sf: atmel: implement power-of-two write/erase funcs
PATCH 1/1 sf: atmel: implement power-of-two write/erase funcs
From: Todor I Mollov <[email protected]>
--- drivers/mtd/spi/atmel.c | 139 +++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 139 insertions(+), 0 deletions(-)
drivers/mtd/spi/atmel.c +139 -0
@@ -196,6 +196,75 @@ static int dataflash_read_fast_at45(struct spi_flash *flash,return spi_flash_read_common (flash, cmd,sizeof (cmd), buf, len); } +/* +* TODO: the two write funcs (_p2/_at45) should get unified ... +*/ +static int dataflash_write_p2 (struct spi_flash *flash, +u32 offset,size_t len,const void *buf) +{ +struct atmel_spi_flash *asf =to_atmel_spi_flash (flash); +unsigned long page_size; +u32 addr = offset; +size_t chunk_len; +size_t actual; +int ret; +u8 cmd[4 ]; + +/* +* TODO: This function currently uses only page buffer #1. We can +* speed this up by using both buffers and loading one buffer while +* the other is being programmed into main memory. +*/ + + page_size = (1 << asf->params->l2_page_size); + + ret =spi_claim_bus (flash->spi); +if (ret) { +debug ("SF: Unable to claim SPI bus\n" ); +return ret; + } + +for (actual =0 ; actual < len; actual += chunk_len) { + chunk_len =min (len - actual, page_size - (addr % page_size)); + +/* Use the same address bits for both commands */ + cmd[0 ] =CMD_AT45_LOAD_BUF1 ; + cmd[1 ] = addr >>16 ; + cmd[2 ] = addr >>8 ; + cmd[3 ] = addr; + + ret =spi_flash_cmd_write (flash->spi, cmd,4 , + buf + actual, chunk_len); +if (ret <0 ) { +debug ("SF: Loading AT45 buffer failed\n" ); + goto out; + } + + cmd[0 ] =CMD_AT45_PROG_BUF1 ; + ret =spi_flash_cmd_write (flash->spi, cmd,4 ,NULL ,0 ); +if (ret <0 ) { +debug ("SF: AT45 page programming failed\n" ); + goto out; + } + + ret =at45_wait_ready (flash,SPI_FLASH_PROG_TIMEOUT ); +if (ret <0 ) { +debug ("SF: AT45 page programming timed out\n" ); + goto out; + } + + addr += chunk_len; + } + +debug ("SF: AT45: Successfully programmed %zu bytes @ 0x%x\n" , + len, offset); + ret =0 ; + +out: +spi_release_bus (flash->spi); +return ret; +} +static int dataflash_write_at45 (struct spi_flash *flash,u32 offset,size_t len,const void *buf) {
@@ -209,6 +278,12 @@ static int dataflash_write_at45(struct spi_flash *flash,int ret;u8 cmd[4 ]; +/* +* TODO: This function currently uses only page buffer #1. We can +* speed this up by using both buffers and loading one buffer while +* the other is being programmed into main memory. +*/ + page_shift = asf->params->l2_page_size; page_size = (1 << page_shift) + (1 << (page_shift -5 )); page_shift++;
@@ -263,6 +338,68 @@ out:return ret; } +/* +* TODO: the two erase funcs (_p2/_at45) should get unified ... +*/ +int dataflash_erase_p2 (struct spi_flash *flash,u32 offset,size_t len) +{ +struct atmel_spi_flash *asf =to_atmel_spi_flash (flash); +unsigned long page_size; + +size_t actual; +int ret; +u8 cmd[4 ]; + +/* +* TODO: This function currently uses page erase only. We can +* probably speed things up by using block and/or sector erase +* when possible. +*/ + + page_size = (1 << asf->params->l2_page_size); + +if (offset % page_size || len % page_size) { +debug ("SF: Erase offset/length not multiple of page size\n" ); +return -1 ; + } + + cmd[0 ] =CMD_AT45_ERASE_PAGE ; + cmd[3 ] =0x00 ; + + ret =spi_claim_bus (flash->spi); +if (ret) { +debug ("SF: Unable to claim SPI bus\n" ); +return ret; + } + +for (actual =0 ; actual < len; actual += page_size) { + cmd[1 ] = offset >>16 ; + cmd[2 ] = offset >>8 ; + + ret =spi_flash_cmd_write (flash->spi, cmd,4 ,NULL ,0 ); +if (ret <0 ) { +debug ("SF: AT45 page erase failed\n" ); + goto out; + } + + ret =at45_wait_ready (flash,SPI_FLASH_PAGE_ERASE_TIMEOUT ); +if (ret <0 ) { +debug ("SF: AT45 page erase timed out\n" ); + goto out; + } + + offset += page_size; + } + +debug ("SF: AT45: Successfully erased %zu bytes @ 0x%x\n" , + len, offset); + ret =0 ; + +out: +spi_release_bus (flash->spi); +return ret; +} +int dataflash_erase_at45 (struct spi_flash *flash,u32 offset,size_t len) {struct atmel_spi_flash *asf =to_atmel_spi_flash (flash);
@@ -382,6 +519,8 @@ struct spi_flash *spi_flash_probe_atmel(struct spi_slave *spi, u8 *idcode) page_size +=1 << (params->l2_page_size -5 ); }else { asf->flash.read = dataflash_read_fast_p2; + asf->flash.write = dataflash_write_p2; + asf->flash.erase = dataflash_erase_p2; }break ;
Signature · 1 line
1.6.2
PATCH 1/1 sf: atmel: implement power-of-two write/erase funcs
Dear Mike Frysinger,
Quoted 8 lines · In message <[email protected]> you wrote:
From: Todor I Mollov <[email protected]>
Signed-off-by Todor I Mollov <[email protected]>Signed-off-by Mike Frysinger <[email protected]>CC Haavard Skinnemoen <[email protected]>--- drivers/mtd/spi/atmel.c | 139 +++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 139 insertions(+), 0 deletions(-)
I understand this is not a bug fix but new stuff, i. e. for the next release?
Best regards,
Wolfgang Denk
Signature · 5 lines
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de The Wright Bothers weren't the first to fly. They were just the first not to crash.
PATCH 1/1 sf: atmel: implement power-of-two write/erase funcs
Quoted 11 lines · On Saturday 04 April 2009 09:39:14 Wolfgang Denk wrote:
Quoted 9 lines · In message Mike Frysinger
From: Todor I Mollov <[email protected]>
Signed-off-by Todor I Mollov <[email protected]>Signed-off-by Mike Frysinger <[email protected]>CC Haavard Skinnemoen <[email protected]>--- drivers/mtd/spi/atmel.c | 139+++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 139 insertions(+), 0 deletions(-)
Quoted 2 lines
I understand this is not a bug fix but new stuff, i. e. for the next release?
correct. i'm posting it for feedback in case someone notices a problem, or the atmel guys want something different. i'll keep it in my sf branch for next release so it doesnt get lost.
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
Url : http://lists.denx.de/pipermail/u-boot/attachments/20090404/43c19143/attachment-0001.pgp
PATCH 1/1 sf: atmel: implement power-of-two write/erase funcs
Quoted 12 lines · On Saturday 04 April 2009 09:39:14 Wolfgang Denk wrote:
Quoted 11 lines
Quoted 9 lines · In message Mike Frysinger
From: Todor I Mollov <[email protected]>
Signed-off-by Todor I Mollov <[email protected]>Signed-off-by Mike Frysinger <[email protected]>CC Haavard Skinnemoen <[email protected]>--- drivers/mtd/spi/atmel.c | 139+++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 139 insertions(+), 0 deletions(-)
Quoted 9 lines
Quoted 2 lines
I understand this is not a bug fix but new stuff, i. e. for the next release?
correct. i'm posting it for feedback in case someone notices a problem, or
the atmel guys want something different. i'll keep it in my sf branch for next release so it doesnt get lost.
-mike
Sorry for the terminology mix-up there. I guess I labeled it as a bug fix since, prior to the patch, attempting to write or erase and at45 flash in powers-of-two mode caused u-boot to crash due to null function pointer call. In my head things that lead to crashes get labeled as bugs.
This occurred in both 2009.01 and 2009.03.
It would probably be a good idea to implement a check in the upper level of the spi flash driver to handle such issues and report them gracefully ;). (It definitely left me scratching my head for a while there until I familiarized myself with the driver structure a bit.)
-Todor