Add a facility allowing processes to inform the VM subsystem they are

critical and should not be killed when pageout is looking for more
memory pages in all the wrong places.

Reviewed by:	arch@
Sponsored by:	St. Bernard Software

5 files changed, 28 insertions, 1 deletions

diff --git a/lib/libc/sys/madvise.2 b/lib/libc/sys/madvise.2
index 27a032f..d783ea0 100644
--- a/lib/libc/sys/madvise.2
+++ b/lib/libc/sys/madvise.2
@@ -117,6 +117,12 @@ system calls.
 Region is not included in a core file.
 .It Dv MADV_CORE
 Include region in a core file.
+.It Dv MADV_PROTECT
+Informs the VM system this process should not be killed when the
+swap space is exhausted.
+The process must have superuser privileges.
+This should be used judiciously in processes that must remain running
+for the system to properly function.
 .El
 .Sh RETURN VALUES
 .Rv -std madvise
@@ -131,6 +137,9 @@ The virtual address range specified by the
 and
 .Fa len
 arguments is not valid.
+.It Bq Er EPERM
+.Dv MADV_PROTECT
+was specified and the process does not have superuser privileges.
 .El
 .Sh SEE ALSO
 .Xr mincore 2 ,
diff --git a/sys/sys/mman.h b/sys/sys/mman.h
index 71ac90c..f8b94cc 100644
--- a/sys/sys/mman.h
+++ b/sys/sys/mman.h
@@ -124,6 +124,7 @@
 #define	MADV_AUTOSYNC	7	/* revert to default flushing strategy */
 #define	MADV_NOCORE	8	/* do not include these pages in a core file */
 #define	MADV_CORE	9	/* revert to including pages in a core file */
+#define	MADV_PROTECT	10	/* protect process from pageout kill */
 
 /*
  * Return bits from mincore
diff --git a/sys/sys/proc.h b/sys/sys/proc.h
index 77464a0..8e25aa0 100644
--- a/sys/sys/proc.h
+++ b/sys/sys/proc.h
@@ -629,6 +629,7 @@ struct proc {
 #define	P_EXEC		0x04000	/* Process called exec. */
 #define	P_THREADED	0x08000	/* Process is using threads. */
 #define	P_CONTINUED	0x10000	/* Proc has continued from a stopped state. */
+#define	P_PROTECTED	0x20000	/* Do not kill on memory overcommit. */
 
 /* flags that control how threads may be suspended for some reason */
 #define	P_STOPPED_SIG		0x20000	/* Stopped due to SIGSTOP/SIGTSTP */
diff --git a/sys/vm/vm_mmap.c b/sys/vm/vm_mmap.c
index 93c8524..d20756f 100644
--- a/sys/vm/vm_mmap.c
+++ b/sys/vm/vm_mmap.c
@@ -752,8 +752,23 @@ madvise(td, uap)
 {
 	vm_offset_t start, end;
 	vm_map_t map;
+	struct proc *p;
+	int error;
 
 	/*
+	 * Check for our special case, advising the swap pager we are
+	 * "immortal."
+	 */
+	if (uap->behav == MADV_PROTECT) {
+		p = td->td_proc;
+		PROC_LOCK(p);
+		error = suser(td);
+		if (error == 0)
+			p->p_flag |= P_PROTECTED;
+		PROC_UNLOCK(p);
+		return (error);
+	}
+	/*
 	 * Check for illegal behavior
 	 */
 	if (uap->behav < 0 || uap->behav > MADV_CORE)
diff --git a/sys/vm/vm_pageout.c b/sys/vm/vm_pageout.c
index dab4ae2..8c07be2 100644
--- a/sys/vm/vm_pageout.c
+++ b/sys/vm/vm_pageout.c
@@ -1184,9 +1184,10 @@ rescan0:
 			if (PROC_TRYLOCK(p) == 0)
 				continue;
 			/*
-			 * if this is a system process, skip it
+			 * If this is a system or protected process, skip it.
 			 */
 			if ((p->p_flag & P_SYSTEM) || (p->p_pid == 1) ||
+			    (p->p_flag & P_PROTECTED) ||
 			    ((p->p_pid < 48) && (vm_swap_size != 0))) {
 				PROC_UNLOCK(p);
 				continue;