Re: [PATCH] Linux-2.5 fix/improve get_pid()

[Linus Torvalds]

On Thu, 8 Aug 2002, Hubertus Franke wrote:
> 
> That is true. All was done under the 16-bit assumption
> My hunch is that the current algorithm might actually work quite well
> for a sparsely populated pid-space (32-bits).

I agree.

So let's just try Andries approach, suggested patch as follows..

(yeah, silly mask and MAX_PID, but since even the kernel uses signed
integers for some of it, this way it never gets close to being an issue).

		Linus

----
--- 1.2/include/linux/threads.h	Tue Feb  5 07:23:04 2002
+++ edited/include/linux/threads.h	Thu Aug  8 14:58:28 2002
@@ -19,6 +19,7 @@
 /*
  * This controls the maximum pid allocated to a process
  */
-#define PID_MAX 0x8000
+#define PID_MASK 0x3fffffff
+#define PID_MAX (PID_MASK+1)
 
 #endif
===== kernel/fork.c 1.57 vs edited =====
--- 1.57/kernel/fork.c	Tue Jul 30 15:49:20 2002
+++ edited/kernel/fork.c	Thu Aug  8 15:00:16 2002
@@ -142,7 +142,7 @@
 		return 0;
 
 	spin_lock(&lastpid_lock);
-	if((++last_pid) & 0xffff8000) {
+	if((++last_pid) & ~PID_MASK) {
 		last_pid = 300;		/* Skip daemons etc. */
 		goto inside;
 	}
@@ -157,7 +157,7 @@
 			   p->tgid == last_pid	||
 			   p->session == last_pid) {
 				if(++last_pid >= next_safe) {
-					if(last_pid & 0xffff8000)
+					if(last_pid & ~PID_MASK)
 						last_pid = 300;
 					next_safe = PID_MAX;
 				}