Merge #214

214: faster zeroing of .bss r=JOE1994 a=JOE1994

This PR updates the code for zeroing the .bss section.
The updated version uses `core::ptr::write_bytes()` which invokes a single **memset**,
compared to the current code which initializes one byte at a time.

The updated version would initialize the .bss section faster than the current version. 

Thank you for reviewing this PR :) :bowing_man: 

Co-authored-by: JOE1994 <joseph942010@gmail.com>

core/src/entry_point/mod.rs

diff --git a/core/src/entry_point/mod.rs b/core/src/entry_point/mod.rs
index 2ab2351..850c213 100644
--- a/core/src/entry_point/mod.rs
+++ b/core/src/entry_point/mod.rs
@@ -104,10 +104,8 @@
     );
 
     // Zero .bss (specified by the linker script).
-    let bss_end = layout_header.bss_start + layout_header.bss_size; // 1 past the end of .bss
-    for i in layout_header.bss_start..bss_end {
-        core::ptr::write(i as *mut u8, 0);
-    }
+    let bss_start = layout_header.bss_start as *mut u8;
+    core::ptr::write_bytes(bss_start, 0u8, layout_header.bss_size);
 
     // TODO: Wait for rustc to have working ROPI-RWPI relocation support, then
     // implement dynamic relocations here. At the moment, rustc does not have