Merge pull request #2 from vcaputo/moar

More candy

56 files changed, 4602 insertions, 453 deletions

diff --git a/Makefile b/Makefile
index c238b5d..13eef36 100644
--- a/Makefile
+++ b/Makefile
@@ -1,13 +1,11 @@
-CFLAGS=-Wall -O3 $(shell pkg-config --cflags libdrm)
-LDFLAGS=-lm $(shell pkg-config --libs libdrm)
+CFLAGS=-D_GNU_SOURCE -Wall -O3 -ffast-math -pthread $(shell pkg-config --cflags libdrm) -I.
+LDFLAGS=-lm -pthread $(shell pkg-config --libs libdrm)
+SRC=rototiller.c modules/ray/*.c modules/roto/roto.c modules/sparkler/*.c drmsetup.c fb.c fps.c util.c
 
-all: rototiller32 rototiller64
+all: rototiller
 
-rototiller32: rototiller32.c
-	$(CC) -o $@ $^ $(CFLAGS) $(LDFLAGS)
-
-rototiller64: rototiller64.c
-	$(CC) -o $@ $^ $(CFLAGS) $(LDFLAGS)
+rototiller: $(SRC) Makefile modules/ray/*.h modules/roto/roto.h modules/sparkler/*.h drmsetup.h fb.h fps.h util.h
+	$(CC) -o $@ $(SRC) $(CFLAGS) $(LDFLAGS)
 
 clean:
-	rm -f rototiller32 rototiller64
+	rm -f rototiller
diff --git a/README b/README
index f21fc6c..e05473e 100644
--- a/README
+++ b/README
@@ -1,5 +1,5 @@
 This is a quick little graphics hack I put together to experiment with libdrm
-and this "dumb buffer" thingy David Arlie added to the kernel back in 2011.
+and this "dumb buffer" thingy David Airlie added to the kernel back in 2011.
 
 If you're like me and miss the pre-KMS days of functional SVGAlib on linux
 where we could write purely software-rendered graphics toys like demos that
diff --git a/drmsetup.c b/drmsetup.c
new file mode 100644
index 0000000..3ed435a
--- /dev/null
+++ b/drmsetup.c
@@ -0,0 +1,147 @@
+/*
+ * Rudimentary drm setup dialog... this is currently a very basic stdio thingy.
+ */
+
+#include <assert.h>
+#include <fcntl.h>
+#include <inttypes.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <xf86drm.h>
+#include <xf86drmMode.h>
+
+#include "util.h"
+
+static const char * encoder_type_name(uint32_t type) {
+	static const char *encoder_types[] = {
+		"None",
+		"DAC",
+		"TMDS",
+		"LVDAC",
+		"VIRTUAL",
+		"DSI"
+	};
+
+	assert(type < nelems(encoder_types));
+
+	return encoder_types[type];
+}
+
+
+static const char * connector_type_name(uint32_t type) {
+	static const char *connector_types[] = {
+		"Unknown",
+		"VGA",
+		"DVII",
+		"DVID",
+		"DVIA",
+		"Composite",
+		"SVIDEO",
+		"LVDS",
+		"Component",
+		"SPinDIN",
+		"DisplayPort",
+		"HDMIA",
+		"HDMIB",
+		"TV",
+		"eDP",
+		"VIRTUAL",
+		"DSI"
+	};
+
+	assert(type < nelems(connector_types));
+
+	return connector_types[type];
+}
+
+
+static const char * connection_type_name(int type) {
+	static const char *connection_types[] = {
+		[1] = "Connected",
+		"Disconnected",
+		"Unknown"
+	};
+
+	assert(type < nelems(connection_types));
+
+	return connection_types[type];
+}
+
+
+/* interactively setup the drm device, store the selections */
+void drm_setup(int *res_drm_fd, uint32_t *res_crtc_id, uint32_t *res_connector_id, drmModeModeInfoPtr *res_mode)
+{
+	int			drm_fd, i;
+	drmVersionPtr		drm_ver;
+	drmModeResPtr		drm_res;
+	drmModeConnectorPtr	drm_con;
+	drmModeEncoderPtr	drm_enc;
+	drmModeCrtcPtr		drm_crtc;
+	char			dev[256];
+	int			connector_num;
+
+	pexit_if(!drmAvailable(),
+		"drm unavailable");
+
+	ask_string(dev, sizeof(dev), "DRM device", "/dev/dri/card0");
+
+	pexit_if((drm_fd = open(dev, O_RDWR)) < 0,
+		"unable to open drm device \"%s\"", dev);
+
+	pexit_if(!(drm_ver = drmGetVersion(drm_fd)),
+		"unable to get drm version");
+
+	printf("\nVersion: %i.%i.%i\nName: \"%.*s\"\nDate: \"%.*s\"\nDescription: \"%.*s\"\n\n",
+		drm_ver->version_major,
+		drm_ver->version_minor,
+		drm_ver->version_patchlevel,
+		drm_ver->name_len,
+		drm_ver->name,
+		drm_ver->date_len,
+		drm_ver->date,
+		drm_ver->desc_len,
+		drm_ver->desc);
+
+	exit_if(!(drm_res = drmModeGetResources(drm_fd)),
+		"unable to get drm resources");
+
+	for (i = 0; i < drm_res->count_connectors; i++) {
+
+		exit_if(!(drm_con = drmModeGetConnector(drm_fd, drm_res->connectors[i])),
+			"unable to get connector %x", (int)drm_res->connectors[i]);
+
+		if (!drm_con->encoder_id) {
+			continue;
+		}
+
+		exit_if(!(drm_enc = drmModeGetEncoder(drm_fd, drm_con->encoder_id)),
+			"unable to get encoder %x", (int)drm_con->encoder_id);
+
+		printf("%i: Connector [%x]: %s [%s%s%s]\n",
+			i, (int)drm_res->connectors[i],
+			connector_type_name(drm_con->connector_type),
+			connection_type_name(drm_con->connection),
+			drm_con->encoder_id ? " via " : "",
+			drm_con->encoder_id ? encoder_type_name(drm_enc->encoder_type) : "");
+			/* TODO show mmWidth/mmHeight? */
+	}
+	ask_num(&connector_num, drm_res->count_connectors, "Connector", 0); // TODO default? 
+
+	exit_if(!(drm_con = drmModeGetConnector(drm_fd, drm_res->connectors[connector_num])),
+		"unable to get connector %x", (int)drm_res->connectors[connector_num]);
+	exit_if(!(drm_enc = drmModeGetEncoder(drm_fd, drm_con->encoder_id)),
+		"unable to get encoder %x", (int)drm_con->encoder_id);
+
+	exit_if(!(drm_crtc = drmModeGetCrtc(drm_fd, drm_enc->crtc_id)),
+		"unable to get crtc %x", (int)drm_enc->crtc_id);
+
+	*res_drm_fd = drm_fd;
+	*res_crtc_id = drm_crtc->crtc_id;
+	*res_connector_id = drm_con->connector_id;
+	*res_mode = &drm_crtc->mode;
+}
diff --git a/drmsetup.h b/drmsetup.h
new file mode 100644
index 0000000..61af2d5
--- /dev/null
+++ b/drmsetup.h
@@ -0,0 +1,10 @@
+#ifndef _DRM_SETUP_H
+#define _DRM_SETUP_H
+
+#include <stddef.h>	/* xf86drmMode.h uses size_t without including stddef.h, sigh */
+#include <stdint.h>
+#include <xf86drmMode.h>
+
+void drm_setup(int *res_drm_fd, uint32_t *res_crtc_id, uint32_t *res_connector_id, drmModeModeInfoPtr *res_mode);
+
+#endif
diff --git a/fb.c b/fb.c
new file mode 100644
index 0000000..39e7feb
--- /dev/null
+++ b/fb.c
@@ -0,0 +1,343 @@
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <pthread.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include <inttypes.h>
+#include <xf86drm.h>
+#include <xf86drmMode.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <sys/ioctl.h>
+#include <sys/mman.h>
+
+#include "fb.h"
+#include "util.h"
+
+/* Copyright (C) 2016 Vito Caputo <vcaputo@pengaru.com> */
+
+
+/* I've used a separate thread for page-flipping duties because the libdrm api
+ * (and related kernel ioctl) for page flips doesn't appear to support queueing
+ * multiple flip requests.  In this use case we aren't interactive and wish to
+ * just accumulate rendered pages until we run out of spare pages, allowing the
+ * renderer to get as far ahead of vsync as possible, and certainly never
+ * blocked waiting for vsync unless there's no spare page available for drawing
+ * into.
+ *
+ * In lieu of a queueing mechanism on the drm fd, we must submit the next page
+ * once the currently submitted page is flipped to - it's at that moment we
+ * won't get EBUSY from the ioctl any longer.  Without a dedicated thread
+ * submitting flip requests and synchronously consuming their flip events,
+ * we're liable to introduce latency in the page flip submission if implemented
+ * in a more opportunistic manner whenever the fb api is entered from the
+ * render loop.
+ *
+ * If the kernel simply let us queue multiple flip requests we could maintain
+ * our submission queue entirely in the drm fd, and get available pages from
+ * the drm event handler once our pool of pages is depleted.  The kernel on
+ * vsync could check the fd to see if another flip is queued and there would be
+ * the least latency possible in submitting the flips - the least likely to
+ * miss a vsync.  This would also elide the need for synchronization in
+ * userspace between the renderer and the flipper thread, since there would no
+ * longer be a flipper thread.
+ *
+ * Let me know if you're aware of a better way with existing mainline drm!
+ */
+
+
+/* Most of fb_page_t is kept private, the public part is
+ * just an fb_fragment_t describing the whole page.
+ */
+typedef struct _fb_page_t _fb_page_t;
+struct _fb_page_t {
+	_fb_page_t	*next;
+	uint32_t	*mmap;
+	size_t		mmap_size;
+	uint32_t	drm_dumb_handle;
+	uint32_t	drm_fb_id;
+	fb_page_t	public_page;
+};
+
+typedef struct fb_t {
+	pthread_t	thread;
+	int		drm_fd;
+	uint32_t	drm_crtc_id;
+
+	_fb_page_t	*active_page;		/* page currently displayed */
+
+	pthread_mutex_t	ready_mutex;
+	pthread_cond_t	ready_cond;
+	_fb_page_t	*ready_pages_head;	/* next pages to flip to */
+	_fb_page_t	*ready_pages_tail;
+
+	pthread_mutex_t	inactive_mutex;
+	pthread_cond_t	inactive_cond;
+	_fb_page_t	*inactive_pages;	/* finished pages available for (re)use */
+
+	unsigned	put_pages_count;
+} fb_t;
+
+#ifndef container_of
+#define container_of(_ptr, _type, _member) \
+	(_type *)((void *)(_ptr) - offsetof(_type, _member))
+#endif
+
+
+/* Synchronize with the page flip by waiting for its event. */
+static inline void fb_drm_flip_wait(fb_t *fb)
+{
+	drmEventContext	drm_ev_ctx = {
+				.version = DRM_EVENT_CONTEXT_VERSION,
+				.vblank_handler = NULL,
+				.page_flip_handler = NULL
+			};
+	drmHandleEvent(fb->drm_fd, &drm_ev_ctx);
+}
+
+
+/* Consumes ready pages queued via fb_page_put(), submits them to drm to flip
+ * on vsync.  Produces inactive pages from those replaced, making them
+ * available to fb_page_get(). */
+static void * fb_flipper_thread(void *_fb)
+{
+	fb_t	*fb = _fb;
+
+	for (;;) {
+		_fb_page_t	*next_active_page;
+		/* wait for a flip req, submit the req page for flip on vsync, wait for it to flip before making the
+		 * active page inactive/available, repeat.
+		 */
+		pthread_mutex_lock(&fb->ready_mutex);
+		while (!fb->ready_pages_head)
+			pthread_cond_wait(&fb->ready_cond, &fb->ready_mutex);
+
+		next_active_page = fb->ready_pages_head;
+		fb->ready_pages_head = next_active_page->next;
+		if (!fb->ready_pages_head)
+			fb->ready_pages_tail = NULL;
+		pthread_mutex_unlock(&fb->ready_mutex);
+
+		/* submit the next active page for page flip on vsync, and wait for it. */
+		pexit_if(drmModePageFlip(fb->drm_fd, fb->drm_crtc_id, next_active_page->drm_fb_id, DRM_MODE_PAGE_FLIP_EVENT, NULL) < 0,
+			"unable to flip page");
+		fb_drm_flip_wait(fb);
+
+		/* now that we're displaying a new page, make the previously active one inactive so rendering can reuse it */
+		pthread_mutex_lock(&fb->inactive_mutex);
+		fb->active_page->next = fb->inactive_pages;
+		fb->inactive_pages = fb->active_page;
+		pthread_cond_signal(&fb->inactive_cond);
+		pthread_mutex_unlock(&fb->inactive_mutex);
+
+		fb->active_page = next_active_page;
+	}
+}
+
+
+/* creates a framebuffer page, which is a coupled drm_fb object and mmap region of memory */
+static void fb_page_new(fb_t *fb, drmModeModeInfoPtr mode)
+{
+	_fb_page_t			*page;
+	struct drm_mode_create_dumb	create_dumb = {
+						.width = mode->hdisplay,
+						.height = mode->vdisplay,
+						.bpp = 32,
+						.flags = 0, // unused,
+					};
+	struct drm_mode_map_dumb	map_dumb = {
+						.pad = 0, // unused
+					};
+	uint32_t			*map, fb_id;
+
+	page = calloc(1, sizeof(_fb_page_t));
+
+	pexit_if(ioctl(fb->drm_fd, DRM_IOCTL_MODE_CREATE_DUMB, &create_dumb) < 0,
+		"unable to create dumb buffer");
+
+	map_dumb.handle = create_dumb.handle;
+	pexit_if(ioctl(fb->drm_fd, DRM_IOCTL_MODE_MAP_DUMB, &map_dumb) < 0,
+		"unable to prepare dumb buffer for mmap");
+	pexit_if(!(map = mmap(NULL, create_dumb.size, PROT_READ|PROT_WRITE, MAP_SHARED, fb->drm_fd, map_dumb.offset)),
+		"unable to mmap dumb buffer");
+	pexit_if(drmModeAddFB(fb->drm_fd, mode->hdisplay, mode->vdisplay, 24, 32, create_dumb.pitch, create_dumb.handle, &fb_id) < 0,
+		"unable to add dumb buffer");
+
+	page->mmap = map;
+	page->mmap_size = create_dumb.size;
+	page->drm_dumb_handle = map_dumb.handle;
+	page->drm_fb_id = fb_id;
+
+	page->public_page.fragment.buf = map;
+	page->public_page.fragment.width = mode->hdisplay;
+	page->public_page.fragment.height = mode->vdisplay;
+	page->public_page.fragment.stride = create_dumb.pitch - (mode->hdisplay * 4);
+
+	page->next = fb->inactive_pages;
+	fb->inactive_pages = page;
+}
+
+
+static void _fb_page_free(fb_t *fb, _fb_page_t *page)
+{
+	struct drm_mode_destroy_dumb	destroy_dumb = {
+						.handle = page->drm_dumb_handle,
+					};
+
+	drmModeRmFB(fb->drm_fd, page->drm_fb_id);
+	munmap(page->mmap, page->mmap_size);
+	ioctl(fb->drm_fd, DRM_IOCTL_MODE_DESTROY_DUMB, &destroy_dumb); // XXX: errors?
+	free(page);
+}
+
+
+/* get the next inactive page from the fb, waiting if necessary. */
+static inline _fb_page_t * _fb_page_get(fb_t *fb)
+{
+	_fb_page_t	*page;
+
+	/* As long as n_pages is >= 3 this won't block unless we're submitting
+	 * pages faster than vhz.
+	 */
+	pthread_mutex_lock(&fb->inactive_mutex);
+	while (!(page = fb->inactive_pages))
+		pthread_cond_wait(&fb->inactive_cond, &fb->inactive_mutex);
+	fb->inactive_pages = page->next;
+	pthread_mutex_unlock(&fb->inactive_mutex);
+
+	page->next = NULL;
+
+	return page;
+}
+
+
+/* public interface */
+fb_page_t * fb_page_get(fb_t *fb)
+{
+	return &(_fb_page_get(fb)->public_page);
+}
+
+
+/* put a page into the fb, queueing for display */
+static inline void _fb_page_put(fb_t *fb, _fb_page_t *page)
+{
+	pthread_mutex_lock(&fb->ready_mutex);
+	if (fb->ready_pages_tail)
+		fb->ready_pages_tail->next = page;
+	else
+		fb->ready_pages_head = page;
+
+	fb->ready_pages_tail = page;
+	pthread_cond_signal(&fb->ready_cond);
+	pthread_mutex_unlock(&fb->ready_mutex);
+}
+
+
+/* public interface */
+
+/* put a page into the fb, queueing for display */
+void fb_page_put(fb_t *fb, fb_page_t *page)
+{
+	fb->put_pages_count++;
+
+	_fb_page_put(fb, container_of(page, _fb_page_t, public_page));
+}
+
+
+/* get (and reset) the current count of put pages */
+void fb_get_put_pages_count(fb_t *fb, unsigned *count)
+{
+	*count = fb->put_pages_count;
+	fb->put_pages_count = 0;
+}
+
+
+/* free the fb and associated resources */
+void fb_free(fb_t *fb)
+{
+	if (fb->active_page) {
+		pthread_cancel(fb->thread);
+		pthread_join(fb->thread, NULL);
+	}
+
+	/* TODO: free all the pages */
+
+	pthread_mutex_destroy(&fb->ready_mutex);
+	pthread_cond_destroy(&fb->ready_cond);
+	pthread_mutex_destroy(&fb->inactive_mutex);
+	pthread_cond_destroy(&fb->inactive_cond);
+
+	free(fb);
+}
+
+
+/* create a new fb instance */
+fb_t * fb_new(int drm_fd, uint32_t crtc_id, uint32_t *connectors, int n_connectors, drmModeModeInfoPtr mode, int n_pages)
+{
+	int		i;
+	_fb_page_t	*page;
+	fb_t		*fb;
+
+	fb = calloc(1, sizeof(fb_t));
+	if (!fb)
+		return NULL;
+
+	fb->drm_fd = drm_fd;
+	fb->drm_crtc_id = crtc_id;
+
+	for (i = 0; i < n_pages; i++)
+		fb_page_new(fb, mode);
+
+	pthread_mutex_init(&fb->ready_mutex, NULL);
+	pthread_cond_init(&fb->ready_cond, NULL);
+	pthread_mutex_init(&fb->inactive_mutex, NULL);
+	pthread_cond_init(&fb->inactive_cond, NULL);
+
+	page = _fb_page_get(fb);
+
+	/* set the video mode, pinning this page, set it as the active page. */
+	if (drmModeSetCrtc(drm_fd, crtc_id, page->drm_fb_id, 0, 0, connectors, n_connectors, mode) < 0) {
+		_fb_page_free(fb, page);
+		fb_free(fb);
+		return NULL;
+	}
+
+	fb->active_page = page;
+
+	/* start up the page flipper thread */
+	pthread_create(&fb->thread, NULL, fb_flipper_thread, fb);
+
+	return fb;
+}
+
+
+/* divide a fragment into n_fragments, storing their values into fragments[],
+ * which is expected to have n_fragments of space. */
+void fb_fragment_divide(fb_fragment_t *fragment, unsigned n_fragments, fb_fragment_t fragments[])
+{
+	unsigned	slice = fragment->height / n_fragments;
+	unsigned	i;
+	void		*buf = fragment->buf;
+	unsigned	pitch = (fragment->width * 4) + fragment->stride;
+	unsigned	y = fragment->y;
+
+	/* This just splits the supplied fragment into even horizontal slices */
+	/* TODO: It probably makes sense to add an fb_fragment_tile() as well, since some rendering
+	 * algorithms benefit from the locality of a tiled fragment.
+	 */
+
+	for (i = 0; i < n_fragments; i++) {
+		fragments[i].buf = buf;
+		fragments[i].x = fragment->x;
+		fragments[i].y = y;
+		fragments[i].width = fragment->width;
+		fragments[i].height = slice;
+		fragments[i].stride = fragment->stride;
+
+		buf += pitch * slice;
+		y += slice;
+	}
+	/* TODO: handle potential fractional tail slice? */
+}
diff --git a/fb.h b/fb.h
new file mode 100644
index 0000000..13f6bcd
--- /dev/null
+++ b/fb.h
@@ -0,0 +1,37 @@
+#ifndef _FB_H
+#define _FB_H
+
+#include <stdint.h>
+#include <sys/types.h>
+#include <xf86drmMode.h> /* for drmModeModeInfoPtr */
+
+/* All renderers should target fb_fragment_t, which may or may not represent
+ * a full-screen mmap.  Helpers are provided for subdividing fragments for
+ * concurrent renderers.
+ */
+typedef struct fb_fragment_t {
+	uint32_t	*buf;		/* pointer to the first pixel in the fragment */
+	unsigned	x, y;		/* absolute coordinates of the upper left corner of this fragment */
+	unsigned	width, height;	/* width and height of this fragment */
+	unsigned	stride;		/* number of bytes from the end of one row to the start of the next */
+} fb_fragment_t;
+
+/* This is a page handle object for page flip submission/life-cycle.
+ * Outside of fb_page_get()/fb_page_put(), you're going to be interested in
+ * fb_fragment_t.  The fragment included here describes the whole page,
+ * it may be divided via fb_fragment_divide().
+ */
+typedef struct fb_page_t {
+	fb_fragment_t	fragment;
+} fb_page_t;
+
+typedef struct fb_t fb_t;
+
+fb_page_t * fb_page_get(fb_t *fb);
+void fb_page_put(fb_t *fb, fb_page_t *page);
+void fb_free(fb_t *fb);
+void fb_get_put_pages_count(fb_t *fb, unsigned *count);
+fb_t * fb_new(int drm_fd, uint32_t crtc_id, uint32_t *connectors, int n_connectors, drmModeModeInfoPtr mode, int n_pages);
+void fb_fragment_divide(fb_fragment_t *fragment, unsigned n_fragments, fb_fragment_t fragments[]);
+
+#endif
diff --git a/fps.c b/fps.c
new file mode 100644
index 0000000..99a3f85
--- /dev/null
+++ b/fps.c
@@ -0,0 +1,46 @@
+#include <signal.h>
+#include <stdio.h>
+#include <sys/time.h>
+
+#include "fb.h"
+#include "util.h"
+
+
+static int	print_fps;
+
+
+static void sigalrm_handler(int signum)
+{
+	print_fps = 1;
+}
+
+
+int fps_setup(void)
+{
+	struct itimerval	interval = { 	
+					.it_interval = { .tv_sec = 1, .tv_usec = 0 },
+					.it_value = { .tv_sec = 1, .tv_usec = 0 },
+				};
+
+	if (signal(SIGALRM, sigalrm_handler) == SIG_ERR)
+		return 0;
+		
+	if (setitimer(ITIMER_REAL, &interval, NULL) < 0)
+		return 0;
+
+	return 1;
+}
+
+
+void fps_print(fb_t *fb)
+{
+	unsigned	n;
+
+	if (!print_fps)
+		return;
+
+	fb_get_put_pages_count(fb, &n);
+	printf("FPS: %u\n", n);
+
+	print_fps = 0;
+}
diff --git a/fps.h b/fps.h
new file mode 100644
index 0000000..1f986c7
--- /dev/null
+++ b/fps.h
@@ -0,0 +1,9 @@
+#ifndef _FPS_H
+#define _FPS_H
+
+#include "fb.h"
+
+int fps_setup(void);
+void fps_print(fb_t *fb);
+
+#endif
diff --git a/modules/ray/ray.c b/modules/ray/ray.c
new file mode 100644
index 0000000..60d08cf
--- /dev/null
+++ b/modules/ray/ray.c
@@ -0,0 +1,161 @@
+#include <stdint.h>
+#include <inttypes.h>
+#include <math.h>
+
+#include "fb.h"
+#include "rototiller.h"
+#include "util.h"
+
+#include "ray_camera.h"
+#include "ray_object.h"
+#include "ray_scene.h"
+#include "ray_threads.h"
+
+/* Copyright (C) 2016 Vito Caputo <vcaputo@pengaru.com> */
+
+/* ray trace a simple scene into the fragment */
+static void ray(fb_fragment_t *fragment)
+{
+	static ray_object_t	objects[] = {
+		{
+			.plane = {
+				.type = RAY_OBJECT_TYPE_PLANE,
+				.surface = {
+					.color = { .x = 0.4, .y = 0.2, .z = 0.5 },
+					.diffuse = 1.0f,
+					.specular = 0.2f,
+				},
+				.normal = { .x = 0.0, .y = 1.0, .z = 0.0 },
+				.distance = -3.2f,
+			}
+		}, {
+			.sphere = {
+				.type = RAY_OBJECT_TYPE_SPHERE,
+				.surface = {
+					.color = { .x = 1.0, .y = 0.0, .z = 0.0 },
+					.diffuse = 1.0f,
+					.specular = 0.05f,
+				},
+				.center = { .x = 0.5, .y = 1.0, .z = 0.0 },
+				.radius = 1.2f,
+			}
+		}, {
+			.sphere = {
+				.type = RAY_OBJECT_TYPE_SPHERE,
+				.surface = {
+					.color = { .x = 0.0, .y = 0.0, .z = 1.0 },
+					.diffuse = 1.0f,
+					.specular = 1.0f,
+				},
+				.center = { .x = -2.0, .y = 1.0, .z = 0.0 },
+				.radius = 0.9f,
+			}
+		}, {
+			.sphere = {
+				.type = RAY_OBJECT_TYPE_SPHERE,
+				.surface = {
+					.color = { .x = 0.0, .y = 1.0, .z = 1.0 },
+					.diffuse = 1.0f,
+					.specular = 1.0f,
+				},
+				.center = { .x = 2.0, .y = -1.0, .z = 0.0 },
+				.radius = 1.0f,
+			}
+		}, {
+			.sphere = {
+				.type = RAY_OBJECT_TYPE_SPHERE,
+				.surface = {
+					.color = { .x = 0.0, .y = 1.0, .z = 0.0 },
+					.diffuse = 1.0f,
+					.specular = 1.0f,
+				},
+				.center = { .x = 0.2, .y = -1.25, .z = 0.0 },
+				.radius = 0.6f,
+			}
+		}, {
+			.light = {
+				.type = RAY_OBJECT_TYPE_LIGHT,
+				.brightness = 1.0,
+				.emitter = {
+					.point.type = RAY_LIGHT_EMITTER_TYPE_POINT,
+					.point.center = { .x = 3.0f, .y = 3.0f, .z = 3.0f },
+					.point.surface = {
+						.color = { .x = 1.0f, .y = 1.0f, .z = 1.0f },
+					},
+				}
+			}
+		}
+	};
+
+	ray_camera_t		camera = {
+					.position = { .x = 0.0, .y = 0.0, .z = 6.0 },
+					.orientation = {
+						.yaw = RAY_EULER_DEGREES(0.0f),
+						.pitch = RAY_EULER_DEGREES(0.0f),
+						.roll = RAY_EULER_DEGREES(180.0f),
+					},
+					.focal_length = 700.0f,
+					.width = fragment->width,
+					.height = fragment->height,
+				};
+
+	static ray_scene_t	scene = {
+					.objects = objects,
+					.n_objects = nelems(objects),
+					.lights = &objects[5],
+					.n_lights = 1,
+					.ambient_color = { .x = 1.0f, .y = 1.0f, .z = 1.0f },
+					.ambient_brightness = .04f,
+				};
+	static int		initialized;
+	static ray_threads_t	*threads;
+	static fb_fragment_t	*fragments;
+	static unsigned		ncpus;
+#if 1
+	/* animated point light source */
+	static double		r;
+
+	r += .02;
+
+	scene.lights[0].light.emitter.point.center.x = cosf(r) * 3.5f;
+	scene.lights[0].light.emitter.point.center.z = sinf(r) * 3.5f;
+	camera.orientation.yaw = sinf(r) / 4;
+	camera.orientation.pitch = sinf(r * 10) / 100;
+	camera.orientation.roll = RAY_EULER_DEGREES(180.0f) + cosf(r) / 10;
+	camera.position.x = cosf(r) / 10;
+	camera.position.z = 4.0f + sinf(r) / 10;
+#endif
+
+	if (!initialized) {
+		initialized = 1;
+		ncpus = get_ncpus();
+
+		if (ncpus > 1) {
+			threads = ray_threads_create(ncpus - 1);
+			fragments = malloc(sizeof(fb_fragment_t) * ncpus);
+		}
+	}
+
+	if (ncpus > 1) {
+		/* Always recompute the fragments[] geometry.
+		 * This way the fragment geometry can change at any moment and things will
+		 * continue functioning, which may prove important later on.
+		 * (imagine things like a preview window, or perhaps composite modules
+		 * which call on other modules supplying virtual fragments of varying dimensions..)
+		 */
+		fb_fragment_divide(fragment, ncpus, fragments);
+	} else {
+		fragments = fragment;
+	}
+
+	ray_scene_render_fragments(&scene, &camera, threads, fragments);
+}
+
+
+rototiller_renderer_t	ray_renderer = {
+	.render = ray,
+	.name = "ray",
+	.description = "Multi-threaded ray tracer",
+	.author = "Vito Caputo <vcaputo@pengaru.com>",
+	.license = "GPLv2",
+};
diff --git a/modules/ray/ray.h b/modules/ray/ray.h
new file mode 100644
index 0000000..d33f96a
--- /dev/null
+++ b/modules/ray/ray.h
@@ -0,0 +1,8 @@
+#ifndef _RAY_RAY_H
+#define _RAY_RAY_H
+
+#include "fb.h"
+
+void ray(fb_fragment_t *fragment);
+
+#endif
diff --git a/modules/ray/ray_3f.h b/modules/ray/ray_3f.h
new file mode 100644
index 0000000..8408abb
--- /dev/null
+++ b/modules/ray/ray_3f.h
@@ -0,0 +1,161 @@
+#ifndef _RAY_3F_H
+#define _RAY_3F_H
+
+#include <math.h>
+
+typedef struct ray_3f_t {
+	float	x, y, z;
+} ray_3f_t;
+
+
+/* return the result of (a + b) */
+static inline ray_3f_t ray_3f_add(ray_3f_t *a, ray_3f_t *b)
+{
+	ray_3f_t	res = {
+				.x = a->x + b->x,
+				.y = a->y + b->y,
+				.z = a->z + b->z,
+			};
+
+	return res;
+}
+
+
+/* return the result of (a - b) */
+static inline ray_3f_t ray_3f_sub(ray_3f_t *a, ray_3f_t *b)
+{
+	ray_3f_t	res = {
+				.x = a->x - b->x,
+				.y = a->y - b->y,
+				.z = a->z - b->z,
+			};
+
+	return res;
+}
+
+
+/* return the result of (-v) */
+static inline ray_3f_t ray_3f_negate(ray_3f_t *v)
+{
+	ray_3f_t	res = {
+				.x = -v->x,
+				.y = -v->y,
+				.z = -v->z,
+			};
+
+	return res;
+}
+
+
+/* return the result of (a * b) */
+static inline ray_3f_t ray_3f_mult(ray_3f_t *a, ray_3f_t *b)
+{
+	ray_3f_t	res = {
+				.x = a->x * b->x,
+				.y = a->y * b->y,
+				.z = a->z * b->z,
+			};
+
+	return res;
+}
+
+
+/* return the result of (v * scalar) */
+static inline ray_3f_t ray_3f_mult_scalar(ray_3f_t *v, float scalar)
+{
+	ray_3f_t	res = {
+				.x = v->x * scalar,
+				.y = v->y * scalar,
+				.z = v->z * scalar,
+			};
+
+	return res;
+}
+
+
+/* return the result of (uv / scalar) */
+static inline ray_3f_t ray_3f_div_scalar(ray_3f_t *v, float scalar)
+{
+	ray_3f_t	res = {
+				.x = v->x / scalar,
+				.y = v->y / scalar,
+				.z = v->z / scalar,
+			};
+
+	return res;
+}
+
+
+/* return the result of (a . b) */
+static inline float ray_3f_dot(ray_3f_t *a, ray_3f_t *b)
+{
+	return a->x * b->x + a->y * b->y + a->z * b->z;
+}
+
+
+/* return the length of the supplied vector */
+static inline float ray_3f_length(ray_3f_t *v)
+{
+	return sqrtf(ray_3f_dot(v, v));
+}
+
+
+/* return the normalized form of the supplied vector */
+static inline ray_3f_t ray_3f_normalize(ray_3f_t *v)
+{
+	ray_3f_t	nv;
+	float		f;
+
+	f = 1.0f / ray_3f_length(v);
+
+	nv.x = f * v->x;
+	nv.y = f * v->y;
+	nv.z = f * v->z;
+
+	return nv;
+}
+
+
+/* return the distance between two arbitrary points */
+static inline float ray_3f_distance(ray_3f_t *a, ray_3f_t *b)
+{
+	return sqrtf(powf(a->x - b->x, 2) + powf(a->y - b->y, 2) + powf(a->z - b->z, 2));
+}
+
+
+/* return the cross product of two unit vectors */
+static inline ray_3f_t ray_3f_cross(ray_3f_t *a, ray_3f_t *b)
+{
+	ray_3f_t	product;
+
+	product.x = a->y * b->z - a->z * b->y;
+	product.y = a->z * b->x - a->x * b->z;
+	product.z = a->x * b->y - a->y * b->x;
+
+	return product;
+}
+
+
+/* return the linearly interpolated vector between the two vectors at point alpha (0-1.0) */
+static inline ray_3f_t ray_3f_lerp(ray_3f_t *a, ray_3f_t *b, float alpha)
+{
+	ray_3f_t	lerp_a, lerp_b;
+
+	lerp_a = ray_3f_mult_scalar(a, 1.0f - alpha);
+	lerp_b = ray_3f_mult_scalar(b, alpha);
+
+	return ray_3f_add(&lerp_a, &lerp_b);
+}
+
+
+/* return the normalized linearly interpolated vector between the two vectors at point alpha (0-1.0) */
+static inline ray_3f_t ray_3f_nlerp(ray_3f_t *a, ray_3f_t *b, float alpha)
+{
+	ray_3f_t	lerp;
+
+	lerp = ray_3f_lerp(a, b, alpha);
+
+	return ray_3f_normalize(&lerp);
+}
+
+#endif
diff --git a/modules/ray/ray_camera.c b/modules/ray/ray_camera.c
new file mode 100644
index 0000000..0703c2e
--- /dev/null
+++ b/modules/ray/ray_camera.c
@@ -0,0 +1,85 @@
+#include "fb.h"
+
+#include "ray_camera.h"
+#include "ray_euler.h"
+
+
+/* Produce a vector from the provided orientation vectors and proportions. */
+static ray_3f_t project_corner(ray_3f_t *forward, ray_3f_t *left, ray_3f_t *up, float focal_length, float horiz, float vert)
+{
+	ray_3f_t	tmp;
+	ray_3f_t	corner;
+
+	corner = ray_3f_mult_scalar(forward, focal_length);
+	tmp = ray_3f_mult_scalar(left, horiz);
+	corner = ray_3f_add(&corner, &tmp);
+	tmp = ray_3f_mult_scalar(up, vert);
+	corner = ray_3f_add(&corner, &tmp);
+
+	return ray_3f_normalize(&corner);
+}
+
+
+/* Produce vectors for the corners of the entire camera frame, used for interpolation. */
+static void project_corners(ray_camera_t *camera, ray_camera_frame_t *frame)
+{
+	ray_3f_t	forward, left, up, right, down;
+	float		half_horiz = (float)camera->width / 2.0f;
+	float		half_vert = (float)camera->height / 2.0f;
+
+	ray_euler_basis(&camera->orientation, &forward, &up, &left);
+	right = ray_3f_negate(&left);
+	down = ray_3f_negate(&up);
+
+	frame->nw = project_corner(&forward, &left, &up, camera->focal_length, half_horiz,  half_vert);
+	frame->ne = project_corner(&forward, &right, &up, camera->focal_length, half_horiz,  half_vert);
+	frame->se = project_corner(&forward, &right, &down, camera->focal_length, half_horiz,  half_vert);
+	frame->sw = project_corner(&forward, &left, &down, camera->focal_length, half_horiz,  half_vert);
+}
+
+
+/* Begin a frame for the fragment of camera projection, initializing frame and ray. */
+void ray_camera_frame_begin(ray_camera_t *camera, fb_fragment_t *fragment, ray_ray_t *ray, ray_camera_frame_t *frame)
+{
+	/* References are kept to the camera, fragment, and ray to be traced.
+	 * The ray is maintained as we step through the frame, that is the
+	 * purpose of this api.
+	 *
+	 * Since the ray direction should be a normalized vector, the obvious
+	 * implementation is a bit costly.  The camera frame api hides this
+	 * detail so we can explore interpolation techniques to potentially
+	 * lessen the per-pixel cost.
+	 */
+	frame->camera = camera;
+	frame->fragment = fragment;
+	frame->ray = ray;
+
+	frame->x = frame->y = 0;
+
+	/* From camera->orientation and camera->focal_length compute the vectors
+	 * through the viewport's corners, and place these normalized vectors
+	 * in frame->(nw,ne,sw,se).
+	 *
+	 * These can than be interpolated between to produce the ray vectors
+	 * throughout the frame's fragment.  The efficient option of linear
+	 * interpolation will not maintain the unit vector length, so to
+	 * produce normalized interpolated directions will require the costly
+	 * normalize function.
+	 *
+	 * I'm hoping a simple length correction table can be used to fixup the
+	 * linearly interpolated vectors to make them unit vectors with just
+	 * scalar multiplication instead of the sqrt of normalize.
+	 */
+	project_corners(camera, frame);
+
+	frame->x_delta = 1.0f / (float)camera->width;
+	frame->y_delta = 1.0f / (float)camera->height;
+	frame->x_alpha = frame->x_delta * (float)fragment->x;
+	frame->y_alpha = frame->y_delta * (float)fragment->y;
+
+	frame->cur_w = ray_3f_nlerp(&frame->nw, &frame->sw, frame->y_alpha);
+	frame->cur_e = ray_3f_nlerp(&frame->ne, &frame->se, frame->y_alpha);
+
+	ray->origin = camera->position;
+	ray->direction = frame->cur_w;
+}
diff --git a/modules/ray/ray_camera.h b/modules/ray/ray_camera.h
new file mode 100644
index 0000000..387f8c5
--- /dev/null
+++ b/modules/ray/ray_camera.h
@@ -0,0 +1,77 @@
+#ifndef _RAY_CAMERA_H
+#define _RAY_CAMERA_H
+
+#include <math.h>
+
+#include "fb.h"
+
+#include "ray_3f.h"
+#include "ray_euler.h"
+#include "ray_ray.h"
+
+
+typedef struct ray_camera_t {
+	ray_3f_t	position;		/* position of camera, the origin of all its rays */
+	ray_euler_t	orientation;		/* orientation of the camera */
+	float		focal_length;		/* controls the field of view */
+	unsigned	width;			/* width of camera viewport in pixels */
+	unsigned	height;			/* height of camera viewport in pixels */
+} ray_camera_t;
+
+
+typedef struct ray_camera_frame_t {
+	ray_camera_t	*camera;		/* the camera supplied to frame_begin() */
+	fb_fragment_t	*fragment;		/* the fragment supplied to frame_begin() */
+	ray_ray_t	*ray;			/* the ray supplied to frame_begin(), which gets updated as we step through the frame. */
+
+	ray_3f_t	nw, ne, sw, se;		/* directions pointing through the corners of the frame fragment */
+	ray_3f_t	cur_w, cur_e;		/* current row's west and east ends */
+	float		x_alpha, y_alpha;	/* interpolation position along the x and y axis */
+	float		x_delta, y_delta;	/* interpolation step delta along the x and y axis */
+	unsigned	x, y;			/* integral position within frame fragment */
+} ray_camera_frame_t;
+
+
+void ray_camera_frame_begin(ray_camera_t *camera, fb_fragment_t *fragment, ray_ray_t *ray, ray_camera_frame_t *frame);
+
+
+/* Step the ray through the frame on the x axis, returns 1 when rays remain on this axis, 0 at the end. */
+/* When 1 is returned, frame->ray is left pointing through the new coordinate. */
+static inline int ray_camera_frame_x_step(ray_camera_frame_t *frame)
+{
+	frame->x++;
+
+	if (frame->x >= frame->fragment->width) {
+		frame->x = 0;
+		frame->x_alpha = frame->x_delta * (float)frame->fragment->x;
+		return 0;
+	}
+
+	frame->x_alpha += frame->x_delta;
+	frame->ray->direction = ray_3f_nlerp(&frame->cur_w, &frame->cur_e, frame->x_alpha);
+
+	return 1;
+}
+
+
+/* Step the ray through the frame on the y axis, returns 1 when rays remain on this axis, 0 at the end. */
+/* When 1 is returned, frame->ray is left pointing through the new coordinate. */
+static inline int ray_camera_frame_y_step(ray_camera_frame_t *frame)
+{
+	frame->y++;
+
+	if (frame->y >= frame->fragment->height) {
+		frame->y = 0;
+		frame->y_alpha = frame->y_delta * (float)frame->fragment->y;
+		return 0;
+	}
+
+	frame->y_alpha += frame->y_delta;
+	frame->cur_w = ray_3f_nlerp(&frame->nw, &frame->sw, frame->y_alpha);
+	frame->cur_e = ray_3f_nlerp(&frame->ne, &frame->se, frame->y_alpha);
+	frame->ray->direction = frame->cur_w;
+
+	return 1;
+}
+
+#endif
diff --git a/modules/ray/ray_color.h b/modules/ray/ray_color.h
new file mode 100644
index 0000000..9fe62c1
--- /dev/null
+++ b/modules/ray/ray_color.h
@@ -0,0 +1,29 @@
+#ifndef _RAY_COLOR_H
+#define _RAY_COLOR_H
+
+#include <stdint.h>
+
+#include "ray_3f.h"
+
+typedef ray_3f_t ray_color_t;
+
+/* convert a vector into a packed, 32-bit rgb pixel value */
+static inline uint32_t ray_color_to_uint32_rgb(ray_color_t color) {
+	uint32_t	pixel;
+
+	/* doing this all per-pixel, ugh. */
+
+	if (color.x > 1.0f) color.x = 1.0f;
+	if (color.y > 1.0f) color.y = 1.0f;
+	if (color.z > 1.0f) color.z = 1.0f;
+
+	pixel = (uint32_t)(color.x * 255.0f);
+	pixel <<= 8;
+	pixel |= (uint32_t)(color.y * 255.0f);
+	pixel <<= 8;
+	pixel |= (uint32_t)(color.z * 255.0f);
+
+	return pixel;
+}
+
+#endif
diff --git a/modules/ray/ray_euler.h b/modules/ray/ray_euler.h
new file mode 100644
index 0000000..86f5221
--- /dev/null
+++ b/modules/ray/ray_euler.h
@@ -0,0 +1,45 @@
+#ifndef _RAY_EULER_H
+#define _RAY_EULER_H
+
+#include <math.h>
+
+#include "ray_3f.h"
+
+
+/* euler angles are convenient for describing orientation */
+typedef struct ray_euler_t {
+	float	pitch;	/* pitch in radiasn */
+	float	yaw;	/* yaw in radians */
+	float	roll;	/* roll in radians */
+} ray_euler_t;
+
+
+/* convenience macro for converting degrees to radians */
+#define RAY_EULER_DEGREES(_deg) \
+	(_deg * (2 * M_PI / 360.0f))
+
+
+/* produce basis vectors from euler angles */
+static inline void ray_euler_basis(ray_euler_t *e, ray_3f_t *forward, ray_3f_t *up, ray_3f_t *left)
+{
+	float	cos_yaw = cosf(e->yaw);
+	float	sin_yaw = sinf(e->yaw);
+	float	cos_roll = cosf(e->roll);
+	float	sin_roll = sinf(e->roll);
+	float	cos_pitch = cosf(e->pitch);
+	float	sin_pitch = sinf(e->pitch);
+
+	forward->x = sin_yaw;
+	forward->y = -sin_pitch * cos_yaw;
+	forward->z = cos_pitch * cos_yaw;
+
+	up->x = -cos_yaw * sin_roll;
+	up->y = -sin_pitch * sin_yaw * sin_roll + cos_pitch * cos_roll;
+	up->z = cos_pitch * sin_yaw * sin_roll + sin_pitch * cos_roll;
+
+	left->x = cos_yaw * cos_roll;
+	left->y = sin_pitch * sin_yaw * cos_roll + cos_pitch * sin_roll;
+	left->z = -cos_pitch * sin_yaw * cos_roll + sin_pitch * sin_roll;
+}
+
+#endif
diff --git a/modules/ray/ray_light_emitter.h b/modules/ray/ray_light_emitter.h
new file mode 100644
index 0000000..3b5509e
--- /dev/null
+++ b/modules/ray/ray_light_emitter.h
@@ -0,0 +1,18 @@
+#ifndef _RAY_LIGHT_EMITTER_H
+#define _RAY_LIGHT_EMITTER_H
+
+#include "ray_object_point.h"
+#include "ray_object_sphere.h"
+
+typedef enum ray_light_emitter_type_t {
+	RAY_LIGHT_EMITTER_TYPE_SPHERE,
+	RAY_LIGHT_EMITTER_TYPE_POINT,
+} ray_light_emitter_type_t;
+
+typedef union ray_light_emitter_t {
+	ray_light_emitter_type_t	type;
+	ray_object_sphere_t		sphere;
+	ray_object_point_t		point;
+} ray_light_emitter_t;
+
+#endif
diff --git a/modules/ray/ray_object.c b/modules/ray/ray_object.c
new file mode 100644
index 0000000..8b324e5
--- /dev/null
+++ b/modules/ray/ray_object.c
@@ -0,0 +1,67 @@
+#include "ray_object.h"
+#include "ray_object_light.h"
+#include "ray_object_plane.h"
+#include "ray_object_point.h"
+#include "ray_object_sphere.h"
+#include "ray_ray.h"
+#include "ray_scene.h"
+#include "ray_surface.h"
+
+
+/* Determine if a ray intersects object.
+ * If the object is intersected, store where along the ray the intersection occurs in res_distance.
+ */
+int ray_object_intersects_ray(ray_object_t *object, ray_ray_t *ray, float *res_distance)
+{
+	switch (object->type) {
+	case RAY_OBJECT_TYPE_SPHERE:
+		return ray_object_sphere_intersects_ray(&object->sphere, ray, res_distance);
+
+	case RAY_OBJECT_TYPE_POINT:
+		return ray_object_point_intersects_ray(&object->point, ray, res_distance);
+
+	case RAY_OBJECT_TYPE_PLANE:
+		return ray_object_plane_intersects_ray(&object->plane, ray, res_distance);
+
+	case RAY_OBJECT_TYPE_LIGHT:
+		return ray_object_light_intersects_ray(&object->light, ray, res_distance);
+	}
+}
+
+
+/* Return the surface normal of object @ point */
+ray_3f_t ray_object_normal(ray_object_t *object, ray_3f_t *point)
+{
+	switch (object->type) {
+	case RAY_OBJECT_TYPE_SPHERE:
+		return ray_object_sphere_normal(&object->sphere, point);
+
+	case RAY_OBJECT_TYPE_POINT:
+		return ray_object_point_normal(&object->point, point);
+
+	case RAY_OBJECT_TYPE_PLANE:
+		return ray_object_plane_normal(&object->plane, point);
+
+	case RAY_OBJECT_TYPE_LIGHT:
+		return ray_object_light_normal(&object->light, point);
+	}
+}
+
+
+/* Return the surface of object @ point */
+ray_surface_t ray_object_surface(ray_object_t *object, ray_3f_t *point)
+{
+	switch (object->type) {
+	case RAY_OBJECT_TYPE_SPHERE:
+		return ray_object_sphere_surface(&object->sphere, point);
+
+	case RAY_OBJECT_TYPE_POINT:
+		return ray_object_point_surface(&object->point, point);
+
+	case RAY_OBJECT_TYPE_PLANE:
+		return ray_object_plane_surface(&object->plane, point);
+
+	case RAY_OBJECT_TYPE_LIGHT:
+		return ray_object_light_surface(&object->light, point);
+	}
+}
diff --git a/modules/ray/ray_object.h b/modules/ray/ray_object.h
new file mode 100644
index 0000000..3dc27d1
--- /dev/null
+++ b/modules/ray/ray_object.h
@@ -0,0 +1,25 @@
+#ifndef _RAY_OBJECT_H
+#define _RAY_OBJECT_H
+
+#include "ray_object_light.h"
+#include "ray_object_plane.h"
+#include "ray_object_point.h"
+#include "ray_object_sphere.h"
+#include "ray_object_type.h"
+#include "ray_ray.h"
+#include "ray_scene.h"
+#include "ray_surface.h"
+
+typedef union ray_object_t {
+	ray_object_type_t	type;
+	ray_object_sphere_t	sphere;
+	ray_object_point_t	point;
+	ray_object_plane_t	plane;
+	ray_object_light_t	light;
+} ray_object_t;
+
+int ray_object_intersects_ray(ray_object_t *object, ray_ray_t *ray, float *res_distance);
+ray_3f_t ray_object_normal(ray_object_t *object, ray_3f_t *point);
+ray_surface_t ray_object_surface(ray_object_t *object, ray_3f_t *point);
+
+#endif
diff --git a/modules/ray/ray_object_light.h b/modules/ray/ray_object_light.h
new file mode 100644
index 0000000..73cf917
--- /dev/null
+++ b/modules/ray/ray_object_light.h
@@ -0,0 +1,60 @@
+#ifndef _RAY_OBJECT_LIGHT_H
+#define _RAY_OBJECT_LIGHT_H
+
+#include "ray_light_emitter.h"
+#include "ray_object_light.h"
+#include "ray_object_point.h"
+#include "ray_object_sphere.h"
+#include "ray_object_type.h"
+#include "ray_ray.h"
+#include "ray_scene.h"
+#include "ray_surface.h"
+
+
+typedef struct ray_object_light_t {
+	ray_object_type_t	type;
+	float			brightness;
+	ray_light_emitter_t	emitter;
+} ray_object_light_t;
+
+
+/* TODO: point is really the only one I've implemented... */
+static inline int ray_object_light_intersects_ray(ray_object_light_t *light, ray_ray_t *ray, float *res_distance)
+{
+	switch (light->emitter.type) {
+	case RAY_LIGHT_EMITTER_TYPE_POINT:
+		return ray_object_point_intersects_ray(&light->emitter.point, ray, res_distance);
+
+	case RAY_LIGHT_EMITTER_TYPE_SPHERE:
+		return ray_object_sphere_intersects_ray(&light->emitter.sphere, ray, res_distance);
+	}
+}
+
+
+static inline ray_3f_t ray_object_light_normal(ray_object_light_t *light, ray_3f_t *point)
+{
+	ray_3f_t	normal;
+
+	/* TODO */
+	switch (light->emitter.type) {
+	case RAY_LIGHT_EMITTER_TYPE_SPHERE:
+		return normal;
+
+	case RAY_LIGHT_EMITTER_TYPE_POINT:
+		return normal;
+	}
+}
+
+
+static inline ray_surface_t ray_object_light_surface(ray_object_light_t *light, ray_3f_t *point)
+{
+	switch (light->emitter.type) {
+	case RAY_LIGHT_EMITTER_TYPE_SPHERE:
+		return ray_object_sphere_surface(&light->emitter.sphere, point);
+
+	case RAY_LIGHT_EMITTER_TYPE_POINT:
+		return ray_object_point_surface(&light->emitter.point, point);
+	}
+}
+
+#endif
diff --git a/modules/ray/ray_object_plane.h b/modules/ray/ray_object_plane.h
new file mode 100644
index 0000000..96f204e
--- /dev/null
+++ b/modules/ray/ray_object_plane.h
@@ -0,0 +1,47 @@
+#ifndef _RAY_OBJECT_PLANE_H
+#define _RAY_OBJECT_PLANE_H
+
+#include "ray_object_type.h"
+#include "ray_ray.h"
+#include "ray_scene.h"
+#include "ray_surface.h"
+
+
+typedef struct ray_object_plane_t {
+	ray_object_type_t	type;
+	ray_surface_t		surface;
+	ray_3f_t		normal;
+	float			distance;
+} ray_object_plane_t;
+
+
+static inline int ray_object_plane_intersects_ray(ray_object_plane_t *plane, ray_ray_t *ray, float *res_distance)
+{
+	float	d = ray_3f_dot(&plane->normal, &ray->direction);
+
+	if (d != 0) {
+		float	distance = -(ray_3f_dot(&plane->normal, &ray->origin) + plane->distance) / d;
+
+		if (distance > 0) {
+			*res_distance = distance;
+
+			return 1;
+		}
+	}
+
+	return 0;
+}
+
+
+static inline ray_3f_t ray_object_plane_normal(ray_object_plane_t *plane, ray_3f_t *point)
+{
+	return plane->normal;
+}
+
+
+static inline ray_surface_t ray_object_plane_surface(ray_object_plane_t *plane, ray_3f_t *point)
+{
+	return plane->surface;
+}
+
+#endif
diff --git a/modules/ray/ray_object_point.h b/modules/ray/ray_object_point.h
new file mode 100644
index 0000000..5c83a68
--- /dev/null
+++ b/modules/ray/ray_object_point.h
@@ -0,0 +1,38 @@
+#ifndef _RAY_OBJECT_POINT_H
+#define _RAY_OBJECT_POINT_H
+
+#include "ray_3f.h"
+#include "ray_object_type.h"
+#include "ray_ray.h"
+#include "ray_scene.h"
+#include "ray_surface.h"
+
+
+typedef struct ray_object_point_t {
+	ray_object_type_t	type;
+	ray_surface_t		surface;
+	ray_3f_t		center;
+} ray_object_point_t;
+
+
+static inline int ray_object_point_intersects_ray(ray_object_point_t *point, ray_ray_t *ray, float *res_distance)
+{
+	/* TODO: determine a ray:point intersection */
+	return 0;
+}
+
+
+static inline ray_3f_t ray_object_point_normal(ray_object_point_t *point, ray_3f_t *_point)
+{
+	ray_3f_t	normal;
+
+	return normal;
+}
+
+
+static inline ray_surface_t ray_object_point_surface(ray_object_point_t *point, ray_3f_t *_point)
+{
+	return point->surface;
+}
+
+#endif
diff --git a/modules/ray/ray_object_sphere.h b/modules/ray/ray_object_sphere.h
new file mode 100644
index 0000000..6786bb7
--- /dev/null
+++ b/modules/ray/ray_object_sphere.h
@@ -0,0 +1,66 @@
+#ifndef _RAY_OBJECT_SPHERE_H
+#define _RAY_OBJECT_SPHERE_H
+
+#include <math.h>
+#include <stdio.h>
+
+#include "ray_3f.h"
+#include "ray_color.h"
+#include "ray_object_type.h"
+#include "ray_ray.h"
+#include "ray_scene.h"
+#include "ray_surface.h"
+
+
+typedef struct ray_object_sphere_t {
+	ray_object_type_t	type;
+	ray_surface_t		surface;
+	ray_3f_t		center;
+	float			radius;
+} ray_object_sphere_t;
+
+
+static inline int ray_object_sphere_intersects_ray(ray_object_sphere_t *sphere, ray_ray_t *ray, float *res_distance)
+{
+	ray_3f_t	v = ray_3f_sub(&ray->origin, &sphere->center);
+	float		b = ray_3f_dot(&v, &ray->direction);
+	float		disc = (sphere->radius * sphere->radius) - ray_3f_dot(&v, &v) + (b * b);
+
+	if (disc > 0) {
+		float	i1, i2;
+
+		disc = sqrtf(disc);
+
+		i1 = b - disc;
+		i2 = b + disc;
+
+		if (i2 > 0 && i1 > 0) {
+			*res_distance = i1;
+			return 1;
+		}
+	}
+
+	return 0;
+}
+
+
+/* return the normal of the surface at the specified point */
+static inline ray_3f_t ray_object_sphere_normal(ray_object_sphere_t *sphere, ray_3f_t *point)
+{
+	ray_3f_t	normal;
+	
+	normal = ray_3f_sub(point, &sphere->center);
+	normal = ray_3f_div_scalar(&normal, sphere->radius);	/* normalize without the sqrt() */
+
+	return normal;
+}
+
+
+/* return the surface of the sphere @ point */
+static inline ray_surface_t ray_object_sphere_surface(ray_object_sphere_t *sphere, ray_3f_t *point)
+{
+	/* uniform solids for now... */
+	return sphere->surface;
+}
+
+#endif
diff --git a/modules/ray/ray_object_type.h b/modules/ray/ray_object_type.h
new file mode 100644
index 0000000..6ce20f5
--- /dev/null
+++ b/modules/ray/ray_object_type.h
@@ -0,0 +1,11 @@
+#ifndef _RAY_OBJECT_TYPE_H
+#define _RAY_OBJECT_TYPE_H
+
+typedef enum ray_object_type_t {
+	RAY_OBJECT_TYPE_SPHERE,
+	RAY_OBJECT_TYPE_POINT,
+	RAY_OBJECT_TYPE_PLANE,
+	RAY_OBJECT_TYPE_LIGHT,
+} ray_object_type_t;
+
+#endif
diff --git a/modules/ray/ray_ray.h b/modules/ray/ray_ray.h
new file mode 100644
index 0000000..91469a2
--- /dev/null
+++ b/modules/ray/ray_ray.h
@@ -0,0 +1,11 @@
+#ifndef _RAY_RAY_H
+#define _RAY_RAY_H
+
+#include "ray_3f.h"
+
+typedef struct ray_ray_t {
+	ray_3f_t	origin;
+	ray_3f_t	direction;
+} ray_ray_t;
+
+#endif
diff --git a/modules/ray/ray_scene.c b/modules/ray/ray_scene.c
new file mode 100644
index 0000000..e44990b
--- /dev/null
+++ b/modules/ray/ray_scene.c
@@ -0,0 +1,188 @@
+#include <stdlib.h>
+#include <math.h>
+
+#include "fb.h"
+
+#include "ray_camera.h"
+#include "ray_color.h"
+#include "ray_object.h"
+#include "ray_ray.h"
+#include "ray_scene.h"
+#include "ray_threads.h"
+
+#define MAX_RECURSION_DEPTH	5
+
+
+static ray_color_t trace_ray(ray_scene_t *scene, ray_ray_t *ray, unsigned depth);
+
+
+/* Determine if the ray is obstructed by an object within the supplied distance, for shadows */
+static inline int ray_is_obstructed(ray_scene_t *scene, ray_ray_t *ray, float distance)
+{
+	unsigned	i;
+
+	for (i = 0; i < scene->n_objects; i++) {
+		float	ood;
+
+		if (scene->objects[i].type == RAY_OBJECT_TYPE_LIGHT)
+			continue;
+
+		if (ray_object_intersects_ray(&scene->objects[i], ray, &ood) &&
+		    ood < distance) {
+			return 1;
+		}
+	}
+
+	return 0;
+}
+
+
+/* Determine the color @ distance on ray on object viewed from origin */
+static inline ray_color_t shade_ray(ray_scene_t *scene, ray_ray_t *ray, ray_object_t *object, float distance, unsigned depth)
+{
+	ray_surface_t	surface;
+	ray_color_t	color;
+	ray_3f_t	rvec = ray_3f_mult_scalar(&ray->direction, distance);
+	ray_3f_t	intersection = ray_3f_sub(&ray->origin, &rvec);
+	ray_3f_t	normal = ray_object_normal(object, &intersection);
+	unsigned	i;
+
+	surface = ray_object_surface(object, &intersection);
+	color = ray_3f_mult_scalar(&scene->ambient_color, scene->ambient_brightness);
+	color = ray_3f_mult(&surface.color, &color);
+
+	/* visit lights for shadows and illumination */
+	for (i = 0; i < scene->n_lights; i++) {
+		ray_3f_t	lvec = ray_3f_sub(&scene->lights[i].light.emitter.point.center, &intersection);
+		float		ldist = ray_3f_length(&lvec);
+		float		lvec_normal_dot;
+		ray_ray_t	shadow_ray;
+
+		lvec = ray_3f_mult_scalar(&lvec, (1.0f / ldist)); /* normalize lvec */
+#if 1
+		/* skip this light if it's obstructed,
+		 * we must shift the origin slightly towards the light to prevent
+		 * spurious self-obstruction at the ray:object intersection */
+		/* negate the light vector so it's pointed at the light rather than from it */
+		shadow_ray.direction = ray_3f_negate(&lvec);
+		shadow_ray.origin = ray_3f_mult_scalar(&shadow_ray.direction, 0.00001f);
+		shadow_ray.origin = ray_3f_add(&shadow_ray.origin, &intersection);
+
+		if (ray_is_obstructed(scene, &shadow_ray, ldist))
+			continue;
+#endif
+		lvec_normal_dot = ray_3f_dot(&normal, &lvec);
+
+		if (lvec_normal_dot > 0) {
+#if 1
+			float		rvec_lvec_dot = ray_3f_dot(&ray->direction, &lvec);
+			ray_color_t	diffuse;
+			ray_color_t	specular;
+
+			diffuse = ray_3f_mult_scalar(&surface.color, lvec_normal_dot);
+			diffuse = ray_3f_mult_scalar(&diffuse, surface.diffuse);
+			color = ray_3f_add(&color, &diffuse);
+
+			/* FIXME: assumes light is a point for its color, and 20 is a constant "Phong exponent",
+			 * which should really be object/surface-specific
+			 */
+			specular = ray_3f_mult_scalar(&scene->lights[i].light.emitter.point.surface.color, powf(rvec_lvec_dot, 20));
+			specular = ray_3f_mult_scalar(&specular, surface.specular);
+			color = ray_3f_add(&color, &specular);
+#else
+			ray_color_t	diffuse;
+
+			diffuse = ray_3f_mult_scalar(&surface.color, lvec_normal_dot);
+			color = ray_3f_add(&color, &diffuse);
+#endif
+		}
+	}
+
+	/* generate a reflection ray */
+#if 1
+	float		dot = ray_3f_dot(&ray->direction, &normal);
+	ray_ray_t	reflected_ray = { .direction = ray_3f_mult_scalar(&normal, dot * 2.0f) };
+	ray_3f_t	reflection;
+
+	reflected_ray.origin = intersection;
+	reflected_ray.direction = ray_3f_sub(&ray->direction, &reflected_ray.direction);
+
+	reflection = trace_ray(scene, &reflected_ray, depth + 1);
+	reflection = ray_3f_mult_scalar(&reflection, surface.specular);
+	color = ray_3f_add(&color, &reflection);
+#endif
+
+	/* TODO: generate a refraction ray */
+
+	return color;
+}
+
+
+static ray_color_t trace_ray(ray_scene_t *scene, ray_ray_t *ray, unsigned depth)
+{
+	ray_object_t	*nearest_object = NULL;
+	float		nearest_object_distance = INFINITY;
+	ray_color_t	color = { .x = 0.0, .y = 0.0, .z = 0.0 };
+	unsigned	i;
+
+	depth++;
+	if (depth > MAX_RECURSION_DEPTH)
+		return color;
+
+	for (i = 0; i < scene->n_objects; i++) {
+		float	distance;
+
+		/* Does this ray intersect object? */
+		if (ray_object_intersects_ray(&scene->objects[i], ray, &distance)) {
+
+			/* Is it the nearest intersection? */
+			if (!nearest_object ||
+			    distance < nearest_object_distance) {
+				nearest_object = &scene->objects[i];
+				nearest_object_distance = distance;
+			}
+		}
+	}
+
+	if (nearest_object)
+		color = shade_ray(scene, ray, nearest_object, nearest_object_distance, depth);
+
+	depth--;
+
+	return color;
+}
+
+
+void ray_scene_render_fragment(ray_scene_t *scene, ray_camera_t *camera, fb_fragment_t *fragment)
+{
+	ray_camera_frame_t	frame;
+	ray_ray_t		ray;
+	uint32_t		*buf = fragment->buf;
+	unsigned		stride = fragment->stride / 4;
+
+	ray_camera_frame_begin(camera, fragment, &ray, &frame);
+	do {
+		do {
+			*buf = ray_color_to_uint32_rgb(trace_ray(scene, &ray, 0));
+			buf++;
+		} while (ray_camera_frame_x_step(&frame));
+
+		buf += stride;
+	} while (ray_camera_frame_y_step(&frame));
+}
+
+/* we expect fragments[threads->n_threads + 1], or fragments[1] when threads == NULL */
+void ray_scene_render_fragments(ray_scene_t *scene, ray_camera_t *camera, ray_threads_t *threads, fb_fragment_t *fragments)
+{
+	unsigned	n_threads = threads ? threads->n_threads + 1 : 1;
+	unsigned	i;
+
+	for (i = 1; i < n_threads; i++)
+		ray_thread_fragment_submit(&threads->threads[i - 1], scene, camera, &fragments[i]);
+
+	/* always render the zero fragment in-line */
+	ray_scene_render_fragment(scene, camera, &fragments[0]);
+
+	for (i = 1; i < n_threads; i++)
+		ray_thread_wait_idle(&threads->threads[i - 1]);
+}
diff --git a/modules/ray/ray_scene.h b/modules/ray/ray_scene.h
new file mode 100644
index 0000000..e9781c6
--- /dev/null
+++ b/modules/ray/ray_scene.h
@@ -0,0 +1,27 @@
+#ifndef _RAY_SCENE_H
+#define _RAY_SCENE_H
+
+#include "fb.h"
+
+#include "ray_camera.h"
+#include "ray_color.h"
+#include "ray_ray.h"
+#include "ray_threads.h"
+
+typedef union ray_object_t ray_object_t;
+
+typedef struct ray_scene_t {
+	ray_object_t	*objects;
+	unsigned	n_objects;
+
+	ray_object_t	*lights;
+	unsigned	n_lights;
+
+	ray_color_t	ambient_color;
+	float		ambient_brightness;
+} ray_scene_t;
+
+void ray_scene_render_fragments(ray_scene_t *scene, ray_camera_t *camera, ray_threads_t *threads, fb_fragment_t *fragments);
+void ray_scene_render_fragment(ray_scene_t *scene, ray_camera_t *camera, fb_fragment_t *fragment);
+
+#endif
diff --git a/modules/ray/ray_surface.h b/modules/ray/ray_surface.h
new file mode 100644
index 0000000..b3e3c68
--- /dev/null
+++ b/modules/ray/ray_surface.h
@@ -0,0 +1,14 @@
+#ifndef _RAY_MATERIAL_H
+#define _RAY_MATERIAL_H
+
+#include "ray_3f.h"
+#include "ray_color.h"
+
+/* Surface properties we expect every object to be able to introspect */
+typedef struct ray_surface_t {
+	ray_color_t	color;
+	float		specular;
+	float		diffuse;
+} ray_surface_t;
+
+#endif
diff --git a/modules/ray/ray_threads.c b/modules/ray/ray_threads.c
new file mode 100644
index 0000000..2369687
--- /dev/null
+++ b/modules/ray/ray_threads.c
@@ -0,0 +1,111 @@
+#include <pthread.h>
+#include <stdlib.h>
+
+#include "fb.h"
+
+#include "ray_scene.h"
+#include "ray_threads.h"
+
+#define BUSY_WAIT_NUM	1000000000	/* How much to spin before sleeping in pthread_cond_wait() */
+
+/* for now assuming x86 */
+#define cpu_relax() \
+	__asm__ __volatile__ ( "pause\n" : : : "memory")
+
+/* This is a very simple/naive implementation, there's certainly room for improvement.
+ *
+ * Without the BUSY_WAIT_NUM spinning this approach seems to leave a fairly
+ * substantial proportion of CPU idle while waiting for the render thread to
+ * complete on my core 2 duo.
+ *
+ * It's probably just latency in getting the render thread woken when the work
+ * is submitted, and since the fragments are split equally the main thread gets
+ * a head start and has to wait when it finishes first.  The spinning is just
+ * an attempt to avoid going to sleep while the render threads finish, there
+ * still needs to be improvement in how the work is submitted.
+ *
+ * I haven't spent much time on optimizing the raytracer yet.
+ */
+
+static void * ray_thread_func(void *_thread)
+{
+	ray_thread_t	*thread = _thread;
+
+	for (;;) {
+		pthread_mutex_lock(&thread->mutex);
+		while (thread->fragment == NULL)
+			pthread_cond_wait(&thread->cond, &thread->mutex);
+
+		ray_scene_render_fragment(thread->scene, thread->camera, thread->fragment);
+		thread->fragment = NULL;
+		pthread_mutex_unlock(&thread->mutex);
+		pthread_cond_signal(&thread->cond);
+	}
+
+	return NULL;
+}
+
+
+void ray_thread_fragment_submit(ray_thread_t *thread, ray_scene_t *scene, ray_camera_t *camera, fb_fragment_t *fragment)
+{
+	pthread_mutex_lock(&thread->mutex);
+	while (thread->fragment != NULL)	/* XXX: never true due to ray_thread_wait_idle() */
+		pthread_cond_wait(&thread->cond, &thread->mutex);
+
+	thread->fragment = fragment;
+	thread->scene = scene;
+	thread->camera = camera;
+
+	pthread_mutex_unlock(&thread->mutex);
+	pthread_cond_signal(&thread->cond);
+}
+
+
+void ray_thread_wait_idle(ray_thread_t *thread)
+{
+	unsigned	n;
+
+	/* Spin before going to sleep, the other thread should not take substantially longer. */
+	for (n = 0; thread->fragment != NULL && n < BUSY_WAIT_NUM; n++)
+		cpu_relax();
+
+	pthread_mutex_lock(&thread->mutex);
+	while (thread->fragment != NULL)
+		pthread_cond_wait(&thread->cond, &thread->mutex);
+	pthread_mutex_unlock(&thread->mutex);
+}
+
+
+ray_threads_t * ray_threads_create(unsigned num)
+{
+	ray_threads_t	*threads;
+	unsigned	i;
+
+	threads = malloc(sizeof(ray_threads_t) + sizeof(ray_thread_t) * num);
+	if (!threads)
+		return NULL;
+
+	for (i = 0; i < num; i++) {
+		pthread_mutex_init(&threads->threads[i].mutex, NULL);
+		pthread_cond_init(&threads->threads[i].cond, NULL);
+		threads->threads[i].fragment = NULL;
+		pthread_create(&threads->threads[i].thread, NULL, ray_thread_func, &threads->threads[i]);
+	}
+	threads->n_threads = num;
+
+	return threads;
+}
+
+
+void ray_threads_destroy(ray_threads_t *threads)
+{
+	unsigned	i;
+
+	for (i = 0; i < threads->n_threads; i++)
+		pthread_cancel(threads->threads[i].thread);
+
+	for (i = 0; i < threads->n_threads; i++)
+		pthread_join(threads->threads[i].thread, NULL);
+
+	free(threads);	
+}
diff --git a/modules/ray/ray_threads.h b/modules/ray/ray_threads.h
new file mode 100644
index 0000000..b4c601d
--- /dev/null
+++ b/modules/ray/ray_threads.h
@@ -0,0 +1,30 @@
+#ifndef _RAY_THREADS_H
+#define _RAY_THREADS_H
+
+#include <pthread.h>
+
+typedef struct ray_scene_t ray_scene_t;
+typedef struct ray_camera_t ray_camera_t;
+typedef struct fb_fragment_t fb_fragment_t;
+
+typedef struct ray_thread_t {
+	pthread_t	thread;
+	pthread_mutex_t	mutex;
+	pthread_cond_t	cond;
+	ray_scene_t	*scene;
+	ray_camera_t	*camera;
+	fb_fragment_t	*fragment;
+} ray_thread_t;
+
+typedef struct ray_threads_t {
+	unsigned	n_threads;
+	ray_thread_t	threads[];
+} ray_threads_t;
+
+
+ray_threads_t * ray_threads_create(unsigned num);
+void ray_threads_destroy(ray_threads_t *threads);
+
+void ray_thread_fragment_submit(ray_thread_t *thread, ray_scene_t *scene, ray_camera_t *camera, fb_fragment_t *fragment);
+void ray_thread_wait_idle(ray_thread_t *thread);
+#endif
diff --git a/modules/roto/roto.c b/modules/roto/roto.c
new file mode 100644
index 0000000..21e68b7
--- /dev/null
+++ b/modules/roto/roto.c
@@ -0,0 +1,231 @@
+#include <stdint.h>
+#include <inttypes.h>
+#include <math.h>
+
+#include "fb.h"
+#include "rototiller.h"
+
+/* Copyright (C) 2016 Vito Caputo <vcaputo@pengaru.com> */
+
+/* Some defines for the fixed-point stuff in render(). */
+#define FIXED_TRIG_LUT_SIZE	4096	/* size of the cos/sin look-up tables */
+#define FIXED_BITS		12	/* fractional bits */
+#define FIXED_EXP		4096	/* 2^FIXED_BITS */
+#define FIXED_COS(_rad)		costab[_rad % FIXED_TRIG_LUT_SIZE]
+#define FIXED_SIN(_rad)		sintab[_rad % FIXED_TRIG_LUT_SIZE]
+#define FIXED_MULT(_a, _b)	((_a * _b) >> FIXED_BITS)
+#define FIXED_NEW(_i)		(_i << FIXED_BITS)
+#define FIXED_TO_INT(_f)	((_f) >> FIXED_BITS)
+
+
+/* Draw a rotating checkered 256x256 texture into fragment. (32-bit version) */
+static void roto32(fb_fragment_t *fragment)
+{
+	static int32_t	costab[FIXED_TRIG_LUT_SIZE], sintab[FIXED_TRIG_LUT_SIZE];
+	static uint8_t	texture[256][256];
+	static int	initialized;
+	static uint32_t	colors[2];
+	static unsigned	r, rr;
+
+	int		y_cos_r, y_sin_r, x_cos_r, x_sin_r, x_cos_r_init, x_sin_r_init, cos_r, sin_r;
+	int		x, y, stride = fragment->stride / 4, width = fragment->width, height = fragment->height;
+	uint8_t		tx, ty; /* 256x256 texture; 8 bit texture indices to modulo via overflow. */
+	uint32_t	*buf = fragment->buf;
+
+	if (!initialized) {
+		int i;
+
+		initialized = 1;
+
+		/* Generate simple checker pattern texture, nothing clever, feel free to play! */
+		/* If you modify texture on every frame instead of only @ initialization you can
+		 * produce some neat output.  These values are indexed into colors[] below. */
+		for (y = 0; y < 128; y++) {
+			for (x = 0; x < 128; x++)
+				texture[y][x] = 1;
+			for (; x < 256; x++)
+				texture[y][x] = 0;
+		}
+		for (; y < 256; y++) {
+			for (x = 0; x < 128; x++)
+				texture[y][x] = 0;
+			for (; x < 256; x++)
+				texture[y][x] = 1;
+		}
+
+		/* Generate fixed-point cos & sin LUTs. */
+		for (i = 0; i < FIXED_TRIG_LUT_SIZE; i++) {
+			costab[i] = ((cos((double)2*M_PI*i/FIXED_TRIG_LUT_SIZE))*FIXED_EXP);
+			sintab[i] = ((sin((double)2*M_PI*i/FIXED_TRIG_LUT_SIZE))*FIXED_EXP);
+		}
+	}
+
+	/* This is all done using fixed-point in the hopes of being faster, and yes assumptions
+	 * are being made WRT the overflow of tx/ty as well, only tested on x86_64. */
+	cos_r = FIXED_COS(r);
+	sin_r = FIXED_SIN(r);
+
+	/* Vary the colors, this is just a mashup of sinusoidal rgb values. */
+	colors[0] =	((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr), FIXED_NEW(127))) + 128) << 16) |
+			((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr / 2), FIXED_NEW(127))) + 128) << 8) |
+			((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr / 3), FIXED_NEW(127))) + 128));
+
+	colors[1] =	((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr / 2), FIXED_NEW(127))) + 128) << 16) |
+			((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr / 2), FIXED_NEW(127))) + 128)) << 8 |
+			((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr), FIXED_NEW(127))) + 128) );
+
+	/* The dimensions are cut in half and negated to center the rotation. */
+	/* The [xy]_{sin,cos}_r variables are accumulators to replace multiplication with addition. */
+	x_cos_r_init = FIXED_MULT(-FIXED_NEW((width / 2)), cos_r);
+	x_sin_r_init = FIXED_MULT(-FIXED_NEW((width / 2)), sin_r);
+
+	y_cos_r = FIXED_MULT(-FIXED_NEW((height / 2)), cos_r);
+	y_sin_r = FIXED_MULT(-FIXED_NEW((height / 2)), sin_r);
+
+	for (y = 0; y < height; y++) {
+
+		x_cos_r = x_cos_r_init;
+		x_sin_r = x_sin_r_init;
+
+		for (x = 0; x < width; x++, buf++) {
+
+			tx = FIXED_TO_INT(x_sin_r - y_cos_r);
+			ty = FIXED_TO_INT(y_sin_r + x_cos_r);
+
+			*buf = colors[texture[ty][tx]];
+
+			x_cos_r += cos_r;
+			x_sin_r += sin_r;
+		}
+
+		buf += stride;
+		y_cos_r += cos_r;
+		y_sin_r += sin_r;
+	}
+
+	// This governs the rotation and color cycle.
+	r += FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr), FIXED_NEW(16)));
+	rr += 2;
+}
+
+
+/* Draw a rotating checkered 256x256 texture into fragment. (64-bit version) */
+static void roto64(fb_fragment_t *fragment)
+{
+	static int32_t	costab[FIXED_TRIG_LUT_SIZE], sintab[FIXED_TRIG_LUT_SIZE];
+	static uint8_t	texture[256][256];
+	static int	initialized;
+	static uint32_t	colors[2];
+	static unsigned	r, rr;
+
+	int		y_cos_r, y_sin_r, x_cos_r, x_sin_r, x_cos_r_init, x_sin_r_init, cos_r, sin_r;
+	int		x, y, stride = fragment->stride / 8, width = fragment->width, height = fragment->height;
+	uint8_t		tx, ty; /* 256x256 texture; 8 bit texture indices to modulo via overflow. */
+	uint64_t	*buf = (uint64_t *)fragment->buf;
+
+	if (!initialized) {
+		int i;
+
+		initialized = 1;
+
+		/* Generate simple checker pattern texture, nothing clever, feel free to play! */
+		/* If you modify texture on every frame instead of only @ initialization you can
+		 * produce some neat output.  These values are indexed into colors[] below. */
+		for (y = 0; y < 128; y++) {
+			for (x = 0; x < 128; x++)
+				texture[y][x] = 1;
+			for (; x < 256; x++)
+				texture[y][x] = 0;
+		}
+		for (; y < 256; y++) {
+			for (x = 0; x < 128; x++)
+				texture[y][x] = 0;
+			for (; x < 256; x++)
+				texture[y][x] = 1;
+		}
+
+		/* Generate fixed-point cos & sin LUTs. */
+		for (i = 0; i < FIXED_TRIG_LUT_SIZE; i++) {
+			costab[i] = ((cos((double)2*M_PI*i/FIXED_TRIG_LUT_SIZE))*FIXED_EXP);
+			sintab[i] = ((sin((double)2*M_PI*i/FIXED_TRIG_LUT_SIZE))*FIXED_EXP);
+		}
+	}
+
+	/* This is all done using fixed-point in the hopes of being faster, and yes assumptions
+	 * are being made WRT the overflow of tx/ty as well, only tested on x86_64. */
+	cos_r = FIXED_COS(r);
+	sin_r = FIXED_SIN(r);
+
+	/* Vary the colors, this is just a mashup of sinusoidal rgb values. */
+	colors[0] =	((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr), FIXED_NEW(127))) + 128) << 16) |
+			((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr / 2), FIXED_NEW(127))) + 128) << 8) |
+			((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr / 3), FIXED_NEW(127))) + 128));
+
+	colors[1] =	((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr / 2), FIXED_NEW(127))) + 128) << 16) |
+			((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr / 2), FIXED_NEW(127))) + 128)) << 8 |
+			((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr), FIXED_NEW(127))) + 128) );
+
+	/* The dimensions are cut in half and negated to center the rotation. */
+	/* The [xy]_{sin,cos}_r variables are accumulators to replace multiplication with addition. */
+	x_cos_r_init = FIXED_MULT(-FIXED_NEW((width / 2)), cos_r);
+	x_sin_r_init = FIXED_MULT(-FIXED_NEW((width / 2)), sin_r);
+
+	y_cos_r = FIXED_MULT(-FIXED_NEW((height / 2)), cos_r);
+	y_sin_r = FIXED_MULT(-FIXED_NEW((height / 2)), sin_r);
+
+	width /= 2;	/* Since we're processing 64-bit words (2 pixels) at a time */
+
+	for (y = 0; y < height; y++) {
+
+		x_cos_r = x_cos_r_init;
+		x_sin_r = x_sin_r_init;
+
+		for (x = 0; x < width; x++, buf++) {
+			uint64_t	p;
+
+			tx = FIXED_TO_INT(x_sin_r - y_cos_r);
+			ty = FIXED_TO_INT(y_sin_r + x_cos_r);
+
+			p = colors[texture[ty][tx]];
+
+			x_cos_r += cos_r;
+			x_sin_r += sin_r;
+
+			tx = FIXED_TO_INT(x_sin_r - y_cos_r);
+			ty = FIXED_TO_INT(y_sin_r + x_cos_r);
+
+			p |= (uint64_t)colors[texture[ty][tx]] << 32;
+			
+			*buf = p;
+
+			x_cos_r += cos_r;
+			x_sin_r += sin_r;
+		}
+
+		buf += stride;
+		y_cos_r += cos_r;
+		y_sin_r += sin_r;
+	}
+
+	// This governs the rotation and color cycle.
+	r += FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr), FIXED_NEW(16)));
+	rr += 2;
+}
+
+
+rototiller_renderer_t	roto32_renderer = {
+	.render = roto32,
+	.name = "roto32",
+	.description = "Tiled texture rotation (32-bit)",
+	.author = "Vito Caputo <vcaputo@pengaru.com>",
+	.license = "GPLv2",
+};
+
+
+rototiller_renderer_t	roto64_renderer = {
+	.render = roto64,
+	.name = "roto64",
+	.description = "Tiled texture rotation (64-bit)",
+	.author = "Vito Caputo <vcaputo@pengaru.com>",
+	.license = "GPLv2",
+};
diff --git a/modules/roto/roto.h b/modules/roto/roto.h
new file mode 100644
index 0000000..84a66a9
--- /dev/null
+++ b/modules/roto/roto.h
@@ -0,0 +1,9 @@
+#ifndef _ROTO_H
+#define _ROTO_H
+
+#include "fb.h"
+
+void roto64(fb_fragment_t *fragment);
+void roto32(fb_fragment_t *fragment);
+
+#endif
diff --git a/modules/sparkler/bsp.c b/modules/sparkler/bsp.c
new file mode 100644
index 0000000..6544993
--- /dev/null
+++ b/modules/sparkler/bsp.c
@@ -0,0 +1,556 @@
+#include <assert.h>
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+
+#include "bsp.h"
+
+
+/* octree-based bsp for faster proximity searches */
+/* meanings:
+ *  octrant = "octo" analog of a quadrant, an octree is a quadtree with an additional dimension (Z/3d)
+ *  bv = bounding volume
+ *  bsp = binary space partition
+ *  occupant = the things being indexed by the bsp (e.g. a particle, or its position)
+ */
+
+
+/* FIXME: these are not tuned at all, and should really all be parameters to bsp_new() instead */
+#define BSP_GROWBY		16
+#define BSP_MAX_OCCUPANTS	64
+#define BSP_MAX_DEPTH		16
+
+#define MAX(_a, _b)	(_a > _b ? _a : _b)
+#define MIN(_a, _b)	(_a < _b ? _a : _b)
+
+
+struct bsp_node_t {
+	v3f_t		center;		/* center point about which the bounding volume's 3d-space is divided */
+	bsp_node_t	*parent;	/* parent bounding volume, NULL when root node */
+	bsp_node_t	*octrants;	/* NULL when a leaf, otherwise an array of 8 bsp_node_t's */
+	list_head_t	occupants;	/* list of occupants in this volume when a leaf node */
+	unsigned	n_occupants;	/* number of ^^ */
+};
+
+
+struct bsp_t {
+	bsp_node_t	root;
+	list_head_t	free;
+};
+
+
+#define OCTRANTS					\
+	octrant(OCT_XL_YL_ZL, (1 << 2 | 1 << 1 | 1))	\
+	octrant(OCT_XR_YL_ZL, (         1 << 1 | 1))	\
+	octrant(OCT_XL_YR_ZL, (1 << 2          | 1))	\
+	octrant(OCT_XR_YR_ZL, (                  1))	\
+	octrant(OCT_XL_YL_ZR, (1 << 2 | 1 << 1    ))	\
+	octrant(OCT_XR_YL_ZR, (         1 << 1    ))	\
+	octrant(OCT_XL_YR_ZR, (1 << 2             ))	\
+	octrant(OCT_XR_YR_ZR, 0)
+
+
+#define octrant(_sym, _val)	_sym = _val,
+typedef enum _octrant_idx_t {
+	OCTRANTS
+} octrant_idx_t;
+#undef octrant
+
+
+
+static inline const char * octstr(octrant_idx_t oidx)
+{
+#define octrant(_sym, _val)	#_sym,
+	static const char	*octrant_strs[] = {
+					OCTRANTS
+				};
+#undef octrant
+
+	return octrant_strs[oidx];
+}
+
+
+static inline void _bsp_print(bsp_node_t *node)
+{
+	static int	depth = 0;
+
+	fprintf(stderr, "%-*s %i: %p\n", depth, " ", depth, node);
+	if (node->octrants) {
+		int	i;
+
+		for (i = 0; i < 8; i++) {
+			fprintf(stderr, "%-*s %i: %s: %p\n", depth, " ", depth, octstr(i), &node->octrants[i]);
+			depth++;
+			_bsp_print(&node->octrants[i]);
+			depth--;
+		}
+	}
+}
+
+
+/* Print a bsp tree to stderr (debugging) */
+void bsp_print(bsp_t *bsp)
+{
+	_bsp_print(&bsp->root);
+}
+
+
+/* Create a new bsp octree. */
+bsp_t * bsp_new(void)
+{
+	bsp_t	*bsp;
+
+	bsp = calloc(1, sizeof(bsp_t));
+	if (!bsp) {
+		return NULL;
+	}
+
+	INIT_LIST_HEAD(&bsp->root.occupants);
+	INIT_LIST_HEAD(&bsp->free);
+
+	return bsp;
+}
+
+
+/* Free a bsp octree */
+void bsp_free(bsp_t *bsp)
+{
+	/* TODO: free everything ... */
+	free(bsp);
+}
+
+
+/* bsp lookup state, encapsulated for preservation across composite
+ * lookup-dependent operations, so they can potentially avoid having
+ * to redo the lookup.  i.e. lookup caching.
+ */
+typedef struct _bsp_lookup_t {
+	int		depth;
+	v3f_t		left;
+	v3f_t		right;
+	bsp_node_t	*bv;
+	octrant_idx_t	oidx;
+} bsp_lookup_t;
+
+/* lookup a position's containing leaf node in the bsp tree, store resultant lookup state in *lookup_res */
+static inline void bsp_lookup_position(bsp_node_t *root, v3f_t *position, bsp_lookup_t *lookup_res)
+{
+	bsp_lookup_t	res = {
+				.bv = root,
+				.depth = 0,
+				.left = v3f_init(-1.0, -1.0, -1.0),	/* TODO: the bsp AABB should be supplied to bsp_new() */
+				.right = v3f_init(1.0, 1.0, 1.0),
+			};
+
+	while (res.bv->octrants) {
+		res.oidx = OCT_XR_YR_ZR;
+		if (position->x <= res.bv->center.x) {
+			res.oidx |= (1 << 2);
+			res.right.x = res.bv->center.x;
+		} else {
+			res.left.x = res.bv->center.x;
+		}
+
+		if (position->y <= res.bv->center.y) {
+			res.oidx |= (1 << 1);
+			res.right.y = res.bv->center.y;
+		} else {
+			res.left.y = res.bv->center.y;
+		}
+
+		if (position->z <= res.bv->center.z) {
+			res.oidx |= 1;
+			res.right.z = res.bv->center.z;
+		} else {
+			res.left.z = res.bv->center.z;
+		}
+
+		res.bv = &res.bv->octrants[res.oidx];
+		res.depth++;
+	}
+
+	*lookup_res = res;
+}
+
+
+/* Add an occupant to a bsp tree, use provided node lookup *l if supplied */
+static inline void _bsp_add_occupant(bsp_t *bsp, bsp_occupant_t *occupant, v3f_t *position, bsp_lookup_t *l)
+{
+	bsp_lookup_t	_lookup;
+
+	/* if no cached lookup result was provided, perform the lookup now. */
+	if (!l) {
+		l = &_lookup;
+		bsp_lookup_position(&bsp->root, position, l);
+	}
+
+	assert(l);
+	assert(l->bv);
+
+	occupant->position = position;
+
+#define map_occupant2octrant(_occupant, _bv, _octrant)		\
+		_octrant = OCT_XR_YR_ZR;			\
+		if (_occupant->position->x <= _bv->center.x) {	\
+			_octrant |= (1 << 2);			\
+		}						\
+		if (_occupant->position->y <= _bv->center.y) {	\
+			_octrant |= (1 << 1);			\
+		}						\
+		if (_occupant->position->z <= _bv->center.z) {	\
+			_octrant |= 1;				\
+		}
+
+	if (l->bv->n_occupants >= BSP_MAX_OCCUPANTS && l->depth < BSP_MAX_DEPTH) {
+		int		i;
+		list_head_t	*t, *_t;
+		bsp_node_t	*bv = l->bv;
+
+		/* bv is full and shallow enough, subdivide it. */
+
+		/* ensure the free list has something for us */
+		if (list_empty(&bsp->free)) {
+			bsp_node_t	*t;
+
+			/* TODO: does using the chunker instead make sense here? */
+			t = calloc(sizeof(bsp_node_t), 8 * BSP_GROWBY);
+			for (i = 0; i < 8 * BSP_GROWBY; i += 8) {
+				list_add(&t[i].occupants, &bsp->free);
+			}
+		}
+
+		/* take an octrants array from the free list */
+		bv->octrants = list_entry(bsp->free.next, bsp_node_t, occupants);
+		list_del(&bv->octrants[0].occupants);
+
+		/* initialize the octrants */
+		for (i = 0; i < 8; i++) {
+			INIT_LIST_HEAD(&bv->octrants[i].occupants);
+			bv->octrants[i].n_occupants = 0;
+			bv->octrants[i].parent = bv;
+			bv->octrants[i].octrants = NULL;
+		}
+
+		/* set the center point in each octrant which places the partitioning hyperplane */
+		/* XXX: note this is pretty unreadable due to reusing the earlier computed values
+		 * where the identical computation is required.
+		 */
+		bv->octrants[OCT_XR_YR_ZR].center.x = (l->right.x - bv->center.x) / 2 + bv->center.x;
+		bv->octrants[OCT_XR_YR_ZR].center.y = (l->right.y - bv->center.y) / 2 + bv->center.y;
+		bv->octrants[OCT_XR_YR_ZR].center.z = (l->right.z - bv->center.z) / 2 + bv->center.z;
+
+		bv->octrants[OCT_XR_YR_ZL].center.x = bv->octrants[OCT_XR_YR_ZR].center.x;
+		bv->octrants[OCT_XR_YR_ZL].center.y = bv->octrants[OCT_XR_YR_ZR].center.y;
+		bv->octrants[OCT_XR_YR_ZL].center.z = (bv->center.z - l->left.z) / 2 + l->left.z;
+
+		bv->octrants[OCT_XR_YL_ZR].center.x = bv->octrants[OCT_XR_YR_ZR].center.x;
+		bv->octrants[OCT_XR_YL_ZR].center.y = (bv->center.y - l->left.y) / 2 + l->left.y;
+		bv->octrants[OCT_XR_YL_ZR].center.z = bv->octrants[OCT_XR_YR_ZR].center.z;
+
+		bv->octrants[OCT_XR_YL_ZL].center.x = bv->octrants[OCT_XR_YR_ZR].center.x;
+		bv->octrants[OCT_XR_YL_ZL].center.y = bv->octrants[OCT_XR_YL_ZR].center.y;
+		bv->octrants[OCT_XR_YL_ZL].center.z = bv->octrants[OCT_XR_YR_ZL].center.z;
+
+		bv->octrants[OCT_XL_YR_ZR].center.x = (bv->center.x - l->left.x) / 2 + l->left.x;
+		bv->octrants[OCT_XL_YR_ZR].center.y = bv->octrants[OCT_XR_YR_ZR].center.y;
+		bv->octrants[OCT_XL_YR_ZR].center.z = bv->octrants[OCT_XR_YR_ZR].center.z;
+
+		bv->octrants[OCT_XL_YR_ZL].center.x = bv->octrants[OCT_XL_YR_ZR].center.x;
+		bv->octrants[OCT_XL_YR_ZL].center.y = bv->octrants[OCT_XR_YR_ZR].center.y;
+		bv->octrants[OCT_XL_YR_ZL].center.z = bv->octrants[OCT_XR_YR_ZL].center.z;
+
+		bv->octrants[OCT_XL_YL_ZR].center.x = bv->octrants[OCT_XL_YR_ZR].center.x;
+		bv->octrants[OCT_XL_YL_ZR].center.y = bv->octrants[OCT_XR_YL_ZR].center.y;
+		bv->octrants[OCT_XL_YL_ZR].center.z = bv->octrants[OCT_XR_YR_ZR].center.z;
+
+		bv->octrants[OCT_XL_YL_ZL].center.x = bv->octrants[OCT_XL_YR_ZR].center.x;
+		bv->octrants[OCT_XL_YL_ZL].center.y = bv->octrants[OCT_XR_YL_ZR].center.y;
+		bv->octrants[OCT_XL_YL_ZL].center.z = bv->octrants[OCT_XR_YR_ZL].center.z;
+
+		/* migrate the occupants into the appropriate octrants */
+		list_for_each_safe(t, _t, &bv->occupants) {
+			octrant_idx_t	oidx;
+			bsp_occupant_t	*o = list_entry(t, bsp_occupant_t, occupants);
+
+			map_occupant2octrant(o, bv, oidx);
+			list_move(t, &bv->octrants[oidx].occupants);
+			o->leaf = &bv->octrants[oidx];
+			bv->octrants[oidx].n_occupants++;
+		}
+		bv->n_occupants = 0;
+
+		/* a new leaf assumes the bv position for the occupant to be added into */
+		map_occupant2octrant(occupant, bv, l->oidx);
+		l->bv = &bv->octrants[l->oidx];
+		l->depth++;
+	}
+
+#undef map_occupant2octrant
+
+	occupant->leaf = l->bv;
+	list_add(&occupant->occupants, &l->bv->occupants);
+	l->bv->n_occupants++;
+
+	assert(occupant->leaf);
+}
+
+
+/* add an occupant to a bsp tree */
+void bsp_add_occupant(bsp_t *bsp, bsp_occupant_t *occupant, v3f_t *position)
+{
+	_bsp_add_occupant(bsp, occupant, position, NULL);
+}
+
+
+/* Delete an occupant from a bsp tree.
+ * Set reservation to prevent potentially freeing a node made empty by our delete that
+ * we have a reference to (i.e. a cached lookup result, see bsp_move_occupant()).
+ */
+static inline void _bsp_delete_occupant(bsp_t *bsp, bsp_occupant_t *occupant, bsp_node_t *reservation)
+{
+	if (occupant->leaf->octrants) {
+		fprintf(stderr, "BUG: deleting occupant(%p) from non-leaf bv(%p)\n", occupant, occupant->leaf);
+	}
+
+	/* delete the occupant */
+	list_del(&occupant->occupants);
+	occupant->leaf->n_occupants--;
+
+	if (list_empty(&occupant->leaf->occupants)) {
+		bsp_node_t	*parent_bv;
+
+		if (occupant->leaf->n_occupants) {
+			fprintf(stderr, "BUG: bv_occupants empty but n_occupants=%u\n", occupant->leaf->n_occupants);
+		}
+
+		/* leaf is now empty, since nodes are allocated as clusters of 8, they aren't freed unless all nodes in the cluster are empty.
+		 * Determine if they're all empty, and free the parent's octrants as a set.
+		 * Repeat this process up the chain of parents, repeatedly converting empty parents into leaf nodes.
+		 * TODO: maybe just use the chunker instead?
+		 */
+
+		for (parent_bv = occupant->leaf->parent; parent_bv && parent_bv != reservation; parent_bv = parent_bv->parent) {
+			int	i;
+
+			/* are _all_ the parent's octrants freeable? */
+			for (i = 0; i < 8; i++) {
+				if (&parent_bv->octrants[i] == reservation ||
+				    parent_bv->octrants[i].octrants ||
+				    !list_empty(&parent_bv->octrants[i].occupants)) {
+					goto _out;
+				}
+			}
+
+			/* "freeing" really just entails putting the octrants cluster of nodes onto the free list */
+			list_add(&parent_bv->octrants[0].occupants, &bsp->free);
+			parent_bv->octrants = NULL;
+		}
+	}
+
+_out:
+	occupant->leaf = NULL;
+}
+
+
+/* Delete an occupant from a bsp tree. */
+void bsp_delete_occupant(bsp_t *bsp, bsp_occupant_t *occupant)
+{
+	_bsp_delete_occupant(bsp, occupant, NULL);
+}
+
+
+/* Move an occupant within a bsp tree to a new position */
+void bsp_move_occupant(bsp_t *bsp, bsp_occupant_t *occupant, v3f_t *position)
+{
+	bsp_lookup_t	lookup_res;
+
+	if (v3f_equal(occupant->position, position)) {
+		return;
+	}
+
+	bsp_lookup_position(&bsp->root, occupant->position, &lookup_res);
+	if (lookup_res.bv == occupant->leaf) {
+		/* leaf unchanged, do nothing past lookup. */
+		occupant->position = position;
+		return;
+	}
+
+	_bsp_delete_occupant(bsp, occupant, lookup_res.bv);
+	_bsp_add_occupant(bsp, occupant, position, &lookup_res);
+}
+
+
+static inline float square(float v)
+{
+	return v * v;
+}
+
+
+typedef enum overlaps_t {
+	OVERLAPS_NONE,		/* objects are completely separated */
+	OVERLAPS_PARTIALLY,	/* objects surfaces one another */
+	OVERLAPS_A_IN_B,	/* first object is fully within the second */
+	OVERLAPS_B_IN_A,	/* second object is fully within the first */
+} overlaps_t;
+
+
+/* Returns wether the axis-aligned bounding box (AABB) overlaps the sphere.
+ * Absolute vs. partial overlaps are distinguished, since it's an important optimization
+ * to know if the sphere falls entirely within one partition of the octree.
+ */
+static inline overlaps_t aabb_overlaps_sphere(v3f_t *aabb_min, v3f_t *aabb_max, v3f_t *sphere_center, float sphere_radius)
+{
+	/* This implementation is based on James Arvo's from Graphics Gems pg. 335 */
+	float	r2 = square(sphere_radius);
+	float	dface = INFINITY;
+	float	dmin = 0;
+	float	dmax = 0;
+	float	a, b;
+
+#define per_dimension(_center, _box_max, _box_min)		\
+	a = square(_center - _box_min);				\
+	b = square(_center - _box_max);				\
+								\
+	dmax += MAX(a, b);					\
+	if (_center >= _box_min && _center <= _box_max) {	\
+		/* sphere center within box */			\
+		dface = MIN(dface, MIN(a, b));			\
+	} else {						\
+		/* sphere center outside the box */		\
+		dface = 0;					\
+		dmin += MIN(a, b);				\
+	}
+
+	per_dimension(sphere_center->x, aabb_max->x, aabb_min->x);
+	per_dimension(sphere_center->y, aabb_max->y, aabb_min->y);
+	per_dimension(sphere_center->z, aabb_max->z, aabb_min->z);
+
+	if (dmax < r2) {
+		/* maximum distance to box smaller than radius, box is inside
+		 * the sphere */
+		return OVERLAPS_A_IN_B;
+	}
+
+	if (dface > r2) {
+		/* sphere center is within box (non-zero dface), and dface is
+		 * greater than sphere diameter, sphere is inside the box. */
+		return OVERLAPS_B_IN_A;
+	}
+
+	if (dmin <= r2) {
+		/* minimum distance from sphere center to box is smaller than
+		 * sphere's radius, surfaces intersect */
+		return OVERLAPS_PARTIALLY;
+	}
+
+	return OVERLAPS_NONE;
+}
+
+
+typedef struct bsp_search_sphere_t {
+	v3f_t	*center;
+	float	radius_min;
+	float	radius_max;
+	void	(*cb)(bsp_t *, list_head_t *, void *);
+	void	*cb_data;
+} bsp_search_sphere_t;
+
+
+static overlaps_t _bsp_search_sphere(bsp_t *bsp, bsp_node_t *node, bsp_search_sphere_t *search, v3f_t *aabb_min, v3f_t *aabb_max)
+{
+	overlaps_t	res;
+	v3f_t		oaabb_min, oaabb_max;
+
+	/* if the radius_max search doesn't overlap aabb_min:aabb_max at all, simply return. */
+	res = aabb_overlaps_sphere(aabb_min, aabb_max, search->center, search->radius_max);
+	if (res == OVERLAPS_NONE) {
+		return res;
+	}
+
+	/* if the radius_max absolutely overlaps the AABB, we must see if the AABB falls entirely within radius_min so we can skip it. */
+	if (res == OVERLAPS_A_IN_B) {
+		res = aabb_overlaps_sphere(aabb_min, aabb_max, search->center, search->radius_min);
+		if (res == OVERLAPS_A_IN_B) {
+			/* AABB is entirely within radius_min, skip it. */
+			return OVERLAPS_NONE;
+		}
+
+		if (res == OVERLAPS_NONE) {
+			/* radius_min didn't overlap, radius_max overlapped aabb 100%, it's entirely within the range. */
+			res = OVERLAPS_A_IN_B;
+		} else {
+			/* radius_min overlapped partially.. */
+			res = OVERLAPS_PARTIALLY;
+		}
+	}
+
+	/* if node is a leaf, call search->cb with the occupants, then return. */
+	if (!node->octrants) {
+		search->cb(bsp, &node->occupants, search->cb_data);
+		return res;
+	}
+
+	/* node is a parent, recur on each octrant with appropriately adjusted aabb_min:aabb_max values */
+	/* if any of the octrants absolutely overlaps the search sphere, skip the others by returning. */
+#define search_octrant(_oid, _aabb_min, _aabb_max)						\
+	res = _bsp_search_sphere(bsp, &node->octrants[_oid], search, _aabb_min, _aabb_max);	\
+	if (res == OVERLAPS_B_IN_A) {								\
+		return res;									\
+	}
+
+	/* OCT_XL_YL_ZL and OCT_XR_YR_ZR AABBs don't require tedious composition */
+	search_octrant(OCT_XL_YL_ZL, aabb_min, &node->center);
+	search_octrant(OCT_XR_YR_ZR, &node->center, aabb_max);
+
+	/* the rest are stitched together requiring temp storage and tedium */
+	v3f_set(&oaabb_min, node->center.x, aabb_min->y, aabb_min->z);
+	v3f_set(&oaabb_max, aabb_max->x, node->center.y, node->center.z);
+	search_octrant(OCT_XR_YL_ZL, &oaabb_min, &oaabb_max);
+
+	v3f_set(&oaabb_min, aabb_min->x, node->center.y, aabb_min->z);
+	v3f_set(&oaabb_max, node->center.x, aabb_max->y, node->center.z);
+	search_octrant(OCT_XL_YR_ZL, &oaabb_min, &oaabb_max);
+
+	v3f_set(&oaabb_min, node->center.x, node->center.y, aabb_min->z);
+	v3f_set(&oaabb_max, aabb_max->x, aabb_max->y, node->center.z);
+	search_octrant(OCT_XR_YR_ZL, &oaabb_min, &oaabb_max);
+
+	v3f_set(&oaabb_min, aabb_min->x, aabb_min->y, node->center.z);
+	v3f_set(&oaabb_max, node->center.x, node->center.y, aabb_max->z);
+	search_octrant(OCT_XL_YL_ZR, &oaabb_min, &oaabb_max);
+
+	v3f_set(&oaabb_min, node->center.x, aabb_min->y, node->center.z);
+	v3f_set(&oaabb_max, aabb_max->x, node->center.y, aabb_max->z);
+	search_octrant(OCT_XR_YL_ZR, &oaabb_min, &oaabb_max);
+
+	v3f_set(&oaabb_min, aabb_min->x, node->center.y, node->center.z);
+	v3f_set(&oaabb_max, node->center.x, aabb_max->y, aabb_max->z);
+	search_octrant(OCT_XL_YR_ZR, &oaabb_min, &oaabb_max);
+
+#undef search_octrant
+
+	/* since early on an OVERLAPS_NONE short-circuits the function, and
+	 * OVERLAPS_ABSOLUTE also causes short-circuits, if we arrive here it's
+	 * a partial overlap
+	 */
+	return OVERLAPS_PARTIALLY;
+}
+
+
+/* search the bsp tree for leaf nodes which intersect the space between radius_min and radius_max of a sphere @ center */
+/* for every leaf node found to intersect the sphere, cb is called with the leaf node's occupants list head */
+/* the callback cb must then further filter the occupants as necessary. */
+void bsp_search_sphere(bsp_t *bsp, v3f_t *center, float radius_min, float radius_max, void (*cb)(bsp_t *, list_head_t *, void *), void *cb_data)
+{
+	bsp_search_sphere_t	search = {
+					.center = center,
+					.radius_min = radius_min,
+					.radius_max = radius_max,
+					.cb = cb,
+					.cb_data = cb_data,
+				};
+	v3f_t			aabb_min = v3f_init(-1.0f, -1.0f, -1.0f);
+	v3f_t			aabb_max = v3f_init(1.0f, 1.0f, 1.0f);
+
+	_bsp_search_sphere(bsp, &bsp->root, &search, &aabb_min, &aabb_max);
+}
diff --git a/modules/sparkler/bsp.h b/modules/sparkler/bsp.h
new file mode 100644
index 0000000..f5ce303
--- /dev/null
+++ b/modules/sparkler/bsp.h
@@ -0,0 +1,28 @@
+#ifndef _BSP_H
+#define _BSP_H
+
+#include <stdint.h>
+
+#include "list.h"
+#include "v3f.h"
+
+typedef struct bsp_t bsp_t;
+typedef struct bsp_node_t bsp_node_t;
+
+/* Embed this in anything you want spatially indexed by the bsp tree. */
+/* TODO: it would be nice to make this opaque, but it's a little annoying. */
+typedef struct bsp_occupant_t {
+	bsp_node_t	*leaf;		/* leaf node containing this occupant */
+	list_head_t	occupants;	/* node on containing leaf node's list of occupants */
+	v3f_t		*position;	/* position of occupant to be partitioned */
+} bsp_occupant_t;
+
+bsp_t * bsp_new(void);
+void bsp_free(bsp_t *bsp);
+void bsp_print(bsp_t *bsp);
+void bsp_add_occupant(bsp_t *bsp, bsp_occupant_t *occupant, v3f_t *position);
+void bsp_delete_occupant(bsp_t *bsp, bsp_occupant_t *occupant);
+void bsp_move_occupant(bsp_t *bsp, bsp_occupant_t *occupant, v3f_t *position);
+void bsp_search_sphere(bsp_t *bsp, v3f_t *center, float radius_min, float radius_max, void (*cb)(bsp_t *, list_head_t *, void *), void *cb_data);
+
+#endif
diff --git a/modules/sparkler/burst.c b/modules/sparkler/burst.c
new file mode 100644
index 0000000..828ca02
--- /dev/null
+++ b/modules/sparkler/burst.c
@@ -0,0 +1,111 @@
+#include <stdlib.h>
+
+#include "bsp.h"
+#include "container.h"
+#include "particle.h"
+#include "particles.h"
+
+
+/* a "burst" (shockwave) particle type */
+/* this doesn't draw anything, it just pushes neighbors away in an increasing radius */
+
+#define BURST_FORCE		0.01f
+#define BURST_MAX_LIFETIME	8
+
+typedef struct _burst_ctxt_t {
+	int	longevity;
+	int	lifetime;
+} burst_ctxt_t;
+
+
+static int burst_init(particles_t *particles, particle_t *p)
+{
+	burst_ctxt_t	*ctxt = p->ctxt;
+
+	ctxt->longevity = ctxt->lifetime = BURST_MAX_LIFETIME;
+	p->props->velocity = 0; /* burst should be stationary */
+	p->props->mass = 0; /* no mass prevents gravity's effects */
+
+	return 1;
+}
+
+
+static inline void thrust_part(particle_t *burst, particle_t *victim, float distance_sq)
+{
+	v3f_t	direction = v3f_sub(&victim->props->position, &burst->props->position);
+
+	/* TODO: normalize is expensive, see about removing these. */
+	direction = v3f_normalize(&direction);
+	victim->props->direction = v3f_add(&victim->props->direction, &direction);
+	victim->props->direction = v3f_normalize(&victim->props->direction);
+
+	victim->props->velocity += BURST_FORCE;
+}
+
+
+typedef struct burst_sphere_t {
+	particle_t	*center;
+	float		radius_min;
+	float		radius_max;
+} burst_sphere_t;
+
+
+static void burst_cb(bsp_t *bsp, list_head_t *occupants, void *_s)
+{
+	burst_sphere_t	*s = _s;
+	bsp_occupant_t	*o;
+	float		rmin_sq = s->radius_min * s->radius_min;
+	float		rmax_sq = s->radius_max * s->radius_max;
+
+	/* XXX: to avoid having a callback per-particle, bsp_occupant_t was
+	 * moved to the public particle, and the particle-specific
+	 * implementations directly perform bsp-accelerated searches.  Another
+	 * wart caused by this is particles_bsp().
+	 */
+	list_for_each_entry(o, occupants, occupants) {
+		particle_t	*p = container_of(o, particle_t, occupant);
+		float		d_sq;
+		
+		if (p == s->center) {
+			/* leave ourselves alone */
+			continue;
+		}
+
+		d_sq = v3f_distance_sq(&s->center->props->position, &p->props->position);
+
+		if (d_sq > rmin_sq && d_sq < rmax_sq) {
+			/* displace the part relative to the burst origin */
+			thrust_part(s->center, p, d_sq);
+		}
+
+	}
+}
+
+
+static particle_status_t burst_sim(particles_t *particles, particle_t *p)
+{
+	burst_ctxt_t	*ctxt = p->ctxt;
+	bsp_t		*bsp = particles_bsp(particles);	/* XXX see note above about bsp_occupant_t */
+	burst_sphere_t	s;
+
+	if (!ctxt->longevity || (ctxt->longevity--) <= 0) {
+		return PARTICLE_DEAD;
+	}
+
+	/* affect neighbors for the shock-wave */
+	s.radius_min = (1.0f - ((float)ctxt->longevity / ctxt->lifetime)) * 0.075f;
+	s.radius_max = s.radius_min + .01f;
+	s.center = p;
+	bsp_search_sphere(bsp, &p->props->position, s.radius_min, s.radius_max, burst_cb, &s);
+
+	return PARTICLE_ALIVE;
+}
+
+
+particle_ops_t	burst_ops = {
+			.context_size = sizeof(burst_ctxt_t),
+			.sim = burst_sim,
+			.init = burst_init,
+			.draw = NULL,
+			.cleanup = NULL,
+		};
diff --git a/modules/sparkler/chunker.c b/modules/sparkler/chunker.c
new file mode 100644
index 0000000..ca072eb
--- /dev/null
+++ b/modules/sparkler/chunker.c
@@ -0,0 +1,225 @@
+#include <assert.h>
+#include <stddef.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <string.h>
+
+#include "chunker.h"
+#include "container.h"
+#include "list.h"
+
+/* Everything associated with the particles tends to be short-lived.
+ *
+ * They come and go frequently in large numbers.  This implements a very basic
+ * chunked allocator which prioritizes efficient allocation and freeing over
+ * low waste of memory.  We malloc chunks at a time, doling out elements from
+ * the chunk sequentially as requested until the chunk is cannot fulfill an
+ * allocation, then we just retire the chunk, add a new chunk and continue.
+ *
+ * When allocations are freed, we simply decrement the refcount for its chunk,
+ * leaving the chunk pinned with holes accumulating until its refcount reaches
+ * zero, at which point the chunk is made available for allocations again.
+ *
+ * This requires a reference to the chunk be returned with every allocation.
+ * It may be possible to reduce the footprint of this by using a relative
+ * offset to the chunk start instead, but that would probably be more harmful
+ * to the alignment.
+ *
+ * This has some similarities to a slab allocator...
+ */
+
+#define CHUNK_ALIGNMENT			8192	/* XXX: this may be unnecessary, callers should be able to ideally size their chunkers */
+#define ALLOC_ALIGNMENT			8	/* allocations within the chunk need to be aligned since their size affects subsequent allocation offsets */
+#define ALIGN(_size, _alignment)	(((_size) + _alignment - 1) & ~(_alignment - 1))
+
+typedef struct chunk_t {
+	chunker_t	*chunker;	/* chunker chunk belongs to */
+	list_head_t	chunks;		/* node on free/pinned list */
+	uint32_t	n_refs;		/* number of references (allocations) to this chunk */
+	unsigned	next_offset;	/* next available offset for allocation */
+	uint8_t		mem[];		/* usable memory from this chunk */
+} chunk_t;
+
+typedef struct allocation_t {
+	chunk_t		*chunk;		/* chunk this allocation came from */
+	uint8_t		mem[];		/* usable memory from this allocation */
+} allocation_t;
+
+struct chunker_t {
+	chunk_t		*chunk;		/* current chunk allocations come from */
+	unsigned	chunk_size;	/* size chunks are allocated in */
+	list_head_t	free_chunks;	/* list of completely free chunks */
+	list_head_t	pinned_chunks;	/* list of chunks pinned because they have an outstanding allocation */
+};
+
+
+/* Add a reference to a chunk. */
+static inline void chunk_ref(chunk_t *chunk)
+{
+	assert(chunk);
+	assert(chunk->chunker);
+
+	chunk->n_refs++;
+
+	assert(chunk->n_refs != 0);
+}
+
+
+/* Remove reference from a chunk, move to free list when no references remain. */
+static inline void chunk_unref(chunk_t *chunk)
+{
+	assert(chunk);
+	assert(chunk->chunker);
+	assert(chunk->n_refs > 0);
+
+	chunk->n_refs--;
+	if (chunk->n_refs == 0) {
+		list_move(&chunk->chunks, &chunk->chunker->free_chunks);
+	}
+}
+
+
+/* Return allocated size of the chunk */
+static inline unsigned chunk_alloc_size(chunker_t *chunker)
+{
+	assert(chunker);
+
+	return (sizeof(chunk_t) + chunker->chunk_size);
+}
+
+
+/* Get a new working chunk, retiring and replacing chunker->chunk. */
+static void chunker_new_chunk(chunker_t *chunker)
+{
+	chunk_t	*chunk;
+
+	assert(chunker);
+
+	if (chunker->chunk) {
+		chunk_unref(chunker->chunk);
+		chunker->chunk = NULL;
+	}
+
+	if (!list_empty(&chunker->free_chunks)) {
+		chunk = list_entry(chunker->free_chunks.next, chunk_t, chunks);
+		list_del(&chunk->chunks);
+	} else {
+		/* No free chunks, must ask libc for memory */
+		chunk = malloc(chunk_alloc_size(chunker));
+	}
+
+	/* Note a chunk is pinned from the moment it's created, and a reference
+	 * is added to represent chunker->chunk, even though no allocations
+	 * occurred yet.
+	 */
+	chunk->n_refs = 1;
+	chunk->next_offset = 0;
+	chunk->chunker = chunker;
+	chunker->chunk = chunk;
+	list_add(&chunk->chunks, &chunker->pinned_chunks);
+}
+
+
+/* Create a new chunker. */
+chunker_t * chunker_new(unsigned chunk_size)
+{
+	chunker_t	*chunker;
+
+	chunker = calloc(1, sizeof(chunker_t));
+	if (!chunker) {
+		return NULL;
+	}
+
+	INIT_LIST_HEAD(&chunker->free_chunks);
+	INIT_LIST_HEAD(&chunker->pinned_chunks);
+
+	/* XXX: chunker->chunk_size does not include the size of the chunk_t container */
+	chunker->chunk_size = ALIGN(chunk_size, CHUNK_ALIGNMENT);
+
+	return chunker;
+}
+
+
+/* Allocate non-zeroed memory from a chunker. */
+void * chunker_alloc(chunker_t *chunker, unsigned size)
+{
+	allocation_t	*allocation;
+
+	assert(chunker);
+	assert(size <= chunker->chunk_size);
+
+	size = ALIGN(sizeof(allocation_t) + size, ALLOC_ALIGNMENT);
+
+	if (!chunker->chunk || size + chunker->chunk->next_offset > chunker->chunk_size) {
+		/* Retire this chunk, time for a new one */
+		chunker_new_chunk(chunker);
+	}
+
+	if (!chunker->chunk) {
+		return NULL;
+	}
+
+	chunk_ref(chunker->chunk);
+	allocation = (allocation_t *)&chunker->chunk->mem[chunker->chunk->next_offset];
+	chunker->chunk->next_offset += size;
+	allocation->chunk = chunker->chunk;
+
+	assert(chunker->chunk->next_offset <= chunker->chunk_size);
+
+	return allocation->mem;
+}
+
+
+/* Free memory allocated from a chunker. */
+void chunker_free(void *ptr)
+{
+	allocation_t	*allocation = container_of(ptr, allocation_t, mem);
+
+	assert(ptr);
+
+	chunk_unref(allocation->chunk);
+}
+
+
+/* Free a chunker and it's associated allocations. */
+void chunker_free_chunker(chunker_t *chunker)
+{
+	chunk_t	*chunk, *_chunk;
+
+	assert(chunker);
+
+	if (chunker->chunk) {
+		chunk_unref(chunker->chunk);
+	}
+
+	assert(list_empty(&chunker->pinned_chunks));
+
+	list_for_each_entry_safe(chunk, _chunk, &chunker->free_chunks, chunks) {
+		free(chunk);
+	}
+
+	free(chunker);
+}
+
+/* TODO: add pinned chunk iterator interface for cache-friendly iterating across
+ * chunk contents.
+ * The idea is that at times when the performance is really important, the
+ * chunks will be full of active particles, because it's the large numbers
+ * which slows us down.  At those times, it's beneficial to not walk linked
+ * lists of structs to process them, instead we just process all the elements
+ * of the chunk as an array and assume everything is active.  The type of
+ * processing being done in this fashion is benign to perform on an unused
+ * element, as long as there's no dangling pointers being dereferenced.  If
+ * there's references, a status field could be maintained in the entry to say
+ * if it's active, then simply skip processing of the inactive elements.  This
+ * tends to be more cache-friendly than chasing pointers.  A linked list
+ * heirarchy of particles is still maintained for the parent:child
+ * relationships under the assumption that some particles will make use of the
+ * tracked descendants, though nothing has been done with it yet.
+ *
+ * The current implementation of the _particle_t is variable length, which precludes
+ * this optimization.  However, breaking out the particle_props_t into a separate
+ * chunker would allow running particles_age() across the props alone directly
+ * within the pinned chunks.  The other passes are still done heirarchically,
+ * and require the full particle context.
+ */
diff --git a/modules/sparkler/chunker.h b/modules/sparkler/chunker.h
new file mode 100644
index 0000000..ac53cec
--- /dev/null
+++ b/modules/sparkler/chunker.h
@@ -0,0 +1,11 @@
+#ifndef _CHUNKER_H
+#define _CHUNKER_H
+
+typedef struct chunker_t chunker_t;
+
+chunker_t * chunker_new(unsigned chunk_size);
+void * chunker_alloc(chunker_t *chunker, unsigned size);
+void chunker_free(void *mem);
+void chunker_free_chunker(chunker_t *chunker);
+
+#endif
diff --git a/modules/sparkler/container.h b/modules/sparkler/container.h
new file mode 100644
index 0000000..a3779e8
--- /dev/null
+++ b/modules/sparkler/container.h
@@ -0,0 +1,11 @@
+#ifndef _CONTAINER_H
+#define _CONTAINER_H
+
+#include <stddef.h>
+
+#ifndef container_of
+#define container_of(_ptr, _type, _member) \
+	(_type *)((void *)(_ptr) - offsetof(_type, _member))
+#endif
+
+#endif
diff --git a/modules/sparkler/draw.h b/modules/sparkler/draw.h
new file mode 100644
index 0000000..58a4a36
--- /dev/null
+++ b/modules/sparkler/draw.h
@@ -0,0 +1,32 @@
+#ifndef _DRAW_H
+#define _DRAW_H
+
+#include <stdint.h>
+
+#include "fb.h"
+
+/* helper for scaling rgb colors and packing them into an pixel */
+static inline uint32_t makergb(uint32_t r, uint32_t g, uint32_t b, float intensity)
+{
+	r = (((float)intensity) * r);
+	g = (((float)intensity) * g);
+	b = (((float)intensity) * b);
+
+	return (((r & 0xff) << 16) | ((g & 0xff) << 8) | (b & 0xff));
+}
+
+static inline int draw_pixel(fb_fragment_t *f, int x, int y, uint32_t pixel)
+{
+	uint32_t	*pixels = f->buf;
+
+	if (y < 0 || y >= f->height || x < 0 || x >= f->width) {
+		return 0;
+	}
+
+	/* FIXME this assumes stride is aligned to 4 */
+	pixels[(y * f->width + (f->stride >> 2)) + x] = pixel;
+
+	return 1;
+}
+
+#endif
diff --git a/modules/sparkler/list.h b/modules/sparkler/list.h
new file mode 100644
index 0000000..48bca36
--- /dev/null
+++ b/modules/sparkler/list.h
@@ -0,0 +1,252 @@
+#ifndef __LIST_H
+#define __LIST_H
+
+/* linux kernel linked list interface */
+
+/*
+ * Simple doubly linked list implementation.
+ *
+ * Some of the internal functions ("__xxx") are useful when
+ * manipulating whole lists rather than single entries, as
+ * sometimes we already know the next/prev entries and we can
+ * generate better code by using them directly rather than
+ * using the generic single-entry routines.
+ */
+
+typedef struct list_head {
+	struct list_head *next, *prev;
+} list_head_t;
+
+#define LIST_HEAD_INIT(name) { &(name), &(name) }
+
+#define LIST_HEAD(name) \
+	struct list_head name = LIST_HEAD_INIT(name)
+
+#define INIT_LIST_HEAD(ptr) do { \
+	(ptr)->next = (ptr); (ptr)->prev = (ptr); \
+} while (0)
+
+/*
+ * Insert a new entry between two known consecutive entries. 
+ *
+ * This is only for internal list manipulation where we know
+ * the prev/next entries already!
+ */
+static inline void __list_add(struct list_head *new,
+			      struct list_head *prev,
+			      struct list_head *next)
+{
+	next->prev = new;
+	new->next = next;
+	new->prev = prev;
+	prev->next = new;
+}
+
+/**
+ * list_add - add a new entry
+ * @new: new entry to be added
+ * @head: list head to add it after
+ *
+ * Insert a new entry after the specified head.
+ * This is good for implementing stacks.
+ */
+static inline void list_add(struct list_head *new, struct list_head *head)
+{
+	__list_add(new, head, head->next);
+}
+
+/**
+ * list_add_tail - add a new entry
+ * @new: new entry to be added
+ * @head: list head to add it before
+ *
+ * Insert a new entry before the specified head.
+ * This is useful for implementing queues.
+ */
+static inline void list_add_tail(struct list_head *new, struct list_head *head)
+{
+	__list_add(new, head->prev, head);
+}
+
+/*
+ * Delete a list entry by making the prev/next entries
+ * point to each other.
+ *
+ * This is only for internal list manipulation where we know
+ * the prev/next entries already!
+ */
+static inline void __list_del(struct list_head *prev, struct list_head *next)
+{
+	next->prev = prev;
+	prev->next = next;
+}
+
+/**
+ * list_del - deletes entry from list.
+ * @entry: the element to delete from the list.
+ * Note: list_empty on entry does not return true after this, the entry is in an undefined state.
+ */
+static inline void list_del(struct list_head *entry)
+{
+	__list_del(entry->prev, entry->next);
+	entry->next = (void *) 0;
+	entry->prev = (void *) 0;
+}
+
+/**
+ * list_del_init - deletes entry from list and reinitialize it.
+ * @entry: the element to delete from the list.
+ */
+static inline void list_del_init(struct list_head *entry)
+{
+	__list_del(entry->prev, entry->next);
+	INIT_LIST_HEAD(entry); 
+}
+
+/**
+ * list_move - delete from one list and add as another's head
+ * @list: the entry to move
+ * @head: the head that will precede our entry
+ */
+static inline void list_move(struct list_head *list, struct list_head *head)
+{
+        __list_del(list->prev, list->next);
+        list_add(list, head);
+}
+
+/**
+ * list_move_tail - delete from one list and add as another's tail
+ * @list: the entry to move
+ * @head: the head that will follow our entry
+ */
+static inline void list_move_tail(struct list_head *list,
+				  struct list_head *head)
+{
+        __list_del(list->prev, list->next);
+        list_add_tail(list, head);
+}
+
+/**
+ * list_empty - tests whether a list is empty
+ * @head: the list to test.
+ */
+static inline int list_empty(struct list_head *head)
+{
+	return head->next == head;
+}
+
+static inline void __list_splice(struct list_head *list,
+				 struct list_head *head)
+{
+	struct list_head *first = list->next;
+	struct list_head *last = list->prev;
+	struct list_head *at = head->next;
+
+	first->prev = head;
+	head->next = first;
+
+	last->next = at;
+	at->prev = last;
+}
+
+/**
+ * list_splice - join two lists
+ * @list: the new list to add.
+ * @head: the place to add it in the first list.
+ */
+static inline void list_splice(struct list_head *list, struct list_head *head)
+{
+	if (!list_empty(list))
+		__list_splice(list, head);
+}
+
+/**
+ * list_splice_init - join two lists and reinitialise the emptied list.
+ * @list: the new list to add.
+ * @head: the place to add it in the first list.
+ *
+ * The list at @list is reinitialised
+ */
+static inline void list_splice_init(struct list_head *list,
+				    struct list_head *head)
+{
+	if (!list_empty(list)) {
+		__list_splice(list, head);
+		INIT_LIST_HEAD(list);
+	}
+}
+
+/**
+ * list_entry - get the struct for this entry
+ * @ptr:	the &struct list_head pointer.
+ * @type:	the type of the struct this is embedded in.
+ * @member:	the name of the list_struct within the struct.
+ */
+#define list_entry(ptr, type, member) \
+	((type *)((char *)(ptr)-(unsigned long)(&((type *)0)->member)))
+
+/**
+ * list_for_each	-	iterate over a list
+ * @pos:	the &struct list_head to use as a loop counter.
+ * @head:	the head for your list.
+ */
+#define list_for_each(pos, head) \
+	for (pos = (head)->next; pos != (head); \
+        	pos = pos->next)
+/**
+ * list_for_each_prev	-	iterate over a list backwards
+ * @pos:	the &struct list_head to use as a loop counter.
+ * @head:	the head for your list.
+ */
+#define list_for_each_prev(pos, head) \
+	for (pos = (head)->prev; pos != (head); \
+        	pos = pos->prev)
+        	
+/**
+ * list_for_each_safe	-	iterate over a list safe against removal of list entry
+ * @pos:	the &struct list_head to use as a loop counter.
+ * @n:		another &struct list_head to use as temporary storage
+ * @head:	the head for your list.
+ */
+#define list_for_each_safe(pos, n, head) \
+	for (pos = (head)->next, n = pos->next; pos != (head); \
+		pos = n, n = pos->next)
+
+/**
+ * list_for_each_entry	-	iterate over list of given type
+ * @pos:	the type * to use as a loop counter.
+ * @head:	the head for your list.
+ * @member:	the name of the list_struct within the struct.
+ */
+#define list_for_each_entry(pos, head, member)				\
+	for (pos = list_entry((head)->next, typeof(*pos), member);	\
+	     &pos->member != (head); 					\
+	     pos = list_entry(pos->member.next, typeof(*pos), member))
+
+/**
+ * list_for_each_entry_prev	-	iterate over list of given type backwards
+ * @pos:	the type * to use as a loop counter.
+ * @head:	the head for your list.
+ * @member:	the name of the list_struct within the struct.
+ */
+#define list_for_each_entry_prev(pos, head, member)				\
+	for (pos = list_entry((head)->prev, typeof(*pos), member);	\
+	     &pos->member != (head); 					\
+	     pos = list_entry(pos->member.prev, typeof(*pos), member))
+
+
+/**
+ * list_for_each_entry_safe - iterate over list of given type safe against removal of list entry
+ * @pos:	the type * to use as a loop counter.
+ * @n:		another type * to use as temporary storage
+ * @head:	the head for your list.
+ * @member:	the name of the list_struct within the struct.
+ */
+#define list_for_each_entry_safe(pos, n, head, member)			\
+	for (pos = list_entry((head)->next, typeof(*pos), member),	\
+		n = list_entry(pos->member.next, typeof(*pos), member);	\
+	     &pos->member != (head); 					\
+	     pos = n, n = list_entry(n->member.next, typeof(*n), member))
+
+
+#endif
diff --git a/modules/sparkler/particle.c b/modules/sparkler/particle.c
new file mode 100644
index 0000000..0e3d2c8
--- /dev/null
+++ b/modules/sparkler/particle.c
@@ -0,0 +1,14 @@
+#include "particle.h"
+
+/* convert a particle to a new type */
+void particle_convert(particles_t *particles, particle_t *p, particle_props_t *props, particle_ops_t *ops)
+{
+	particle_cleanup(particles, p);
+	if (props) {
+		*p->props = *props;
+	}
+	if (ops) {
+		p->ops = ops;
+	}
+	particle_init(particles, p);
+}
diff --git a/modules/sparkler/particle.h b/modules/sparkler/particle.h
new file mode 100644
index 0000000..95c117e
--- /dev/null
+++ b/modules/sparkler/particle.h
@@ -0,0 +1,79 @@
+#ifndef _PARTICLE_H
+#define _PARTICLE_H
+
+#include "bsp.h"
+#include "fb.h"
+#include "v3f.h"
+
+typedef struct particle_props_t {
+	v3f_t		position;	/* position in 3d space */
+	v3f_t		direction;	/* trajectory in 3d space */
+	float		velocity;	/* linear velocity */
+	float		mass;		/* mass of particle */
+	float		drag;		/* drag of particle */
+	int		of_use:1;	/* are these properties of use/meaningful? */
+} particle_props_t;
+
+typedef enum particle_status_t {
+	PARTICLE_ALIVE,
+	PARTICLE_DEAD
+} particle_status_t;
+
+typedef struct particle_t particle_t;
+typedef struct particles_t particles_t;
+
+typedef struct particle_ops_t {
+	unsigned		context_size;								/* size of the particle context (0 for none) */
+	int			(*init)(particles_t *, particle_t *);					/* initialize the particle, called after allocating context (optional) */
+	void			(*cleanup)(particles_t *, particle_t *);				/* cleanup function, called before freeing context (optional) */
+	particle_status_t	(*sim)(particles_t *, particle_t *);					/* simulate the particle for another cycle (required) */
+	void			(*draw)(particles_t *, particle_t *, int, int, fb_fragment_t *);	/* draw the particle, 3d->2d projection has been done already (optional) */
+} particle_ops_t;
+
+struct particle_t {
+	bsp_occupant_t		occupant;	/* occupant node in the bsp tree */
+	particle_props_t	*props;
+	particle_ops_t		*ops;
+	void			*ctxt;
+};
+
+
+//#define rand_within_range(_min, _max) ((rand() % (_max - _min)) + _min)
+// the style of random number generator used by c libraries has less entropy in the lower bits meaning one shouldn't just use modulo, while this is slower, the results do seem a little different.
+#define rand_within_range(_min, _max) (int)(((float)_min) + ((float)rand() / (float)RAND_MAX) * (_max - _min))
+
+#define INHERIT_OPS	NULL
+#define INHERIT_PROPS	NULL
+
+
+static inline int particle_init(particles_t *particles, particle_t *p) {
+	if (p->ops->init) {
+		return p->ops->init(particles, p);
+	}
+
+	return 1;
+}
+
+
+static inline void particle_cleanup(particles_t *particles, particle_t *p) {
+	if (p->ops->cleanup) {
+		p->ops->cleanup(particles, p);
+	}
+}
+
+
+static inline particle_status_t particle_sim(particles_t *particles, particle_t *p) {
+	return p->ops->sim(particles, p);
+}
+
+
+static inline void particle_draw(particles_t *particles, particle_t *p, int x, int y, fb_fragment_t *f) {
+	if (p->ops->draw) {
+		p->ops->draw(particles, p, x, y, f);
+	}
+}
+
+
+void particle_convert(particles_t *particles, particle_t *p, particle_props_t *props, particle_ops_t *ops);
+
+#endif
diff --git a/modules/sparkler/particles.c b/modules/sparkler/particles.c
new file mode 100644
index 0000000..ee23972
--- /dev/null
+++ b/modules/sparkler/particles.c
@@ -0,0 +1,338 @@
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/types.h>
+#include <unistd.h>
+#include <math.h>
+#include <stdlib.h>
+#include <time.h>
+
+#include "fb.h"
+
+#include "chunker.h"
+#include "container.h"
+#include "bsp.h"
+#include "list.h"
+#include "particle.h"
+#include "particles.h"
+#include "v3f.h"
+
+#define ZCONST	0.4f
+
+/* private particle with all the particles bookkeeping... */
+typedef struct _particle_t {
+	list_head_t		siblings;	/* sibling particles */
+	list_head_t		children;	/* children particles */
+
+	particle_props_t	props;		/* we reference this in the public particle, I might change
+						 * the way props are allocated so coding everything to use a
+						 * reference for now.  It may make sense to have props allocated
+						 * separately via their own chunker, and perform some mass operations
+						 * against the list of chunks rather than chasing the pointers of
+						 * the particle heirarchy. TODO
+						 */
+	particle_t		public;		/* the public particle_t is embedded */
+
+	uint8_t			context[];	/* particle type-specific context [public.ops.context_size] */
+} _particle_t;
+
+struct particles_t {
+	chunker_t		*chunker;	/* chunker for variably-sized particle allocation (includes context) */
+	list_head_t		active;		/* top-level active list of particles heirarchy */
+	bsp_t			*bsp;		/* bsp spatial index of the particles */
+};
+
+
+/* create a new particle system */
+particles_t * particles_new(void)
+{
+	particles_t	*particles;
+
+	particles = calloc(1, sizeof(particles_t));
+	if (!particles) {
+		return NULL;
+	}
+
+	particles->chunker = chunker_new(sizeof(_particle_t) * 128);
+	if (!particles->chunker) {
+		return NULL;
+	}
+
+	particles->bsp = bsp_new();
+	if (!particles->bsp) {
+		return NULL;
+	}
+
+	INIT_LIST_HEAD(&particles->active);
+
+	return particles;
+}
+
+
+/* TODO: add a public interface for destroying particles?  for now we just return PARTICLE_DEAD in the sim */
+static inline void _particles_free_particle(particles_t *particles, _particle_t *p)
+{
+	assert(p);
+
+	particle_cleanup(particles, &p->public);
+	chunker_free(p);
+}
+
+
+static inline void _particles_free(particles_t *particles, list_head_t *list)
+{
+	_particle_t	*p, *_p;
+
+	assert(particles);
+	assert(list);
+
+	list_for_each_entry_safe(p, _p, list, siblings) {
+		_particles_free(particles, &p->children);
+		_particles_free_particle(particles, p);
+	}
+}
+
+
+/* free up all the particles */
+void particles_free(particles_t *particles)
+{
+	assert(particles);
+
+	_particles_free(particles, &particles->active);
+}
+
+
+/* reclaim a dead particle, moving it to the free list */
+static void particles_reap_particle(particles_t *particles, _particle_t *particle)
+{
+	assert(particles);
+	assert(particle);
+
+	if (!list_empty(&particle->children)) {
+		/* adopt any orphaned children using the global parts list */
+		list_splice(&particle->children, &particles->active);
+	}
+
+	list_del(&particle->siblings);
+	bsp_delete_occupant(particles->bsp, &particle->public.occupant);
+	_particles_free_particle(particles, particle);
+}
+
+
+/* add a particle to the specified list */
+static inline int _particles_add_particle(particles_t *particles, list_head_t *list, particle_props_t *props, particle_ops_t *ops)
+{
+	_particle_t	*p;
+
+	assert(particles);
+	assert(ops);
+	assert(list);
+
+	p = chunker_alloc(particles->chunker, sizeof(_particle_t) + ops->context_size);
+	if (!p) {
+		return 0;
+	}
+
+	INIT_LIST_HEAD(&p->children);
+	INIT_LIST_HEAD(&p->siblings);
+
+	/* inherit the parent's properties and ops if they're not explicitly provided */
+	if (props) {
+		p->props = *props;
+	} else {
+		p->props.of_use = 0;
+	}
+
+	p->public.props = &p->props;
+	p->public.ops = ops;
+
+	if (ops->context_size) {
+		p->public.ctxt = p->context;
+	}
+
+	if (!particle_init(particles, &p->public)) {
+		/* XXX FIXME this shouldn't be normal, we don't want to allocate
+		 * particles that cannot be initialized.  the rockets today set a cap
+		 * by failing initialization, that's silly. */
+		chunker_free(p);
+		return 0;
+	}
+
+	p->public.props->of_use = 1;
+	list_add(&p->siblings, list);
+	bsp_add_occupant(particles->bsp, &p->public.occupant, &p->props.position);
+
+	return 1;
+}
+
+
+/* add a new "top-level" particle of the specified props and ops taking from the provided parts list */
+int particles_add_particle(particles_t *particles, particle_props_t *props, particle_ops_t *ops)
+{
+	assert(particles);
+
+	return _particles_add_particle(particles, &particles->active, props, ops);
+}
+
+
+/* spawn a new child particle from a parent, initializing it via inheritance if desired */
+void particles_spawn_particle(particles_t *particles, particle_t *parent, particle_props_t *props, particle_ops_t *ops)
+{
+	_particle_t	*p = container_of(parent, _particle_t, public);
+
+	assert(particles);
+	assert(parent);
+
+	_particles_add_particle(particles, &p->children, props ? props : parent->props, ops ? ops : parent->ops);
+}
+
+
+/* plural version of particle_add(); adds multiple "top-level" particles of uniform props and ops */
+void particles_add_particles(particles_t *particles, particle_props_t *props, particle_ops_t *ops, int num)
+{
+	int	i;
+
+	assert(particles);
+
+	for (i = 0; i < num; i++) {
+		_particles_add_particle(particles, &particles->active, props, ops);
+	}
+}
+
+
+/* Simple accessor to get the bsp pointer, the bsp is special because we don't want to do
+ * callbacks per-occupant, so the bsp_occupant_t and search functions are used directly by
+ * the per-particle code needing nearest-neighbor search.  that requires an accessor since
+ * particles_t is opaque.  This seemed less shitty than opening up particles_t.
+ */
+bsp_t * particles_bsp(particles_t *particles)
+{
+	assert(particles);
+	assert(particles->bsp);
+
+	return particles->bsp;
+}
+
+
+static inline void _particles_draw(particles_t *particles, list_head_t *list, fb_fragment_t *fragment)
+{
+	float		w2 = fragment->width / 2, h2 = fragment->height / 2;
+	_particle_t	*p;
+
+	assert(particles);
+	assert(list);
+	assert(fragment);
+
+	list_for_each_entry(p, list, siblings) {
+		int	x, y;
+
+		/* project the 3d coordinates onto the 2d plane */
+		x = ((float)(p->props.position.x / (p->props.position.z - ZCONST)) * w2) + w2;
+		y = ((float)(p->props.position.y / (p->props.position.z - ZCONST)) * h2) + h2;
+
+		particle_draw(particles, &p->public, x, y, fragment);
+		_particles_draw(particles, &p->children, fragment);
+	}
+}
+
+
+/* draw all of the particles, currently called in heirarchical order */
+void particles_draw(particles_t *particles, fb_fragment_t *fragment)
+{
+	assert(particles);
+
+	_particles_draw(particles, &particles->active, fragment);
+}
+
+
+static inline particle_status_t _particles_sim(particles_t *particles, list_head_t *list)
+{
+	particle_status_t	ret = PARTICLE_DEAD, s;
+	_particle_t		*p, *_p;
+
+	assert(particles);
+	assert(list);
+
+	list_for_each_entry_safe(p, _p, list, siblings) {
+		if ((s = particle_sim(particles, &p->public)) == PARTICLE_ALIVE) {
+			ret = PARTICLE_ALIVE;
+
+			if (_particles_sim(particles, &p->children) == PARTICLE_ALIVE) {
+				ret = PARTICLE_ALIVE;
+			}
+		} else {
+			particles_reap_particle(particles, p);
+		}
+	}
+
+	return ret;
+}
+
+
+/* simulate the particles, call the sim method of every particle in the heirarchy, this is what makes the particles dynamic */
+/* if any paticle is still living, we return PARTICLE_ALIVE, to inform the caller when everything's dead */
+particle_status_t particles_sim(particles_t *particles)
+{
+	assert(particles);
+
+	return _particles_sim(particles, &particles->active);
+}
+
+
+static inline void _particles_age(particles_t *particles, list_head_t *list)
+{
+	_particle_t	*p;
+
+	assert(particles);
+	assert(list);
+
+	/* TODO: since this *only* involves the properties struct, if they were
+	 * allocated from a separate slab containing only properties, it'd be
+	 * more efficient to iterate across property arrays and skip inactive
+	 * entries.  This heirarchical pointer-chasing recursion isn't
+	 * particularly good for cache utilization.
+	 */
+	list_for_each_entry(p, list, siblings) {
+#if 1
+		if (p->props.mass > 0.0f) {
+			/* gravity, TODO: mass isn't applied. */
+			static v3f_t	gravity = v3f_init(0.0f, -1.0f, 0.0f);
+			v3f_t		g;
+
+			g = v3f_mult_scalar(&gravity, 0.08f);
+			p->props.direction = v3f_add(&p->props.direction, &g);
+			p->props.direction = v3f_normalize(&p->props.direction);
+		}
+#endif
+
+#if 1
+		/* some drag/resistance proportional to velocity TODO: integrate mass */
+		if (p->props.velocity > 0.0f) {
+			p->props.velocity -= ((p->props.velocity * p->props.velocity * p->props.drag));
+			if (p->props.velocity < 0.0f) {
+				p->props.velocity = 0;
+			}
+		}
+#endif
+
+		/* regular movement */
+		if (p->props.velocity > 0.0f) {
+			v3f_t	movement = v3f_mult_scalar(&p->props.direction, p->props.velocity);
+
+			p->props.position = v3f_add(&p->props.position, &movement);
+			bsp_move_occupant(particles->bsp, &p->public.occupant, &p->props.position);
+		}
+
+		_particles_age(particles, &p->children);
+	}
+}
+
+
+/* advance time for all the particles (move them), this doesn't currently invoke any part-specific helpers, it's just applying
+ * physics-type stuff, moving particles according to their velocities, directions, mass, drag, gravity etc... */
+void particles_age(particles_t *particles)
+{
+	assert(particles);
+
+	_particles_age(particles, &particles->active);
+}
diff --git a/modules/sparkler/particles.h b/modules/sparkler/particles.h
new file mode 100644
index 0000000..689934b
--- /dev/null
+++ b/modules/sparkler/particles.h
@@ -0,0 +1,21 @@
+#ifndef _PARTICLES_H
+#define _PARTICLES_H
+
+#include "bsp.h"
+#include "fb.h"
+#include "list.h"
+#include "particle.h"
+
+typedef struct particles_t particles_t;
+
+particles_t * particles_new(void);
+void particles_draw(particles_t *particles, fb_fragment_t *fragment);
+particle_status_t particles_sim(particles_t *particles);
+void particles_age(particles_t *particles);
+void particles_free(particles_t *particles);
+int particles_add_particle(particles_t *particles, particle_props_t *props, particle_ops_t *ops);
+void particles_spawn_particle(particles_t *particles, particle_t *parent, particle_props_t *props, particle_ops_t *ops);
+void particles_add_particles(particles_t *particles, particle_props_t *props, particle_ops_t *ops, int num);
+bsp_t * particles_bsp(particles_t *particles);
+
+#endif
diff --git a/modules/sparkler/rocket.c b/modules/sparkler/rocket.c
new file mode 100644
index 0000000..6b9dc5e
--- /dev/null
+++ b/modules/sparkler/rocket.c
@@ -0,0 +1,144 @@
+#include <stdlib.h>
+
+#include "draw.h"
+#include "particle.h"
+#include "particles.h"
+
+/* a "rocket" particle type */
+#define ROCKET_MAX_DECAY_RATE	20
+#define ROCKET_MIN_DECAY_RATE	2
+#define ROCKET_MAX_LIFETIME	500
+#define ROCKET_MIN_LIFETIME	300
+#define ROCKETS_MAX		20
+#define ROCKETS_XPLODE_MIN_SIZE	2000
+#define ROCKETS_XPLODE_MAX_SIZE	8000
+
+extern particle_ops_t burst_ops;
+extern particle_ops_t spark_ops;
+extern particle_ops_t xplode_ops;
+
+static unsigned rockets_cnt;
+
+typedef struct rocket_ctxt_t {
+	int		decay_rate;
+	int		longevity;
+	v3f_t		wander;
+	float		last_velocity;	/* cache velocity to sense violent accelerations and explode when they happen */
+} rocket_ctxt_t;
+
+
+static int rocket_init(particles_t *particles, particle_t *p)
+{
+	rocket_ctxt_t	*ctxt = p->ctxt;
+
+	if (rockets_cnt >= ROCKETS_MAX) {
+		return 0;
+	}
+	rockets_cnt++;
+
+	ctxt->decay_rate = rand_within_range(ROCKET_MIN_DECAY_RATE, ROCKET_MAX_DECAY_RATE);
+	ctxt->longevity = rand_within_range(ROCKET_MIN_LIFETIME, ROCKET_MAX_LIFETIME);
+
+	ctxt->wander.x = (float)(rand_within_range(0, 628) - 314) / 10000.0f;
+	ctxt->wander.y = (float)(rand_within_range(0, 628) - 314) / 10000.0f;
+	ctxt->wander.z = (float)(rand_within_range(0, 628) - 314) / 10000.0f;
+	ctxt->wander = v3f_normalize(&ctxt->wander);
+
+	ctxt->last_velocity = p->props->velocity;
+	p->props->drag = 0.4;
+	p->props->mass = 0.8;
+
+	return 1;
+}
+
+
+static particle_status_t rocket_sim(particles_t *particles, particle_t *p)
+{
+	rocket_ctxt_t	*ctxt = p->ctxt;
+	int		i, n_sparks;
+
+	if (!ctxt->longevity ||
+	    (ctxt->longevity -= ctxt->decay_rate) <= 0 ||
+	    p->props->velocity - ctxt->last_velocity > p->props->velocity * .05) {	/* explode if accelerated too hard (burst) */
+		int	n_xplode;
+		/* on death we explode */
+
+		ctxt->longevity = 0;
+
+		/* add a burst shockwave particle at our location
+		 * TODO: need way to supply particle-type-specific parameters at spawn (burst size should derive from n_xplode)
+		 */
+		particles_spawn_particle(particles, p, NULL, &burst_ops);
+
+		/* add a bunch of new explosion particles */
+		/* TODO: also particle-type-specific parameters, colors!  rocket bursts should be able to vary the color. */
+		n_xplode = rand_within_range(ROCKETS_XPLODE_MIN_SIZE, ROCKETS_XPLODE_MAX_SIZE);
+		for (i = 0; i < n_xplode; i++) {
+			particle_props_t	props = *p->props;
+			particle_ops_t		*ops = &xplode_ops;
+
+			props.direction.x = ((float)(rand_within_range(0, 314159 * 2) - 314159) / 100000.0);
+			props.direction.y = ((float)(rand_within_range(0, 314159 * 2) - 314159) / 100000.0);
+			props.direction.z = ((float)(rand_within_range(0, 314159 * 2) - 314159) / 100000.0);
+			props.direction = v3f_normalize(&props.direction);
+
+			//props->velocity = ((float)rand_within_range(100, 200) / 100000.0);
+			props.velocity = ((float)rand_within_range(100, 300) / 100000.0);
+			particles_spawn_particle(particles, p, &props, ops);
+		}
+		return PARTICLE_DEAD;
+	}
+
+#if 1
+	/* FIXME: this isn't behaving as intended */
+	p->props->direction = v3f_add(&p->props->direction, &ctxt->wander);
+	p->props->direction = v3f_normalize(&p->props->direction);
+#endif
+	p->props->velocity += .00003;
+
+	/* spray some sparks behind the rocket */
+	n_sparks = rand_within_range(10, 40);
+	for (i = 0; i < n_sparks; i++) {
+		particle_props_t	props = *p->props;
+
+		props.direction = v3f_negate(&props.direction);
+
+		props.direction.x += (float)(rand_within_range(0, 40) - 20) / 100.0;
+		props.direction.y += (float)(rand_within_range(0, 40) - 20) / 100.0;
+		props.direction.z += (float)(rand_within_range(0, 40) - 20) / 100.0;
+		props.direction = v3f_normalize(&props.direction);
+
+		props.velocity = (float)rand_within_range(10, 50) / 100000.0;
+		particles_spawn_particle(particles, p, &props, &spark_ops);
+	}
+
+	ctxt->last_velocity = p->props->velocity;
+
+	return PARTICLE_ALIVE;
+}
+
+
+static void rocket_draw(particles_t *particles, particle_t *p, int x, int y, fb_fragment_t *f)
+{
+	rocket_ctxt_t	*ctxt = p->ctxt;
+
+	if (!draw_pixel(f, x, y, 0xff0000)) {
+		/* kill off parts that wander off screen */
+		ctxt->longevity = 0;
+	}
+}
+
+
+static void rocket_cleanup(particles_t *particles, particle_t *p)
+{
+	rockets_cnt--;
+}
+
+
+particle_ops_t	rocket_ops = {
+			.context_size = sizeof(rocket_ctxt_t),
+			.sim = rocket_sim,
+			.init = rocket_init,
+			.draw = rocket_draw,
+			.cleanup = rocket_cleanup,
+		};
diff --git a/modules/sparkler/simple.c b/modules/sparkler/simple.c
new file mode 100644
index 0000000..e453e46
--- /dev/null
+++ b/modules/sparkler/simple.c
@@ -0,0 +1,113 @@
+#include <stdlib.h>
+
+#include "draw.h"
+#include "particle.h"
+#include "particles.h"
+
+
+/* a "simple" particle type */
+#define SIMPLE_MAX_DECAY_RATE	20
+#define SIMPLE_MIN_DECAY_RATE	2
+#define SIMPLE_MAX_LIFETIME	110
+#define SIMPLE_MIN_LIFETIME	30
+#define SIMPLE_MAX_SPAWN	15
+#define SIMPLE_MIN_SPAWN	2
+
+extern particle_ops_t rocket_ops;
+
+typedef struct _simple_ctxt_t {
+	int		decay_rate;
+	int		longevity;
+	int		lifetime;
+} simple_ctxt_t;
+
+
+static int simple_init(particles_t *particles, particle_t *p)
+{
+	simple_ctxt_t	*ctxt = p->ctxt;
+
+	ctxt->decay_rate = rand_within_range(SIMPLE_MIN_DECAY_RATE, SIMPLE_MAX_DECAY_RATE);
+	ctxt->lifetime = ctxt->longevity = rand_within_range(SIMPLE_MIN_LIFETIME, SIMPLE_MAX_LIFETIME);
+
+	if (!p->props->of_use) {
+		/* everything starts from the bottom center */
+		p->props->position.x = 0;
+		p->props->position.y = 0;
+		p->props->position.z = 0;
+
+		/* TODO: direction random-ish within the range of a narrow upward facing cone */
+		p->props->direction.x = (float)(rand_within_range(0, 6) - 3) * .1f;
+		p->props->direction.y = 1.0f + (float)(rand_within_range(0, 6) - 3) * .1f;
+		p->props->direction.z = (float)(rand_within_range(0, 6) - 3) * .1f;
+		p->props->direction = v3f_normalize(&p->props->direction);
+
+		p->props->velocity = (float)rand_within_range(300, 800) / 100000.0;
+
+		p->props->drag = 0.03;
+		p->props->mass = 0.3;
+		p->props->of_use = 1;
+	} /* else { we've been given properties, manipulate them or run with them? } */
+
+	return 1;
+}
+
+
+static particle_status_t simple_sim(particles_t *particles, particle_t *p)
+{
+	simple_ctxt_t	*ctxt = p->ctxt;
+
+	/* a particle is free to manipulate its children list when aging, but not itself or its siblings */
+	/* return PARTICLE_DEAD to remove kill yourself, do not age children here, the age pass will recurse
+	 * into children and age them independently _after_ their parents have been aged
+	 */
+	if (!ctxt->longevity || (ctxt->longevity -= ctxt->decay_rate) <= 0) {
+		ctxt->longevity = 0;
+		return PARTICLE_DEAD;
+	}
+
+	/* create particles inheriting our type based on some silly conditions, with some tweaks to their direction */
+	if (ctxt->longevity == 42 || (ctxt->longevity > 500 && !(ctxt->longevity % 50))) {
+		int	i, num = rand_within_range(SIMPLE_MIN_SPAWN, SIMPLE_MAX_SPAWN);
+
+		for (i = 0; i < num; i++) {
+			particle_props_t	props = *p->props;
+			particle_ops_t		*ops = INHERIT_OPS;
+
+			if (i == (SIMPLE_MAX_SPAWN - 2)) {
+				ops = &rocket_ops;
+				props.velocity = (float)rand_within_range(60, 100) / 1000000.0;
+			} else {
+				props.velocity = (float)rand_within_range(30, 100) / 10000.0;
+			}
+
+			props.direction.x += (float)(rand_within_range(0, 315 * 2) - 315) / 100.0;
+			props.direction.y += (float)(rand_within_range(0, 315 * 2) - 315) / 100.0;
+			props.direction.z += (float)(rand_within_range(0, 315 * 2) - 315) / 100.0;
+			props.direction = v3f_normalize(&props.direction);
+
+			particles_spawn_particle(particles, p, &props, ops); // XXX
+		}
+	}
+
+	return PARTICLE_ALIVE;
+}
+
+
+static void simple_draw(particles_t *particles, particle_t *p, int x, int y, fb_fragment_t *f)
+{
+	simple_ctxt_t	*ctxt = p->ctxt;
+
+	if (!draw_pixel(f, x, y, makergb(0xff, 0xff, 0xff, ((float)ctxt->longevity / ctxt->lifetime)))) {
+		/* immediately kill off stars that wander off screen */
+		ctxt->longevity = 0;
+	}
+}
+
+
+particle_ops_t	simple_ops = {
+			.context_size = sizeof(simple_ctxt_t),
+			.sim = simple_sim,
+			.init = simple_init,
+			.draw = simple_draw,
+			.cleanup = NULL,
+		};
diff --git a/modules/sparkler/spark.c b/modules/sparkler/spark.c
new file mode 100644
index 0000000..ea68ac2
--- /dev/null
+++ b/modules/sparkler/spark.c
@@ -0,0 +1,63 @@
+#include <stdlib.h>
+
+#include "draw.h"
+#include "particle.h"
+#include "particles.h"
+
+/* a "spark" particle type, emitted from behind rockets */
+#define SPARK_MAX_DECAY_RATE	20
+#define SPARK_MIN_DECAY_RATE	2
+#define SPARK_MAX_LIFETIME	150
+#define SPARK_MIN_LIFETIME	1
+
+typedef struct _spark_ctxt_t {
+	int		decay_rate;
+	int		longevity;
+	int		lifetime;
+} spark_ctxt_t;
+
+
+static int spark_init(particles_t *particles, particle_t *p)
+{
+	spark_ctxt_t	*ctxt = p->ctxt;
+
+	p->props->drag = 20.0;
+	p->props->mass = 0.1;
+	ctxt->decay_rate = rand_within_range(SPARK_MIN_DECAY_RATE, SPARK_MAX_DECAY_RATE);
+	ctxt->lifetime = ctxt->longevity = rand_within_range(SPARK_MIN_LIFETIME, SPARK_MAX_LIFETIME);
+
+	return 1;
+}
+
+
+static particle_status_t spark_sim(particles_t *particles, particle_t *p)
+{
+	spark_ctxt_t	*ctxt = p->ctxt;
+
+	if (!ctxt->longevity || (ctxt->longevity -= ctxt->decay_rate) <= 0) {
+		ctxt->longevity = 0;
+		return PARTICLE_DEAD;
+	}
+
+	return PARTICLE_ALIVE;
+}
+
+
+static void spark_draw(particles_t *particles, particle_t *p, int x, int y, fb_fragment_t *f)
+{
+	spark_ctxt_t	*ctxt = p->ctxt;
+
+	if (!draw_pixel(f, x, y, makergb(0xff, 0xa0, 0x20, ((float)ctxt->longevity / ctxt->lifetime)))) {
+		/* offscreen */
+		ctxt->longevity = 0;
+	}
+}
+
+
+particle_ops_t	spark_ops = {
+			.context_size = sizeof(spark_ctxt_t),
+			.sim = spark_sim,
+			.init = spark_init,
+			.draw = spark_draw,
+			.cleanup = NULL,
+		};
diff --git a/modules/sparkler/sparkler.c b/modules/sparkler/sparkler.c
new file mode 100644
index 0000000..de2123e
--- /dev/null
+++ b/modules/sparkler/sparkler.c
@@ -0,0 +1,53 @@
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+#include <time.h>
+#include <unistd.h>
+
+#include "fb.h"
+#include "rototiller.h"
+#include "util.h"
+
+#include "particles.h"
+
+/* particle system gadget (C) Vito Caputo <vcaputo@pengaru.com> 2/15/2014 */
+/* 1/10/2015 added octree bsp (though not yet leveraged) */
+/* 11/25/2016 refactor and begun adapting to rototiller */
+
+#define INIT_PARTS 100
+
+extern particle_ops_t	simple_ops;
+
+
+/* Render a 3D particle system */
+static void sparkler(fb_fragment_t *fragment)
+{
+	static particles_t	*particles;
+	static int		initialized;
+	uint32_t		*buf = fragment->buf;
+
+	if (!initialized) {
+		srand(time(NULL) + getpid());
+
+		particles = particles_new();
+		particles_add_particles(particles, NULL, &simple_ops, INIT_PARTS);
+
+		initialized = 1;
+	}
+
+	particles_age(particles);
+	memset(buf, 0, ((fragment->width << 2) + fragment->stride) * fragment->height);
+
+	particles_draw(particles, fragment);
+	particles_sim(particles);
+	particles_add_particles(particles, NULL, &simple_ops, INIT_PARTS / 4);
+}
+
+
+rototiller_renderer_t	sparkler_renderer = {
+	.render = sparkler,
+	.name = "sparkler",
+	.description = "Particle system with spatial interactions",
+	.author = "Vito Caputo <vcaputo@pengaru.com>",
+	.license = "GPLv2",
+};
diff --git a/modules/sparkler/sparkler.h b/modules/sparkler/sparkler.h
new file mode 100644
index 0000000..3beb610
--- /dev/null
+++ b/modules/sparkler/sparkler.h
@@ -0,0 +1,8 @@
+#ifndef _SPARKLER_H
+#define _SPARKLER_H
+
+#include "fb.h"
+
+void sparkler(fb_fragment_t *fragment);
+
+#endif
diff --git a/modules/sparkler/v3f.h b/modules/sparkler/v3f.h
new file mode 100644
index 0000000..8bf7e24
--- /dev/null
+++ b/modules/sparkler/v3f.h
@@ -0,0 +1,157 @@
+#ifndef _V3F_H
+#define _V3F_H
+
+#include <math.h>
+
+typedef struct v3f_t {
+	float	x, y, z;
+} v3f_t;
+
+#define v3f_set(_v3f, _x, _y, _z)	\
+	(_v3f)->x = _x;			\
+	(_v3f)->y = _y;			\
+	(_v3f)->z = _z;
+
+#define v3f_init(_x, _y, _z)		\
+	{				\
+		.x = _x,		\
+		.y = _y,		\
+		.z = _z,		\
+	}
+
+/* return if a and b are equal */
+static inline int v3f_equal(v3f_t *a, v3f_t *b)
+{
+	return (a->x == b->x && a->y == b->y && a->z == b->z);
+}
+
+
+/* return the result of (a + b) */
+static inline v3f_t v3f_add(v3f_t *a, v3f_t *b)
+{
+	v3f_t	res = v3f_init(a->x + b->x, a->y + b->y, a->z + b->z);
+
+	return res;
+}
+
+
+/* return the result of (a - b) */
+static inline v3f_t v3f_sub(v3f_t *a, v3f_t *b)
+{
+	v3f_t	res = v3f_init(a->x - b->x, a->y - b->y, a->z - b->z);
+
+	return res;
+}
+
+
+/* return the result of (-v) */
+static inline v3f_t v3f_negate(v3f_t *v)
+{
+	v3f_t	res = v3f_init(-v->x, -v->y, -v->z);
+
+	return res;
+}
+
+
+/* return the result of (a * b) */
+static inline v3f_t v3f_mult(v3f_t *a, v3f_t *b)
+{
+	v3f_t	res = v3f_init(a->x * b->x, a->y * b->y, a->z * b->z);
+
+	return res;
+}
+
+
+/* return the result of (v * scalar) */
+static inline v3f_t v3f_mult_scalar(v3f_t *v, float scalar)
+{
+	v3f_t	res = v3f_init( v->x * scalar, v->y * scalar, v->z * scalar);
+
+	return res;
+}
+
+
+/* return the result of (uv / scalar) */
+static inline v3f_t v3f_div_scalar(v3f_t *v, float scalar)
+{
+	v3f_t	res = v3f_init(v->x / scalar, v->y / scalar, v->z / scalar);
+
+	return res;
+}
+
+
+/* return the result of (a . b) */
+static inline float v3f_dot(v3f_t *a, v3f_t *b)
+{
+	return a->x * b->x + a->y * b->y + a->z * b->z;
+}
+
+
+/* return the length of the supplied vector */
+static inline float v3f_length(v3f_t *v)
+{
+	return sqrtf(v3f_dot(v, v));
+}
+
+
+/* return the normalized form of the supplied vector */
+static inline v3f_t v3f_normalize(v3f_t *v)
+{
+	v3f_t	nv;
+	float	f;
+
+	f = 1.0f / v3f_length(v);
+
+	v3f_set(&nv, f * v->x, f * v->y, f * v->z);
+
+	return nv;
+}
+
+
+/* return the distance squared between two arbitrary points */
+static inline float v3f_distance_sq(v3f_t *a, v3f_t *b)
+{
+	return powf(a->x - b->x, 2) + powf(a->y - b->y, 2) + powf(a->z - b->z, 2);
+}
+
+
+/* return the distance between two arbitrary points */
+/* (consider using v3f_distance_sq() instead if possible, sqrtf() is slow) */
+static inline float v3f_distance(v3f_t *a, v3f_t *b)
+{
+	return sqrtf(v3f_distance_sq(a, b));
+}
+
+
+/* return the cross product of two unit vectors */
+static inline v3f_t v3f_cross(v3f_t *a, v3f_t *b)
+{
+	v3f_t	product = v3f_init(a->y * b->z - a->z * b->y, a->z * b->x - a->x * b->z, a->x * b->y - a->y * b->x);
+
+	return product;
+}
+
+
+/* return the linearly interpolated vector between the two vectors at point alpha (0-1.0) */
+static inline v3f_t v3f_lerp(v3f_t *a, v3f_t *b, float alpha)
+{
+	v3f_t	lerp_a, lerp_b;
+
+	lerp_a = v3f_mult_scalar(a, 1.0f - alpha);
+	lerp_b = v3f_mult_scalar(b, alpha);
+
+	return v3f_add(&lerp_a, &lerp_b);
+}
+
+
+/* return the normalized linearly interpolated vector between the two vectors at point alpha (0-1.0) */
+static inline v3f_t v3f_nlerp(v3f_t *a, v3f_t *b, float alpha)
+{
+	v3f_t	lerp;
+
+	lerp = v3f_lerp(a, b, alpha);
+
+	return v3f_normalize(&lerp);
+}
+
+#endif
diff --git a/modules/sparkler/xplode.c b/modules/sparkler/xplode.c
new file mode 100644
index 0000000..24a436e
--- /dev/null
+++ b/modules/sparkler/xplode.c
@@ -0,0 +1,82 @@
+#include <stdlib.h>
+
+#include "draw.h"
+#include "particle.h"
+#include "particles.h"
+
+/* a "xplode" particle type, emitted by rockets in large numbers at the end of their lifetime */
+#define XPLODE_MAX_DECAY_RATE	10
+#define XPLODE_MIN_DECAY_RATE	5
+#define XPLODE_MAX_LIFETIME	150
+#define XPLODE_MIN_LIFETIME	5
+
+extern particle_ops_t spark_ops;
+particle_ops_t	xplode_ops;
+
+typedef struct _xplode_ctxt_t {
+	int		decay_rate;
+	int		longevity;
+	int		lifetime;
+} xplode_ctxt_t;
+
+
+static int xplode_init(particles_t *particles, particle_t *p)
+{
+	xplode_ctxt_t	*ctxt = p->ctxt;
+
+	ctxt->decay_rate = rand_within_range(XPLODE_MIN_DECAY_RATE, XPLODE_MAX_DECAY_RATE);
+	ctxt->lifetime = ctxt->longevity = rand_within_range(XPLODE_MIN_LIFETIME, XPLODE_MAX_LIFETIME);
+
+	p->props->drag = 10.9;
+	p->props->mass = 0.3;
+
+	return 1;
+}
+
+
+static particle_status_t xplode_sim(particles_t *particles, particle_t *p)
+{
+	xplode_ctxt_t	*ctxt = p->ctxt;
+
+	if (!ctxt->longevity || (ctxt->longevity -= ctxt->decay_rate) <= 0) {
+		ctxt->longevity = 0;
+		return PARTICLE_DEAD;
+	}
+
+	/* litter some small sparks behind the explosion particle */
+	if (!(ctxt->lifetime % 30)) {
+		particle_props_t	props = *p->props;
+
+		props.velocity = (float)rand_within_range(10, 50) / 10000.0;
+		particles_spawn_particle(particles, p, &props, &xplode_ops);
+	}
+
+	return PARTICLE_ALIVE;
+}
+
+
+static void xplode_draw(particles_t *particles, particle_t *p, int x, int y, fb_fragment_t *f)
+{
+	xplode_ctxt_t	*ctxt = p->ctxt;
+	uint32_t	color;
+
+	if (ctxt->longevity == ctxt->lifetime) {
+		color = makergb(0xff, 0xff, 0xa0, 1.0);
+	} else {
+		color = makergb(0xff, 0xff, 0x00, ((float)ctxt->longevity / ctxt->lifetime));
+	}
+
+	if (!draw_pixel(f, x, y, color)) {
+		/* offscreen */
+		ctxt->longevity = 0;
+	}
+}
+
+
+particle_ops_t	xplode_ops = {
+			.context_size = sizeof(xplode_ctxt_t),
+			.sim = xplode_sim,
+			.init = xplode_init,
+			.draw = xplode_draw,
+			.cleanup = NULL,
+		};
diff --git a/rototiller.c b/rototiller.c
new file mode 100644
index 0000000..b1c8c36
--- /dev/null
+++ b/rototiller.c
@@ -0,0 +1,85 @@
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <inttypes.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <xf86drm.h>
+#include <xf86drmMode.h>
+
+#include "drmsetup.h"
+#include "fb.h"
+#include "fps.h"
+#include "rototiller.h"
+#include "util.h"
+
+/* Copyright (C) 2016 Vito Caputo <vcaputo@pengaru.com> */
+
+#define NUM_FB_PAGES	3
+/* ^ By triple-buffering, we can have a page tied up being displayed, another
+ * tied up submitted and waiting for vsync, and still not block on getting
+ * another page so we can begin rendering another frame before vsync.  With
+ * just two pages we end up twiddling thumbs until the vsync arrives.
+ */
+
+extern rototiller_renderer_t	roto32_renderer;
+extern rototiller_renderer_t	roto64_renderer;
+extern rototiller_renderer_t	ray_renderer;
+extern rototiller_renderer_t	sparkler_renderer;
+
+static rototiller_renderer_t	*renderers[] = {
+	&roto32_renderer,
+	&roto64_renderer,
+	&ray_renderer,
+	&sparkler_renderer,
+};
+
+
+static void renderer_select(int *renderer)
+{
+	int	i;
+
+	puts("");
+	for (i = 0; i < nelems(renderers); i++) {
+		printf("%i: %s - %s\n", i, renderers[i]->name, renderers[i]->description);
+	}
+
+	ask_num(renderer, nelems(renderers) - 1, "Renderer", 0);
+}
+
+
+int main(int argc, const char *argv[])
+{
+	int			drm_fd;
+	drmModeModeInfoPtr	drm_mode;
+	uint32_t		drm_crtc_id;
+	uint32_t		drm_connector_id;
+	fb_t			*fb;
+	int			renderer;
+
+	drm_setup(&drm_fd, &drm_crtc_id, &drm_connector_id, &drm_mode);
+	renderer_select(&renderer);
+
+	exit_if(!(fb = fb_new(drm_fd, drm_crtc_id, &drm_connector_id, 1, drm_mode, NUM_FB_PAGES)),
+		"unable to create fb");
+
+	pexit_if(!fps_setup(),
+		"unable to setup fps counter");
+
+	for (;;) {
+		fb_page_t	*page;
+
+		fps_print(fb);
+
+		page = fb_page_get(fb);
+		renderers[renderer]->render(&page->fragment);
+		fb_page_put(fb, page);
+	}
+
+	fb_free(fb);
+	close(drm_fd);
+
+	return EXIT_SUCCESS;
+}
diff --git a/rototiller.h b/rototiller.h
new file mode 100644
index 0000000..0ff2c3f
--- /dev/null
+++ b/rototiller.h
@@ -0,0 +1,12 @@
+#ifndef _ROTOTILLER_H
+#define _ROTOTILLER_H
+
+typedef struct rototiller_renderer_t {
+	void	(*render)(fb_fragment_t *);
+	char	*name;
+	char	*description;
+	char	*author;
+	char	*license;
+} rototiller_renderer_t;
+
+#endif
diff --git a/rototiller32.c b/rototiller32.c
deleted file mode 100644
index 13b762f..0000000
--- a/rototiller32.c
+++ /dev/null
@@ -1,214 +0,0 @@
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <fcntl.h>
-#include <stdio.h>
-#include <errno.h>
-#include <stdlib.h>
-#include <string.h>
-#include <stdint.h>
-#include <inttypes.h>
-#include <xf86drm.h>
-#include <xf86drmMode.h>
-#include <sys/ioctl.h>
-#include <sys/mman.h>
-#include <math.h>
-
-/* Copyright (C) 2016 Vito Caputo <vcaputo@pengaru.com> */
-
-#define exit_if(_cond, _fmt, ...) \
-	if (_cond) { \
-		fprintf(stderr, "Fatal error: " _fmt "\n", ##__VA_ARGS__); \
-		exit(EXIT_FAILURE); \
-	}
-
-#define pexit_if(_cond, _fmt, ...) \
-	exit_if(_cond, _fmt ": %s", ##__VA_ARGS__, strerror(errno))
-
-/* Some defines for the fixed-point stuff in render(). */
-#define FIXED_TRIG_LUT_SIZE	4096	/* size of the cos/sin look-up tables */
-#define FIXED_BITS		12	/* fractional bits */
-#define FIXED_EXP		4096	/* 2^FIXED_BITS */
-#define FIXED_COS(_rad)		costab[_rad % FIXED_TRIG_LUT_SIZE]
-#define FIXED_SIN(_rad)		sintab[_rad % FIXED_TRIG_LUT_SIZE]
-#define FIXED_MULT(_a, _b)	((_a * _b) >> FIXED_BITS)
-#define FIXED_NEW(_i)		(_i << FIXED_BITS)
-#define FIXED_TO_INT(_f)	((_f) >> FIXED_BITS)
-
-/* Draw a rotating checkered 256x256 texture into next_page. */
-static void render(uint32_t *current_page, uint32_t *next_page, int width, int height, int pitch) {
-	static int32_t	costab[FIXED_TRIG_LUT_SIZE], sintab[FIXED_TRIG_LUT_SIZE];
-	static uint8_t	texture[256][256];
-	static int	initialized;
-	static uint32_t	colors[2];
-	static unsigned	r, rr;
-
-	int		y_cos_r, y_sin_r, x_cos_r, x_sin_r, x_cos_r_init, x_sin_r_init, cos_r, sin_r;
-	int		x, y, stride;
-	uint8_t		tx, ty; /* 256x256 texture; 8 bit texture indices to modulo via overflow. */
-
-	if (!initialized) {
-		int i;
-
-		initialized = 1;
-
-		/* Generate simple checker pattern texture, nothing clever, feel free to play! */
-		/* If you modify texture on every frame instead of only @ initialization you can
-		 * produce some neat output.  These values are indexed into colors[] below. */
-		for (y = 0; y < 128; y++) {
-			for (x = 0; x < 128; x++)
-				texture[y][x] = 1;
-			for (; x < 256; x++)
-				texture[y][x] = 0;
-		}
-		for (; y < 256; y++) {
-			for (x = 0; x < 128; x++)
-				texture[y][x] = 0;
-			for (; x < 256; x++)
-				texture[y][x] = 1;
-		}
-
-		/* Generate fixed-point cos & sin LUTs. */
-		for (i = 0; i < FIXED_TRIG_LUT_SIZE; i++) {
-			costab[i] = ((cos((double)2*M_PI*i/FIXED_TRIG_LUT_SIZE))*FIXED_EXP);
-			sintab[i] = ((sin((double)2*M_PI*i/FIXED_TRIG_LUT_SIZE))*FIXED_EXP);
-		}
-	}
-
-	pitch /= 4; /* pitch is number of bytes in a row, scale it to uint32_t units. */
-	stride = (pitch - width); /* stride is number of words from row end to start of next row */
-
-	/* This is all done using fixed-point in the hopes of being faster, and yes assumptions
-	 * are being made WRT the overflow of tx/ty as well, only tested on x86_64. */
-	cos_r = FIXED_COS(r);
-	sin_r = FIXED_SIN(r);
-
-	/* Vary the colors, this is just a mashup of sinusoidal rgb values. */
-	colors[0] =	((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr), FIXED_NEW(127))) + 128) << 16) |
-			((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr / 2), FIXED_NEW(127))) + 128) << 8) |
-			((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr / 3), FIXED_NEW(127))) + 128));
-
-	colors[1] =	((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr / 2), FIXED_NEW(127))) + 128) << 16) |
-			((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr / 2), FIXED_NEW(127))) + 128)) << 8 |
-			((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr), FIXED_NEW(127))) + 128) );
-
-	/* The dimensions are cut in half and negated to center the rotation. */
-	/* The [xy]_{sin,cos}_r variables are accumulators to replace multiplication with addition. */
-	x_cos_r_init = FIXED_MULT(-FIXED_NEW((width / 2)), cos_r);
-	x_sin_r_init = FIXED_MULT(-FIXED_NEW((width / 2)), sin_r);
-
-	y_cos_r = FIXED_MULT(-FIXED_NEW((height / 2)), cos_r);
-	y_sin_r = FIXED_MULT(-FIXED_NEW((height / 2)), sin_r);
-
-	for (y = 0; y < height; y++) {
-
-		x_cos_r = x_cos_r_init;
-		x_sin_r = x_sin_r_init;
-
-		for (x = 0; x < width; x++, next_page++) {
-
-			tx = FIXED_TO_INT(x_sin_r - y_cos_r);
-			ty = FIXED_TO_INT(y_sin_r + x_cos_r);
-
-			*next_page = colors[texture[ty][tx]];
-
-			x_cos_r += cos_r;
-			x_sin_r += sin_r;
-		}
-
-		next_page += stride;
-		y_cos_r += cos_r;
-		y_sin_r += sin_r;
-	}
-
-	// This governs the rotation and color cycle.
-	r += FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr), FIXED_NEW(16)));
-	rr += 2;
-}
-
-
-
-int main(int argc, const char *argv[]) {
-	int			drm_fd;
-	drmModeResPtr		drm_res;
-	drmModeConnectorPtr	drm_con;
-	uint32_t		*fb_maps[2], drm_fbs[2];
-	unsigned		page = 0, next_page;
-
-	pexit_if(!drmAvailable(),
-		"drm unavailable");
-
-	/* FIXME: use drmOpen(), requires digging to see what you're supposed to supply it for name. */
-	pexit_if((drm_fd = open("/dev/dri/card0", O_RDWR)) < 0,
-		"unable to open drm device");
-
-	/* this requires root, but doesn't seem necessary for what's being done here, which is a bit surprising. */
-//	pexit_if(drmSetMaster(drm_fd) < 0,
-//		"unable to set master");
-
-	exit_if(!(drm_res = drmModeGetResources(drm_fd)),
-		"unable to get drm resources");
-
-	exit_if(drm_res->count_connectors < 1 ||
-		!(drm_con = drmModeGetConnector(drm_fd, drm_res->connectors[0])),
-		"unable to get first connector");
-
-	/* create double-buffers */
-	struct drm_mode_create_dumb create_dumb = {
-		.width = drm_con->modes[0].hdisplay,
-		.height = drm_con->modes[0].vdisplay,
-		.bpp = 32,
-		.flags = 0, // unused,
-	};
-	pexit_if(ioctl(drm_fd, DRM_IOCTL_MODE_CREATE_DUMB, &create_dumb) < 0,
-		"unable to create dumb buffer A");
-
-	struct drm_mode_map_dumb map_dumb = {
-		.handle = create_dumb.handle,
-		.pad = 0, // unused
-	};
-	pexit_if(ioctl(drm_fd, DRM_IOCTL_MODE_MAP_DUMB, &map_dumb) < 0,
-		"unable to prepare dumb buffer A for mmap");
-	pexit_if(!(fb_maps[0] = mmap(NULL, create_dumb.size, PROT_READ|PROT_WRITE, MAP_SHARED, drm_fd, map_dumb.offset)),
-		"unable to mmap dumb buffer A");
-
-	pexit_if(drmModeAddFB(drm_fd, create_dumb.width, create_dumb.height, 24, create_dumb.bpp, create_dumb.pitch, create_dumb.handle, &drm_fbs[0]) < 0,
-		"unable to add dumb buffer A as fb");
-
-	/* second one... */
-	pexit_if(ioctl(drm_fd, DRM_IOCTL_MODE_CREATE_DUMB, &create_dumb) < 0,
-		"unable to create dumb buffer B");
-	pexit_if(drmModeAddFB(drm_fd, create_dumb.width, create_dumb.height, 24, create_dumb.bpp, create_dumb.pitch, create_dumb.handle, &drm_fbs[1]) < 0,
-		"unable to add dumb buffer B as fb");
-
-	map_dumb.handle = create_dumb.handle;
-
-	pexit_if(ioctl(drm_fd, DRM_IOCTL_MODE_MAP_DUMB, &map_dumb) < 0,
-		"unable to prepare dumb buffer B for mmap");
-	pexit_if(!(fb_maps[1] = mmap(NULL, create_dumb.size, PROT_READ|PROT_WRITE, MAP_SHARED, drm_fd, map_dumb.offset)),
-		"unable to mmap dumb buffer B");
-
-	/* make the current page the visible one */
-	pexit_if(drmModeSetCrtc(drm_fd, drm_res->crtcs[0], drm_fbs[page], 0, 0, drm_res->connectors, 1, drm_con->modes) < 0,
-		"unable to configure crtc");
-
-	drmEventContext drm_ev_ctx = {
-		.version = DRM_EVENT_CONTEXT_VERSION,
-		.vblank_handler = NULL,
-		.page_flip_handler = NULL
-	};
-
-	// now the rendering & page-flipping loop */
-	for (;;page = next_page) {
-		next_page = (page + 1) % 2;
-
-		/* render next page */
-		render(fb_maps[page], fb_maps[next_page], create_dumb.width, create_dumb.height, create_dumb.pitch);
-
-		/* flip synchronously */
-		pexit_if(drmModePageFlip(drm_fd, drm_res->crtcs[0], drm_fbs[next_page], DRM_MODE_PAGE_FLIP_EVENT, NULL) < 0,
-			"unable to flip page %u to %u", page, next_page);
-		drmHandleEvent(drm_fd, &drm_ev_ctx);
-	}
-
-	return EXIT_SUCCESS;
-}
diff --git a/rototiller64.c b/rototiller64.c
deleted file mode 100644
index f273be8..0000000
--- a/rototiller64.c
+++ /dev/null
@@ -1,229 +0,0 @@
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <fcntl.h>
-#include <stdio.h>
-#include <errno.h>
-#include <stdlib.h>
-#include <string.h>
-#include <stdint.h>
-#include <inttypes.h>
-#include <xf86drm.h>
-#include <xf86drmMode.h>
-#include <sys/ioctl.h>
-#include <sys/mman.h>
-#include <math.h>
-
-/* Copyright (C) 2016 Vito Caputo <vcaputo@pengaru.com> */
-
-#define exit_if(_cond, _fmt, ...) \
-	if (_cond) { \
-		fprintf(stderr, "Fatal error: " _fmt "\n", ##__VA_ARGS__); \
-		exit(EXIT_FAILURE); \
-	}
-
-#define pexit_if(_cond, _fmt, ...) \
-	exit_if(_cond, _fmt ": %s", ##__VA_ARGS__, strerror(errno))
-
-/* Some defines for the fixed-point stuff in render(). */
-#define FIXED_TRIG_LUT_SIZE	4096	/* size of the cos/sin look-up tables */
-#define FIXED_BITS		12	/* fractional bits */
-#define FIXED_EXP		4096	/* 2^FIXED_BITS */
-#define FIXED_COS(_rad)		costab[_rad % FIXED_TRIG_LUT_SIZE]
-#define FIXED_SIN(_rad)		sintab[_rad % FIXED_TRIG_LUT_SIZE]
-#define FIXED_MULT(_a, _b)	((_a * _b) >> FIXED_BITS)
-#define FIXED_NEW(_i)		(_i << FIXED_BITS)
-#define FIXED_TO_INT(_f)	((_f) >> FIXED_BITS)
-
-/* Draw a rotating checkered 256x256 texture into next_page. */
-static void render(uint32_t *current_page, uint32_t *next_page, int width, int height, int pitch) {
-	static int32_t	costab[FIXED_TRIG_LUT_SIZE], sintab[FIXED_TRIG_LUT_SIZE];
-	static uint8_t	texture[256][256];
-	static int	initialized;
-	static uint32_t	colors[2];
-	static unsigned	r, rr;
-
-	int		y_cos_r, y_sin_r, x_cos_r, x_sin_r, x_cos_r_init, x_sin_r_init, cos_r, sin_r;
-	int		x, y, stride;
-	uint8_t		tx, ty; /* 256x256 texture; 8 bit texture indices to modulo via overflow. */
-	uint64_t	*_next_page = (uint64_t *)next_page;
-
-	if (!initialized) {
-		int i;
-
-		initialized = 1;
-
-		/* Generate simple checker pattern texture, nothing clever, feel free to play! */
-		/* If you modify texture on every frame instead of only @ initialization you can
-		 * produce some neat output.  These values are indexed into colors[] below. */
-		for (y = 0; y < 128; y++) {
-			for (x = 0; x < 128; x++)
-				texture[y][x] = 1;
-			for (; x < 256; x++)
-				texture[y][x] = 0;
-		}
-		for (; y < 256; y++) {
-			for (x = 0; x < 128; x++)
-				texture[y][x] = 0;
-			for (; x < 256; x++)
-				texture[y][x] = 1;
-		}
-
-		/* Generate fixed-point cos & sin LUTs. */
-		for (i = 0; i < FIXED_TRIG_LUT_SIZE; i++) {
-			costab[i] = ((cos((double)2*M_PI*i/FIXED_TRIG_LUT_SIZE))*FIXED_EXP);
-			sintab[i] = ((sin((double)2*M_PI*i/FIXED_TRIG_LUT_SIZE))*FIXED_EXP);
-		}
-	}
-
-	pitch /= 4; /* pitch is number of bytes in a row, scale it to uint32_t units. */
-	stride = (pitch - width); /* stride is number of words from row end to start of next row */
-
-	/* This is all done using fixed-point in the hopes of being faster, and yes assumptions
-	 * are being made WRT the overflow of tx/ty as well, only tested on x86_64. */
-	cos_r = FIXED_COS(r);
-	sin_r = FIXED_SIN(r);
-
-	/* Vary the colors, this is just a mashup of sinusoidal rgb values. */
-	colors[0] =	((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr), FIXED_NEW(127))) + 128) << 16) |
-			((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr / 2), FIXED_NEW(127))) + 128) << 8) |
-			((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr / 3), FIXED_NEW(127))) + 128));
-
-	colors[1] =	((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr / 2), FIXED_NEW(127))) + 128) << 16) |
-			((FIXED_TO_INT(FIXED_MULT(FIXED_COS(rr / 2), FIXED_NEW(127))) + 128)) << 8 |
-			((FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr), FIXED_NEW(127))) + 128) );
-
-	/* The dimensions are cut in half and negated to center the rotation. */
-	/* The [xy]_{sin,cos}_r variables are accumulators to replace multiplication with addition. */
-	x_cos_r_init = FIXED_MULT(-FIXED_NEW((width / 2)), cos_r);
-	x_sin_r_init = FIXED_MULT(-FIXED_NEW((width / 2)), sin_r);
-
-	y_cos_r = FIXED_MULT(-FIXED_NEW((height / 2)), cos_r);
-	y_sin_r = FIXED_MULT(-FIXED_NEW((height / 2)), sin_r);
-
-	width /= 2;
-	stride /= 2;
-
-	for (y = 0; y < height; y++) {
-
-		x_cos_r = x_cos_r_init;
-		x_sin_r = x_sin_r_init;
-
-		for (x = 0; x < width; x++, _next_page++) {
-			uint64_t	p;
-
-			tx = FIXED_TO_INT(x_sin_r - y_cos_r);
-			ty = FIXED_TO_INT(y_sin_r + x_cos_r);
-
-			p = colors[texture[ty][tx]];
-
-			x_cos_r += cos_r;
-			x_sin_r += sin_r;
-
-			tx = FIXED_TO_INT(x_sin_r - y_cos_r);
-			ty = FIXED_TO_INT(y_sin_r + x_cos_r);
-
-			p |= (uint64_t)colors[texture[ty][tx]] << 32;
-			
-			*_next_page = p;
-
-			x_cos_r += cos_r;
-			x_sin_r += sin_r;
-		}
-
-		_next_page += stride;
-		y_cos_r += cos_r;
-		y_sin_r += sin_r;
-	}
-
-	// This governs the rotation and color cycle.
-	r += FIXED_TO_INT(FIXED_MULT(FIXED_SIN(rr), FIXED_NEW(16)));
-	rr += 2;
-}
-
-
-
-int main(int argc, const char *argv[]) {
-	int			drm_fd;
-	drmModeResPtr		drm_res;
-	drmModeConnectorPtr	drm_con;
-	uint32_t		*fb_maps[2], drm_fbs[2];
-	unsigned		page = 0, next_page;
-
-	pexit_if(!drmAvailable(),
-		"drm unavailable");
-
-	/* FIXME: use drmOpen(), requires digging to see what you're supposed to supply it for name. */
-	pexit_if((drm_fd = open("/dev/dri/card0", O_RDWR)) < 0,
-		"unable to open drm device");
-
-	/* this requires root, but doesn't seem necessary for what's being done here, which is a bit surprising. */
-//	pexit_if(drmSetMaster(drm_fd) < 0,
-//		"unable to set master");
-
-	exit_if(!(drm_res = drmModeGetResources(drm_fd)),
-		"unable to get drm resources");
-
-	exit_if(drm_res->count_connectors < 1 ||
-		!(drm_con = drmModeGetConnector(drm_fd, drm_res->connectors[0])),
-		"unable to get first connector");
-
-	/* create double-buffers */
-	struct drm_mode_create_dumb create_dumb = {
-		.width = drm_con->modes[0].hdisplay,
-		.height = drm_con->modes[0].vdisplay,
-		.bpp = 32,
-		.flags = 0, // unused,
-	};
-	pexit_if(ioctl(drm_fd, DRM_IOCTL_MODE_CREATE_DUMB, &create_dumb) < 0,
-		"unable to create dumb buffer A");
-
-	struct drm_mode_map_dumb map_dumb = {
-		.handle = create_dumb.handle,
-		.pad = 0, // unused
-	};
-	pexit_if(ioctl(drm_fd, DRM_IOCTL_MODE_MAP_DUMB, &map_dumb) < 0,
-		"unable to prepare dumb buffer A for mmap");
-	pexit_if(!(fb_maps[0] = mmap(NULL, create_dumb.size, PROT_READ|PROT_WRITE, MAP_SHARED, drm_fd, map_dumb.offset)),
-		"unable to mmap dumb buffer A");
-
-	pexit_if(drmModeAddFB(drm_fd, create_dumb.width, create_dumb.height, 24, create_dumb.bpp, create_dumb.pitch, create_dumb.handle, &drm_fbs[0]) < 0,
-		"unable to add dumb buffer A as fb");
-
-	/* second one... */
-	pexit_if(ioctl(drm_fd, DRM_IOCTL_MODE_CREATE_DUMB, &create_dumb) < 0,
-		"unable to create dumb buffer B");
-	pexit_if(drmModeAddFB(drm_fd, create_dumb.width, create_dumb.height, 24, create_dumb.bpp, create_dumb.pitch, create_dumb.handle, &drm_fbs[1]) < 0,
-		"unable to add dumb buffer B as fb");
-
-	map_dumb.handle = create_dumb.handle;
-
-	pexit_if(ioctl(drm_fd, DRM_IOCTL_MODE_MAP_DUMB, &map_dumb) < 0,
-		"unable to prepare dumb buffer B for mmap");
-	pexit_if(!(fb_maps[1] = mmap(NULL, create_dumb.size, PROT_READ|PROT_WRITE, MAP_SHARED, drm_fd, map_dumb.offset)),
-		"unable to mmap dumb buffer B");
-
-	/* make the current page the visible one */
-	pexit_if(drmModeSetCrtc(drm_fd, drm_res->crtcs[0], drm_fbs[page], 0, 0, drm_res->connectors, 1, drm_con->modes) < 0,
-		"unable to configure crtc");
-
-	drmEventContext drm_ev_ctx = {
-		.version = DRM_EVENT_CONTEXT_VERSION,
-		.vblank_handler = NULL,
-		.page_flip_handler = NULL
-	};
-
-	// now the rendering & page-flipping loop */
-	for (;;page = next_page) {
-		next_page = (page + 1) % 2;
-
-		/* render next page */
-		render(fb_maps[page], fb_maps[next_page], create_dumb.width, create_dumb.height, create_dumb.pitch);
-
-		/* flip synchronously */
-		pexit_if(drmModePageFlip(drm_fd, drm_res->crtcs[0], drm_fbs[next_page], DRM_MODE_PAGE_FLIP_EVENT, NULL) < 0,
-			"unable to flip page %u to %u", page, next_page);
-		drmHandleEvent(drm_fd, &drm_ev_ctx);
-	}
-
-	return EXIT_SUCCESS;
-}
diff --git a/util.c b/util.c
new file mode 100644
index 0000000..0e5825a
--- /dev/null
+++ b/util.c
@@ -0,0 +1,60 @@
+#include <limits.h>
+#include <stdio.h>
+#include <string.h>
+#include <unistd.h>
+
+#include "util.h"
+
+#define SYSFS_CPU	"/sys/devices/system/cpu/cpu"
+#define MAXCPUS		1024
+
+unsigned get_ncpus(void)
+{
+	char		path[cstrlen(SYSFS_CPU "1024") + 1];
+	unsigned	n;
+
+	for (n = 0; n < MAXCPUS; n++) {
+		snprintf(path, sizeof(path), "%s%u", SYSFS_CPU, n);
+		if (access(path, F_OK) == -1)
+			break;
+	}
+
+	return n == 0 ? 1 : n;
+}
+
+
+static void query(const char *prompt, const char *def, char *buf, int len)
+{
+	buf[0] = '\0';
+
+	printf("%s [%s]: ", prompt, def);
+	fflush(stdout);
+
+	fgets(buf, len, stdin);
+	if (buf[0] == '\0' || buf[0] == '\n') {
+		snprintf(buf, len, "%s", def);
+	} else if(strchr(buf, '\n')) {
+		*strchr(buf, '\n') = '\0';
+	}
+}
+
+
+void ask_string(char *buf, int len, const char *prompt, const char *def)
+{
+	query(prompt, def, buf, len);
+}
+
+
+void ask_num(int *res, int max, const char *prompt, int def)
+{
+	char	buf[21], buf2[256];
+	int	num;
+
+	snprintf(buf, sizeof(buf), "%i", def);
+	do {
+		query(prompt, buf, buf2, sizeof(buf2));
+		num = atoi(buf2); /* TODO: errors (strtol)*/
+	} while (num > max);
+
+	*res = num;
+}
diff --git a/util.h b/util.h
new file mode 100644
index 0000000..549b529
--- /dev/null
+++ b/util.h
@@ -0,0 +1,28 @@
+#ifndef _UTIL_H
+#define _UTIL_H
+
+#include <errno.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#define exit_if(_cond, _fmt, ...) \
+	if (_cond) { \
+		fprintf(stderr, "Fatal error: " _fmt "\n", ##__VA_ARGS__); \
+		exit(EXIT_FAILURE); \
+	}
+
+#define pexit_if(_cond, _fmt, ...) \
+	exit_if(_cond, _fmt ": %s", ##__VA_ARGS__, strerror(errno))
+
+#define nelems(_array) \
+	(sizeof(_array) / sizeof(_array[0]))
+
+#define cstrlen(_str) \
+	(sizeof(_str) - 1)
+
+unsigned get_ncpus(void);
+void ask_string(char *buf, int len, const char *prompt, const char *def);
+void ask_num(int *res, int max, const char *prompt, int def);
+
+#endif /* _UTIL_H */