{
struct ivtv_stream *s = &itv->streams[itv->cur_pio_stream];
struct ivtv_buffer *buf;
- struct list_head *p;
int i = 0;
IVTV_DEBUG_HI_DMA("ivtv_pio_work_handler\n");
return;
}
IVTV_DEBUG_HI_DMA("Process PIO %s\n", s->name);
- buf = list_entry(s->q_dma.list.next, struct ivtv_buffer, list);
- list_for_each(p, &s->q_dma.list) {
- struct ivtv_buffer *buf = list_entry(p, struct ivtv_buffer, list);
+ list_for_each_entry(buf, &s->q_dma.list, list) {
u32 size = s->sg_processing[i].size & 0x3ffff;
/* Copy the data from the card to the buffer */
{
struct ivtv *itv = s->itv;
struct ivtv_buffer *buf;
- struct list_head *p;
u32 bytes_needed = 0;
u32 offset, size;
u32 UVoffset = 0, UVsize = 0;
/* got the buffers, now fill in sg_pending */
buf = list_entry(s->q_predma.list.next, struct ivtv_buffer, list);
memset(buf->buf, 0, 128);
- list_for_each(p, &s->q_predma.list) {
- struct ivtv_buffer *buf = list_entry(p, struct ivtv_buffer, list);
-
+ list_for_each_entry(buf, &s->q_predma.list, list) {
if (skip_bufs-- > 0)
continue;
s->sg_pending[idx].dst = buf->dma_handle;
if (buf)
buf->bytesused += s->dma_last_offset;
if (buf && s->type == IVTV_DEC_STREAM_TYPE_VBI) {
- list_for_each(p, &s->q_dma.list) {
- buf = list_entry(p, struct ivtv_buffer, list);
-
+ list_for_each_entry(buf, &s->q_dma.list, list) {
/* Parse and Groom VBI Data */
s->q_dma.bytesused -= buf->bytesused;
ivtv_process_vbi_data(itv, buf, 0, s->type);
{
struct ivtv *itv = s->itv;
struct ivtv_buffer *buf;
- struct list_head *p;
u32 y_size = itv->params.height * itv->params.width;
u32 uv_offset = offset + IVTV_YUV_BUFFER_UV_OFFSET;
int y_done = 0;
int idx = 0;
IVTV_DEBUG_HI_DMA("DEC PREPARE DMA %s: %08x %08x\n", s->name, s->q_predma.bytesused, offset);
- buf = list_entry(s->q_predma.list.next, struct ivtv_buffer, list);
- list_for_each(p, &s->q_predma.list) {
- struct ivtv_buffer *buf = list_entry(p, struct ivtv_buffer, list);
-
+ list_for_each_entry(buf, &s->q_predma.list, list) {
/* YUV UV Offset from Y Buffer */
if (s->type == IVTV_DEC_STREAM_TYPE_YUV && !y_done && bytes_written >= y_size) {
offset = uv_offset;
we just drop the old requests when there are already three
requests queued. */
if (s->sg_pending_size > 2) {
- struct list_head *p;
- list_for_each(p, &s->q_predma.list) {
- struct ivtv_buffer *buf = list_entry(p, struct ivtv_buffer, list);
+ struct ivtv_buffer *buf;
+ list_for_each_entry(buf, &s->q_predma.list, list)
ivtv_buf_sync_for_cpu(s, buf);
- }
ivtv_queue_move(s, &s->q_predma, NULL, &s->q_free, 0);
s->sg_pending_size = 0;
}