From 4038ec5662a5b1ebe652e9170069e3f2f79c7e71 Mon Sep 17 00:00:00 2001
From: Miquel Sabaté Solà <mikisabate@gmail.com>
Date: Sat, 24 Aug 2024 00:05:47 +0200
Subject: Bootstrap the project
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

For now the base layout has been defined and we have a binary that is
properly reached through openSBI.

Signed-off-by: Miquel Sabaté Solà <mikisabate@gmail.com>
---
 kernel/fbos.ld.S | 52 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 kernel/head.S    |  6 ++++++
 kernel/main.c    | 12 ++++++++++++
 3 files changed, 70 insertions(+)
 create mode 100644 kernel/fbos.ld.S
 create mode 100644 kernel/head.S
 create mode 100644 kernel/main.c

(limited to 'kernel')

diff --git a/kernel/fbos.ld.S b/kernel/fbos.ld.S
new file mode 100644
index 0000000..f288161
--- /dev/null
+++ b/kernel/fbos.ld.S
@@ -0,0 +1,52 @@
+#include <fbos/mm.h>
+
+SECTIONS {
+	// Ensure that the image starts at the very exact address SBI expects it to.
+	. = LINK_ADDR;
+
+	// You would usually want to separate the head section into its own thing
+	// instead of clumping it into the main `.text` one. Well, I'm no expert on
+	// linker configuration, so patches are welcome :)
+	.text : {
+		// The very first thing has to be the `_start` function, which is where
+		// SBI will jump into. Afterwards comes the rest of `.text.head`.
+		_start = .;
+		*(.text.head)
+
+		// Aligning it to a full page is maybe a bit too much considering how
+		// small `.text.head` really is. I just saw this same thing on the Linux
+		// Kernel and it felt clean.
+		. = ALIGN(PAGE_SIZE);
+
+		// From now on the rest of the `.text` could just be a continuation, but
+		// I further split it into `.text.kernel` so the jump from head isn't
+		// that large. That is, we put first the very core of the kernel, and
+		// the rest can go wherever.
+		__kernel_text_start = .;
+		*(.text.kernel)
+		__kernel_text_end = .;
+
+		// And the rest.
+    	*(.text)
+	}
+
+	// In total we would reserve two full blown pages for the kernel code, which
+	// is definitely too much, but whatever, memory is cheap.
+	. = ALIGN(PAGE_SIZE);
+
+	.data : {
+    	*(.data)
+	}
+
+	. = ALIGN(8);
+
+	.rodata : {
+    	*(.rodata)
+	}
+
+	. = ALIGN(8);
+
+	.bss : {
+    	*(.bss)
+	}
+}
diff --git a/kernel/head.S b/kernel/head.S
new file mode 100644
index 0000000..ba155cb
--- /dev/null
+++ b/kernel/head.S
@@ -0,0 +1,6 @@
+.global _start
+.section .text.head
+
+_start:
+	// TODO
+	call start_kernel
diff --git a/kernel/main.c b/kernel/main.c
new file mode 100644
index 0000000..325f858
--- /dev/null
+++ b/kernel/main.c
@@ -0,0 +1,12 @@
+#include <fbos/init.h>
+
+/*
+ * This is the main entry point of the kernel after head.S is done. This
+ * function can (and will) assume that everything has been reset and that we can
+ * start the whole thing.
+ */
+__noreturn __kernel void start_kernel(void)
+{
+	for (;;)
+		;
+}
-- 
cgit v1.2.3