From 13e108abd242c415e1171ae5f896b5f9bf79a3af Mon Sep 17 00:00:00 2001 From: Alan Cox Date: Sun, 16 Sep 2018 00:44:14 +0100 Subject: [PATCH] simple: explain why we don't worry about disk read versus memory clear On the Z80 and many other 8bit micros a sequential series of disk reads (especially from CF) are as fast as a memory clear or very close to it. The TRS80 is perhaps the obvious exception. --- Kernel/simple.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/Kernel/simple.c b/Kernel/simple.c index ea66f6d4..7e225bf7 100644 --- a/Kernel/simple.c +++ b/Kernel/simple.c @@ -19,6 +19,11 @@ * on switches. It makes no sense to support it here because to do that * well we want to support a simple first fit allocator for swap ranges * so our swap isn't huge and empty. + * + * The fact we don't do this for 8bit systems may seem weird, but on most + * of the systems supported a contiguous series of disk reads of 512 + * byte blocks isn't *that* much slower than a memory zero. (inir + sector + * setup versus ldir on Z80 for example) */ #include -- 2.34.1