svn commit: r368006 - head/sys/kern

[Kyle Evans]

Author: kevans
Date: Wed Nov 25 01:08:57 2020
New Revision: 368006
URL: https://svnweb.freebsd.org/changeset/base/368006

Log:
  kern: never restart syscalls calling closefp(), e.g. close(2)
  
  All paths leading into closefp() will either replace or remove the fd from
  the filedesc table, and closefp() will call fo_close methods that can and do
  currently sleep without regard for the possibility of an ERESTART. This can
  be dangerous in multithreaded applications as another thread could have
  opened another file in its place that is subsequently operated on upon
  restart.
  
  The following are seemingly the only ones that will pass back ERESTART
  in-tree:
  - sockets (SO_LINGER)
  - fusefs
  - nfsclient
  
  Reviewed by:	jilles, kib
  MFC after:	1 week
  Differential Revision:	https://reviews.freebsd.org/D27310

Modified:
  head/sys/kern/kern_descrip.c

Modified: head/sys/kern/kern_descrip.c
==============================================================================
--- head/sys/kern/kern_descrip.c	Wed Nov 25 00:10:54 2020	(r368005)
+++ head/sys/kern/kern_descrip.c	Wed Nov 25 01:08:57 2020	(r368006)
@@ -1279,6 +1279,15 @@ closefp(struct filedesc *fdp, int fd, struct file *fp,
 	FILEDESC_XUNLOCK(fdp);
 
 	error = closef(fp, td);
+
+	/*
+	 * All paths leading up to closefp() will have already removed or
+	 * replaced the fd in the filedesc table, so a restart would not
+	 * operate on the same file.
+	 */
+	if (error == ERESTART)
+		error = EINTR;
+
 	if (holdleaders) {
 		FILEDESC_XLOCK(fdp);
 		fdp->fd_holdleaderscount--;