A cross. Sort of. Also, remember that program is 32 words per PIO, not SM.

src/main.c

@@ -15,11 +15,11 @@
 #include "pico/stdlib.h"
 
 #include <stdint.h>
+#include <stdio.h>
+#include <string.h>
 
 #include "vga.pio.h"
 
-#include <stdio.h>
-
 #define FRAMES 2
 
 #define HPIXEL 640
@@ -108,34 +108,56 @@ void setup_clocks(void)
 	clock_configure(clk_peri, 0, CLOCKS_CLK_PERI_CTRL_AUXSRC_VALUE_CLK_SYS, 126 * MHZ, 126 * MHZ);
 }
 
+static bool dma_ready = false;
 static int dma_channel;
 
-void dma_handler()
+void new_frame_handler(void)
 {
+	// restart DMA
+	if (dma_ready)
+		dma_channel_set_read_addr(dma_channel, &frames[0].data[0], true);
+
+	pio_interrupt_clear(pio0_hw, 1);
+}
+
+void dma_handler(void)
+{
+	// acknoweledge DMA
 	dma_hw->ints0 = 1 << dma_channel;
-	dma_channel_set_read_addr(dma_channel, &frames[0].data[5120], true);
+}
-	// printf(".");
+
+void set_pixel(uint16_t x, uint16_t y)
+{
+	frames[0].data[y * 80 + x / 8] |= (1 << x % 8);
 }
 
 int main()
 {
 	setup_clocks();
 	stdio_init_all();
-	sleep_ms(5000);
+	sleep_ms(1000);
 
-	// "draw" an horizontal dotted line (?)
+	// clear frame buffer
-	for (unsigned int c = 0; c < 5; c++)
+	memset(&frames[0].data[0], 0, sizeof(frames[0].data));
-	{
+	// horizontal line
-		frames[0].data[80 * 64 + 60 + c] = 0x55;
+	for (uint16_t x = 400; x < 620; ++x)
-	}
+		set_pixel(x, 64);
+	// vertical line
+	for (uint16_t y = 20; y < 460; ++y)
+		set_pixel(400, y);
 
 	// PIO and state machines
 	PIO pio = pio0;
 
+	// Prepare frame interrupt
+	pio_set_irq1_source_enabled(pio, pis_interrupt1, true);
+	irq_set_exclusive_handler(PIO0_IRQ_1, new_frame_handler);
+	irq_set_enabled(PIO0_IRQ_1, true);
+
 	// VGA HSYNC program
 	uint offset1 = pio_add_program(pio, &vga_hsync_program);
 	uint sm1 = pio_claim_unused_sm(pio, true);
-	printf("Starting VGA sync machine on %u...\n", sm1);
+	printf("Starting VGA hsync machine on %u...\n", sm1);
 	vga_hsync_program_init(pio, sm1, offset1);
 	printf("Feeding horizontal sync...\n");
 	// low pulse is X pixels long, and SM runs at 4x of pixel clock
@@ -154,8 +176,12 @@ int main()
 	printf("OK\n");
 
 	// VGA pixel program
+
+	printf("pio_add_program...\n");
 	uint offset0 = pio_add_program(pio, &vga_free_run_program);
-	uint sm0 = pio_claim_unused_sm(pio, true);
+	printf("pio_claim_unused_sm...\n");
+	uint sm0 = pio_claim_unused_sm(pio, false);
+	printf("sm0 = %u\n", sm0);
 	printf("Starting VGA pixel machine on %u...\n", sm0);
 	vga_pixel_program_init(pio, sm0, offset0);
 	pio_sm_put_blocking(pio, sm0, 640 - 1);
@@ -173,12 +199,13 @@ int main()
 	irq_set_exclusive_handler(DMA_IRQ_0, dma_handler);
 	irq_set_enabled(DMA_IRQ_0, true);
 
-	dma_channel_configure(dma_channel, &dma_config, &pio0_hw->txf[2], &frames[0].data[0], 80, true);
+	dma_channel_configure(dma_channel, &dma_config, &pio0_hw->txf[2], &frames[0].data[0], 38400,
+	                      false);
 
-	// dma_handler();
+	dma_ready = true;
 
 	printf("Start!\n");
 
 	for (;;)
-		;
+		tight_loop_contents();
 }

src/vga.pio

@@ -19,8 +19,7 @@ entrypoint_vga_hsync:
     pull
 
 loop:
-    nop
+    irq set 0 [1]
-    irq set 0
 
     mov x, isr
     set pins, 0
@@ -31,8 +30,7 @@ hsync_pulse:
     set pins, 1
 hsync_idle:
     jmp x-- hsync_idle
-    nop
+    jmp loop [1]
-    jmp loop
 
 
 .program vga_vsync
@@ -56,4 +54,6 @@ vsync_idle:
     set pins, 1
     jmp x-- vsync_idle
 
+    irq set 1
+
     jmp loop