* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue @ 2006-08-04 12:32 Sam Song 2006-08-04 13:44 ` Wolfgang Denk 2006-08-07 7:24 ` Zang Roy-r61911 0 siblings, 2 replies; 31+ messages in thread From: Sam Song @ 2006-08-04 12:32 UTC (permalink / raw) To: u-boot Hello Ed Okerson, Recently, Roy Zang released MPC7448 HPC-II platform support to this list and found a CFI Flash driver problem - North-bridge chip TSI108 working as little-endian but write to Flash is byte-swapped in a wrong way. The workaround is to enable CFG_FLASH_USE_BUFFER_WRITE in little endian setting. However, CFG_FLASH_USE_BUFFER_WRITE should have nothing to do with little endian if I am right. So I suspect flash writing with little endian could have some problems. Once I removed little endian stuff in flash_add_byte(), CFI Driver was funtional for MPC7448 HPC-II with/without CFG_FLASH_USE_BUFFER_WRITE. So I'd like to make sure whether my fix is a general thing or just a luck. Thanks, Sam ___________________________________________________________ Mp3???-??????? http://music.yahoo.com.cn/?source=mail_mailbox_footer -------------- next part -------------- A non-text attachment was scrubbed... Name: cfi.diff Type: text/x-patch Size: 1490 bytes Desc: pat1625328208 Url : http://lists.denx.de/pipermail/u-boot/attachments/20060804/b5ce54e5/attachment.bin ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-04 12:32 [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue Sam Song @ 2006-08-04 13:44 ` Wolfgang Denk 2006-08-07 10:34 ` Sam Song 2006-08-07 11:04 ` Yuli Barcohen 2006-08-07 7:24 ` Zang Roy-r61911 1 sibling, 2 replies; 31+ messages in thread From: Wolfgang Denk @ 2006-08-04 13:44 UTC (permalink / raw) To: u-boot Dear Sam, in message <20060804123231.77800.qmail@web15908.mail.cnb.yahoo.com> you wrote: > > Once I removed little endian stuff in > flash_add_byte(), CFI Driver was funtional > for MPC7448 HPC-II with/without > CFG_FLASH_USE_BUFFER_WRITE. So I'd like to > make sure whether my fix is a general thing > or just a luck. Ummm... did you try running the resulting code on any real LE system? > static void flash_add_byte (flash_info_t * info, cfiword_t * cword, uchar c) > { > -#if defined(__LITTLE_ENDIAN) > - unsigned short w; > - unsigned int l; > - unsigned long long ll; > -#endif ... I understand that removing this code will break all real little- endian systems. > diff --git a/include/asm-ppc/processor.h b/include/asm-ppc/processor.h > index ea5b0d2..3d3c96f 100644 > --- a/include/asm-ppc/processor.h > +++ b/include/asm-ppc/processor.h > @@ -12,6 +12,12 @@ #include <linux/config.h> > #include <asm/ptrace.h> > #include <asm/types.h> > > +#ifdef CONFIG_MPC7448HPC2 > +#ifndef __LITTLE_ENDIAN > +#define __LITTLE_ENDIAN > +#endif /* __LITTLE_ENDIAN ** USI-SS */ > +#endif /* CONFIG_MPC7448HPC2 */ > + And this is definitely inappropirate, too. All PowerPC systems we have so far are BE, including yours, event hough it performs funny (read: broken) memory accesses. Defining __LITTLE_ENDIAN on a big-endian system is fundamentally broken. I will never accept such a patch. Best regards, Wolfgang Denk -- Software Engineering: Embedded and Realtime Systems, Embedded Linux Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de Crash programs fail because they are based on the theory that, with nine women pregnant, you can get a baby a month. - Wernher von Braun ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-04 13:44 ` Wolfgang Denk @ 2006-08-07 10:34 ` Sam Song 2006-08-07 11:04 ` Yuli Barcohen 1 sibling, 0 replies; 31+ messages in thread From: Sam Song @ 2006-08-07 10:34 UTC (permalink / raw) To: u-boot Wolfgang Denk <wd@denx.de> wrote: > Ummm... did you try running the resulting code on > any real LE system? No. I have no such a platform. I'd like to get any feedback from LE real test. > I understand that removing this code will break > all real little- > endian systems. If so, then drop my fixes. The puzzle to me is the difference between TSI108 working as LE mode and a real LE case. > > +#ifdef CONFIG_MPC7448HPC2 > > +#ifndef __LITTLE_ENDIAN > > +#define __LITTLE_ENDIAN [snip] > Defining __LITTLE_ENDIAN on a big-endian system > is fundamentally > broken. I will never accept such a patch. Ummm, it's just my RFC hack. Should haven't released it to list. Thanks, Sam ___________________________________________________________ Mp3???-??????? http://music.yahoo.com.cn/?source=mail_mailbox_footer ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-04 13:44 ` Wolfgang Denk 2006-08-07 10:34 ` Sam Song @ 2006-08-07 11:04 ` Yuli Barcohen 2006-08-07 11:43 ` Stefan Roese 2006-08-08 1:59 ` Sam Song 1 sibling, 2 replies; 31+ messages in thread From: Yuli Barcohen @ 2006-08-07 11:04 UTC (permalink / raw) To: u-boot Sounds like it's the (in)famous byte lanes swapping issue. It's so common that Linux MTD code even includes support for this. There are even more problematic flashes like Spansion S70GL256M which consists of two parts, one big-endian and another little-endian. I had to support the lanes swapping (and even the "mixed-endian") on many customers' boards. I've attached to this mail the CFI driver version which I use. It works well on many PowerPC boards but I've got no little-endian system to test it (my tree is not fully merged with official GIT right now so I'm not sending a patch, I can prepare one if you think the approach is OK). -- ======================================================================== Yuli Barcohen | Phone +972-9-765-1788 | Software Project Leader yuli at arabellasw.com | Fax +972-9-765-7494 | Arabella Software, Israel ======================================================================== -------------- next part -------------- /* * (C) Copyright 2002-2004 * Brad Kemp, Seranoa Networks, Brad.Kemp at seranoa.com * * Copyright (C) 2003, 2006 Arabella Software Ltd. * Yuli Barcohen <yuli@arabellasw.com> * Modified to work with AMD flashes * Added support for byte lanes swap * Added support for 32-bit chips consisting of two 16-bit devices * (for example, S70GL256M00) * * Copyright (C) 2004 * Ed Okerson * Modified to work with little-endian systems. * * See file CREDITS for list of people who contributed to this * project. * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public License as * published by the Free Software Foundation; either version 2 of * the License, or (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, * MA 02111-1307 USA * * History * 01/20/2004 - combined variants of original driver. * 01/22/2004 - Write performance enhancements for parallel chips (Tolunay) * 01/23/2004 - Support for x8/x16 chips (Rune Raknerud) * 01/27/2004 - Little endian support Ed Okerson * * Tested Architectures * Port Width Chip Width # of banks Flash Chip Board * 32 16 1 28F128J3 seranoa/eagle * 64 16 1 28F128J3 seranoa/falcon * */ /* The DEBUG define must be before common to enable debugging */ /* #define DEBUG */ #include <common.h> #ifdef CFG_FLASH_CFI_DRIVER #include <watchdog.h> #include <asm/processor.h> #include <asm/byteorder.h> #include <environment.h> /* * This file implements a Common Flash Interface (CFI) driver for U-Boot. * The width of the port and the width of the chips are determined at initialization. * These widths are used to calculate the address for access CFI data structures. * It has been tested on an Intel Strataflash implementation and AMD 29F016D. * * References * JEDEC Standard JESD68 - Common Flash Interface (CFI) * JEDEC Standard JEP137-A Common Flash Interface (CFI) ID Codes * Intel Application Note 646 Common Flash Interface (CFI) and Command Sets * Intel 290667-008 3 Volt Intel StrataFlash Memory datasheet * * TODO * * Use Primary Extended Query table (PRI) and Alternate Algorithm Query * Table (ALT) to determine if protection is available * * Add support for other command sets Use the PRI and ALT to determine command set * Verify erase and program timeouts. */ #if defined(__LITTLE_ENDIAN) && !defined(CFG_FLASH_CFI_SWAP) #define CFG_FLASH_CFI_SWAP #endif #ifndef CFG_FLASH_BANKS_LIST #define CFG_FLASH_BANKS_LIST { CFG_FLASH_BASE } #endif #define FLASH_CMD_CFI 0x98 #define FLASH_CMD_READ_ID 0x90 #define FLASH_CMD_RESET 0xff #define FLASH_CMD_BLOCK_ERASE 0x20 #define FLASH_CMD_ERASE_CONFIRM 0xD0 #define FLASH_CMD_WRITE 0x40 #define FLASH_CMD_PROTECT 0x60 #define FLASH_CMD_PROTECT_SET 0x01 #define FLASH_CMD_PROTECT_CLEAR 0xD0 #define FLASH_CMD_CLEAR_STATUS 0x50 #define FLASH_CMD_WRITE_TO_BUFFER 0xE8 #define FLASH_CMD_WRITE_BUFFER_CONFIRM 0xD0 #define FLASH_STATUS_DONE 0x80 #define FLASH_STATUS_ESS 0x40 #define FLASH_STATUS_ECLBS 0x20 #define FLASH_STATUS_PSLBS 0x10 #define FLASH_STATUS_VPENS 0x08 #define FLASH_STATUS_PSS 0x04 #define FLASH_STATUS_DPS 0x02 #define FLASH_STATUS_R 0x01 #define FLASH_STATUS_PROTECT 0x01 #define AMD_CMD_RESET 0xF0 #define AMD_CMD_WRITE 0xA0 #define AMD_CMD_ERASE_START 0x80 #define AMD_CMD_ERASE_SECTOR 0x30 #define AMD_CMD_UNLOCK_START 0xAA #define AMD_CMD_UNLOCK_ACK 0x55 #define AMD_CMD_WRITE_TO_BUFFER 0x25 #define AMD_CMD_WRITE_BUFFER_CONFIRM 0x29 #define AMD_STATUS_TOGGLE 0x40 #define AMD_STATUS_ERROR 0x20 #define AMD_ADDR_ERASE_START ((info->portwidth == FLASH_CFI_8BIT) ? 0xAAA : 0x555) #define AMD_ADDR_START ((info->portwidth == FLASH_CFI_8BIT) ? 0xAAA : 0x555) #define AMD_ADDR_ACK ((info->portwidth == FLASH_CFI_8BIT) ? 0x555 : 0x2AA) #define FLASH_OFFSET_CFI 0x55 #define FLASH_OFFSET_CFI_RESP 0x10 #define FLASH_OFFSET_PRIMARY_VENDOR 0x13 #define FLASH_OFFSET_EXT_QUERY_T_P_ADDR 0x15 /* extended query table primary addr */ #define FLASH_OFFSET_WTOUT 0x1F #define FLASH_OFFSET_WBTOUT 0x20 #define FLASH_OFFSET_ETOUT 0x21 #define FLASH_OFFSET_CETOUT 0x22 #define FLASH_OFFSET_WMAX_TOUT 0x23 #define FLASH_OFFSET_WBMAX_TOUT 0x24 #define FLASH_OFFSET_EMAX_TOUT 0x25 #define FLASH_OFFSET_CEMAX_TOUT 0x26 #define FLASH_OFFSET_SIZE 0x27 #define FLASH_OFFSET_INTERFACE 0x28 #define FLASH_OFFSET_BUFFER_SIZE 0x2A #define FLASH_OFFSET_NUM_ERASE_REGIONS 0x2C #define FLASH_OFFSET_ERASE_REGIONS 0x2D #define FLASH_OFFSET_PROTECT 0x02 #define FLASH_OFFSET_USER_PROTECTION 0x85 #define FLASH_OFFSET_INTEL_PROTECTION 0x81 #define FLASH_MAN_CFI 0x01000000 #define CFI_CMDSET_NONE 0 #define CFI_CMDSET_INTEL_EXTENDED 1 #define CFI_CMDSET_AMD_STANDARD 2 #define CFI_CMDSET_INTEL_STANDARD 3 #define CFI_CMDSET_AMD_EXTENDED 4 #define CFI_CMDSET_MITSU_STANDARD 256 #define CFI_CMDSET_MITSU_EXTENDED 257 #define CFI_CMDSET_SST 258 #ifdef CFG_FLASH_CFI_AMD_RESET /* needed for STM_ID_29W320DB on UC100 */ # undef FLASH_CMD_RESET # define FLASH_CMD_RESET AMD_CMD_RESET /* use AMD-Reset instead */ #endif typedef union { unsigned char c; unsigned short w; unsigned long l; unsigned long long ll; } cfiword_t; typedef union { volatile unsigned char *cp; volatile unsigned short *wp; volatile unsigned long *lp; volatile unsigned long long *llp; } cfiptr_t; #define NUM_ERASE_REGIONS 4 /* use CFG_MAX_FLASH_BANKS_DETECT if defined */ #ifdef CFG_MAX_FLASH_BANKS_DETECT static ulong bank_base[CFG_MAX_FLASH_BANKS_DETECT] = CFG_FLASH_BANKS_LIST; flash_info_t flash_info[CFG_MAX_FLASH_BANKS_DETECT]; /* FLASH chips info */ #else static ulong bank_base[CFG_MAX_FLASH_BANKS] = CFG_FLASH_BANKS_LIST; flash_info_t flash_info[CFG_MAX_FLASH_BANKS]; /* FLASH chips info */ #endif /* * Check if chip width is defined. If not, start detecting with 8bit. */ #ifndef CFG_FLASH_CFI_WIDTH #define CFG_FLASH_CFI_WIDTH FLASH_CFI_8BIT #endif /*----------------------------------------------------------------------- * Functions */ typedef unsigned long flash_sect_t; static void flash_add_byte (flash_info_t * info, cfiword_t * cword, uchar c); static void flash_make_cmd (flash_info_t * info, uchar cmd, void *cmdbuf); static void flash_write_cmd (flash_info_t * info, flash_sect_t sect, uint offset, uchar cmd); static void flash_unlock_seq (flash_info_t * info, flash_sect_t sect); static int flash_isequal (flash_info_t * info, flash_sect_t sect, uint offset, uchar cmd); static int flash_isset (flash_info_t * info, flash_sect_t sect, uint offset, uchar cmd); static int flash_toggle (flash_info_t * info, flash_sect_t sect, uint offset, uchar cmd); static int flash_detect_cfi (flash_info_t * info); static int flash_write_cfiword (flash_info_t * info, ulong dest, cfiword_t cword); static int flash_full_status_check (flash_info_t * info, flash_sect_t sector, ulong tout, char *prompt); ulong flash_get_size (ulong base, int banknum); #if defined(CFG_ENV_IS_IN_FLASH) || defined(CFG_ENV_ADDR_REDUND) || (CFG_MONITOR_BASE >= CFG_FLASH_BASE) static flash_info_t *flash_get_info(ulong base); #endif #ifdef CFG_FLASH_USE_BUFFER_WRITE static int flash_write_cfibuffer (flash_info_t * info, ulong dest, uchar * cp, int len); #endif /*----------------------------------------------------------------------- * create an address based on the offset and the port width */ inline uchar *flash_make_addr (flash_info_t * info, flash_sect_t sect, uint offset) { return ((uchar *) (info->start[sect] + (offset * info->portwidth))); } #ifdef DEBUG /*----------------------------------------------------------------------- * Debug support */ void print_longlong (char *str, unsigned long long data) { int i; char *cp; cp = (unsigned char *) &data; for (i = 0; i < 8; i++) sprintf (&str[i * 2], "%2.2x", *cp++); } static void flash_printqry (flash_info_t * info, flash_sect_t sect) { cfiptr_t cptr; int x, y; for (x = 0; x < 0x40; x += 16U / info->portwidth) { cptr.cp = flash_make_addr (info, sect, x + FLASH_OFFSET_CFI_RESP); debug ("%p : ", cptr.cp); for (y = 0; y < 16; y++) { debug ("%2.2x ", cptr.cp[y]); } debug (" "); for (y = 0; y < 16; y++) { if (cptr.cp[y] >= 0x20 && cptr.cp[y] <= 0x7e) { debug ("%c", cptr.cp[y]); } else { debug ("."); } } debug ("\n"); } } #endif /*----------------------------------------------------------------------- * read a character@a port width address */ inline uchar flash_read_uchar (flash_info_t * info, uint offset) { uchar *cp; cp = flash_make_addr (info, 0, offset); #if defined(CFG_FLASH_CFI_SWAP) return (cp[0]); #else return (cp[info->portwidth - 1]); #endif } /*----------------------------------------------------------------------- * read a short word by swapping for ppc format. */ ushort flash_read_ushort (flash_info_t * info, flash_sect_t sect, uint offset) { uchar *addr; ushort retval; #ifdef DEBUG int x; #endif addr = flash_make_addr (info, sect, offset); #ifdef DEBUG debug ("ushort addr is at %p info->portwidth = %d\n", addr, info->portwidth); for (x = 0; x < 2 * info->portwidth; x++) { debug ("addr[%x] = 0x%x\n", x, addr[x]); } #endif #if defined(CFG_FLASH_CFI_SWAP) retval = ((addr[(info->portwidth)] << 8) | addr[0]); #else retval = ((addr[(2 * info->portwidth) - 1] << 8) | addr[info->portwidth - 1]); #endif debug ("retval = 0x%x\n", retval); return retval; } /*----------------------------------------------------------------------- * read a long word by picking the least significant byte of each maiximum * port size word. Swap for ppc format. */ ulong flash_read_long (flash_info_t * info, flash_sect_t sect, uint offset) { uchar *addr; ulong retval; #ifdef DEBUG int x; #endif addr = flash_make_addr (info, sect, offset); #ifdef DEBUG debug ("long addr is at %p info->portwidth = %d\n", addr, info->portwidth); for (x = 0; x < 4 * info->portwidth; x++) { debug ("addr[%x] = 0x%x\n", x, addr[x]); } #endif #if defined(CFG_FLASH_CFI_SWAP) retval = (addr[0] << 16) | (addr[(info->portwidth)] << 24) | (addr[(2 * info->portwidth)]) | (addr[(3 * info->portwidth)] << 8); #else retval = (addr[(2 * info->portwidth) - 1] << 24) | (addr[(info->portwidth) - 1] << 16) | (addr[(4 * info->portwidth) - 1] << 8) | addr[(3 * info->portwidth) - 1]; #endif return retval; } /*----------------------------------------------------------------------- */ unsigned long flash_init (void) { unsigned long size = 0; int i; #ifdef CFG_FLASH_PROTECTION char *s = getenv("unlock"); #endif /* Init: no FLASHes known */ for (i = 0; i < CFG_MAX_FLASH_BANKS; ++i) { flash_info[i].flash_id = FLASH_UNKNOWN; size += flash_info[i].size = flash_get_size (bank_base[i], i); if (flash_info[i].flash_id == FLASH_UNKNOWN) { #ifndef CFG_FLASH_QUIET_TEST printf ("## Unknown FLASH on Bank %d - Size = 0x%08lx = %ld MB\n", i, flash_info[i].size, flash_info[i].size << 20); #endif /* CFG_FLASH_QUIET_TEST */ } #ifdef CFG_FLASH_PROTECTION else if ((s != NULL) && (strcmp(s, "yes") == 0)) { /* * Only the U-Boot image and it's environment is protected, * all other sectors are unprotected (unlocked) if flash * hardware protection is used (CFG_FLASH_PROTECTION) and * the environment variable "unlock" is set to "yes". */ if (flash_info[i].legacy_unlock) { int k; /* * Disable legacy_unlock temporarily, since * flash_real_protect would relock all other sectors * again otherwise. */ flash_info[i].legacy_unlock = 0; /* * Legacy unlocking (e.g. Intel J3) -> unlock only one * sector. This will unlock all sectors. */ flash_real_protect (&flash_info[i], 0, 0); flash_info[i].legacy_unlock = 1; /* * Manually mark other sectors as unlocked (unprotected) */ for (k = 1; k < flash_info[i].sector_count; k++) flash_info[i].protect[k] = 0; } else { /* * No legancy unlocking -> unlock all sectors */ flash_protect (FLAG_PROTECT_CLEAR, flash_info[i].start[0], flash_info[i].start[0] + flash_info[i].size - 1, &flash_info[i]); } } #endif /* CFG_FLASH_PROTECTION */ } /* Monitor protection ON by default */ #if (CFG_MONITOR_BASE >= CFG_FLASH_BASE) flash_protect (FLAG_PROTECT_SET, CFG_MONITOR_BASE, CFG_MONITOR_BASE + monitor_flash_len - 1, flash_get_info(CFG_MONITOR_BASE)); #endif /* Environment protection ON by default */ #ifdef CFG_ENV_IS_IN_FLASH flash_protect (FLAG_PROTECT_SET, CFG_ENV_ADDR, CFG_ENV_ADDR + CFG_ENV_SECT_SIZE - 1, flash_get_info(CFG_ENV_ADDR)); #endif /* Redundant environment protection ON by default */ #ifdef CFG_ENV_ADDR_REDUND flash_protect (FLAG_PROTECT_SET, CFG_ENV_ADDR_REDUND, CFG_ENV_ADDR_REDUND + CFG_ENV_SIZE_REDUND - 1, flash_get_info(CFG_ENV_ADDR_REDUND)); #endif return (size); } /*----------------------------------------------------------------------- */ #if defined(CFG_ENV_IS_IN_FLASH) || defined(CFG_ENV_ADDR_REDUND) || (CFG_MONITOR_BASE >= CFG_FLASH_BASE) static flash_info_t *flash_get_info(ulong base) { int i; flash_info_t * info = 0; for (i = 0; i < CFG_MAX_FLASH_BANKS; i ++) { info = & flash_info[i]; if (info->size && info->start[0] <= base && base <= info->start[0] + info->size - 1) break; } return i == CFG_MAX_FLASH_BANKS ? 0 : info; } #endif /*----------------------------------------------------------------------- */ int flash_erase (flash_info_t * info, int s_first, int s_last) { int rcode = 0; int prot; flash_sect_t sect; if (info->flash_id != FLASH_MAN_CFI) { puts ("Can't erase unknown flash type - aborted\n"); return 1; } if ((s_first < 0) || (s_first > s_last)) { puts ("- no sectors to erase\n"); return 1; } prot = 0; for (sect = s_first; sect <= s_last; ++sect) { if (info->protect[sect]) { prot++; } } if (prot) { printf ("- Warning: %d protected sectors will not be erased!\n", prot); } else { putc ('\n'); } for (sect = s_first; sect <= s_last; sect++) { if (info->protect[sect] == 0) { /* not protected */ switch (info->vendor) { case CFI_CMDSET_INTEL_STANDARD: case CFI_CMDSET_INTEL_EXTENDED: flash_write_cmd (info, sect, 0, FLASH_CMD_CLEAR_STATUS); flash_write_cmd (info, sect, 0, FLASH_CMD_BLOCK_ERASE); flash_write_cmd (info, sect, 0, FLASH_CMD_ERASE_CONFIRM); break; case CFI_CMDSET_AMD_STANDARD: case CFI_CMDSET_AMD_EXTENDED: flash_unlock_seq (info, sect); flash_write_cmd (info, sect, AMD_ADDR_ERASE_START, AMD_CMD_ERASE_START); flash_unlock_seq (info, sect); flash_write_cmd (info, sect, 0, AMD_CMD_ERASE_SECTOR); break; default: debug ("Unkown flash vendor %d\n", info->vendor); break; } if (flash_full_status_check (info, sect, info->erase_blk_tout, "erase")) { rcode = 1; } else putc ('.'); } } puts (" done\n"); return rcode; } /*----------------------------------------------------------------------- */ void flash_print_info (flash_info_t * info) { int i; if (info->flash_id != FLASH_MAN_CFI) { puts ("missing or unknown FLASH type\n"); return; } printf ("CFI conformant FLASH (%d x %d)", (info->portwidth << 3), (info->chipwidth << 3)); printf (" Size: %ld MB in %d Sectors\n", info->size >> 20, info->sector_count); printf (" Erase timeout %ld ms, write timeout %ld ms, buffer write timeout %ld ms, buffer size %d\n", info->erase_blk_tout, info->write_tout, info->buffer_write_tout, info->buffer_size); puts (" Sector Start Addresses:"); for (i = 0; i < info->sector_count; ++i) { #ifdef CFG_FLASH_EMPTY_INFO int k; int size; int erased; volatile unsigned long *flash; /* * Check if whole sector is erased */ if (i != (info->sector_count - 1)) size = info->start[i + 1] - info->start[i]; else size = info->start[0] + info->size - info->start[i]; erased = 1; flash = (volatile unsigned long *) info->start[i]; size = size >> 2; /* divide by 4 for longword access */ for (k = 0; k < size; k++) { if (*flash++ != 0xffffffff) { erased = 0; break; } } if ((i % 5) == 0) printf ("\n"); /* print empty and read-only info */ printf (" %08lX%s%s", info->start[i], erased ? " E" : " ", info->protect[i] ? "RO " : " "); #else /* ! CFG_FLASH_EMPTY_INFO */ if ((i % 5) == 0) printf ("\n "); printf (" %08lX%s", info->start[i], info->protect[i] ? " (RO)" : " "); #endif } putc ('\n'); return; } /*----------------------------------------------------------------------- * Copy memory to flash, returns: * 0 - OK * 1 - write timeout * 2 - Flash not erased */ int write_buff (flash_info_t * info, uchar * src, ulong addr, ulong cnt) { ulong wp; ulong cp; int aln; cfiword_t cword; int i, rc; #ifdef CFG_FLASH_USE_BUFFER_WRITE int buffered_size; #endif /* get lower aligned address */ /* get lower aligned address */ wp = (addr & ~(info->portwidth - 1)); /* handle unaligned start */ if ((aln = addr - wp) != 0) { cword.l = 0; cp = wp; for (i = 0; i < aln; ++i, ++cp) flash_add_byte (info, &cword, (*(uchar *) cp)); for (; (i < info->portwidth) && (cnt > 0); i++) { flash_add_byte (info, &cword, *src++); cnt--; cp++; } for (; (cnt == 0) && (i < info->portwidth); ++i, ++cp) flash_add_byte (info, &cword, (*(uchar *) cp)); if ((rc = flash_write_cfiword (info, wp, cword)) != 0) return rc; wp = cp; } /* handle the aligned part */ #ifdef CFG_FLASH_USE_BUFFER_WRITE buffered_size = (info->portwidth / info->chipwidth); buffered_size *= info->buffer_size; while (cnt >= info->portwidth) { /* prohibit buffer write when buffer_size is 1 */ if (info->buffer_size == 1) { cword.l = 0; for (i = 0; i < info->portwidth; i++) flash_add_byte (info, &cword, *src++); if ((rc = flash_write_cfiword (info, wp, cword)) != 0) return rc; wp += info->portwidth; cnt -= info->portwidth; continue; } /* write buffer until next buffered_size aligned boundary */ i = buffered_size - (wp % buffered_size); if (i > cnt) i = cnt; if ((rc = flash_write_cfibuffer (info, wp, src, i)) != ERR_OK) return rc; i -= i & (info->portwidth - 1); wp += i; src += i; cnt -= i; } #else while (cnt >= info->portwidth) { cword.l = 0; for (i = 0; i < info->portwidth; i++) { flash_add_byte (info, &cword, *src++); } if ((rc = flash_write_cfiword (info, wp, cword)) != 0) return rc; wp += info->portwidth; cnt -= info->portwidth; } #endif /* CFG_FLASH_USE_BUFFER_WRITE */ if (cnt == 0) { return (0); } /* * handle unaligned tail bytes */ cword.l = 0; for (i = 0, cp = wp; (i < info->portwidth) && (cnt > 0); ++i, ++cp) { flash_add_byte (info, &cword, *src++); --cnt; } for (; i < info->portwidth; ++i, ++cp) { flash_add_byte (info, &cword, (*(uchar *) cp)); } return flash_write_cfiword (info, wp, cword); } /*----------------------------------------------------------------------- */ #ifdef CFG_FLASH_PROTECTION int flash_real_protect (flash_info_t * info, long sector, int prot) { int retcode = 0; flash_write_cmd (info, sector, 0, FLASH_CMD_CLEAR_STATUS); flash_write_cmd (info, sector, 0, FLASH_CMD_PROTECT); if (prot) flash_write_cmd (info, sector, 0, FLASH_CMD_PROTECT_SET); else flash_write_cmd (info, sector, 0, FLASH_CMD_PROTECT_CLEAR); if ((retcode = flash_full_status_check (info, sector, info->erase_blk_tout, prot ? "protect" : "unprotect")) == 0) { info->protect[sector] = prot; /* * On some of Intel's flash chips (marked via legacy_unlock) * unprotect unprotects all locking. */ if ((prot == 0) && (info->legacy_unlock)) { flash_sect_t i; for (i = 0; i < info->sector_count; i++) { if (info->protect[i]) flash_real_protect (info, i, 1); } } } return retcode; } /*----------------------------------------------------------------------- * flash_read_user_serial - read the OneTimeProgramming cells */ void flash_read_user_serial (flash_info_t * info, void *buffer, int offset, int len) { uchar *src; uchar *dst; dst = buffer; src = flash_make_addr (info, 0, FLASH_OFFSET_USER_PROTECTION); flash_write_cmd (info, 0, 0, FLASH_CMD_READ_ID); memcpy (dst, src + offset, len); flash_write_cmd (info, 0, 0, info->cmd_reset); } /* * flash_read_factory_serial - read the device Id from the protection area */ void flash_read_factory_serial (flash_info_t * info, void *buffer, int offset, int len) { uchar *src; src = flash_make_addr (info, 0, FLASH_OFFSET_INTEL_PROTECTION); flash_write_cmd (info, 0, 0, FLASH_CMD_READ_ID); memcpy (buffer, src + offset, len); flash_write_cmd (info, 0, 0, info->cmd_reset); } #endif /* CFG_FLASH_PROTECTION */ /* * flash_is_busy - check to see if the flash is busy * This routine checks the status of the chip and returns true if the chip is busy */ static int flash_is_busy (flash_info_t * info, flash_sect_t sect) { int retval; switch (info->vendor) { case CFI_CMDSET_INTEL_STANDARD: case CFI_CMDSET_INTEL_EXTENDED: retval = !flash_isset (info, sect, 0, FLASH_STATUS_DONE); break; case CFI_CMDSET_AMD_STANDARD: case CFI_CMDSET_AMD_EXTENDED: retval = flash_toggle (info, sect, 0, AMD_STATUS_TOGGLE); break; default: retval = 0; } debug ("flash_is_busy: %d\n", retval); return retval; } /*----------------------------------------------------------------------- * wait for XSR.7 to be set. Time out with an error if it does not. * This routine does not set the flash to read-array mode. */ static int flash_status_check (flash_info_t * info, flash_sect_t sector, ulong tout, char *prompt) { ulong start; #if CFG_HZ != 1000 tout *= CFG_HZ/1000; #endif /* Wait for command completion */ start = get_timer (0); while (flash_is_busy (info, sector)) { if (get_timer (start) > tout) { printf ("Flash %s timeout at address %lx data %lx\n", prompt, info->start[sector], flash_read_long (info, sector, 0)); flash_write_cmd (info, sector, 0, info->cmd_reset); return ERR_TIMOUT; } udelay (1); /* also triggers watchdog */ } return ERR_OK; } /*----------------------------------------------------------------------- * Wait for XSR.7 to be set, if it times out print an error, otherwise do a full status check. * This routine sets the flash to read-array mode. */ static int flash_full_status_check (flash_info_t * info, flash_sect_t sector, ulong tout, char *prompt) { int retcode; retcode = flash_status_check (info, sector, tout, prompt); switch (info->vendor) { case CFI_CMDSET_INTEL_EXTENDED: case CFI_CMDSET_INTEL_STANDARD: if ((retcode == ERR_OK) && !flash_isequal (info, sector, 0, FLASH_STATUS_DONE)) { retcode = ERR_INVAL; printf ("Flash %s error at address %lx\n", prompt, info->start[sector]); if (flash_isset (info, sector, 0, FLASH_STATUS_ECLBS | FLASH_STATUS_PSLBS)) { puts ("Command Sequence Error.\n"); } else if (flash_isset (info, sector, 0, FLASH_STATUS_ECLBS)) { puts ("Block Erase Error.\n"); retcode = ERR_NOT_ERASED; } else if (flash_isset (info, sector, 0, FLASH_STATUS_PSLBS)) { puts ("Locking Error\n"); } if (flash_isset (info, sector, 0, FLASH_STATUS_DPS)) { puts ("Block locked.\n"); retcode = ERR_PROTECTED; } if (flash_isset (info, sector, 0, FLASH_STATUS_VPENS)) puts ("Vpp Low Error.\n"); } flash_write_cmd (info, sector, 0, info->cmd_reset); break; default: break; } return retcode; } /*----------------------------------------------------------------------- */ static void flash_add_byte (flash_info_t * info, cfiword_t * cword, uchar c) { #if defined(__LITTLE_ENDIAN) unsigned short w; unsigned int l; unsigned long long ll; #endif switch (info->portwidth) { case FLASH_CFI_8BIT: cword->c = c; break; case FLASH_CFI_16BIT: #if defined(__LITTLE_ENDIAN) w = c; w <<= 8; cword->w = (cword->w >> 8) | w; #else cword->w = (cword->w << 8) | c; #endif break; case FLASH_CFI_32BIT: #if defined(__LITTLE_ENDIAN) l = c; l <<= 24; cword->l = (cword->l >> 8) | l; #else cword->l = (cword->l << 8) | c; #endif break; case FLASH_CFI_64BIT: #if defined(__LITTLE_ENDIAN) ll = c; ll <<= 56; cword->ll = (cword->ll >> 8) | ll; #else cword->ll = (cword->ll << 8) | c; #endif break; } } /*----------------------------------------------------------------------- * make a proper sized command based on the port and chip widths */ static void flash_make_cmd (flash_info_t * info, uchar cmd, void *cmdbuf) { int i; uchar *cp = (uchar *) cmdbuf; #if defined(CFG_FLASH_CFI_SWAP) for (i = info->portwidth; i > 0; i--) #else for (i = 1; i <= info->portwidth; i++) #endif *cp++ = (i & (info->chipwidth - 1)) ? '\0' : cmd; #ifdef CFG_FLASH_CFI_2x16 if ((info->portwidth == FLASH_CFI_32BIT) && (info->chipwidth == FLASH_CFI_BY16)) { uchar tmp; cp = (uchar *) cmdbuf; tmp = cp[1]; cp[1] = cp[2]; cp[2] = tmp; } #endif /* CFG_FLASH_CFI_2x16 */ } /* * Write a proper sized command to the correct address */ static void flash_write_cmd (flash_info_t * info, flash_sect_t sect, uint offset, uchar cmd) { volatile cfiptr_t addr; cfiword_t cword; addr.cp = flash_make_addr (info, sect, offset); flash_make_cmd (info, cmd, &cword); switch (info->portwidth) { case FLASH_CFI_8BIT: debug ("fwc addr %p cmd %x %x 8bit x %d bit\n", addr.cp, cmd, cword.c, info->chipwidth << CFI_FLASH_SHIFT_WIDTH); *addr.cp = cword.c; #ifdef CONFIG_BLACKFIN asm("ssync;"); #endif break; case FLASH_CFI_16BIT: debug ("fwc addr %p cmd %x %4.4x 16bit x %d bit\n", addr.wp, cmd, cword.w, info->chipwidth << CFI_FLASH_SHIFT_WIDTH); *addr.wp = cword.w; #ifdef CONFIG_BLACKFIN asm("ssync;"); #endif break; case FLASH_CFI_32BIT: debug ("fwc addr %p cmd %x %8.8lx 32bit x %d bit\n", addr.lp, cmd, cword.l, info->chipwidth << CFI_FLASH_SHIFT_WIDTH); *addr.lp = cword.l; #ifdef CONFIG_BLACKFIN asm("ssync;"); #endif break; case FLASH_CFI_64BIT: #ifdef DEBUG { char str[20]; print_longlong (str, cword.ll); debug ("fwrite addr %p cmd %x %s 64 bit x %d bit\n", addr.llp, cmd, str, info->chipwidth << CFI_FLASH_SHIFT_WIDTH); } #endif *addr.llp = cword.ll; #ifdef CONFIG_BLACKFIN asm("ssync;"); #endif break; } } static void flash_unlock_seq (flash_info_t * info, flash_sect_t sect) { flash_write_cmd (info, sect, AMD_ADDR_START, AMD_CMD_UNLOCK_START); flash_write_cmd (info, sect, AMD_ADDR_ACK, AMD_CMD_UNLOCK_ACK); } /*----------------------------------------------------------------------- */ static int flash_isequal (flash_info_t * info, flash_sect_t sect, uint offset, uchar cmd) { cfiptr_t cptr; cfiword_t cword; int retval; cptr.cp = flash_make_addr (info, sect, offset); flash_make_cmd (info, cmd, &cword); debug ("is= cmd %x(%c) addr %p ", cmd, cmd, cptr.cp); switch (info->portwidth) { case FLASH_CFI_8BIT: debug ("is= %x %x\n", cptr.cp[0], cword.c); retval = (cptr.cp[0] == cword.c); break; case FLASH_CFI_16BIT: debug ("is= %4.4x %4.4x\n", cptr.wp[0], cword.w); retval = (cptr.wp[0] == cword.w); break; case FLASH_CFI_32BIT: debug ("is= %8.8lx %8.8lx\n", cptr.lp[0], cword.l); retval = (cptr.lp[0] == cword.l); break; case FLASH_CFI_64BIT: #ifdef DEBUG { char str1[20]; char str2[20]; print_longlong (str1, cptr.llp[0]); print_longlong (str2, cword.ll); debug ("is= %s %s\n", str1, str2); } #endif retval = (cptr.llp[0] == cword.ll); break; default: retval = 0; break; } return retval; } /*----------------------------------------------------------------------- */ static int flash_isset (flash_info_t * info, flash_sect_t sect, uint offset, uchar cmd) { cfiptr_t cptr; cfiword_t cword; int retval; cptr.cp = flash_make_addr (info, sect, offset); flash_make_cmd (info, cmd, &cword); switch (info->portwidth) { case FLASH_CFI_8BIT: retval = ((cptr.cp[0] & cword.c) == cword.c); break; case FLASH_CFI_16BIT: retval = ((cptr.wp[0] & cword.w) == cword.w); break; case FLASH_CFI_32BIT: retval = ((cptr.lp[0] & cword.l) == cword.l); break; case FLASH_CFI_64BIT: retval = ((cptr.llp[0] & cword.ll) == cword.ll); break; default: retval = 0; break; } return retval; } /*----------------------------------------------------------------------- */ static int flash_toggle (flash_info_t * info, flash_sect_t sect, uint offset, uchar cmd) { cfiptr_t cptr; cfiword_t cword; int retval; cptr.cp = flash_make_addr (info, sect, offset); flash_make_cmd (info, cmd, &cword); switch (info->portwidth) { case FLASH_CFI_8BIT: retval = ((cptr.cp[0] & cword.c) != (cptr.cp[0] & cword.c)); break; case FLASH_CFI_16BIT: retval = ((cptr.wp[0] & cword.w) != (cptr.wp[0] & cword.w)); break; case FLASH_CFI_32BIT: retval = ((cptr.lp[0] & cword.l) != (cptr.lp[0] & cword.l)); break; case FLASH_CFI_64BIT: retval = ((cptr.llp[0] & cword.ll) != (cptr.llp[0] & cword.ll)); break; default: retval = 0; break; } return retval; } /*----------------------------------------------------------------------- * detect if flash is compatible with the Common Flash Interface (CFI) * http://www.jedec.org/download/search/jesd68.pdf * */ static int flash_detect_cfi (flash_info_t * info) { debug ("flash detect cfi\n"); for (info->portwidth = CFG_FLASH_CFI_WIDTH; info->portwidth <= FLASH_CFI_64BIT; info->portwidth <<= 1) { for (info->chipwidth = FLASH_CFI_BY8; info->chipwidth <= info->portwidth; info->chipwidth <<= 1) { flash_write_cmd (info, 0, 0, info->cmd_reset); flash_write_cmd (info, 0, FLASH_OFFSET_CFI, FLASH_CMD_CFI); if (flash_isequal (info, 0, FLASH_OFFSET_CFI_RESP, 'Q') && flash_isequal (info, 0, FLASH_OFFSET_CFI_RESP + 1, 'R') && flash_isequal (info, 0, FLASH_OFFSET_CFI_RESP + 2, 'Y')) { info->interface = flash_read_ushort (info, 0, FLASH_OFFSET_INTERFACE); debug ("device interface is %d\n", info->interface); debug ("found port %d chip %d ", info->portwidth, info->chipwidth); debug ("port %d bits chip %d bits\n", info->portwidth << CFI_FLASH_SHIFT_WIDTH, info->chipwidth << CFI_FLASH_SHIFT_WIDTH); return 1; } } } debug ("not found\n"); return 0; } /* * The following code cannot be run from FLASH! * */ ulong flash_get_size (ulong base, int banknum) { flash_info_t *info = &flash_info[banknum]; int i, j; flash_sect_t sect_cnt; unsigned long sector; unsigned long tmp; int size_ratio; uchar num_erase_regions; int erase_region_size; int erase_region_count; #ifdef CFG_FLASH_PROTECTION int ext_addr; info->legacy_unlock = 0; #endif info->start[0] = base; if (flash_detect_cfi (info)) { info->vendor = flash_read_ushort (info, 0, FLASH_OFFSET_PRIMARY_VENDOR); #ifdef DEBUG flash_printqry (info, 0); #endif switch (info->vendor) { case CFI_CMDSET_INTEL_STANDARD: case CFI_CMDSET_INTEL_EXTENDED: default: info->cmd_reset = FLASH_CMD_RESET; #ifdef CFG_FLASH_PROTECTION /* read legacy lock/unlock bit from intel flash */ ext_addr = flash_read_ushort (info, 0, FLASH_OFFSET_EXT_QUERY_T_P_ADDR); info->legacy_unlock = flash_read_uchar (info, ext_addr + 5) & 0x08; #endif break; case CFI_CMDSET_AMD_STANDARD: case CFI_CMDSET_AMD_EXTENDED: info->cmd_reset = AMD_CMD_RESET; break; } debug ("manufacturer is %d\n", info->vendor); size_ratio = info->portwidth / info->chipwidth; /* if the chip is x8/x16 reduce the ratio by half */ if ((info->interface == FLASH_CFI_X8X16) && (info->chipwidth == FLASH_CFI_BY8)) { size_ratio >>= 1; } num_erase_regions = flash_read_uchar (info, FLASH_OFFSET_NUM_ERASE_REGIONS); debug ("size_ratio %d port %d bits chip %d bits\n", size_ratio, info->portwidth << CFI_FLASH_SHIFT_WIDTH, info->chipwidth << CFI_FLASH_SHIFT_WIDTH); debug ("found %d erase regions\n", num_erase_regions); sect_cnt = 0; sector = base; for (i = 0; i < num_erase_regions; i++) { if (i > NUM_ERASE_REGIONS) { printf ("%d erase regions found, only %d used\n", num_erase_regions, NUM_ERASE_REGIONS); break; } tmp = flash_read_long (info, 0, FLASH_OFFSET_ERASE_REGIONS + i * 4); erase_region_size = (tmp & 0xffff) ? ((tmp & 0xffff) * 256) : 128; tmp >>= 16; erase_region_count = (tmp & 0xffff) + 1; debug ("erase_region_count = %d erase_region_size = %d\n", erase_region_count, erase_region_size); for (j = 0; j < erase_region_count; j++) { info->start[sect_cnt] = sector; sector += (erase_region_size * size_ratio); /* * Only read protection status from supported devices (intel...) */ switch (info->vendor) { case CFI_CMDSET_INTEL_EXTENDED: case CFI_CMDSET_INTEL_STANDARD: info->protect[sect_cnt] = flash_isset (info, sect_cnt, FLASH_OFFSET_PROTECT, FLASH_STATUS_PROTECT); break; default: info->protect[sect_cnt] = 0; /* default: not protected */ } sect_cnt++; } } info->sector_count = sect_cnt; /* multiply the size by the number of chips */ info->size = (1 << flash_read_uchar (info, FLASH_OFFSET_SIZE)) * size_ratio; info->buffer_size = (1 << flash_read_ushort (info, 0, FLASH_OFFSET_BUFFER_SIZE)); tmp = 1 << flash_read_uchar (info, FLASH_OFFSET_ETOUT); info->erase_blk_tout = (tmp * (1 << flash_read_uchar (info, FLASH_OFFSET_EMAX_TOUT))); tmp = (1 << flash_read_uchar (info, FLASH_OFFSET_WBTOUT)) * (1 << flash_read_uchar (info, FLASH_OFFSET_WBMAX_TOUT)); info->buffer_write_tout = tmp / 1000 + (tmp % 1000 ? 1 : 0); /* round up when converting to ms */ tmp = (1 << flash_read_uchar (info, FLASH_OFFSET_WTOUT)) * (1 << flash_read_uchar (info, FLASH_OFFSET_WMAX_TOUT)); info->write_tout = tmp / 1000 + (tmp % 1000 ? 1 : 0); /* round up when converting to ms */ info->flash_id = FLASH_MAN_CFI; if ((info->interface == FLASH_CFI_X8X16) && (info->chipwidth == FLASH_CFI_BY8)) { info->portwidth >>= 1; /* XXX - Need to test on x8/x16 in parallel. */ } } flash_write_cmd (info, 0, 0, info->cmd_reset); return (info->size); } /* loop through the sectors from the highest address * when the passed address is greater or equal to the sector address * we have a match */ static flash_sect_t find_sector (flash_info_t * info, ulong addr) { flash_sect_t sector; for (sector = info->sector_count - 1; sector >= 0; sector--) { if (addr >= info->start[sector]) break; } return sector; } /*----------------------------------------------------------------------- */ static int flash_write_cfiword (flash_info_t * info, ulong dest, cfiword_t cword) { cfiptr_t ctladdr; cfiptr_t cptr; int flag; ctladdr.cp = flash_make_addr (info, 0, 0); cptr.cp = (uchar *) dest; /* Check if Flash is (sufficiently) erased */ switch (info->portwidth) { case FLASH_CFI_8BIT: flag = ((cptr.cp[0] & cword.c) == cword.c); break; case FLASH_CFI_16BIT: flag = ((cptr.wp[0] & cword.w) == cword.w); break; case FLASH_CFI_32BIT: flag = ((cptr.lp[0] & cword.l) == cword.l); break; case FLASH_CFI_64BIT: flag = ((cptr.llp[0] & cword.ll) == cword.ll); break; default: return 2; } if (!flag) return 2; /* Disable interrupts which might cause a timeout here */ flag = disable_interrupts (); switch (info->vendor) { case CFI_CMDSET_INTEL_EXTENDED: case CFI_CMDSET_INTEL_STANDARD: flash_write_cmd (info, 0, 0, FLASH_CMD_CLEAR_STATUS); flash_write_cmd (info, 0, 0, FLASH_CMD_WRITE); break; case CFI_CMDSET_AMD_EXTENDED: case CFI_CMDSET_AMD_STANDARD: flash_unlock_seq (info, 0); flash_write_cmd (info, 0, AMD_ADDR_START, AMD_CMD_WRITE); break; } switch (info->portwidth) { case FLASH_CFI_8BIT: cptr.cp[0] = cword.c; break; case FLASH_CFI_16BIT: cptr.wp[0] = cword.w; break; case FLASH_CFI_32BIT: cptr.lp[0] = cword.l; break; case FLASH_CFI_64BIT: cptr.llp[0] = cword.ll; break; } /* re-enable interrupts if necessary */ if (flag) enable_interrupts (); return flash_full_status_check (info, find_sector (info, dest), info->write_tout, "write"); } #ifdef CFG_FLASH_USE_BUFFER_WRITE static int flash_write_cfibuffer (flash_info_t * info, ulong dest, uchar * cp, int len) { flash_sect_t sector; int cnt; int retcode; volatile cfiptr_t src; volatile cfiptr_t dst; switch (info->vendor) { case CFI_CMDSET_INTEL_STANDARD: case CFI_CMDSET_INTEL_EXTENDED: src.cp = cp; dst.cp = (uchar *) dest; sector = find_sector (info, dest); flash_write_cmd (info, sector, 0, FLASH_CMD_CLEAR_STATUS); flash_write_cmd (info, sector, 0, FLASH_CMD_WRITE_TO_BUFFER); if ((retcode = flash_status_check (info, sector, info->buffer_write_tout, "write to buffer")) == ERR_OK) { /* reduce the number of loops by the width of the port */ switch (info->portwidth) { case FLASH_CFI_8BIT: cnt = len; break; case FLASH_CFI_16BIT: cnt = len >> 1; break; case FLASH_CFI_32BIT: cnt = len >> 2; break; case FLASH_CFI_64BIT: cnt = len >> 3; break; default: return ERR_INVAL; break; } flash_write_cmd (info, sector, 0, (uchar) cnt - 1); while (cnt-- > 0) { switch (info->portwidth) { case FLASH_CFI_8BIT: *dst.cp++ = *src.cp++; break; case FLASH_CFI_16BIT: *dst.wp++ = *src.wp++; break; case FLASH_CFI_32BIT: *dst.lp++ = *src.lp++; break; case FLASH_CFI_64BIT: *dst.llp++ = *src.llp++; break; default: return ERR_INVAL; break; } } flash_write_cmd (info, sector, 0, FLASH_CMD_WRITE_BUFFER_CONFIRM); retcode = flash_full_status_check (info, sector, info->buffer_write_tout, "buffer write"); } return retcode; case CFI_CMDSET_AMD_STANDARD: case CFI_CMDSET_AMD_EXTENDED: src.cp = cp; dst.cp = (uchar *) dest; sector = find_sector (info, dest); flash_unlock_seq(info,0); flash_write_cmd (info, sector, 0, AMD_CMD_WRITE_TO_BUFFER); switch (info->portwidth) { case FLASH_CFI_8BIT: cnt = len; flash_write_cmd (info, sector, 0, (uchar) cnt - 1); while (cnt-- > 0) *dst.cp++ = *src.cp++; break; case FLASH_CFI_16BIT: cnt = len >> 1; flash_write_cmd (info, sector, 0, (uchar) cnt - 1); while (cnt-- > 0) *dst.wp++ = *src.wp++; break; case FLASH_CFI_32BIT: cnt = len >> 2; flash_write_cmd (info, sector, 0, (uchar) cnt - 1); while (cnt-- > 0) *dst.lp++ = *src.lp++; break; case FLASH_CFI_64BIT: cnt = len >> 3; flash_write_cmd (info, sector, 0, (uchar) cnt - 1); while (cnt-- > 0) *dst.llp++ = *src.llp++; break; default: return ERR_INVAL; } flash_write_cmd (info, sector, 0, AMD_CMD_WRITE_BUFFER_CONFIRM); retcode = flash_full_status_check (info, sector, info->buffer_write_tout, "buffer write"); return retcode; default: debug ("Unknown Command Set\n"); return ERR_INVAL; } } #endif /* CFG_FLASH_USE_BUFFER_WRITE */ #endif /* CFG_FLASH_CFI */ ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-07 11:04 ` Yuli Barcohen @ 2006-08-07 11:43 ` Stefan Roese 2006-08-08 1:59 ` Sam Song 1 sibling, 0 replies; 31+ messages in thread From: Stefan Roese @ 2006-08-07 11:43 UTC (permalink / raw) To: u-boot Hi Yuli, On Monday 07 August 2006 13:04, Yuli Barcohen wrote: > Sounds like it's the (in)famous byte lanes swapping issue. It's so > common that Linux MTD code even includes support for this. There are > even more problematic flashes like Spansion S70GL256M which consists of > two parts, one big-endian and another little-endian. I had to support > the lanes swapping (and even the "mixed-endian") on many customers' > boards. I've attached to this mail the CFI driver version which I > use. It works well on many PowerPC boards but I've got no little-endian > system to test it (my tree is not fully merged with official GIT right > now so I'm not sending a patch, I can prepare one if you think the > approach is OK). Yes, please send a patch, so we can better see the changes to the official version. Thanks. Best regards, Stefan ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-07 11:04 ` Yuli Barcohen 2006-08-07 11:43 ` Stefan Roese @ 2006-08-08 1:59 ` Sam Song 2006-08-08 2:15 ` Zang Roy-r61911 1 sibling, 1 reply; 31+ messages in thread From: Sam Song @ 2006-08-08 1:59 UTC (permalink / raw) To: u-boot Yuli Barcohen <yuli@arabellasw.com> wrote: > Sounds like it's the (in)famous byte lanes > swapping issue. It's so [snip] > system to test it (my tree is not fully merged with > official GIT right > now so I'm not sending a patch, I can prepare one if > you think the > approach is OK). Indeed, I applied this CFI Driver on MPC7448HPC2 and it did work. But a small puzzle is that CFG_FLASH_CFI_SWAP shouldn't be tied to __LITTLE_ENDIAN. According to my test, they two can be set separetely. diff --git a/include/configs/taiga.h b/include/configs/taiga.h index 8d4e9ad..878f268 100644 --- a/include/configs/taiga.h +++ b/include/configs/taiga.h @@ -500,6 +500,7 @@ #define CFG_BOOTMAPSZ (8<<20) /* Init */ #define CFG_FLASH_CFI 1 #define CFG_FLASH_CFI_DRIVER 1 +#define CFG_FLASH_CFI_SWAP #define CFG_FLASH_USE_BUFFER_WRITE 1 /* Disable CFG_FLASH_USE_BUFFER_WRITE is OK */ #define CFG_MAX_FLASH_BANKS 1 Thanks a lot, Sam __________________________________________________ ??????????????? http://cn.mail.yahoo.com -------------- next part -------------- A non-text attachment was scrubbed... Name: cfi_debug Type: application/octet-stream Size: 8094 bytes Desc: 1820405661-cfi_debug Url : http://lists.denx.de/pipermail/u-boot/attachments/20060808/3cccad4d/attachment.obj ^ permalink raw reply related [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 1:59 ` Sam Song @ 2006-08-08 2:15 ` Zang Roy-r61911 2006-08-08 4:54 ` Sam Song 0 siblings, 1 reply; 31+ messages in thread From: Zang Roy-r61911 @ 2006-08-08 2:15 UTC (permalink / raw) To: u-boot > Indeed, I applied this CFI Driver on MPC7448HPC2 and it did > work. But a small puzzle is that CFG_FLASH_CFI_SWAP shouldn't > be tied to __LITTLE_ENDIAN. According to my test, they two > can be set separetely. Please try the patch I posted yesterday. define CFG_FLASH_TSI_SWAP_BYTE in your board config file. > > diff --git a/include/configs/taiga.h > b/include/configs/taiga.h Where do you get this file? It should be mpc7448hpc2.h. taiga.h is our previous header file. ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 2:15 ` Zang Roy-r61911 @ 2006-08-08 4:54 ` Sam Song 2006-08-08 5:18 ` Zang Roy-r61911 0 siblings, 1 reply; 31+ messages in thread From: Sam Song @ 2006-08-08 4:54 UTC (permalink / raw) To: u-boot Zang Roy-r61911 <tie-fei.zang@freescale.com> wrote: > Please try the patch I posted yesterday. > define CFG_FLASH_TSI_SWAP_BYTE in your board config > file. It should work as expected. But I think we could work out a more general CFI driver. So Yuli's patch is still acceptable. What do you think? > > diff --git a/include/configs/taiga.h > > b/include/configs/taiga.h > Where do you get this file? It should be > mpc7448hpc2.h. > taiga.h is our previous header file. I shorten mpc7448hpc2 as taiga for few typing in my git tree. It's content could be the same as yours. Thanks, Sam ___________________________________________________________ ????????-3.5G???20M??? http://cn.mail.yahoo.com ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 4:54 ` Sam Song @ 2006-08-08 5:18 ` Zang Roy-r61911 2006-08-08 10:52 ` Sam Song 0 siblings, 1 reply; 31+ messages in thread From: Zang Roy-r61911 @ 2006-08-08 5:18 UTC (permalink / raw) To: u-boot > > Please try the patch I posted yesterday. > > define CFG_FLASH_TSI_SWAP_BYTE in your board config file. > > It should work as expected. But I think we could work out a > more general CFI driver. So Yuli's patch is still acceptable. > What do you think? Where is the patch? I do not object. Our final goal is to enable the general flash driver work on the board. Roy ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 5:18 ` Zang Roy-r61911 @ 2006-08-08 10:52 ` Sam Song 2006-08-08 11:00 ` Yuli Barcohen 2006-08-09 1:32 ` Zang Roy-r61911 0 siblings, 2 replies; 31+ messages in thread From: Sam Song @ 2006-08-08 10:52 UTC (permalink / raw) To: u-boot Zang Roy-r61911 <tie-fei.zang@freescale.com> wrote: > Where is the patch? I do not object. > Our final goal is to enable the general flash driver > work on the board. Sorry, I misused the word patch:-). The "patch" is Yuli's cfi_flash.c in his attached file. See your mail archive. It would be there. I tested that it was OK but still need your confirmation, I am afraid. Thanks, Sam ___________________________________________________________ Mp3???-??????? http://music.yahoo.com.cn/?source=mail_mailbox_footer ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 10:52 ` Sam Song @ 2006-08-08 11:00 ` Yuli Barcohen 2006-08-08 11:10 ` Li Yang-r58472 2006-08-09 6:14 ` Zang Roy-r61911 2006-08-09 1:32 ` Zang Roy-r61911 1 sibling, 2 replies; 31+ messages in thread From: Yuli Barcohen @ 2006-08-08 11:00 UTC (permalink / raw) To: u-boot >>>>> Sam Song writes: Zang> Where is the patch? I do not object. Our final goal is to Zang> enable the general flash driver work on the board. Sam> Sorry, I misused the word patch:-). The "patch" is Yuli's Sam> cfi_flash.c in his attached file. See your mail archive. It Sam> would be there. I tested that it was OK but still need your Sam> confirmation, I am afraid. OK, the patch is attached. As we know, it works on big-endian systems. On little-endian ones, it should change nothing. -- ======================================================================== Yuli Barcohen | Phone +972-9-765-1788 | Software Project Leader yuli at arabellasw.com | Fax +972-9-765-7494 | Arabella Software, Israel ======================================================================== -------------- next part -------------- --- ../GIT/drivers/cfi_flash.c 2006-06-06 09:48:51.000000000 +0300 +++ drivers/cfi_flash.c 2006-07-26 17:16:37.000000000 +0300 @@ -2,9 +2,12 @@ * (C) Copyright 2002-2004 * Brad Kemp, Seranoa Networks, Brad.Kemp at seranoa.com * - * Copyright (C) 2003 Arabella Software Ltd. + * Copyright (C) 2003, 2006 Arabella Software Ltd. * Yuli Barcohen <yuli@arabellasw.com> * Modified to work with AMD flashes + * Added support for byte lanes swap + * Added support for 32-bit chips consisting of two 16-bit devices + * (for example, S70GL256M00) * * Copyright (C) 2004 * Ed Okerson @@ -45,10 +48,13 @@ /* #define DEBUG */ #include <common.h> + +#ifdef CFG_FLASH_CFI_DRIVER + +#include <watchdog.h> #include <asm/processor.h> #include <asm/byteorder.h> #include <environment.h> -#ifdef CFG_FLASH_CFI_DRIVER /* * This file implements a Common Flash Interface (CFI) driver for U-Boot. @@ -71,6 +77,10 @@ * Verify erase and program timeouts. */ +#if defined(__LITTLE_ENDIAN) && !defined(CFG_FLASH_CFI_SWAP) +#define CFG_FLASH_CFI_SWAP +#endif + #ifndef CFG_FLASH_BANKS_LIST #define CFG_FLASH_BANKS_LIST { CFG_FLASH_BASE } #endif @@ -268,7 +278,7 @@ inline uchar flash_read_uchar (flash_inf uchar *cp; cp = flash_make_addr (info, 0, offset); -#if defined(__LITTLE_ENDIAN) +#if defined(CFG_FLASH_CFI_SWAP) return (cp[0]); #else return (cp[info->portwidth - 1]); @@ -295,7 +305,7 @@ ushort flash_read_ushort (flash_info_t * debug ("addr[%x] = 0x%x\n", x, addr[x]); } #endif -#if defined(__LITTLE_ENDIAN) +#if defined(CFG_FLASH_CFI_SWAP) retval = ((addr[(info->portwidth)] << 8) | addr[0]); #else retval = ((addr[(2 * info->portwidth) - 1] << 8) | @@ -327,7 +337,7 @@ ulong flash_read_long (flash_info_t * in debug ("addr[%x] = 0x%x\n", x, addr[x]); } #endif -#if defined(__LITTLE_ENDIAN) +#if defined(CFG_FLASH_CFI_SWAP) retval = (addr[0] << 16) | (addr[(info->portwidth)] << 24) | (addr[(2 * info->portwidth)]) | (addr[(3 * info->portwidth)] << 8); #else @@ -892,12 +902,22 @@ static void flash_make_cmd (flash_info_t int i; uchar *cp = (uchar *) cmdbuf; -#if defined(__LITTLE_ENDIAN) +#if defined(CFG_FLASH_CFI_SWAP) for (i = info->portwidth; i > 0; i--) #else for (i = 1; i <= info->portwidth; i++) #endif *cp++ = (i & (info->chipwidth - 1)) ? '\0' : cmd; +#ifdef CFG_FLASH_CFI_2x16 + if ((info->portwidth == FLASH_CFI_32BIT) && (info->chipwidth == FLASH_CFI_BY16)) + { + uchar tmp; + cp = (uchar *) cmdbuf; + tmp = cp[1]; + cp[1] = cp[2]; + cp[2] = tmp; + } +#endif /* CFG_FLASH_CFI_2x16 */ } /* ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 11:00 ` Yuli Barcohen @ 2006-08-08 11:10 ` Li Yang-r58472 2006-08-08 11:25 ` Yuli Barcohen 2006-08-09 6:14 ` Zang Roy-r61911 1 sibling, 1 reply; 31+ messages in thread From: Li Yang-r58472 @ 2006-08-08 11:10 UTC (permalink / raw) To: u-boot +#if defined(__LITTLE_ENDIAN) && !defined(CFG_FLASH_CFI_SWAP) +#define CFG_FLASH_CFI_SWAP +#endif + [snip] -#if defined(__LITTLE_ENDIAN) +#if defined(CFG_FLASH_CFI_SWAP) Why use CFG_FLASH_CFI_SWAP instead of __LITTLE_ENDIAN. It seems that they are equivalent. Best Regards, Leo > -----Original Message----- > From: u-boot-users-bounces at lists.sourceforge.net > [mailto:u-boot-users-bounces at lists.sourceforge.net] On Behalf Of Yuli Barcohen > Sent: Tuesday, August 08, 2006 7:00 PM > To: u-boot-users at lists.sourceforge.net > Cc: Sam Song; Wolfgang Denk > Subject: Re: [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue > > >>>>> Sam Song writes: > > Zang> Where is the patch? I do not object. Our final goal is to > Zang> enable the general flash driver work on the board. > > Sam> Sorry, I misused the word patch:-). The "patch" is Yuli's > Sam> cfi_flash.c in his attached file. See your mail archive. It > Sam> would be there. I tested that it was OK but still need your > Sam> confirmation, I am afraid. > > OK, the patch is attached. As we know, it works on big-endian systems. > On little-endian ones, it should change nothing. > > -- > ======================================================================== > Yuli Barcohen | Phone +972-9-765-1788 | Software Project Leader > yuli at arabellasw.com | Fax +972-9-765-7494 | Arabella Software, Israel > ======================================================================== ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 11:10 ` Li Yang-r58472 @ 2006-08-08 11:25 ` Yuli Barcohen 2006-08-08 11:42 ` Li Yang-r58472 2006-08-08 13:10 ` Sam Song 0 siblings, 2 replies; 31+ messages in thread From: Yuli Barcohen @ 2006-08-08 11:25 UTC (permalink / raw) To: u-boot >>>>> Li Yang writes: Li> +#if defined(__LITTLE_ENDIAN) && !defined(CFG_FLASH_CFI_SWAP) Li> +#define CFG_FLASH_CFI_SWAP Li> +#endif Li> + [snip] Li> -#if defined(__LITTLE_ENDIAN) Li> +#if defined(CFG_FLASH_CFI_SWAP) Li> Why use CFG_FLASH_CFI_SWAP instead of __LITTLE_ENDIAN. It seems Li> that they are equivalent. No, they aren't. First of all, on big-endian systems __LITTLE_ENDIAN is never defined and on little-endian systems it's always defined automatically while CFG_FLASH_CFI_SWAP must be defined manually if your board has such a flash. On little-endian systems, the driver should work without the need to define CFG_FLASH_CFI_SWAP. Also, there are places in the code for __LITTLE_ENDIAN only, not for CFG_FLASH_CFI_SWAP. -- ======================================================================== Yuli Barcohen | Phone +972-9-765-1788 | Software Project Leader yuli at arabellasw.com | Fax +972-9-765-7494 | Arabella Software, Israel ======================================================================== ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 11:25 ` Yuli Barcohen @ 2006-08-08 11:42 ` Li Yang-r58472 2006-08-09 6:22 ` Zang Roy-r61911 2006-08-08 13:10 ` Sam Song 1 sibling, 1 reply; 31+ messages in thread From: Li Yang-r58472 @ 2006-08-08 11:42 UTC (permalink / raw) To: u-boot Got it. It's a little bit misleading to read your patch alone. Your patch is mostly the same with Roy's latest patch, but with a more generic name. So, I prefer your version. :) Best Regards, Leo > -----Original Message----- > From: Yuli Barcohen [mailto:yuli at arabellasw.com] > Sent: Tuesday, August 08, 2006 7:26 PM > To: Li Yang-r58472 > Cc: u-boot-users at lists.sourceforge.net; Sam Song; Wolfgang Denk > Subject: RE: [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue > > >>>>> Li Yang writes: > > Li> +#if defined(__LITTLE_ENDIAN) && !defined(CFG_FLASH_CFI_SWAP) > Li> +#define CFG_FLASH_CFI_SWAP > Li> +#endif > Li> + [snip] > Li> -#if defined(__LITTLE_ENDIAN) > Li> +#if defined(CFG_FLASH_CFI_SWAP) > > Li> Why use CFG_FLASH_CFI_SWAP instead of __LITTLE_ENDIAN. It seems > Li> that they are equivalent. > > No, they aren't. First of all, on big-endian systems __LITTLE_ENDIAN is > never defined and on little-endian systems it's always defined > automatically while CFG_FLASH_CFI_SWAP must be defined manually if your > board has such a flash. On little-endian systems, the driver should work > without the need to define CFG_FLASH_CFI_SWAP. Also, there are places in > the code for __LITTLE_ENDIAN only, not for CFG_FLASH_CFI_SWAP. > > -- > ======================================================================== > Yuli Barcohen | Phone +972-9-765-1788 | Software Project Leader > yuli at arabellasw.com | Fax +972-9-765-7494 | Arabella Software, Israel > ======================================================================== ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 11:42 ` Li Yang-r58472 @ 2006-08-09 6:22 ` Zang Roy-r61911 0 siblings, 0 replies; 31+ messages in thread From: Zang Roy-r61911 @ 2006-08-09 6:22 UTC (permalink / raw) To: u-boot > Got it. It's a little bit misleading to read your patch alone. > > Your patch is mostly the same with Roy's latest patch, but > with a more generic name. So, I prefer your version. :) Ummm... the name is better for other board. I had thought there were few boards to get into this issue except mpc7448hpc2. Roy ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 11:25 ` Yuli Barcohen 2006-08-08 11:42 ` Li Yang-r58472 @ 2006-08-08 13:10 ` Sam Song 2006-08-08 13:31 ` Yuli Barcohen 1 sibling, 1 reply; 31+ messages in thread From: Sam Song @ 2006-08-08 13:10 UTC (permalink / raw) To: u-boot Yuli Barcohen <yuli@arabellasw.com> wrote: > No, they aren't. First of all, on big-endian systems > __LITTLE_ENDIAN is > never defined and on little-endian systems it's > always defined > automatically while CFG_FLASH_CFI_SWAP must be > defined manually if your > board has such a flash. On little-endian systems, > the driver should work > without the need to define CFG_FLASH_CFI_SWAP. Also, > there are places in > the code for __LITTLE_ENDIAN only, not for > CFG_FLASH_CFI_SWAP. Seems your explaination is not the same as your patch:-). I hope I weren't clear this time. In your patch, CFG_FLASH_CFI_SWAP is automatically defined in __LITTLE_ENDIAN case. So on __LITTLE_ENDIAN system, CFG_FLASH_CFI_SWAP MUST BE ENABLED by default. +#if defined(__LITTLE_ENDIAN) && !defined(CFG_FLASH_CFI_SWAP) +#define CFG_FLASH_CFI_SWAP +#endif Regards, Sam ___________________________________________________________ Mp3???-??????? http://music.yahoo.com.cn/?source=mail_mailbox_footer ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 13:10 ` Sam Song @ 2006-08-08 13:31 ` Yuli Barcohen 2006-08-08 14:44 ` Sam Song 0 siblings, 1 reply; 31+ messages in thread From: Yuli Barcohen @ 2006-08-08 13:31 UTC (permalink / raw) To: u-boot >>>>> Sam Song writes: Yuli> No, they aren't. First of all, on big-endian systems Yuli> __LITTLE_ENDIAN is never defined and on little-endian systems Yuli> it's always defined automatically while CFG_FLASH_CFI_SWAP Yuli> must be defined manually if your board has such a flash. On Yuli> little-endian systems, the driver should work without the need Yuli> to define CFG_FLASH_CFI_SWAP. Also, there are places in the Yuli> code for __LITTLE_ENDIAN only, not for CFG_FLASH_CFI_SWAP. Sam> Seems your explaination is not the same as your patch:-). Yes, the explanation is in English and the patch is in C:) The meaning is the same though. Sam> I hope I weren't clear this time. Sam> In your patch, CFG_FLASH_CFI_SWAP is automatically defined in Sam> __LITTLE_ENDIAN case. So on __LITTLE_ENDIAN system, Sam> CFG_FLASH_CFI_SWAP MUST BE ENABLED by default. So it's exactly what I explained: little-endian systems work automatically without the need to define CFG_FLASH_CFI_SWAP. I meant, you don't have to #define CFG_FLASH_CFI_SWAP in the board configuration file to have working little-endian system. It's defined automatically when needed. Sorry if it wasn't clear enough. Sam> +#if defined(__LITTLE_ENDIAN) && !defined(CFG_FLASH_CFI_SWAP) Sam> +#define CFG_FLASH_CFI_SWAP Sam> +#endif If it makes things simpler, it's possible to replace #ifdef CFG_FLASH_CFI_SWAP by #if defined(__LITTLE_ENDIAN) || defined(CFG_FLASH_CFI_SWAP) and remove the above mentioned definition of CFG_FLASH_CFI_SWAP. -- ======================================================================== Yuli Barcohen | Phone +972-9-765-1788 | Software Project Leader yuli at arabellasw.com | Fax +972-9-765-7494 | Arabella Software, Israel ======================================================================== ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 13:31 ` Yuli Barcohen @ 2006-08-08 14:44 ` Sam Song 0 siblings, 0 replies; 31+ messages in thread From: Sam Song @ 2006-08-08 14:44 UTC (permalink / raw) To: u-boot Yuli Barcohen <yuli@arabellasw.com> wrote: > Yes, the explanation is in English and the patch is > in C:) The meaning is the same though. Well said:-) > So it's exactly what I explained: little-endian > systems work Oops, it is clear to me now. Keep all the code like that. Thanks, Sam ___________________________________________________________ ??????-3.5G???20M?? http://cn.mail.yahoo.com/ ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 11:00 ` Yuli Barcohen 2006-08-08 11:10 ` Li Yang-r58472 @ 2006-08-09 6:14 ` Zang Roy-r61911 2006-08-10 15:09 ` Sam Song 1 sibling, 1 reply; 31+ messages in thread From: Zang Roy-r61911 @ 2006-08-09 6:14 UTC (permalink / raw) To: u-boot > Zang> Where is the patch? I do not object. Our final goal is to > Zang> enable the general flash driver work on the board. > > Sam> Sorry, I misused the word patch:-). The "patch" is Yuli's > Sam> cfi_flash.c in his attached file. See your mail archive. It > Sam> would be there. I tested that it was OK but still need your > Sam> confirmation, I am afraid. > > OK, the patch is attached. As we know, it works on big-endian systems. > On little-endian ones, it should change nothing. > This patch is similar to my previous one. it works on my mpc7448hpc2 (tagia) board. I appreciate the method and the name you import CFG_FLASH_CFI_SWAP. I hope it can deal with similar boards. Wolfgang, Could you apply it in your public git tree? Roy ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-09 6:14 ` Zang Roy-r61911 @ 2006-08-10 15:09 ` Sam Song 0 siblings, 0 replies; 31+ messages in thread From: Sam Song @ 2006-08-10 15:09 UTC (permalink / raw) To: u-boot Zang Roy-r61911 <tie-fei.zang@freescale.com> wrote: > This patch is similar to my previous one. it works > on my mpc7448hpc2 (tagia) board. > I appreciate the method and the name you import > CFG_FLASH_CFI_SWAP. [snip] > Wolfgang, > Could you apply it in your public git tree? > Roy > Seems it's better to create the patch in a new thread sending to the list. Besides, it is necessary to add an entry for CFG_FLASH_CFI_SWAP in README. Thanks, Sam __________________________________________________ ??????????????? http://cn.mail.yahoo.com ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 10:52 ` Sam Song 2006-08-08 11:00 ` Yuli Barcohen @ 2006-08-09 1:32 ` Zang Roy-r61911 1 sibling, 0 replies; 31+ messages in thread From: Zang Roy-r61911 @ 2006-08-09 1:32 UTC (permalink / raw) To: u-boot > > Where is the patch? I do not object. > > Our final goal is to enable the general flash driver work on the > > board. > > Sorry, I misused the word patch:-). The "patch" is Yuli's > cfi_flash.c in his attached file. See your mail archive. It > would be there. I tested that it was OK but still need your > confirmation, I am afraid. I will try it. While it's better to provide patch. We can see the difference. Roy ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-04 12:32 [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue Sam Song 2006-08-04 13:44 ` Wolfgang Denk @ 2006-08-07 7:24 ` Zang Roy-r61911 2006-08-08 2:12 ` Sam Song 1 sibling, 1 reply; 31+ messages in thread From: Zang Roy-r61911 @ 2006-08-07 7:24 UTC (permalink / raw) To: u-boot > > Hello Ed Okerson, > > Recently, Roy Zang released MPC7448 HPC-II platform support > to this list and found a CFI Flash driver problem - > North-bridge chip TSI108 working as little-endian but write > to Flash is byte-swapped in a wrong way. > The workaround is to enable > CFG_FLASH_USE_BUFFER_WRITE in little endian setting. However, > CFG_FLASH_USE_BUFFER_WRITE should have nothing to do with > little endian if I am right. So I suspect flash writing with > little endian could have some problems. Flash writing on a pure little endian system should be right. CFG_FLASH_USE_BUFFER_WRITE could not solve the issue on mpc7448hpc2 board. Just as I mentioned before., "For handling unaligned head and tail bytes, the byte swap issue also exists. For example => cp.b 400002 fff80002 12 (from ram to flash)." There will be some adjustment for my mpc7448hpc2 patch. Now I am struggled with the tsi108 config read exception. I hope I can repost the whole patch again in this week. For cfi_flash.c driver, I add the following patch. diff --git a/drivers/cfi_flash.c b/drivers/cfi_flash.c index fd0a186..f621bc1 100644 --- a/drivers/cfi_flash.c +++ b/drivers/cfi_flash.c @@ -268,7 +268,7 @@ inline uchar flash_read_uchar (flash_inf uchar *cp; cp = flash_make_addr (info, 0, offset); -#if defined(__LITTLE_ENDIAN) +#if defined(__LITTLE_ENDIAN) || defined(CFG_FLASH_TSI_SWAP_BYTE) return (cp[0]); #else return (cp[info->portwidth - 1]); @@ -295,7 +295,7 @@ #ifdef DEBUG debug ("addr[%x] = 0x%x\n", x, addr[x]); } #endif -#if defined(__LITTLE_ENDIAN) +#if defined(__LITTLE_ENDIAN) || defined(CFG_FLASH_TSI_SWAP_BYTE) retval = ((addr[(info->portwidth)] << 8) | addr[0]); #else retval = ((addr[(2 * info->portwidth) - 1] << 8) | @@ -327,7 +327,7 @@ #ifdef DEBUG debug ("addr[%x] = 0x%x\n", x, addr[x]); } #endif -#if defined(__LITTLE_ENDIAN) +#if defined(__LITTLE_ENDIAN) || defined(CFG_FLASH_TSI_SWAP_BYTE) retval = (addr[0] << 16) | (addr[(info->portwidth)] << 24) | (addr[(2 * info->portwidth)]) | (addr[(3 * info->portwidth)] << 8); #else @@ -596,7 +596,6 @@ #ifdef CFG_FLASH_USE_BUFFER_WRITE int buffered_size; #endif /* get lower aligned address */ - /* get lower aligned address */ wp = (addr & ~(info->portwidth - 1)); /* handle unaligned start */ @@ -892,7 +891,7 @@ static void flash_make_cmd (flash_info_t int i; uchar *cp = (uchar *) cmdbuf; -#if defined(__LITTLE_ENDIAN) +#if defined(__LITTLE_ENDIAN) || defined(CFG_FLASH_TSI_SWAP_BYTE) for (i = info->portwidth; i > 0; i--) #else for (i = 1; i <= info->portwidth; i++) ^ permalink raw reply related [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-07 7:24 ` Zang Roy-r61911 @ 2006-08-08 2:12 ` Sam Song 2006-08-08 2:30 ` Zang Roy-r61911 0 siblings, 1 reply; 31+ messages in thread From: Sam Song @ 2006-08-08 2:12 UTC (permalink / raw) To: u-boot Zang Roy-r61911 <tie-fei.zang@freescale.com>wrote: > Flash writing on a pure little endian system should > be right. Out of interest, what the difference between a pure little endian system and TSI108 HLP working as LE mode? Could we set TSI108 HLP working as BE? Thanks a lot, Sam ___________________________________________________________ Mp3???-??????? http://music.yahoo.com.cn/?source=mail_mailbox_footer ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 2:12 ` Sam Song @ 2006-08-08 2:30 ` Zang Roy-r61911 2006-08-08 4:59 ` Sam Song 0 siblings, 1 reply; 31+ messages in thread From: Zang Roy-r61911 @ 2006-08-08 2:30 UTC (permalink / raw) To: u-boot > > Flash writing on a pure little endian system should be right. > > Out of interest, Why? It just begins. If everything is OK, what should we do? what the difference between a pure little > endian system and TSI108 HLP working as LE mode? > A pure little endian system might be X86 processor + little endian connected flash. > Could we set TSI108 HLP working as BE? we can set register HLP_DATA_SWAP_CTRL for data swap for tsi108 HLP port. Please refer 7.6 Data Swapping chapter of the data sheet. Roy ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 2:30 ` Zang Roy-r61911 @ 2006-08-08 4:59 ` Sam Song 2006-08-08 5:13 ` Zang Roy-r61911 0 siblings, 1 reply; 31+ messages in thread From: Sam Song @ 2006-08-08 4:59 UTC (permalink / raw) To: u-boot Zang Roy-r61911 <tie-fei.zang@freescale.com> wrote: > A pure little endian system might be > X86 processor + little endian connected flash. It reminded me of the fact there is probably a LE system in our LAB. I will play with it when available. > > Could we set TSI108 HLP working as BE? > we can set register HLP_DATA_SWAP_CTRL for data swap > for tsi108 HLP port. Please refer 7.6 Data Swapping > chapter of the data sheet. Thanks. I will check that. Sam ___________________________________________________________ ??????-3.5G???20M?? http://cn.mail.yahoo.com/ ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 4:59 ` Sam Song @ 2006-08-08 5:13 ` Zang Roy-r61911 2006-08-08 11:00 ` Sam Song 0 siblings, 1 reply; 31+ messages in thread From: Zang Roy-r61911 @ 2006-08-08 5:13 UTC (permalink / raw) To: u-boot > > A pure little endian system might be > > X86 processor + little endian connected flash. > > It reminded me of the fact there is probably a LE system in > our LAB. I will play with it when available. That's good. > > > > Could we set TSI108 HLP working as BE? > > we can set register HLP_DATA_SWAP_CTRL for data swap for tsi108 HLP > > port. Please refer 7.6 Data Swapping chapter of the data sheet. > > Thanks. I will check that. I have tried all the combinations, it does not seem workable for our issue. Roy ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 5:13 ` Zang Roy-r61911 @ 2006-08-08 11:00 ` Sam Song 0 siblings, 0 replies; 31+ messages in thread From: Sam Song @ 2006-08-08 11:00 UTC (permalink / raw) To: u-boot Zang Roy-r61911 <tie-fei.zang@freescale.com> wrote: > I have tried all the combinations, it does not seem > workable for our issue. That's fine. I meant to make such a try. Well, you save me from that. Thanks a lot, Sam ___________________________________________________________ Mp3???-??????? http://music.yahoo.com.cn/?source=mail_mailbox_footer ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue @ 2006-08-08 11:21 Li Yang-r58472 2006-08-08 13:47 ` Yuli Barcohen 0 siblings, 1 reply; 31+ messages in thread From: Li Yang-r58472 @ 2006-08-08 11:21 UTC (permalink / raw) To: u-boot Please ignore my last question. Lately, I find out in your .c file that there are still some places using __LITTLE_ENDIAN. However, I still have a question that in which condition CFG_FLASH_CFI_SWAP is required to set? Best Regards, Leo > -----Original Message----- > From: Li Yang-r58472 > Sent: Tuesday, August 08, 2006 7:11 PM > To: 'Yuli Barcohen'; u-boot-users at lists.sourceforge.net > Cc: Sam Song; Wolfgang Denk > Subject: RE: [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue > > +#if defined(__LITTLE_ENDIAN) && !defined(CFG_FLASH_CFI_SWAP) > +#define CFG_FLASH_CFI_SWAP > +#endif > + > [snip] > -#if defined(__LITTLE_ENDIAN) > +#if defined(CFG_FLASH_CFI_SWAP) > > Why use CFG_FLASH_CFI_SWAP instead of __LITTLE_ENDIAN. It seems that they are > equivalent. > > Best Regards, > Leo > > -----Original Message----- > > From: u-boot-users-bounces at lists.sourceforge.net > > [mailto:u-boot-users-bounces at lists.sourceforge.net] On Behalf Of Yuli Barcohen > > Sent: Tuesday, August 08, 2006 7:00 PM > > To: u-boot-users at lists.sourceforge.net > > Cc: Sam Song; Wolfgang Denk > > Subject: Re: [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue > > > > >>>>> Sam Song writes: > > > > Zang> Where is the patch? I do not object. Our final goal is to > > Zang> enable the general flash driver work on the board. > > > > Sam> Sorry, I misused the word patch:-). The "patch" is Yuli's > > Sam> cfi_flash.c in his attached file. See your mail archive. It > > Sam> would be there. I tested that it was OK but still need your > > Sam> confirmation, I am afraid. > > > > OK, the patch is attached. As we know, it works on big-endian systems. > > On little-endian ones, it should change nothing. > > > > -- > > ======================================================================== > > Yuli Barcohen | Phone +972-9-765-1788 | Software Project Leader > > yuli at arabellasw.com | Fax +972-9-765-7494 | Arabella Software, Israel > > ======================================================================== ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 11:21 Li Yang-r58472 @ 2006-08-08 13:47 ` Yuli Barcohen 2006-08-08 19:51 ` Tolunay Orkun 0 siblings, 1 reply; 31+ messages in thread From: Yuli Barcohen @ 2006-08-08 13:47 UTC (permalink / raw) To: u-boot >>>>> Li Yang writes: Li> Please ignore my last question. Lately, I find out in your .c Li> file that there are still some places using __LITTLE_ENDIAN. Li> However, I still have a question that in which condition Li> CFG_FLASH_CFI_SWAP is required to set? As I mentioned, it's for byte lanes swapping. If you connect a 16-bit flash chip to a PPC, you need to connect only two byte lanes of the PPC data bus i.e. lines D0-D15. For PPC, D0 is MSB and D15 is LSB. For the flash, D0 is LSB and D15 is MSB. So, you can connect D0 to D15, D1 to D14, and so on. Another possibility D0-D7 of PPC to D7-D0 of the flash and D8-D15 of PPC to D15-D8 of the flash. The latter case is the case of swapped byte lanes because PPC's most significant byte goes always to D0-D7 and thus to the least significant byte of the flash. So, in the latter case, you have to define CFG_FLASH_CFI_SWAP. Without CFG_FLASH_CFI_SWAP, any flash command (which is always single byte) will go to PPC's D8-D15 and consequently to the most significant byte of the flash. This won't work because the commands must go to the flash's least significant byte. Of course, PPC is only an example, the same problem can occur with other CPUs too, it only depends on the bus architecture. -- ======================================================================== Yuli Barcohen | Phone +972-9-765-1788 | Software Project Leader yuli at arabellasw.com | Fax +972-9-765-7494 | Arabella Software, Israel ======================================================================== ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 13:47 ` Yuli Barcohen @ 2006-08-08 19:51 ` Tolunay Orkun 2006-08-09 11:59 ` Yuli Barcohen 0 siblings, 1 reply; 31+ messages in thread From: Tolunay Orkun @ 2006-08-08 19:51 UTC (permalink / raw) To: u-boot Yuli Barcohen wrote: > As I mentioned, it's for byte lanes swapping. If you connect a 16-bit > flash chip to a PPC, you need to connect only two byte lanes of the PPC > data bus i.e. lines D0-D15. For PPC, D0 is MSB and D15 is LSB. For the > flash, D0 is LSB and D15 is MSB. So, you can connect D0 to D15, D1 to > D14, and so on. This is the correct way of interfacing flash chip per PowerPC documentation. > Another possibility D0-D7 of PPC to D7-D0 of the flash and D8-D15 of PPC > to D15-D8 of the flash. The latter case is the case of swapped byte lanes > because PPC's most significant byte goes always to D0-D7 and thus to the > least significant byte of the flash. This is really not the correct way to interface the Flash. It seems like hardware designer goofed up and asking the software guys to fix his/her mess. > So, in the latter case, you have to define CFG_FLASH_CFI_SWAP. Without > CFG_FLASH_CFI_SWAP, any flash command (which is always single byte) will > go to PPC's D8-D15 and consequently to the most significant byte of the > flash. This won't work because the commands must go to the flash's least > significant byte. Of course, PPC is only an example, the same problem > can occur with other CPUs too, it only depends on the bus architecture. Really it depends on the designer. The hardware designer can scramble the data bus in many more ways. Best regards, Tolunay ^ permalink raw reply [flat|nested] 31+ messages in thread
* [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue 2006-08-08 19:51 ` Tolunay Orkun @ 2006-08-09 11:59 ` Yuli Barcohen 0 siblings, 0 replies; 31+ messages in thread From: Yuli Barcohen @ 2006-08-09 11:59 UTC (permalink / raw) To: u-boot >>>>> Tolunay Orkun writes: Yuli> As I mentioned, it's for byte lanes swapping. If you connect a Yuli> 16-bit flash chip to a PPC, you need to connect only two byte Yuli> lanes of the PPC data bus i.e. lines D0-D15. For PPC, D0 is Yuli> MSB and D15 is LSB. For the flash, D0 is LSB and D15 is Yuli> MSB. So, you can connect D0 to D15, D1 to D14, and so on. Tolunay> This is the correct way of interfacing flash chip per Tolunay> PowerPC documentation. In most cases, yes, though there is no single "correct" way. Yuli> Another possibility D0-D7 of PPC to D7-D0 of the flash and Yuli> D8-D15 of PPC to D15-D8 of the flash. The latter case is the Yuli> case of swapped byte lanes because PPC's most significant byte Yuli> goes always to D0-D7 and thus to the least significant byte of Yuli> the flash. Tolunay> This is really not the correct way to interface the Tolunay> Flash. It seems like hardware designer goofed up and asking Tolunay> the software guys to fix his/her mess. It depends. In most cases, it's a designer's mistake. However, I saw it on roughly half of all boards I worked with so probably it can be declared a "feature." Linux MTD includes support for it. In some cases, it's not a mistake but the only way to connect the flash. For example, there are many flash chips which can work in both 8-bit and 16-bit modes. If you want to have both options on your board (probably, you're building an evaluation board or reference design), you have to swap the byte lanes if you want to preserve glueless PPC-flash interface. Yuli> So, in the latter case, you have to define Yuli> CFG_FLASH_CFI_SWAP. Without CFG_FLASH_CFI_SWAP, any flash Yuli> command (which is always single byte) will go to PPC's D8-D15 Yuli> and consequently to the most significant byte of the Yuli> flash. This won't work because the commands must go to the Yuli> flash's least significant byte. Of course, PPC is only an Yuli> example, the same problem can occur with other CPUs too, it Yuli> only depends on the bus architecture. Tolunay> Really it depends on the designer. The hardware designer Tolunay> can scramble the data bus in many more ways. Because of this, I think that using #ifdef __LITTLE_ENDIAN is not a good idea but I preserved the existing algorithm because I have no little-endian systems to test possible changes. -- ======================================================================== Yuli Barcohen | Phone +972-9-765-1788 | Software Project Leader yuli at arabellasw.com | Fax +972-9-765-7494 | Arabella Software, Israel ======================================================================== ^ permalink raw reply [flat|nested] 31+ messages in thread
end of thread, other threads:[~2006-08-10 15:09 UTC | newest] Thread overview: 31+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2006-08-04 12:32 [U-Boot-Users] [RFC] CFI Driver Little-Endian write Issue Sam Song 2006-08-04 13:44 ` Wolfgang Denk 2006-08-07 10:34 ` Sam Song 2006-08-07 11:04 ` Yuli Barcohen 2006-08-07 11:43 ` Stefan Roese 2006-08-08 1:59 ` Sam Song 2006-08-08 2:15 ` Zang Roy-r61911 2006-08-08 4:54 ` Sam Song 2006-08-08 5:18 ` Zang Roy-r61911 2006-08-08 10:52 ` Sam Song 2006-08-08 11:00 ` Yuli Barcohen 2006-08-08 11:10 ` Li Yang-r58472 2006-08-08 11:25 ` Yuli Barcohen 2006-08-08 11:42 ` Li Yang-r58472 2006-08-09 6:22 ` Zang Roy-r61911 2006-08-08 13:10 ` Sam Song 2006-08-08 13:31 ` Yuli Barcohen 2006-08-08 14:44 ` Sam Song 2006-08-09 6:14 ` Zang Roy-r61911 2006-08-10 15:09 ` Sam Song 2006-08-09 1:32 ` Zang Roy-r61911 2006-08-07 7:24 ` Zang Roy-r61911 2006-08-08 2:12 ` Sam Song 2006-08-08 2:30 ` Zang Roy-r61911 2006-08-08 4:59 ` Sam Song 2006-08-08 5:13 ` Zang Roy-r61911 2006-08-08 11:00 ` Sam Song -- strict thread matches above, loose matches on Subject: below -- 2006-08-08 11:21 Li Yang-r58472 2006-08-08 13:47 ` Yuli Barcohen 2006-08-08 19:51 ` Tolunay Orkun 2006-08-09 11:59 ` Yuli Barcohen
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox