svn commit: r320608 - releng/11.1/sys/cam

Kenneth D. Merry ken at FreeBSD.org
Mon Jul 3 18:20:46 UTC 2017


Author: ken
Date: Mon Jul  3 18:20:45 2017
New Revision: 320608
URL: https://svnweb.freebsd.org/changeset/base/320608

Log:
  Merge r320602 from stable/11 into releng/11.1:
    ------------------------------------------------------------------------
    r320602 | ken | 2017-07-03 09:34:21 -0600 (Mon, 03 Jul 2017) | 45 lines
  
    MFC r320421:
  
      ------------------------------------------------------------------------
      r320421 | ken | 2017-06-27 13:26:02 -0600 (Tue, 27 Jun 2017) | 37 lines
  
      Fix a panic in camperiphfree().
  
      If a peripheral driver (e.g. da, sa, cd) is added or removed from the
      peripheral driver list while an unrelated peripheral driver instance (e.g.
      da0, sa5, cd2) is going away and is inside camperiphfree(), we could
      dereference an invalid pointer.
  
      When peripheral drivers are added or removed (see periphdriver_register()
      and periphdriver_unregister()), the peripheral driver array is resized
      and existing entries are moved.
  
      Although we hold the topology lock while we traverse the peripheral driver
      list, we retain a pointer to the location of the peripheral driver pointer
      and then drop the topology lock.  So we are still vulnerable to the list
      getting moved around while the lock is dropped.
  
      To solve the problem, cache a copy of the peripheral driver pointer.  If
      its storage location in the list changes while we have the lock dropped, it
      won't have any effect.
  
      This doesn't solve the issue that peripheral drivers ("da", "cd", as opposed
      to individual instances like "da0", "cd0") are not generally part of a
      reference counting scheme to guard against deregistering them while there
      are instances active.  The caller (generally the person unloading a module)
      has to be aware of active drivers and not unload something that is in use.
  
      sys/cam/cam_periph.c:
      	In camperiphfree(), cache a pointer to the peripheral driver
      	instance to avoid holding a pointer to an invalid memory location
      	in the event that the peripheral driver list changes while we have
      	the topology lock dropped.
  
      PR:		kern/219701
      Submitted by:	avg
      Sponsored by:	Spectra Logic
  
      ------------------------------------------------------------------------
    ------------------------------------------------------------------------
  
  Approved by:	re (gjb)

Modified:
  releng/11.1/sys/cam/cam_periph.c
Directory Properties:
  releng/11.1/   (props changed)

Modified: releng/11.1/sys/cam/cam_periph.c
==============================================================================
--- releng/11.1/sys/cam/cam_periph.c	Mon Jul  3 18:07:09 2017	(r320607)
+++ releng/11.1/sys/cam/cam_periph.c	Mon Jul  3 18:20:45 2017	(r320608)
@@ -654,6 +654,7 @@ static void
 camperiphfree(struct cam_periph *periph)
 {
 	struct periph_driver **p_drv;
+	struct periph_driver *drv;
 
 	cam_periph_assert(periph, MA_OWNED);
 	KASSERT(periph->periph_allocating == 0, ("%s%d: freed while allocating",
@@ -666,6 +667,15 @@ camperiphfree(struct cam_periph *periph)
 		printf("camperiphfree: attempt to free non-existant periph\n");
 		return;
 	}
+	/*
+	 * Cache a pointer to the periph_driver structure.  If a
+	 * periph_driver is added or removed from the array (see
+	 * periphdriver_register()) while we drop the toplogy lock
+	 * below, p_drv may change.  This doesn't protect against this
+	 * particular periph_driver going away.  That will require full
+	 * reference counting in the periph_driver infrastructure.
+	 */
+	drv = *p_drv;
 
 	/*
 	 * We need to set this flag before dropping the topology lock, to
@@ -701,8 +711,8 @@ camperiphfree(struct cam_periph *periph)
 	 */
 	xpt_lock_buses();
 
-	TAILQ_REMOVE(&(*p_drv)->units, periph, unit_links);
-	(*p_drv)->generation++;
+	TAILQ_REMOVE(&drv->units, periph, unit_links);
+	drv->generation++;
 
 	xpt_remove_periph(periph);
 


More information about the svn-src-all mailing list