From: Alan Cox <alan@linux.intel.com>
Date: Thu, 21 Jan 2016 11:53:52 +0000 (+0000)
Subject: syscall_proc: document that dofork might do two things
X-Git-Url: https://git.ndcode.org/public/gitweb.cgi?a=commitdiff_plain;h=3cd4d2e958385f8709afaca6eaaeb045127380ca;p=FUZIX.git

syscall_proc: document that dofork might do two things

On banked platforms it'll probably return twice for now anyway. On some others
the child will return to user space directly on a different kernel stack.
---

diff --git a/Kernel/syscall_proc.c b/Kernel/syscall_proc.c
index 0acfa56f..213d3173 100644
--- a/Kernel/syscall_proc.c
+++ b/Kernel/syscall_proc.c
@@ -373,10 +373,18 @@ arg_t _fork(void)
 		return -1;
 
 	irq = di();
-	// we're going to run our child process next, so mark this process as being ready to run
+	/*
+	 * We're going to run our child process next, so mark this process as
+	 * being ready to run
+	 */
 	udata.u_ptab->p_status = P_READY;
-	// kick off the new process (the bifurcation happens inside here, we returns in both 
-	// the child and parent contexts)
+	/*
+	 * Kick off the new process (the bifurcation happens inside here, we
+	 * *MAY* returns in both the child and parent contexts, however in a
+	 * non error case the child may also directly return to userspace
+	 * with the return code of 0 and not return from here. Do not assume
+	 * you can execute any child code reliably beyond this call
+	 */
 	r = dofork(new_process);
 #ifdef DEBUG
 	kprintf("Dofork %x (n %x)returns %d\n", udata.u_ptab,