net: downloads at tens of megabits, not one

Measured against the host fetching the same 67 MB from Alpine's CDN at
11.5 MB/s, SchOS managed about 120 KB/s. Three causes:

- The receive window reopened when data entered the socket buffer,
  before the reader drained it; lwIP then refused segments and the
  sender waited on retransmission timers. It reopens on read now.
- The e1000 receive ring held 16 frames, so a full window arriving back
  to back lost everything past the sixteenth. 256 now.
- Without window scaling the window stops at 64 KB, one window per
  round trip. Scaling on, 256 KB window, 512 KB socket ring (a power of
  two, which the free-running indices need), copied in blocks.

Now 5.8 MB/s for the same file, and apk update in about two seconds.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01S1giNvrHTrPMPw7V6x1kQ4

kernel/net/lwip/lwipopts.h

@@ -35,10 +35,12 @@
 #define MEMP_NUM_UDP_PCB            8
 #define MEMP_NUM_TCP_PCB            16
 #define MEMP_NUM_TCP_PCB_LISTEN     8
-#define MEMP_NUM_TCP_SEG            64
+#define MEMP_NUM_TCP_SEG            256   // a window held out of order
 #define MEMP_NUM_ARP_QUEUE          16
 #define MEMP_NUM_SYS_TIMEOUT        16
-#define PBUF_POOL_SIZE              48
+// Enough for a full receive window held out of order, plus the frames in
+// flight from the driver: a window of TCP_WND below, in pool-sized pieces.
+#define PBUF_POOL_SIZE              256
 #define PBUF_POOL_BUFSIZE           1536   // one Ethernet frame plus room
 
 // Protocols. IPv4 only: the driver is proven for it, the QEMU test network is
@@ -77,7 +79,17 @@
 
 #define TCP_MSS                     1460
 #define TCP_SND_BUF                 (8 * TCP_MSS)
-#define TCP_WND                     (8 * TCP_MSS)
+// The receive window, and with it the download speed: one window per round
+// trip. Eight segments held a transfer from a CDN to well under a megabyte a
+// second; thirty-two, the most that fits in TCP's 16-bit window field, held
+// it to 1.5 MB/s at 25 ms -- an eighth of a 100 Mbit line, measured against
+// the host fetching the same file. Window scaling (RFC 7323) lifts the 64 KB
+// ceiling: the field is shifted by TCP_RCV_SCALE, so 256 KB is advertised as
+// 16384 << 4. Every socket's buffer in kernel/net/inet.c is larger than this,
+// which is what makes the window an honest promise -- see inet_recv_cb.
+#define LWIP_WND_SCALE              1
+#define TCP_RCV_SCALE               4
+#define TCP_WND                     (256 * 1024)
 #define TCP_SND_QUEUELEN            ((4 * TCP_SND_BUF) / TCP_MSS)
 #define TCP_LISTEN_BACKLOG          1