From 075c87d52a4d90c0f00f8ba7ba849677771a9be3 Mon Sep 17 00:00:00 2001
From: salmonik <salmonik@tuta.io>
Date: Fri, 31 Jul 2026 18:08:45 +0200
Subject: vt: defer scrolls and batch framebuffer pushes
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Streaming output scrolled one framebuffer push per line (each a
~3-8 MB write-only copy).  The vt now scrolls the logical grid
instantly in RAM and defers the display shift: N pending scrolls are
pushed together as one shift_up_n — one push regardless of batch size.

- vt: PENDING_SCROLLS counter; scroll_up shifts CELLS only; flush
  renders nothing while pending (the diff would otherwise re-rasterise
  the whole screen per line); sync_scroll() does one gop::shift_up_n +
  RENDERED mirror + normal diff flush (which re-blits just the newly
  exposed rows and the cursor).
- Sync triggers: batch of 16 rows, ~40 ms of stream pause (100 Hz
  TICKS), any CSI sequence (full-screen apps like the editor render
  immediately), clear_all/set_cursor_visible, interactive input
  (tty read_line/read_byte), and the shell before each prompt.
- gop: shift_up → shift_up_n(n) (one shadow memcpy + one push).

Verified in QEMU: mmap/hexdump streaming renders in coherent batches
with the prompt syncing at the end; clear wipes clean; the editor
scrolls and redraws immediately (45-line file, gutter + status bar
intact, cursor correct); save/exit/cat round-trip unchanged.  Release
and dev builds pass.
---
 CONTEXT.md | 24 ++++++++++++++++--------
 1 file changed, 16 insertions(+), 8 deletions(-)

(limited to 'CONTEXT.md')

diff --git a/CONTEXT.md b/CONTEXT.md
index 32a6eca..8fa4025 100644
--- a/CONTEXT.md
+++ b/CONTEXT.md
@@ -163,14 +163,22 @@ Single Rust crate (`x86_64-unknown-uefi` target).  Module breakdown:
 ## Known issues / perf notes
 
 - **Scroll cost** — scrolling still shifts the whole framebuffer (a
-  ~3-8 MB pixel copy per row), but a RAM backbuffer (`gop::SHADOW`)
-  makes it fast on real hardware: every raster write goes to the shadow
-  too, and scrolling is a cached shadow memcpy + one write-only push to
-  the display.  The display is never *read* — that was the killer, since
-  real-hardware GOP framebuffers are uncacheable/write-combining and
-  uncached reads cost a bus round-trip per cache line.  Still on the
-  someday list: a pan-based ring so scrolling never touches the whole
-  framebuffer at all.
+  ~3-8 MB pixel copy per row), but two mechanisms make it fast on real
+  hardware:
+  1. a RAM backbuffer (`gop::SHADOW`): every raster write goes to the
+     shadow too; scrolling is a cached shadow memcpy + one write-only
+     push to the display — the display is never *read* (uncached reads
+     were the killer, since real-hardware GOP framebuffers are
+     uncacheable/write-combining).
+  2. deferred scroll batching (`vt::PENDING_SCROLLS`): the logical grid
+     scrolls instantly in RAM while the framebuffer shift+push is
+     deferred; N pending scrolls cost one push.  Bursts of streaming
+     output (hexdump/cat) push every ~16 rows (or after ~40 ms of
+     pause), and the shell syncs before each prompt.  Any CSI sequence
+     (full-screen apps like the editor) or interactive input pushes
+     immediately.
+  Still on the someday list: a pan-based ring so scrolling never touches
+  the whole framebuffer at all.
 - **Editor redraw** re-blits changed cells per keystroke (fine at 1280×800);
   the "optimize drawing someday" item.
 - **ramfs only**: edited files live in memory, lost on reboot until disk
-- 
cgit v1.2.3