#include <linux/mm.h>
#include <linux/time.h>
#include <linux/list.h>
-#include <linux/semaphore.h>
#include "omap_dmm_tiler.h"
#include "omap_dmm_priv.h"
return 0;
}
-irqreturn_t omap_dmm_irq_handler(int irq, void *arg)
+static void release_engine(struct refill_engine *engine)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&list_lock, flags);
+ list_add(&engine->idle_node, &omap_dmm->idle_head);
+ spin_unlock_irqrestore(&list_lock, flags);
+
+ atomic_inc(&omap_dmm->engine_counter);
+ wake_up_interruptible(&omap_dmm->engine_queue);
+}
+
+static irqreturn_t omap_dmm_irq_handler(int irq, void *arg)
{
struct dmm *dmm = arg;
uint32_t status = readl(dmm->base + DMM_PAT_IRQSTATUS);
writel(status, dmm->base + DMM_PAT_IRQSTATUS);
for (i = 0; i < dmm->num_engines; i++) {
- if (status & DMM_IRQSTAT_LST)
+ if (status & DMM_IRQSTAT_LST) {
wake_up_interruptible(&dmm->engines[i].wait_for_refill);
+ if (dmm->engines[i].async)
+ release_engine(&dmm->engines[i]);
+ }
+
status >>= 8;
}
{
struct dmm_txn *txn = NULL;
struct refill_engine *engine = NULL;
+ int ret;
+ unsigned long flags;
+
- down(&dmm->engine_sem);
+ /* wait until an engine is available */
+ ret = wait_event_interruptible(omap_dmm->engine_queue,
+ atomic_add_unless(&omap_dmm->engine_counter, -1, 0));
+ if (ret)
+ return ERR_PTR(ret);
/* grab an idle engine */
- spin_lock(&list_lock);
+ spin_lock_irqsave(&list_lock, flags);
if (!list_empty(&dmm->idle_head)) {
engine = list_entry(dmm->idle_head.next, struct refill_engine,
idle_node);
list_del(&engine->idle_node);
}
- spin_unlock(&list_lock);
+ spin_unlock_irqrestore(&list_lock, flags);
BUG_ON(!engine);
* Add region to DMM transaction. If pages or pages[i] is NULL, then the
* corresponding slot is cleared (ie. dummy_pa is programmed)
*/
-static int dmm_txn_append(struct dmm_txn *txn, struct pat_area *area,
+static void dmm_txn_append(struct dmm_txn *txn, struct pat_area *area,
struct page **pages, uint32_t npages, uint32_t roll)
{
dma_addr_t pat_pa = 0;
int columns = (1 + area->x1 - area->x0);
int rows = (1 + area->y1 - area->y0);
int i = columns*rows;
- u32 *lut = omap_dmm->lut + (engine->tcm->lut_id * omap_dmm->lut_width *
- omap_dmm->lut_height) +
- (area->y0 * omap_dmm->lut_width) + area->x0;
pat = alloc_dma(txn, sizeof(struct pat), &pat_pa);
page_to_phys(pages[n]) : engine->dmm->dummy_pa;
}
- /* fill in lut with new addresses */
- for (i = 0; i < rows; i++, lut += omap_dmm->lut_width)
- memcpy(lut, &data[i*columns], columns * sizeof(u32));
-
txn->last_pat = pat;
- return 0;
+ return;
}
/**
goto cleanup;
}
+ /* mark whether it is async to denote list management in IRQ handler */
+ engine->async = wait ? false : true;
+
/* kick reload */
writel(engine->refill_pa,
dmm->base + reg[PAT_DESCR][engine->id]);
}
cleanup:
- spin_lock(&list_lock);
- list_add(&engine->idle_node, &dmm->idle_head);
- spin_unlock(&list_lock);
+ /* only place engine back on list if we are done with it */
+ if (ret || wait)
+ release_engine(engine);
- up(&omap_dmm->engine_sem);
return ret;
}
txn = dmm_txn_init(omap_dmm, area->tcm);
if (IS_ERR_OR_NULL(txn))
- return PTR_ERR(txn);
+ return -ENOMEM;
tcm_for_each_slice(slice, *area, area_s) {
struct pat_area p_area = {
.x1 = slice.p1.x, .y1 = slice.p1.y,
};
- ret = dmm_txn_append(txn, &p_area, pages, npages, roll);
- if (ret)
- goto fail;
+ dmm_txn_append(txn, &p_area, pages, npages, roll);
roll += tcm_sizeof(slice);
}
ret = dmm_txn_commit(txn, wait);
-fail:
return ret;
}
struct tiler_block *block = kzalloc(sizeof(*block), GFP_KERNEL);
u32 min_align = 128;
int ret;
+ unsigned long flags;
BUG_ON(!validfmt(fmt));
}
/* add to allocation list */
- spin_lock(&list_lock);
+ spin_lock_irqsave(&list_lock, flags);
list_add(&block->alloc_node, &omap_dmm->alloc_head);
- spin_unlock(&list_lock);
+ spin_unlock_irqrestore(&list_lock, flags);
return block;
}
{
struct tiler_block *block = kzalloc(sizeof(*block), GFP_KERNEL);
int num_pages = (size + PAGE_SIZE - 1) >> PAGE_SHIFT;
+ unsigned long flags;
if (!block)
- return 0;
+ return ERR_PTR(-ENOMEM);
block->fmt = TILFMT_PAGE;
return ERR_PTR(-ENOMEM);
}
- spin_lock(&list_lock);
+ spin_lock_irqsave(&list_lock, flags);
list_add(&block->alloc_node, &omap_dmm->alloc_head);
- spin_unlock(&list_lock);
+ spin_unlock_irqrestore(&list_lock, flags);
return block;
}
int tiler_release(struct tiler_block *block)
{
int ret = tcm_free(&block->area);
+ unsigned long flags;
if (block->area.tcm)
dev_err(omap_dmm->dev, "failed to release block\n");
- spin_lock(&list_lock);
+ spin_lock_irqsave(&list_lock, flags);
list_del(&block->alloc_node);
- spin_unlock(&list_lock);
+ spin_unlock_irqrestore(&list_lock, flags);
kfree(block);
return ret;
* Utils
*/
-/* calculate the tiler space address of a pixel in a view orientation */
-static u32 tiler_get_address(u32 orient, enum tiler_fmt fmt, u32 x, u32 y)
+/* calculate the tiler space address of a pixel in a view orientation...
+ * below description copied from the display subsystem section of TRM:
+ *
+ * When the TILER is addressed, the bits:
+ * [28:27] = 0x0 for 8-bit tiled
+ * 0x1 for 16-bit tiled
+ * 0x2 for 32-bit tiled
+ * 0x3 for page mode
+ * [31:29] = 0x0 for 0-degree view
+ * 0x1 for 180-degree view + mirroring
+ * 0x2 for 0-degree view + mirroring
+ * 0x3 for 180-degree view
+ * 0x4 for 270-degree view + mirroring
+ * 0x5 for 270-degree view
+ * 0x6 for 90-degree view
+ * 0x7 for 90-degree view + mirroring
+ * Otherwise the bits indicated the corresponding bit address to access
+ * the SDRAM.
+ */
+static u32 tiler_get_address(enum tiler_fmt fmt, u32 orient, u32 x, u32 y)
{
u32 x_bits, y_bits, tmp, x_mask, y_mask, alignment;
x_mask = MASK(x_bits);
y_mask = MASK(y_bits);
- if (x < 0 || x > x_mask || y < 0 || y > y_mask)
+ if (x < 0 || x > x_mask || y < 0 || y > y_mask) {
+ DBG("invalid coords: %u < 0 || %u > %u || %u < 0 || %u > %u",
+ x, x, x_mask, y, y, y_mask);
return 0;
+ }
/* account for mirroring */
if (orient & MASK_X_INVERT)
{
BUG_ON(!validfmt(block->fmt));
- return TILVIEW_8BIT + tiler_get_address(0, block->fmt,
+ return TILVIEW_8BIT + tiler_get_address(block->fmt, 0,
block->area.p0.x * geom[block->fmt].slot_w,
block->area.p0.y * geom[block->fmt].slot_h);
}
+dma_addr_t tiler_tsptr(struct tiler_block *block, uint32_t orient,
+ uint32_t x, uint32_t y)
+{
+ struct tcm_pt *p = &block->area.p0;
+ BUG_ON(!validfmt(block->fmt));
+
+ return tiler_get_address(block->fmt, orient,
+ (p->x * geom[block->fmt].slot_w) + x,
+ (p->y * geom[block->fmt].slot_h) + y);
+}
+
void tiler_align(enum tiler_fmt fmt, uint16_t *w, uint16_t *h)
{
BUG_ON(!validfmt(fmt));
*h = round_up(*h, geom[fmt].slot_h);
}
-uint32_t tiler_stride(enum tiler_fmt fmt)
+uint32_t tiler_stride(enum tiler_fmt fmt, uint32_t orient)
{
BUG_ON(!validfmt(fmt));
- return 1 << (CONT_WIDTH_BITS + geom[fmt].y_shft);
+ if (orient & MASK_XY_FLIP)
+ return 1 << (CONT_HEIGHT_BITS + geom[fmt].x_shft);
+ else
+ return 1 << (CONT_WIDTH_BITS + geom[fmt].y_shft);
}
size_t tiler_size(enum tiler_fmt fmt, uint16_t w, uint16_t h)
return round_up(geom[fmt].cpp * w, PAGE_SIZE) * h;
}
-bool dmm_is_initialized(void)
+bool dmm_is_available(void)
{
return omap_dmm ? true : false;
}
{
struct tiler_block *block, *_block;
int i;
+ unsigned long flags;
if (omap_dmm) {
/* free all area regions */
- spin_lock(&list_lock);
+ spin_lock_irqsave(&list_lock, flags);
list_for_each_entry_safe(block, _block, &omap_dmm->alloc_head,
alloc_node) {
list_del(&block->alloc_node);
kfree(block);
}
- spin_unlock(&list_lock);
+ spin_unlock_irqrestore(&list_lock, flags);
for (i = 0; i < omap_dmm->num_lut; i++)
if (omap_dmm->tcm && omap_dmm->tcm[i])
kfree(omap_dmm->engines);
if (omap_dmm->refill_va)
- dma_free_coherent(omap_dmm->dev,
+ dma_free_writecombine(omap_dmm->dev,
REFILL_BUFFER_SIZE * omap_dmm->num_engines,
omap_dmm->refill_va,
omap_dmm->refill_pa);
if (omap_dmm->dummy_page)
__free_page(omap_dmm->dummy_page);
- vfree(omap_dmm->lut);
-
if (omap_dmm->irq > 0)
free_irq(omap_dmm->irq, omap_dmm);
{
int ret = -EFAULT, i;
struct tcm_area area = {0};
- u32 hwinfo, pat_geom, lut_table_size;
+ u32 hwinfo, pat_geom;
struct resource *mem;
omap_dmm = kzalloc(sizeof(*omap_dmm), GFP_KERNEL);
INIT_LIST_HEAD(&omap_dmm->alloc_head);
INIT_LIST_HEAD(&omap_dmm->idle_head);
+ init_waitqueue_head(&omap_dmm->engine_queue);
+
/* lookup hwmod data - base address and irq */
mem = platform_get_resource(dev, IORESOURCE_MEM, 0);
if (!mem) {
omap_dmm->container_width = 256;
omap_dmm->container_height = 128;
+ atomic_set(&omap_dmm->engine_counter, omap_dmm->num_engines);
+
/* read out actual LUT width and height */
pat_geom = readl(omap_dmm->base + DMM_PAT_GEOMETRY);
omap_dmm->lut_width = ((pat_geom >> 16) & 0xF) << 5;
*/
writel(0x7e7e7e7e, omap_dmm->base + DMM_PAT_IRQENABLE_SET);
- lut_table_size = omap_dmm->lut_width * omap_dmm->lut_height *
- omap_dmm->num_lut;
-
- omap_dmm->lut = vmalloc(lut_table_size * sizeof(*omap_dmm->lut));
- if (!omap_dmm->lut) {
- dev_err(&dev->dev, "could not allocate lut table\n");
- ret = -ENOMEM;
- goto fail;
- }
-
omap_dmm->dummy_page = alloc_page(GFP_KERNEL | __GFP_DMA32);
if (!omap_dmm->dummy_page) {
dev_err(&dev->dev, "could not allocate dummy page\n");
omap_dmm->dummy_pa = page_to_phys(omap_dmm->dummy_page);
/* alloc refill memory */
- omap_dmm->refill_va = dma_alloc_coherent(&dev->dev,
+ omap_dmm->refill_va = dma_alloc_writecombine(&dev->dev,
REFILL_BUFFER_SIZE * omap_dmm->num_engines,
&omap_dmm->refill_pa, GFP_KERNEL);
if (!omap_dmm->refill_va) {
goto fail;
}
- sema_init(&omap_dmm->engine_sem, omap_dmm->num_engines);
for (i = 0; i < omap_dmm->num_engines; i++) {
omap_dmm->engines[i].id = i;
omap_dmm->engines[i].dmm = omap_dmm;
.p1.y = omap_dmm->container_height - 1,
};
- for (i = 0; i < lut_table_size; i++)
- omap_dmm->lut[i] = omap_dmm->dummy_pa;
-
/* initialize all LUTs to dummy page entries */
for (i = 0; i < omap_dmm->num_lut; i++) {
area.tcm = omap_dmm->tcm[i];