeh_pe_state_mark(pe, EEH_PE_RECOVERING);
-
 	eeh_pe_update_time_stamp(pe);
 	pe->freeze_count++;
 	if (pe->freeze_count > eeh_max_freezes) {
@@ -963,6 +988,12 @@ void eeh_handle_normal_event(struct eeh_pe *pe)
 			return;
 		}
 	}
+
+	/*
+	 * Clean up any PEs without devices. While marked as EEH_PE_RECOVERYING
+	 * we don't want to modify the PE tree structure so we do it here.
+	 */
+	eeh_pe_cleanup(pe);
 	eeh_pe_state_clear(pe, EEH_PE_RECOVERING, true);
 }
 
@@ -1035,6 +1066,7 @@ void eeh_handle_special_event(void)
 		 */
 		if (rc == EEH_NEXT_ERR_FROZEN_PE ||
 		    rc == EEH_NEXT_ERR_FENCED_PHB) {
+			eeh_pe_state_mark(pe, EEH_PE_RECOVERING);
 			eeh_handle_normal_event(pe);
 		} else {
 			pci_lock_rescan_remove();