[U-Boot-Users] NAND flash support
Hi,
Correct me if I'm wrong but u-boot only support 512 byte page size nand flash components. Is there going to be any nand flash driver support for 2048 bytes (8-bit data bus) page size components?
Thanks, Terence.
In message OFCE5A7B44.FA9F0802-ON482570AB.00300B9C-482570AB.003050F6@applera.com you wrote:
Correct me if I'm wrong but u-boot only support 512 byte page size nand flash components. Is there going to be any nand flash driver support for 2048 bytes (8-bit data bus) page size components?
The #testing-NAND branch ofthe U-Boot GIT repository contains a major rewrite of NAND support. Unfortunately we don't make any progress with merging it into the main branch because nobody seems to be testing or using it.
Best regards,
Wolfgang Denk
On 10/31/05, Wolfgang Denk wd@denx.de wrote:
The #testing-NAND branch ofthe U-Boot GIT repository contains a major rewrite of NAND support. Unfortunately we don't make any progress with merging it into the main branch because nobody seems to be testing or using it.
I've been trying to use the new NAND code on custom board based on MPC8272 with Samsung 128Mb NAND chip (K9K1G08U0M). So far I had no problems with read/dump functionality. I've implemented the nand_erase wrapper in the include/nand.h and after restoring the most of nand_wait in drivers/nand/nand_base.c the erase worked for me. The only thing that I haven't succeeded to enable is the write. When I try to write relatively large chunks of data, there is a bus fault just after I see the message "NAND write: device 0 offset 0, size 1048576 ... 1048576 bytes written: OK". Besides, the blocks that were written are marked bad.
The relevant parts of my board configuration are: #define CFG_MAX_NAND_DEVICE 1 #define CFG_NAND_BASE 0xff000000 #define CONFIG_NEW_NAND_CODE 1 #define CONFIG_MTD_DEBUG 1 #define CONFIG_MTD_DEBUG_VERBOSE 1
The functions for HW access are the same as I use in Linux mtd driver which works fine.
-- Sincerely Yours, Mike.
On Tue, Nov 01, 2005 at 09:46:33AM +0200, Mike Rapoport wrote:
On 10/31/05, Wolfgang Denk wd@denx.de wrote:
The #testing-NAND branch ofthe U-Boot GIT repository contains a major rewrite of NAND support. Unfortunately we don't make any progress with merging it into the main branch because nobody seems to be testing or using it.
I've been trying to use the new NAND code on custom board based on MPC8272 with Samsung 128Mb NAND chip (K9K1G08U0M). So far I had no problems with read/dump functionality. I've implemented the nand_erase wrapper in the include/nand.h and after restoring the most of nand_wait in drivers/nand/nand_base.c the erase worked for me.
Could you please provide a patch?
The only thing that I haven't succeeded to enable is the write. When I try to write relatively large chunks of data, there is a bus fault just after I see the message "NAND write: device 0 offset 0, size 1048576 ... 1048576 bytes written: OK". Besides, the blocks that were written are marked bad.
The relevant parts of my board configuration are: #define CFG_MAX_NAND_DEVICE 1 #define CFG_NAND_BASE 0xff000000 #define CONFIG_NEW_NAND_CODE 1 #define CONFIG_MTD_DEBUG 1 #define CONFIG_MTD_DEBUG_VERBOSE 1
The functions for HW access are the same as I use in Linux mtd driver which works fine.
What version of MTD code are you using in your Linux kernel? I'm just updating U-Boot code to be based on what's contained in 2.6.14, which _may_ address some problems people were reporting before. I hope to get time to send update next week.
Thanks, ladis
On 11/1/05, Ladislav Michl ladis@linux-mips.org wrote:
Could you please provide a patch?
I've tried to make it as clean as possible, but there still may be some junk I used for debugging.
What version of MTD code are you using in your Linux kernel? I'm just updating U-Boot code to be based on what's contained in 2.6.14, which _may_ address some problems people were reporting before. I hope to get time to send update next week.
I use kernel 2.6.12.3 and except a wrong ID for 512Mb NAND flashe, the MTD code contained there works fine for me.
-- Sincerely Yours, Mike.
On Tue, Nov 01, 2005 at 04:22:28PM +0200, Mike Rapoport wrote:
I've tried to make it as clean as possible, but there still may be some junk I used for debugging.
Thanks a lot. Updated version (with timeout handling) follows - generated against testing-NAND branch. Wolfgang, could you apply?
I also got another board with NAND chip for testing and I was able to reproduce your problem with with bus error. I'll look at it as soon as I get some spare time.
Signed-off-by: Ladislav Michl ladis@linux-mips.org
CHANGELOG: * Add hook to NAND erase and implement nand_wait function. Patch by Mike Rapoport, 1 Nov 2005
diff --git a/common/Makefile b/common/Makefile --- a/common/Makefile +++ b/common/Makefile @@ -37,7 +37,7 @@ COBJS = main.o ACEX1K.o altera.o bedbug. cmd_i2c.o cmd_ide.o cmd_immap.o cmd_itest.o cmd_jffs2.o \ cmd_load.o cmd_log.o \ cmd_mem.o cmd_mii.o cmd_misc.o cmd_mmc.o \ - cmd_nand.o cmd_net.o cmd_nvedit.o \ + cmd_nand.o cmd_nand_new.o cmd_net.o cmd_nvedit.o \ cmd_pci.o cmd_pcmcia.o cmd_portio.o \ cmd_reginfo.o cmd_reiser.o cmd_scsi.o cmd_spi.o cmd_universe.o cmd_usb.o cmd_vfd.o \ command.o console.o devices.o dlmalloc.o docecc.o \ diff --git a/drivers/nand/nand_base.c b/drivers/nand/nand_base.c --- a/drivers/nand/nand_base.c +++ b/drivers/nand/nand_base.c @@ -189,7 +189,11 @@ static void nand_release_device (struct spin_unlock (&this->chip_lock); } #else -#define nand_release_device(mtd) do {} while(0) +static void nand_release_device (struct mtd_info *mtd) +{ + struct nand_chip *this = mtd->priv; + this->select_chip(mtd, -1); /* De-select the NAND device */ +} #endif
/** @@ -831,8 +835,34 @@ static int nand_wait(struct mtd_info *mt #else static int nand_wait(struct mtd_info *mtd, struct nand_chip *this, int state) { - /* TODO */ - return 0; + unsigned long timeo; + + if (state == FL_ERASING) + timeo = CFG_HZ * 400; + else + timeo = CFG_HZ * 20; + + if ((state == FL_ERASING) && (this->options & NAND_IS_AND)) + this->cmdfunc(mtd, NAND_CMD_STATUS_MULTI, -1, -1); + else + this->cmdfunc(mtd, NAND_CMD_STATUS, -1, -1); + + reset_timer_masked(); + + while (1) { + if (get_timer_masked() > timeo) + return 0; + + if (this->dev_ready) { + if (this->dev_ready(mtd)) + break; + } else { + if (this->read_byte(mtd) & NAND_STATUS_READY) + break; + } + } + + return this->read_byte(mtd); } #endif
diff --git a/include/bmp_logo.h b/include/bmp_logo.h diff --git a/include/nand.h b/include/nand.h --- a/include/nand.h +++ b/include/nand.h @@ -50,7 +50,14 @@ static inline int nand_block_isbad(nand_
static inline int nand_erase(nand_info_t *info, ulong off, ulong size) { - return 0; /* FIXME */ + struct erase_info instr; + + instr.mtd = info; + instr.addr = off; + instr.len = size; + instr.callback = 0; + + return info->erase(info, &instr); }
#endif
participants (4)
-
Ladislav Michl -
Mike Rapoport -
Terence Soh -
Wolfgang Denk