mirror of
https://github.com/FEX-Emu/linux.git
synced 2024-12-15 05:11:32 +00:00
mxc_nand: introduce mxc_do_addr_cycle
This factors the address cycle to a seperate function. This becomes useful in a later patch where we can simplify the command processing by making use of this function. Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This commit is contained in:
parent
06ecb04ac5
commit
a3e65b64d5
@ -658,6 +658,52 @@ static void mxc_nand_select_chip(struct mtd_info *mtd, int chip)
|
||||
}
|
||||
}
|
||||
|
||||
static void mxc_do_addr_cycle(struct mtd_info *mtd, int column, int page_addr)
|
||||
{
|
||||
struct nand_chip *nand_chip = mtd->priv;
|
||||
struct mxc_nand_host *host = nand_chip->priv;
|
||||
|
||||
/* Write out column address, if necessary */
|
||||
if (column != -1) {
|
||||
/*
|
||||
* MXC NANDFC can only perform full page+spare or
|
||||
* spare-only read/write. When the upper layers
|
||||
* layers perform a read/write buf operation,
|
||||
* we will used the saved column adress to index into
|
||||
* the full page.
|
||||
*/
|
||||
send_addr(host, 0, page_addr == -1);
|
||||
if (host->pagesize_2k)
|
||||
/* another col addr cycle for 2k page */
|
||||
send_addr(host, 0, false);
|
||||
}
|
||||
|
||||
/* Write out page address, if necessary */
|
||||
if (page_addr != -1) {
|
||||
/* paddr_0 - p_addr_7 */
|
||||
send_addr(host, (page_addr & 0xff), false);
|
||||
|
||||
if (host->pagesize_2k) {
|
||||
if (mtd->size >= 0x10000000) {
|
||||
/* paddr_8 - paddr_15 */
|
||||
send_addr(host, (page_addr >> 8) & 0xff, false);
|
||||
send_addr(host, (page_addr >> 16) & 0xff, true);
|
||||
} else
|
||||
/* paddr_8 - paddr_15 */
|
||||
send_addr(host, (page_addr >> 8) & 0xff, true);
|
||||
} else {
|
||||
/* One more address cycle for higher density devices */
|
||||
if (mtd->size >= 0x4000000) {
|
||||
/* paddr_8 - paddr_15 */
|
||||
send_addr(host, (page_addr >> 8) & 0xff, false);
|
||||
send_addr(host, (page_addr >> 16) & 0xff, true);
|
||||
} else
|
||||
/* paddr_8 - paddr_15 */
|
||||
send_addr(host, (page_addr >> 8) & 0xff, true);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* Used by the upper layer to write command to NAND Flash for
|
||||
* different operations to be carried out on NAND Flash */
|
||||
static void mxc_nand_command(struct mtd_info *mtd, unsigned command,
|
||||
@ -746,46 +792,7 @@ static void mxc_nand_command(struct mtd_info *mtd, unsigned command,
|
||||
|
||||
/* Write out the command to the device. */
|
||||
send_cmd(host, command, useirq);
|
||||
|
||||
/* Write out column address, if necessary */
|
||||
if (column != -1) {
|
||||
/*
|
||||
* MXC NANDFC can only perform full page+spare or
|
||||
* spare-only read/write. When the upper layers
|
||||
* layers perform a read/write buf operation,
|
||||
* we will used the saved column adress to index into
|
||||
* the full page.
|
||||
*/
|
||||
send_addr(host, 0, page_addr == -1);
|
||||
if (host->pagesize_2k)
|
||||
/* another col addr cycle for 2k page */
|
||||
send_addr(host, 0, false);
|
||||
}
|
||||
|
||||
/* Write out page address, if necessary */
|
||||
if (page_addr != -1) {
|
||||
/* paddr_0 - p_addr_7 */
|
||||
send_addr(host, (page_addr & 0xff), false);
|
||||
|
||||
if (host->pagesize_2k) {
|
||||
if (mtd->size >= 0x10000000) {
|
||||
/* paddr_8 - paddr_15 */
|
||||
send_addr(host, (page_addr >> 8) & 0xff, false);
|
||||
send_addr(host, (page_addr >> 16) & 0xff, true);
|
||||
} else
|
||||
/* paddr_8 - paddr_15 */
|
||||
send_addr(host, (page_addr >> 8) & 0xff, true);
|
||||
} else {
|
||||
/* One more address cycle for higher density devices */
|
||||
if (mtd->size >= 0x4000000) {
|
||||
/* paddr_8 - paddr_15 */
|
||||
send_addr(host, (page_addr >> 8) & 0xff, false);
|
||||
send_addr(host, (page_addr >> 16) & 0xff, true);
|
||||
} else
|
||||
/* paddr_8 - paddr_15 */
|
||||
send_addr(host, (page_addr >> 8) & 0xff, true);
|
||||
}
|
||||
}
|
||||
mxc_do_addr_cycle(mtd, column, page_addr);
|
||||
|
||||
/* Command post-processing step */
|
||||
switch (command) {
|
||||
|
Loading…
Reference in New Issue
Block a user