Update 4700 with actual list, note about file.

diff --git a/2019-kvm-forum/4700-fast-zero-implementation.html b/2019-kvm-forum/4700-fast-zero-implementation.html
index fd8be8e..fbee89d 100644 (file)
--- a/2019-kvm-forum/4700-fast-zero-implementation.html
+++ b/2019-kvm-forum/4700-fast-zero-implementation.html
@@ -17,7 +17,21 @@
 plugins()
 filters("1")
 
-document.getElementById("filter1-error").classList.add("highlighted")
+document.getElementById("plugin-data").classList.add("highlighted")
+document.getElementById("plugin-full").classList.add("highlighted")
+document.getElementById("plugin-memory").classList.add("highlighted")
+document.getElementById("plugin-nbd").classList.add("highlighted")
+document.getElementById("plugin-null").classList.add("highlighted")
+document.getElementById("plugin-ocaml").classList.add("highlighted")
+document.getElementById("plugin-sh").classList.add("highlighted")
+document.getElementById("plugin-rust").classList.add("highlighted")
+
+document.getElementById("filter1-blocksize").classList.add("highlighted")
+document.getElementById("filter1-cache").classList.add("highlighted")
+document.getElementById("filter1-cow").classList.add("highlighted")
+document.getElementById("filter1-delay").classList.add("highlighted")
+document.getElementById("filter1-log").classList.add("highlighted")
+document.getElementById("filter1-nozero").classList.add("highlighted")
+document.getElementById("filter1-truncate").classList.add("highlighted")
 </script>
-XXX populate full list of changed filters/plugins
 </p>

diff --git a/2019-kvm-forum/notes-04-fast-zero b/2019-kvm-forum/notes-04-fast-zero
index bb28d7f..59f2029 100644 (file)
--- a/2019-kvm-forum/notes-04-fast-zero
+++ b/2019-kvm-forum/notes-04-fast-zero
@@ -146,6 +146,12 @@ released in 1.14).  Among other things, the nbdkit changes to the
 nozero filter added the parameters I used in my demo for controlling
 whether to advertise and/or honor the semantics of the new flag.
 
+[if time:] Note that the file plugin was not touched in the initial
+patches. This is because accurate support is harder than it looks:
+both fallocate(FALLOC_FL_ZERO_RANGE) and ioctl(BLKZEROOUT) can trigger
+fallbacks to slow writes, so we would need kernel support for new
+interfaces that guarantee fast failure.
+
 * segue: XXX
 slide 4800-? Or just sentence leading into Rich's demos?