pipe: PIPE_MAX 2->4 (5-stage pipelines) + fix cross-yield byte corruption

Bump PIPE_MAX to 4 so a|b|c|d|e (4 pipes) works. This stays within the
single-byte buffer-offset math (idx*64+pos <= 3*64+63 = 255) and the fd
space ($F0..$F7, clear of $FF console).

The bump exposed a data-corruption bug that also affected the 2-pipe case
(just invisibly - a wc-only test can't see mangled bytes): pipe_write kept
the byte-to-write in the SHARED wPipeByte global across its SchedYield
(buffer full), so a concurrent pipe op clobbered it and the writer then
stored the wrong byte. Now the byte is held in D across the yield, and
pipe_bufptr no longer clobbers D; pipe_read/pipe_write also push their idx
across SchedYield rather than assume the yield preserves registers.

Verified: count N | cat now streams EXACT content (no 'linn'/'llne'
corruption); count 60 | cat | cat | wc = 60 180 1671; 5-stage
count 4 | cat | cat | cat | cat prints line 1..4; SIGPIPE (count 200|true)
and count 100|wc still fine, no hangs.

1 files changed, 22 insertions, 9 deletions

diff --git a/src/pipe.asm b/src/pipe.asm
index f9473a4..1a77c5e 100644
--- a/src/pipe.asm
+++ b/src/pipe.asm
@@ -108,19 +108,23 @@ pipe_close::
     dec [hl]
     ret
 
-; pipe_bufptr - C = idx, A = pos -> HL = &wPipeBuf[idx*PIPE_BUFSZ + pos]. keeps BC.
-; (idx*64 + pos <= 255, so the whole offset fits in one byte.)
+; pipe_bufptr - C = idx, A = pos -> HL = &wPipeBuf[idx*PIPE_BUFSZ + pos].
+; (idx*64 + pos <= 255, so the whole offset fits in one byte.) Preserves B,C,D.
 pipe_bufptr:
     ld e, a                             ; E = pos
     ld a, c
     swap a
     add a
     add a                               ; A = idx*64
-    add e                               ; + pos
+    add e                               ; + pos (0..255)
     ld e, a
-    ld d, 0
     ld hl, wPipeBuf
-    add hl, de
+    ld a, l
+    add e
+    ld l, a
+    ld a, h
+    adc 0
+    ld h, a
     ret
 
 ; pipe_read - A = read fd -> A = byte, CF set on EOF. Blocks while empty w/ writer.
@@ -140,7 +144,9 @@ pipe_read::
     ld a, [hl]
     or a
     jr z, .eof                          ; none -> EOF
-    call SchedYield                     ; block: let the writer run
+    push bc                             ; block: let the writer run (keep idx)
+    call SchedYield
+    pop bc
     jr .retry
 .data
     ld hl, wPipeHead
@@ -175,6 +181,9 @@ pipe_write::
     sub PIPE_FD_BASE
     srl a
     ld c, a                             ; C = idx
+    ld a, [wPipeByte]
+    ld d, a                             ; D = byte (must survive yields; wPipeByte
+                                        ;   is shared and gets clobbered by others)
 .retry
     ld hl, wPipeCnt
     ld b, 0
@@ -187,15 +196,19 @@ pipe_write::
     ld a, [hl]
     or a
     jr z, .epipe                        ; none -> SIGPIPE
-    call SchedYield                     ; block: let the reader drain
+    push bc                             ; block: let the reader drain (keep idx+byte)
+    push de
+    call SchedYield
+    pop de
+    pop bc
     jr .retry
 .space
     ld hl, wPipeTail
     add hl, bc
     ld a, [hl]                          ; tail
     push af
-    call pipe_bufptr                    ; A=tail, C=idx -> HL = &buf byte
-    ld a, [wPipeByte]
+    call pipe_bufptr                    ; A=tail, C=idx -> HL (preserves D)
+    ld a, d                             ; the byte
     ld [hl], a
     pop af                              ; tail
     inc a