Skip to content

L
linux-elphel
Commit c92588c9 authored by Mikhail Karpenko's avatar Mikhail Karpenko
Browse files
Options

Frame alignment tests look fine

parent e41643e0
Show whitespace changes
Inline Side-by-side
Showing with 129 additions and 342 deletions
src/drivers/ata/ahci_elphel.c
View file @ c92588c9
......@@ -389,6 +389,8 @@ static void elphel_qc_prep(struct ata_queued_cmd *qc)
#define PHY_BLOCK_SIZE 512
//#define PHY_BLOCK_SIZE 4096
#define JPEG_MARKER_LEN 2
/** The size in bytes of JPEG marker length field */
#define JPEG_SIZE_LEN 2
#define SG_TBL_SZ 256
/** Include REM buffer to total size calculation */
#define INCLUDE_REM 1
......@@ -477,13 +479,17 @@ static int circbuf_get_ptr_tst(int sensor_port, size_t offset, size_t len, struc
if (use_preset == 0) {
if (jpg_0_sz != 0)
jpg_0_sz = jpg_0_sz % (DATA_BUFF_SIZE - 1);
if (jpg_1_sz != 0)
if (jpg_1_sz != 0) {
jpg_0_sz -= (jpg_0_sz % ALIGNMENT_SIZE);
jpg_1_sz = jpg_1_sz % (DATA_BUFF_SIZE - 1);
}
} else if (use_preset == 1) {
if (g_jpg_0_sz != 0)
jpg_0_sz = jpg_0_sz % g_jpg_0_sz;
if (g_jpg_1_sz != 0)
if (g_jpg_1_sz != 0) {
jpg_0_sz -= (jpg_0_sz % ALIGNMENT_SIZE);
jpg_1_sz = jpg_1_sz % g_jpg_1_sz;
}
} else if (use_preset == 2) {
jpg_0_sz = g_jpg_0_sz;
jpg_1_sz = g_jpg_1_sz;
......@@ -536,14 +542,6 @@ static int check_chunks(struct fvec *vects)
dev_err(NULL, "ERROR: unaligned write from slot %d, length %u\n", i, vects[i].iov_len);
ret = -1;
}
if ((vects[i].iov_dma % ALIGNMENT_ADDR) != 0) {
dev_err(NULL, "ERROR: unaligned DMA address in slot %d: 0x%x\n", i, vects[i].iov_dma);
ret = -1;
}
}
if ((i == CHUNK_ALIGN_0 || i == CHUNK_ALIGN_1) && vects[i].iov_len > 2 * ALIGNMENT_SIZE) {
dev_err(NULL, "ERROR: alignment buffer %d overflow\n", i);
ret = -1;
}
}
if ((sz % PHY_BLOCK_SIZE) != 0) {
......@@ -688,37 +686,9 @@ static inline unsigned char *vectrpos(struct fvec *vec, size_t offset)
{
return (unsigned char *)vec->iov_base + (vec->iov_len - offset);
}
/** Check if bus address will be properly aligned after moving vector forward by @e len bytes */
static inline int is_addr_aligned(const struct fvec *vec, size_t len)
{
int ret = 0;
if (vec->iov_len >= len) {
if (((vec->iov_dma + len) % ALIGNMENT_ADDR) == 0)
ret = 1;
}
return ret;
}
static size_t align_address(struct fvec *dest, struct fvec *src, size_t len)
{
size_t cut_bytes = (src->iov_dma + len) % ALIGNMENT_ADDR;
size_t pad_len = ALIGNMENT_SIZE - cut_bytes;
if (pad_len <= len) {
len -= pad_len;
}
if (len != 0) {
len += cut_bytes;
}
/* set marker length */
app15[3] = pad_len - JPEG_MARKER_LEN;
printk(KERN_DEBUG ">>> copy %u bytes from APP15 to other buffer\n", pad_len);
vectcpy(dest, app15, pad_len);
return len;
}
static void align_frame(struct device *dev, struct elphel_ahci_priv *dpriv)
{
int i;
int is_delayed = 0;
unsigned char *src;
size_t len, total_sz, data_len;
......@@ -778,332 +748,153 @@ static void align_frame(struct device *dev, struct elphel_ahci_priv *dpriv)
vectcpy(cbuff, chunks[CHUNK_HEADER].iov_base, len);
vectshrink(&chunks[CHUNK_HEADER], chunks[CHUNK_HEADER].iov_len);
/* check if we have enough data for further processing */
len = chunks[CHUNK_DATA_0].iov_len + chunks[CHUNK_DATA_1].iov_len + chunks[CHUNK_TRAILER].iov_len;
/* check if there is enough data to continue - JPEG data length can be too short */
len = get_size_from(chunks, CHUNK_DATA_0, 0, EXCLUDE_REM);
if (len < PHY_BLOCK_SIZE) {
data_len = cbuff->iov_len % PHY_BLOCK_SIZE;
dev_dbg(dev, "JPEG data size is too small: %u; unaligned data length: %u\n", len, data_len);
if (data_len + len <= fbuffs->rem_buff.iov_len) {
/* place data to REM buffer and continue */
src = vectrpos(cbuff, data_len);
printk(KERN_DEBUG ">>> 5.0.1 copy %u bytes from COMMON to REM buffer \n", len);
vectcpy(&chunks[CHUNK_REM], src, data_len);
vectshrink(cbuff, data_len);
printk(KERN_DEBUG ">>> 5.0.1 copy %u bytes from DATA_0 to REM buffer \n", len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_0].iov_base, chunks[CHUNK_DATA_0].iov_len);
vectshrink(&chunks[CHUNK_DATA_0], chunks[CHUNK_DATA_0].iov_len);
printk(KERN_DEBUG ">>> 5.0.1 copy %u bytes from DATA_1 to REM buffer \n", len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_1].iov_base, chunks[CHUNK_DATA_1].iov_len);
vectshrink(&chunks[CHUNK_DATA_1], chunks[CHUNK_DATA_1].iov_len);
printk(KERN_DEBUG ">>> 5.0.1 copy %u bytes from TRAILER to REM buffer \n", len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
} else {
/* the REM buffer is too short, process some buffers to align COMMON buffer to sector boundary */
dev_dbg(dev, "REM buffer is too short, align COMMON to sector boundary\n");
data_len = PHY_BLOCK_SIZE - data_len;
if (data_len >= chunks[CHUNK_DATA_0].iov_len) {
/* copy whole DATA_0 to COMMON */
printk(KERN_DEBUG ">>> 5.0.2 copy %u bytes from DATA_0 to COMMON buffer \n", chunks[CHUNK_DATA_0].iov_len);
size_t num = align_bytes_num(cbuff->iov_len, PHY_BLOCK_SIZE);
if (len >= num) {
/* there is enough data to align common buffer to sector boundary */
if (num >= chunks[CHUNK_DATA_0].iov_len) {
printk(KERN_DEBUG ">>> copy %u bytes from DATA_0 to common buffer\n", chunks[CHUNK_DATA_0].iov_len);
vectcpy(cbuff, chunks[CHUNK_DATA_0].iov_base, chunks[CHUNK_DATA_0].iov_len);
data_len -= chunks[CHUNK_DATA_0].iov_len;
num -= chunks[CHUNK_DATA_0].iov_len;
vectshrink(&chunks[CHUNK_DATA_0], chunks[CHUNK_DATA_0].iov_len);
} else {
/* copy a part of DATA_0 to COMMON */
printk(KERN_DEBUG ">>> 5.0.2 copy %u bytes from DATA_0 to COMMON buffer \n", data_len);
src = vectrpos(&chunks[CHUNK_DATA_0], data_len);
vectcpy(cbuff, src, data_len);
vectshrink(&chunks[CHUNK_DATA_0], data_len);
data_len = 0;
}
if (data_len != 0) {
if (data_len >= chunks[CHUNK_DATA_1].iov_len) {
/* copy whole DATA_1 to COMMON */
printk(KERN_DEBUG ">>> 5.0.2 copy %u bytes from DATA_1 to COMMON buffer \n", chunks[CHUNK_DATA_1].iov_len);
vectcpy(cbuff, chunks[CHUNK_DATA_1].iov_base, chunks[CHUNK_DATA_1].iov_len);
data_len -= chunks[CHUNK_DATA_1].iov_len;
vectshrink(&chunks[CHUNK_DATA_1], chunks[CHUNK_DATA_1].iov_len);
} else {
/* copy a part of DATA_1 to COMMON */
printk(KERN_DEBUG ">>> 5.0.2 copy %u bytes from DATA_1 to COMMON buffer \n", data_len);
src = vectrpos(&chunks[CHUNK_DATA_1], data_len);
vectcpy(cbuff, src, data_len);
vectshrink(&chunks[CHUNK_DATA_1], data_len);
data_len = 0;
}
if (data_len != 0) {
/* copy a part of TRAILER to COMMON */
printk(KERN_DEBUG ">>> 5.0.2 copy %u bytes from TRAILER to COMMON buffer \n", data_len);
src = vectrpos(&chunks[CHUNK_TRAILER], data_len);
vectcpy(cbuff, src, data_len);
vectshrink(&chunks[CHUNK_TRAILER], data_len);
data_len = 0;
}
}
/* COMMON buffer is aligned to sector boundary, copy all other data to REM buffer */
if (chunks[CHUNK_DATA_0].iov_len != 0) {
printk(KERN_DEBUG ">>> 5.0.3 copy %u bytes from DATA_0 to REM buffer \n", chunks[CHUNK_DATA_0].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_0].iov_base, chunks[CHUNK_DATA_0].iov_len);
vectshrink(&chunks[CHUNK_DATA_0], chunks[CHUNK_DATA_0].iov_len);
}
if (chunks[CHUNK_DATA_1].iov_len != 0) {
printk(KERN_DEBUG ">>> 5.0.3 copy %u bytes from DATA_1 to REM buffer \n", chunks[CHUNK_DATA_1].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_1].iov_base, chunks[CHUNK_DATA_1].iov_len);
vectshrink(&chunks[CHUNK_DATA_1], chunks[CHUNK_DATA_1].iov_len);
}
if (chunks[CHUNK_TRAILER].iov_len != 0) {
printk(KERN_DEBUG ">>> 5.0.3 copy %u bytes from TRAILER to REM buffer \n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
}
}
return;
}
/* there is enough data in data buffers, align common buffer */
len = align_bytes_num(cbuff->iov_len, ALIGNMENT_SIZE);
if (likely(len != 0)) {
dev_dbg(dev, "align COMMON buffer, add %u bytes from other buffers\n", len);
if ((len + ALIGNMENT_ADDR) <= chunks[CHUNK_DATA_0].iov_len) {
printk(KERN_DEBUG ">>> check alignment of DMA address: 0x%x\n", chunks[CHUNK_DATA_0].iov_dma);
if (is_addr_aligned(&chunks[CHUNK_DATA_0], len) == 0) {
/* we are going to copy =len= bytes from DATA_0 buffer to align COMMON buffer to
* ALIGNMNET_SIZE boundary which will make the starting address of DATA_0 buffer
* unaligned to ALIGNMENT_ADDR boundary, fix this by moving additional bytes from
* the beginning of DATA_0 buffer and padding COMMON buffer with APP15 marker */
len = align_address(cbuff, &chunks[CHUNK_DATA_0], len);
}
printk(KERN_DEBUG ">>> 5.1.1 copy %u bytes from DATA_0 to common buffer \n", len);
vectcpy(cbuff, chunks[CHUNK_DATA_0].iov_base, len);
vectmov(&chunks[CHUNK_DATA_0], len);
} else {
/* handle special case: JPEG data rolls over circular buffer boundary and the
* data chunk in the end of the buffer is less than we need for alignment; copy
* all data from first chunk, discard it, copy remainder from second chunk
* and from trailer if needed
*/
if (len <= (chunks[CHUNK_DATA_0].iov_len + chunks[CHUNK_DATA_1].iov_len + chunks[CHUNK_TRAILER].iov_len + ALIGNMENT_ADDR)) {
printk(KERN_DEBUG ">>> 5.2 copy %u bytes from DATA_0 to common buffer\n", chunks[CHUNK_DATA_0].iov_len);
vectcpy(cbuff, chunks[CHUNK_DATA_0].iov_base, chunks[CHUNK_DATA_0].iov_len);
len -= chunks[CHUNK_DATA_0].iov_len;
vectshrink(&chunks[CHUNK_DATA_0], chunks[CHUNK_DATA_0].iov_len);
if (chunks[CHUNK_DATA_1].iov_len >= len) {
printk(KERN_DEBUG ">>> check alignment of DMA address: 0x%x\n", chunks[CHUNK_DATA_1].iov_dma);
if (is_addr_aligned(&chunks[CHUNK_DATA_1], len) == 0) {
len = align_address(cbuff, &chunks[CHUNK_DATA_1], len);
}
printk(KERN_DEBUG ">>> 5.2.4 copy %u bytes from DATA_1 to common buffer\n", len);
vectcpy(cbuff, chunks[CHUNK_DATA_1].iov_base, len);
vectmov(&chunks[CHUNK_DATA_1], len);
len = 0;
} else {
printk(KERN_DEBUG ">>> 5.2.5 copy %u bytes from DATA_1 to common buffer\n", chunks[CHUNK_DATA_1].iov_len);
printk(KERN_DEBUG ">>> copy %u bytes from DATA_0 to common buffer\n", num);
src = vectrpos(&chunks[CHUNK_DATA_0], num);
vectcpy(cbuff, chunks[CHUNK_DATA_0].iov_base, num);
vectshrink(&chunks[CHUNK_DATA_0], num);
num = 0;
}
if (num >= chunks[CHUNK_DATA_1].iov_len) {
printk(KERN_DEBUG ">>> copy %u bytes from DATA_1 to common buffer\n", chunks[CHUNK_DATA_1].iov_len);
vectcpy(cbuff, chunks[CHUNK_DATA_1].iov_base, chunks[CHUNK_DATA_1].iov_len);
len -= chunks[CHUNK_DATA_1].iov_len;
num -= chunks[CHUNK_DATA_1].iov_len;
vectshrink(&chunks[CHUNK_DATA_1], chunks[CHUNK_DATA_1].iov_len);
}
if (chunks[CHUNK_TRAILER].iov_len >= len) {
printk(KERN_DEBUG ">>> 5.2.6 copy %u bytes from TRAILER to common buffer\n", len);
vectcpy(cbuff, chunks[CHUNK_TRAILER].iov_base, len);
vectmov(&chunks[CHUNK_TRAILER], len);
len = 0;
} else {
printk(KERN_DEBUG ">>> 5.2.7 copy %u bytes from TRAILER to common buffer\n", chunks[CHUNK_TRAILER].iov_len);
printk(KERN_DEBUG ">>> copy %u bytes from DATA_1 to common buffer\n", num);
src = vectrpos(&chunks[CHUNK_DATA_1], num);
vectcpy(cbuff, chunks[CHUNK_DATA_1].iov_base, num);
vectshrink(&chunks[CHUNK_DATA_1], num);
num = 0;
}
if (num >= chunks[CHUNK_TRAILER].iov_len) {
printk(KERN_DEBUG ">>> copy %u bytes from TRAILER to common buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(cbuff, chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
len -= chunks[CHUNK_TRAILER].iov_len;
num -= chunks[CHUNK_TRAILER].iov_len;
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
} else {
printk(KERN_DEBUG ">>> copy %u bytes from TRAILER to common buffer\n", num);
src = vectrpos(&chunks[CHUNK_TRAILER], num);
vectcpy(cbuff, chunks[CHUNK_TRAILER].iov_base, num);
vectshrink(&chunks[CHUNK_TRAILER], num);
num = 0;
}
} else {
/* JPEG data is too short, place all to remainder buffer */
/* there is not enough data to align common buffer to sector boundary, truncate common buffer */
data_len = cbuff->iov_len % PHY_BLOCK_SIZE;
src = vectrpos(cbuff, data_len);
printk(KERN_DEBUG ">>> 5.2.1 copy %u bytes from COMMON to REM buffer\n", data_len);
printk(KERN_DEBUG ">>> copy %u bytes from COMMON to REM buffer\n", data_len);
vectcpy(&chunks[CHUNK_REM], src, data_len);
vectshrink(cbuff, data_len);
printk(KERN_DEBUG ">>> 5.2.1 copy %u bytes from DATA_0 to REM buffer\n", chunks[CHUNK_DATA_0].iov_len);
}
printk(KERN_DEBUG ">>> copy %u bytes from DATA_0 to REM buffer\n", chunks[CHUNK_DATA_0].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_0].iov_base, chunks[CHUNK_DATA_0].iov_len);
vectshrink(&chunks[CHUNK_DATA_0], chunks[CHUNK_DATA_0].iov_len);
printk(KERN_DEBUG ">>> 5.2.1 copy %u bytes from DATA_1 to REM buffer\n", chunks[CHUNK_DATA_1].iov_len);
printk(KERN_DEBUG ">>> copy %u bytes from DATA_1 to REM buffer\n", chunks[CHUNK_DATA_1].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_1].iov_base, chunks[CHUNK_DATA_1].iov_len);
vectshrink(&chunks[CHUNK_DATA_1], chunks[CHUNK_DATA_1].iov_len);
printk(KERN_DEBUG ">>> 5.2.1 copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
printk(KERN_DEBUG ">>> copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
return;
}
/* align common buffer to ALIGNMENT boundary */
len = align_bytes_num(cbuff->iov_len, ALIGNMENT_SIZE);
if (len < JPEG_MARKER_LEN + JPEG_SIZE_LEN) {
/* the number of bytes needed for alignment is less than the length of the marker itself, increase the number of stuffing bytes */
len += ALIGNMENT_SIZE;
}
}
dev_dbg(dev, "total number of stuffing bytes in APP15 marker: %u\n", len);
app15[3] = len - JPEG_MARKER_LEN;
vectcpy(cbuff, app15, len);
/* JPEG header and Exif data is aligned to ALIGNMENT_SIZE boundary in a common buffer now,
* the entire frame should be aligned to physical sector boundary and remainder should be
* copied to a buffer for recording during next frame */
/* align frame to sector size boundary; total size could have changed by the moment - recalculate */
total_sz = get_size_from(chunks, 0, 0, INCLUDE_REM);
len = total_sz % PHY_BLOCK_SIZE;
chunks[CHUNK_ALIGN_0].iov_base = (unsigned char *)cbuff->iov_base + cbuff->iov_len;
chunks[CHUNK_ALIGN_0].iov_len = 0;
dev_dbg(dev, "total frame size: %u, unaligned bytes to sector boundary: %u\n", total_sz, len);
if (likely(chunks[CHUNK_DATA_1].iov_len == 0)) {
/* JPEG data is not split, align just one buffer to physical sector boundary if it is not already aligned */
if (chunks[CHUNK_DATA_0].iov_len + chunks[CHUNK_TRAILER].iov_len >= len) {
if (likely(len >= chunks[CHUNK_TRAILER].iov_len)) {
data_len = len - chunks[CHUNK_TRAILER].iov_len;
src = vectrpos(&chunks[CHUNK_DATA_0], data_len);
printk(KERN_DEBUG ">>> 6.1 copy %u bytes from DATA_0 to remainder buffer\n", data_len);
vectcpy(&chunks[CHUNK_REM], src, data_len);
vectshrink(&chunks[CHUNK_DATA_0], data_len);
printk(KERN_DEBUG ">>> 6.1 copy %u bytes from TRAILER to remainder buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
} else {
/* align DATA_0 buffer, break trailing marker and place first part to ALIGN_0 buffer and second part to REM */
data_len = chunks[CHUNK_DATA_0].iov_len % ALIGNMENT_SIZE;
printk(KERN_DEBUG ">>> 6.1.2 copy %u bytes from DATA_0 to ALIGN_0 buffer\n", data_len);
src = vectrpos(&chunks[CHUNK_DATA_0], data_len);
vectcpy(&chunks[CHUNK_ALIGN_0], src, data_len);
vectshrink(&chunks[CHUNK_DATA_0], data_len);
if (len == 0)
len = chunks[CHUNK_TRAILER].iov_len;
printk(KERN_DEBUG ">>> 6.1.2 copy %u bytes from TRAILER to ALIGN_0 buffer\n", len);
vectcpy(&chunks[CHUNK_ALIGN_0], chunks[CHUNK_TRAILER].iov_base, len);
vectmov(&chunks[CHUNK_TRAILER], len);
printk(KERN_DEBUG ">>> 6.1.2 copy %u bytes from TRAILER to remainder buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
}
} else {
/* the number of bytes needed to align to sector boundary is greater than the number of bytes left
* in DATA_0 and TRAILER buffers, realign COMMON buffer and delay DATA_0 and TRAILING buffers recording */
data_len = cbuff->iov_len % PHY_BLOCK_SIZE;
printk(KERN_DEBUG ">>> 6.1.3 copy %u bytes from COMMON to remainder buffer\n", data_len);
vectcpy(&chunks[CHUNK_REM], cbuff->iov_base, data_len);
vectshrink(cbuff, data_len);
printk(KERN_DEBUG ">>> 6.1.3 copy %u bytes from DATA_0 to remainder buffer\n", chunks[CHUNK_DATA_0].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_0].iov_base, chunks[CHUNK_DATA_0].iov_len);
vectshrink(&chunks[CHUNK_DATA_0], chunks[CHUNK_DATA_0].iov_len);
printk(KERN_DEBUG ">>> 6.1.3 copy %u bytes from TRAILER to remainder buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
}
} else {
printk(KERN_DEBUG ">>> 6.2 <<<\n");
/* JPEG data rolls over circular buffer boundary */
struct kvec data_rem;
data_len = chunks[CHUNK_DATA_0].iov_len % ALIGNMENT_SIZE;
size_t align_len = ALIGNMENT_SIZE - data_len;
chunks[CHUNK_ALIGN_1].iov_base = (unsigned char *)chunks[CHUNK_ALIGN_0].iov_base + chunks[CHUNK_ALIGN_0].iov_len;
chunks[CHUNK_ALIGN_1].iov_len = 0;
if (chunks[CHUNK_DATA_1].iov_len >= len + align_len) {
/* DATA_1 buffer contains enough bytes to align frame to sector boundary,
* align DATA_0 buffer to ALIGNMENT_SIZE boundary only */
if (data_len != 0) {
printk(KERN_DEBUG ">>> 6.2 copy %u bytes from DATA_0 to ALIGN_0 buffer\n", data_len);
src = vectrpos(&chunks[CHUNK_DATA_0], data_len);
vectcpy(&chunks[CHUNK_ALIGN_0], src, data_len);
vectshrink(&chunks[CHUNK_DATA_0], data_len);
printk(KERN_DEBUG ">>> check alignment of DMA address: 0x%x\n", chunks[CHUNK_DATA_1].iov_dma);
if (is_addr_aligned(&chunks[CHUNK_DATA_1], align_len) == 0) {
align_len = align_address(&chunks[CHUNK_ALIGN_0], &chunks[CHUNK_DATA_1], align_len);
}
printk(KERN_DEBUG ">>> 6.2 copy %u bytes from DATA_1 to ALIGN_0 buffer\n", align_len);
vectcpy(&chunks[CHUNK_ALIGN_0], chunks[CHUNK_DATA_1].iov_base, align_len);
vectmov(&chunks[CHUNK_DATA_1], align_len);
/* adjust ALIGN_1 position */
chunks[CHUNK_ALIGN_1].iov_base = (unsigned char *)chunks[CHUNK_ALIGN_0].iov_base + chunks[CHUNK_ALIGN_0].iov_len;
chunks[CHUNK_ALIGN_1].iov_len = 0;
}
} else {
/* there is no enough data in the second JPEG data buffer, delay its recording and
* align first buffer to physical sector boundary */
data_len = (cbuff->iov_len + chunks[CHUNK_DATA_0].iov_len) % PHY_BLOCK_SIZE;
printk(KERN_DEBUG ">>> 6.2.2 align to data_len: %u\n", data_len);
if (data_len <= chunks[CHUNK_DATA_0].iov_len) {
dev_dbg(dev, "number of bytes crossing sector boundary: %u\n", len);
if (len != 0) {
if (len >= (chunks[CHUNK_DATA_1].iov_len + chunks[CHUNK_TRAILER].iov_len)) {
/* current frame is not split or the second part of JPEG data is too short */
data_len = len - chunks[CHUNK_DATA_1].iov_len - chunks[CHUNK_TRAILER].iov_len;
src = vectrpos(&chunks[CHUNK_DATA_0], data_len);
printk(KERN_DEBUG ">>> 6.2.2 copy %u bytes from DATA_0 to REM buffer\n", data_len);
printk(KERN_DEBUG ">>> 5.1 copy %u bytes from DATA_0 to REM buffer\n", data_len);
vectcpy(&chunks[CHUNK_REM], src, data_len);
vectshrink(&chunks[CHUNK_DATA_0], data_len);
} else {
data_len -= chunks[CHUNK_DATA_0].iov_len;
src = vectrpos(cbuff, data_len);
printk(KERN_DEBUG ">>> 6.2.2 copy %u bytes from COMMON to REM buffer\n", data_len);
vectcpy(&chunks[CHUNK_REM], cbuff->iov_base, data_len);
vectshrink(cbuff, data_len);
printk(KERN_DEBUG ">>> 6.2.2 copy %u bytes from DATA_0 to REM buffer\n", chunks[CHUNK_DATA_0].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_0].iov_base, chunks[CHUNK_DATA_0].iov_len);
vectshrink(&chunks[CHUNK_DATA_0], chunks[CHUNK_DATA_0].iov_len);
/* COMMON buffer has been updated, adjust ALIGN_0 and ALIGN_1 position */
chunks[CHUNK_ALIGN_1].iov_base = (unsigned char *)chunks[CHUNK_ALIGN_0].iov_base + chunks[CHUNK_ALIGN_0].iov_len;
chunks[CHUNK_ALIGN_1].iov_len = 0;
chunks[CHUNK_ALIGN_1].iov_base = (unsigned char *)chunks[CHUNK_ALIGN_0].iov_base + chunks[CHUNK_ALIGN_0].iov_len;
chunks[CHUNK_ALIGN_1].iov_len = 0;
}
printk(KERN_DEBUG ">>> 6.2.2 copy %u bytes from DATA_1 to REM buffer\n", chunks[CHUNK_DATA_1].iov_len);
printk(KERN_DEBUG ">>> 5.1 copy %u bytes from DATA_1 to REM buffer\n", chunks[CHUNK_DATA_1].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_1].iov_base, chunks[CHUNK_DATA_1].iov_len);
vectshrink(&chunks[CHUNK_DATA_1], chunks[CHUNK_DATA_1].iov_len);
printk(KERN_DEBUG ">>> 6.2.2 copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
printk(KERN_DEBUG ">>> 5.1 copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
}
printk(KERN_DEBUG ">>> 6.3 <<<\n");
/* the total length could have changed by the moment, recalculate */
total_sz = get_size_from(chunks, 0, 0, INCLUDE_REM);
len = total_sz % PHY_BLOCK_SIZE;
dev_dbg(dev, "total frame size: %u, unaligned bytes to sector boundary: %u\n", total_sz, len);
if (chunks[CHUNK_DATA_1].iov_len + chunks[CHUNK_TRAILER].iov_len > len) {
if (len >= chunks[CHUNK_TRAILER].iov_len) {
/* DATA_1 and TRAILER buffers contain enough data to align the current frame to sector boundary */
} else if (len >= chunks[CHUNK_TRAILER].iov_len) {
/* there is enough data in second part to align the frame */
data_len = len - chunks[CHUNK_TRAILER].iov_len;
printk(KERN_DEBUG ">>> 6.3.1 copy %u bytes from DATA_1 to REM buffer\n", data_len);
src = vectrpos(&chunks[CHUNK_DATA_1], data_len);
printk(KERN_DEBUG ">>> 5.2 copy %u bytes from DATA_1 to REM buffer\n", data_len);
vectcpy(&chunks[CHUNK_REM], src, data_len);
vectshrink(&chunks[CHUNK_DATA_1], data_len);
printk(KERN_DEBUG ">>> 6.3.1 copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
} else if (len != 0) {
/* break trailing marker and copy the remaining data to REM buffer, align DATA_1 and TRAILER parts in
* ALIGN_1 buffer */
data_len = chunks[CHUNK_DATA_1].iov_len % ALIGNMENT_SIZE;
printk(KERN_DEBUG ">>> 6.3.2 copy %u bytes from DATA_1 to ALIGN_1 buffer\n", data_len);
src = vectrpos(&chunks[CHUNK_DATA_1], data_len);
vectcpy(&chunks[CHUNK_ALIGN_1], chunks[CHUNK_DATA_1].iov_base, data_len);
vectshrink(&chunks[CHUNK_DATA_1], data_len);
printk(KERN_DEBUG ">>> 6.3.2 copy %u bytes from TRAILER to ALIGN_1 buffer\n", len);
vectcpy(&chunks[CHUNK_ALIGN_1], chunks[CHUNK_TRAILER].iov_base, len);
vectmov(&chunks[CHUNK_TRAILER], len);
printk(KERN_DEBUG ">>> 6.3.2 copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
printk(KERN_DEBUG ">>> 5.2 copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
} else {
/* the frame is aligned to sector boundary, but DATA_1 and TRAILER buffers may be
* unaligned to ALIGNMENT_SIZE boundary */
data_len = chunks[CHUNK_DATA_1].iov_len % ALIGNMENT_SIZE;
if (data_len >= chunks[CHUNK_TRAILER].iov_len) {
printk(KERN_DEBUG ">>> 6.3.3 copy %u bytes from DATA_1 to ALIGN_1 buffer\n", data_len);
src = vectrpos(&chunks[CHUNK_DATA_1], data_len);
vectcpy(&chunks[CHUNK_ALIGN_1], src, data_len);
vectshrink(&chunks[CHUNK_DATA_1], data_len);
printk(KERN_DEBUG ">>> 6.3.3 copy %u bytes from TRAILER to ALIGN_1 buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_ALIGN_1], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
/* the trailing marker is split by sector boundary, copy (PHY_BLOCK_SIZE - 1) bytes from
* JPEG data block(s) to remainder buffer and then add trailing marker */
data_len = PHY_BLOCK_SIZE - (chunks[CHUNK_TRAILER].iov_len - len);
if (data_len >= chunks[CHUNK_DATA_1].iov_len) {
size_t cut_len = data_len - chunks[CHUNK_DATA_1].iov_len;
src = vectrpos(&chunks[CHUNK_DATA_0], cut_len);
printk(KERN_DEBUG ">>> 5.3 copy %u bytes from DATA_0 to REM buffer\n", cut_len);
vectcpy(&chunks[CHUNK_REM], src, cut_len);
vectshrink(&chunks[CHUNK_DATA_0], cut_len);
printk(KERN_DEBUG ">>> 5.3 copy %u bytes from DATA_1 to REM buffer\n", chunks[CHUNK_DATA_1].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_DATA_1].iov_base, chunks[CHUNK_DATA_1].iov_len);
vectshrink(&chunks[CHUNK_DATA_1], chunks[CHUNK_DATA_1].iov_len);
printk(KERN_DEBUG ">>> 5.3 copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
} else {
/* break trailing marker and copy the remaining data to REM buffer */
data_len = chunks[CHUNK_DATA_1].iov_len % ALIGNMENT_SIZE - len;
printk(KERN_DEBUG ">>> 6.3.4 copy %u bytes from DATA_1 to ALIGN_1 buffer\n", data_len);
src = vectrpos(&chunks[CHUNK_DATA_1], data_len);
vectcpy(&chunks[CHUNK_ALIGN_1], chunks[CHUNK_DATA_1].iov_base, data_len);
printk(KERN_DEBUG ">>> 5.3 copy %u bytes from DATA_1 to REM buffer\n", data_len);
vectcpy(&chunks[CHUNK_REM], src, data_len);
vectshrink(&chunks[CHUNK_DATA_1], data_len);
printk(KERN_DEBUG ">>> 6.3.4 copy %u bytes from TRAILER to ALIGN_1 buffer\n", len);
vectcpy(&chunks[CHUNK_ALIGN_1], chunks[CHUNK_TRAILER].iov_base, len);
vectmov(&chunks[CHUNK_TRAILER], len);
printk(KERN_DEBUG ">>> 6.3.4 copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
printk(KERN_DEBUG ">>> 5.3 copy %u bytes from TRAILER to REM buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_REM], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
}
}
} else {
/* the frame is aligned to sector boundary but some buffers may be not */
chunks[CHUNK_ALIGN].iov_base = vectrpos(cbuff, 0);
chunks[CHUNK_ALIGN].iov_len = 0;
if (chunks[CHUNK_DATA_1].iov_len == 0) {
data_len = chunks[CHUNK_DATA_0].iov_len % ALIGNMENT_SIZE;
src = vectrpos(&chunks[CHUNK_DATA_0], data_len);
printk(KERN_DEBUG ">>> 5.4 copy %u bytes from DATA_0 to ALIGN buffer\n", data_len);
vectcpy(&chunks[CHUNK_ALIGN], src, data_len);
vectshrink(&chunks[CHUNK_DATA_0], data_len);
} else {
data_len = chunks[CHUNK_DATA_1].iov_len % ALIGNMENT_SIZE;
src = vectrpos(&chunks[CHUNK_DATA_1], data_len);
printk(KERN_DEBUG ">>> 5.4 copy %u bytes from DATA_1 to ALIGN buffer\n", data_len);
vectcpy(&chunks[CHUNK_ALIGN], src, data_len);
vectshrink(&chunks[CHUNK_DATA_1], data_len);
}
printk(KERN_DEBUG ">>> 5.4 copy %u bytes from TRAILER to ALIGN buffer\n", chunks[CHUNK_TRAILER].iov_len);
vectcpy(&chunks[CHUNK_ALIGN], chunks[CHUNK_TRAILER].iov_base, chunks[CHUNK_TRAILER].iov_len);
vectshrink(&chunks[CHUNK_TRAILER], chunks[CHUNK_TRAILER].iov_len);
}
/* debug sanity check, should not happen */
......@@ -1252,10 +1043,10 @@ static int init_buffers(struct device *dev, struct frame_buffers *buffs)
if (dma_mapping_error(dev, buffs->common_buff.iov_dma))
goto err_common_dma;
buffs->rem_buff.iov_base = kmalloc(PHY_BLOCK_SIZE, GFP_KERNEL);
buffs->rem_buff.iov_base = kmalloc(2 * PHY_BLOCK_SIZE, GFP_KERNEL);
if (!buffs->rem_buff.iov_base)
goto err_remainder;
buffs->rem_buff.iov_len = PHY_BLOCK_SIZE;
buffs->rem_buff.iov_len = 2 * PHY_BLOCK_SIZE;
/* debug code follows */
g_jpg_data_0 = kzalloc(DATA_BUFF_SIZE, GFP_KERNEL);
......@@ -1460,7 +1251,6 @@ static ssize_t rawdev_write(struct device *dev, ///<
printk(KERN_DEBUG ">>> bytes received from exif driver: %u\n", rcvd);
if (rcvd > 0 && rcvd < buffs->exif_buff.iov_len)
// print_hex_dump_bytes("", DUMP_PREFIX_OFFSET, buffs->exif_buff.iov_base, rcvd);
printk(KERN_DEBUG ">>>\tskipped");
chunks[CHUNK_EXIF].iov_len = rcvd;
// rcvd = jpeghead_get_data(fdata.sensor_port, buffs->jpheader_buff.iov_base, buffs->jpheader_buff.iov_len, 0);
......
src/drivers/ata/ahci_elphel.h
View file @ c92588c9
......@@ -41,13 +41,11 @@
* trailing marker, and pointer to a buffer containing the remainder of a
* frame. Ten chunks of data in total.
* @todo Fix description */
#define MAX_DATA_CHUNKS 10
#define MAX_DATA_CHUNKS 9
/** Default port number */
#define DEFAULT_PORT_NUM 0
/** Align buffers length to this amount of bytes */
#define ALIGNMENT_SIZE 32
/** Address alignment boundary */
#define ALIGNMENT_ADDR 2
/** This structure holds raw device buffer pointers */
struct drv_pointers {
......@@ -88,10 +86,9 @@ enum {
CHUNK_HEADER,
CHUNK_COMMON,
CHUNK_DATA_0,
CHUNK_ALIGN_0,
CHUNK_DATA_1,
CHUNK_ALIGN_1,
CHUNK_TRAILER,
CHUNK_ALIGN,
CHUNK_REM
};
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment