From 1afc9d3fcaa7c5992659bb8b69f639b746dda2bc Mon Sep 17 00:00:00 2001
From: Robert Morris <rtm@nephron.lcs.mit.edu>
Date: Thu, 5 Aug 2010 21:16:55 -0400
Subject: add some comments find out the hard way why user and kernel must have
 separate segment descriptors

---
 bootasm.S | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

(limited to 'bootasm.S')

diff --git a/bootasm.S b/bootasm.S
index f6af255..56175ce 100644
--- a/bootasm.S
+++ b/bootasm.S
@@ -51,8 +51,10 @@ seta20.2:
   orl     $CR0_PE, %eax
   movl    %eax, %cr0
   
-  # Jump to next instruction, but in 32-bit code segment.
-  # Switches processor into 32-bit mode.
+  # This ljmp is how you load the CS (Code Segment) register.
+  # SEG_ASM produces segment descriptors with the 32-bit mode
+  # flag set (the D flag), so addresses and word operands will
+  # default to 32 bits after this jump.
   ljmp    $(SEG_KCODE<<3), $start32
 
 .code32                       # Assemble for 32-bit mode
-- 
cgit v1.2.3