move nand drivers to src/flash/nand/
[openocd.git] / src / flash / nand / davinci.c
diff --git a/src/flash/nand/davinci.c b/src/flash/nand/davinci.c
new file mode 100644 (file)
index 0000000..72cd378
--- /dev/null
@@ -0,0 +1,756 @@
+/***************************************************************************
+ *   Copyright (C) 2009 by David Brownell                                  *
+ *                                                                         *
+ *   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.             *
+ ***************************************************************************/
+
+/*
+ * DaVinci family NAND controller support for OpenOCD.
+ *
+ * This driver uses hardware ECC (1-bit or 4-bit) unless
+ * the chip is accessed in "raw" mode.
+ */
+
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include "arm_nandio.h"
+
+
+enum ecc {
+       HWECC1,         /* all controllers support 1-bit ECC */
+       HWECC4,         /* newer chips also have 4-bit ECC hardware */
+       HWECC4_INFIX,   /* avoid this layout, except maybe for boot code */
+};
+
+struct davinci_nand {
+       struct target   *target;
+
+       uint8_t         chipsel;        /* chipselect 0..3 == CS2..CS5 */
+       uint8_t         eccmode;
+
+       /* Async EMIF controller base */
+       uint32_t                aemif;
+
+       /* NAND chip addresses */
+       uint32_t                data;           /* without CLE or ALE */
+       uint32_t                cmd;            /* with CLE */
+       uint32_t                addr;           /* with ALE */
+
+       /* write acceleration */
+       struct arm_nand_data    io;
+
+       /* page i/o for the relevant flavor of hardware ECC */
+       int (*read_page)(struct nand_device *nand, uint32_t page,
+                       uint8_t *data, uint32_t data_size, uint8_t *oob, uint32_t oob_size);
+       int (*write_page)(struct nand_device *nand, uint32_t page,
+                       uint8_t *data, uint32_t data_size, uint8_t *oob, uint32_t oob_size);
+};
+
+#define NANDFCR                0x60            /* flash control register */
+#define NANDFSR                0x64            /* flash status register */
+#define NANDFECC       0x70            /* 1-bit ECC data, CS0, 1st of 4 */
+#define NAND4BITECCLOAD        0xbc            /* 4-bit ECC, load saved values */
+#define NAND4BITECC    0xc0            /* 4-bit ECC data, 1st of 4 */
+#define NANDERRADDR    0xd0            /* 4-bit ECC err addr, 1st of 2 */
+#define NANDERRVAL     0xd8            /* 4-bit ECC err value, 1st of 2 */
+
+static int halted(struct target *target, const char *label)
+{
+       if (target->state == TARGET_HALTED)
+               return true;
+
+       LOG_ERROR("Target must be halted to use NAND controller (%s)", label);
+       return false;
+}
+
+static int davinci_init(struct nand_device *nand)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+       uint32_t nandfcr;
+
+       if (!halted(target, "init"))
+               return ERROR_NAND_OPERATION_FAILED;
+
+       /* We require something else to have configured AEMIF to talk
+        * to NAND chip in this range (including timings and width).
+        */
+       target_read_u32(target, info->aemif + NANDFCR, &nandfcr);
+       if (!(nandfcr & (1 << info->chipsel))) {
+               LOG_ERROR("chip address %08" PRIx32 " not NAND-enabled?", info->data);
+               return ERROR_NAND_OPERATION_FAILED;
+       }
+
+       /* REVISIT verify:  AxCR must be in 8-bit mode, since that's all we
+        * tested.  16 bit support should work too; but not with 4-bit ECC.
+        */
+
+       return ERROR_OK;
+}
+
+static int davinci_reset(struct nand_device *nand)
+{
+       return ERROR_OK;
+}
+
+static int davinci_nand_ready(struct nand_device *nand, int timeout)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+       uint32_t nandfsr;
+
+       /* NOTE: return code is zero/error, else success; not ERROR_* */
+
+       if (!halted(target, "ready"))
+               return 0;
+
+       do {
+               target_read_u32(target, info->aemif + NANDFSR, &nandfsr);
+
+               if (nandfsr & 0x01)
+                       return 1;
+
+               alive_sleep(1);
+       } while (timeout-- > 0);
+
+       return 0;
+}
+
+static int davinci_command(struct nand_device *nand, uint8_t command)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+
+       if (!halted(target, "command"))
+               return ERROR_NAND_OPERATION_FAILED;
+
+       target_write_u8(target, info->cmd, command);
+       return ERROR_OK;
+}
+
+static int davinci_address(struct nand_device *nand, uint8_t address)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+
+       if (!halted(target, "address"))
+               return ERROR_NAND_OPERATION_FAILED;
+
+       target_write_u8(target, info->addr, address);
+       return ERROR_OK;
+}
+
+static int davinci_write_data(struct nand_device *nand, uint16_t data)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+
+       if (!halted(target, "write_data"))
+               return ERROR_NAND_OPERATION_FAILED;
+
+       target_write_u8(target, info->data, data);
+       return ERROR_OK;
+}
+
+static int davinci_read_data(struct nand_device *nand, void *data)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+
+       if (!halted(target, "read_data"))
+               return ERROR_NAND_OPERATION_FAILED;
+
+       target_read_u8(target, info->data, data);
+       return ERROR_OK;
+}
+
+/* REVISIT a bit of native code should let block reads be MUCH faster */
+
+static int davinci_read_block_data(struct nand_device *nand,
+               uint8_t *data, int data_size)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+       uint32_t nfdata = info->data;
+       uint32_t tmp;
+
+       if (!halted(target, "read_block"))
+               return ERROR_NAND_OPERATION_FAILED;
+
+       while (data_size >= 4) {
+               target_read_u32(target, nfdata, &tmp);
+
+               data[0] = tmp;
+               data[1] = tmp >> 8;
+               data[2] = tmp >> 16;
+               data[3] = tmp >> 24;
+
+               data_size -= 4;
+               data += 4;
+       }
+
+       while (data_size > 0) {
+               target_read_u8(target, nfdata, data);
+
+               data_size -= 1;
+               data += 1;
+       }
+
+       return ERROR_OK;
+}
+
+static int davinci_write_block_data(struct nand_device *nand,
+               uint8_t *data, int data_size)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+       uint32_t nfdata = info->data;
+       uint32_t tmp;
+       int status;
+
+       if (!halted(target, "write_block"))
+               return ERROR_NAND_OPERATION_FAILED;
+
+       /* try the fast way first */
+       status = arm_nandwrite(&info->io, data, data_size);
+       if (status != ERROR_NAND_NO_BUFFER)
+               return status;
+
+       /* else do it slowly */
+       while (data_size >= 4) {
+               tmp = le_to_h_u32(data);
+               target_write_u32(target, nfdata, tmp);
+
+               data_size -= 4;
+               data += 4;
+       }
+
+       while (data_size > 0) {
+               target_write_u8(target, nfdata, *data);
+
+               data_size -= 1;
+               data += 1;
+       }
+
+       return ERROR_OK;
+}
+
+static int davinci_write_page(struct nand_device *nand, uint32_t page,
+               uint8_t *data, uint32_t data_size, uint8_t *oob, uint32_t oob_size)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       uint8_t *ooballoc = NULL;
+       int status;
+
+       if (!nand->device)
+               return ERROR_NAND_DEVICE_NOT_PROBED;
+       if (!halted(info->target, "write_page"))
+               return ERROR_NAND_OPERATION_FAILED;
+
+       /* Always write both data and OOB ... we are not "raw" I/O! */
+       if (!data) {
+               LOG_ERROR("Missing NAND data; try 'nand raw_access enable'\n");
+               return ERROR_NAND_OPERATION_FAILED;
+       }
+
+       /* If we're not given OOB, write 0xff where we don't write ECC codes. */
+       switch (nand->page_size) {
+       case 512:
+               oob_size = 16;
+               break;
+       case 2048:
+               oob_size = 64;
+               break;
+       case 4096:
+               oob_size = 128;
+               break;
+       default:
+               return ERROR_NAND_OPERATION_FAILED;
+       }
+       if (!oob) {
+               ooballoc = malloc(oob_size);
+               if (!ooballoc)
+                       return ERROR_NAND_OPERATION_FAILED;
+               oob = ooballoc;
+               memset(oob, 0x0ff, oob_size);
+       }
+
+       /* REVISIT avoid wasting SRAM:  unless nand->use_raw is set,
+        * use 512 byte chunks.  Read side support will often want
+        * to include oob_size ...
+        */
+       info->io.chunk_size = nand->page_size;
+
+       status = info->write_page(nand, page, data, data_size, oob, oob_size);
+       free(ooballoc);
+       return status;
+}
+
+static int davinci_read_page(struct nand_device *nand, uint32_t page,
+               uint8_t *data, uint32_t data_size, uint8_t *oob, uint32_t oob_size)
+{
+       struct davinci_nand *info = nand->controller_priv;
+
+       if (!nand->device)
+               return ERROR_NAND_DEVICE_NOT_PROBED;
+       if (!halted(info->target, "read_page"))
+               return ERROR_NAND_OPERATION_FAILED;
+
+       return info->read_page(nand, page, data, data_size, oob, oob_size);
+}
+
+static void davinci_write_pagecmd(struct nand_device *nand, uint8_t cmd, uint32_t page)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+       int page3 = nand->address_cycles - (nand->page_size == 512);
+
+       /* write command ({page,otp}x{read,program} */
+       target_write_u8(target, info->cmd, cmd);
+
+       /* column address (beginning-of-page) */
+       target_write_u8(target, info->addr, 0);
+       if (nand->page_size > 512)
+               target_write_u8(target, info->addr, 0);
+
+       /* page address */
+       target_write_u8(target, info->addr, page);
+       target_write_u8(target, info->addr, page >> 8);
+       if (page3)
+               target_write_u8(target, info->addr, page >> 16);
+       if (page3 == 2)
+               target_write_u8(target, info->addr, page >> 24);
+}
+
+static int davinci_writepage_tail(struct nand_device *nand,
+               uint8_t *oob, uint32_t oob_size)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+       uint8_t status;
+
+       if (oob_size)
+               davinci_write_block_data(nand, oob, oob_size);
+
+       /* non-cachemode page program */
+       target_write_u8(target, info->cmd, NAND_CMD_PAGEPROG);
+
+       if (!davinci_nand_ready(nand, 100))
+               return ERROR_NAND_OPERATION_TIMEOUT;
+
+       if (nand_read_status(nand, &status) != ERROR_OK) {
+               LOG_ERROR("couldn't read status");
+               return ERROR_NAND_OPERATION_FAILED;
+       }
+
+       if (status & NAND_STATUS_FAIL) {
+               LOG_ERROR("write operation failed, status: 0x%02x", status);
+               return ERROR_NAND_OPERATION_FAILED;
+       }
+
+       return ERROR_OK;
+}
+
+/*
+ * All DaVinci family chips support 1-bit ECC on a per-chipselect basis.
+ */
+static int davinci_write_page_ecc1(struct nand_device *nand, uint32_t page,
+               uint8_t *data, uint32_t data_size, uint8_t *oob, uint32_t oob_size)
+{
+       unsigned oob_offset;
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+       const uint32_t fcr_addr = info->aemif + NANDFCR;
+       const uint32_t ecc1_addr = info->aemif + NANDFECC + (4 * info->chipsel);
+       uint32_t fcr, ecc1;
+
+       /* Write contiguous ECC bytes starting at specified offset.
+        * NOTE: Linux reserves twice as many bytes as we need; and
+        * for 16-bit OOB, those extra bytes are discontiguous.
+        */
+       switch (nand->page_size) {
+       case 512:
+               oob_offset = 0;
+               break;
+       case 2048:
+               oob_offset = 40;
+               break;
+       default:
+               oob_offset = 80;
+               break;
+       }
+
+       davinci_write_pagecmd(nand, NAND_CMD_SEQIN, page);
+
+       /* scrub any old ECC state */
+       target_read_u32(target, ecc1_addr, &ecc1);
+
+       target_read_u32(target, fcr_addr, &fcr);
+       fcr |= 1 << (8 + info->chipsel);
+
+       do {
+               /* set "start csX 1bit ecc" bit */
+               target_write_u32(target, fcr_addr, fcr);
+
+               /* write 512 bytes */
+               davinci_write_block_data(nand, data, 512);
+               data += 512;
+               data_size -= 512;
+
+               /* read the ecc, pack to 3 bytes, and invert so the ecc
+                * in an erased block is correct
+                */
+               target_read_u32(target, ecc1_addr, &ecc1);
+               ecc1 = (ecc1 & 0x0fff) | ((ecc1 & 0x0fff0000) >> 4);
+               ecc1 = ~ecc1;
+
+               /* save correct ECC code into oob data */
+               oob[oob_offset++] = (uint8_t)(ecc1);
+               oob[oob_offset++] = (uint8_t)(ecc1 >> 8);
+               oob[oob_offset++] = (uint8_t)(ecc1 >> 16);
+
+       } while (data_size);
+
+       /* write OOB into spare area */
+       return davinci_writepage_tail(nand, oob, oob_size);
+}
+
+/*
+ * Preferred "new style" ECC layout for use with 4-bit ECC.  This somewhat
+ * slows down large page reads done with error correction (since the OOB
+ * is read first, so its ECC data can be used incrementally), but the
+ * manufacturer bad block markers are safe.  Contrast:  old "infix" style.
+ */
+static int davinci_write_page_ecc4(struct nand_device *nand, uint32_t page,
+               uint8_t *data, uint32_t data_size, uint8_t *oob, uint32_t oob_size)
+{
+       static const uint8_t ecc512[] = {
+               0, 1, 2, 3, 4, /* 5== mfr badblock */
+               6, 7, /* 8..12 for BBT or JFFS2 */ 13, 14, 15,
+       };
+       static const uint8_t ecc2048[] = {
+               24, 25, 26, 27, 28, 29, 30, 31, 32, 33,
+               34, 35, 36, 37, 38, 39, 40, 41, 42, 43,
+               44, 45, 46, 47, 48, 49, 50, 51, 52, 53,
+               54, 55, 56, 57, 58, 59, 60, 61, 62, 63,
+       };
+       static const uint8_t ecc4096[] = {
+                48,  49,  50,  51,  52,  53,  54,  55,  56,  57,
+                58,  59,  60,  61,  62,  63,  64,  65,  66,  67,
+                68,  69,  70,  71,  72,  73,  74,  75,  76,  77,
+                78,  79,  80,  81,  82,  83,  84,  85,  86,  87,
+                88,  89,  90,  91,  92,  93,  94,  95,  96,  97,
+                98,  99, 100, 101, 102, 103, 104, 105, 106, 107,
+               108, 109, 110, 111, 112, 113, 114, 115, 116, 117,
+               118, 119, 120, 121, 122, 123, 124, 125, 126, 127,
+       };
+
+       struct davinci_nand *info = nand->controller_priv;
+       const uint8_t *l;
+       struct target *target = info->target;
+       const uint32_t fcr_addr = info->aemif + NANDFCR;
+       const uint32_t ecc4_addr = info->aemif + NAND4BITECC;
+       uint32_t fcr, ecc4;
+
+       /* Use the same ECC layout Linux uses.  For small page chips
+        * it's a bit cramped.
+        *
+        * NOTE:  at this writing, 4KB pages have issues in Linux
+        * because they need more than 64 bytes of ECC data, which
+        * the standard ECC logic can't handle.
+        */
+       switch (nand->page_size) {
+       case 512:
+               l = ecc512;
+               break;
+       case 2048:
+               l = ecc2048;
+               break;
+       default:
+               l = ecc4096;
+               break;
+       }
+
+       davinci_write_pagecmd(nand, NAND_CMD_SEQIN, page);
+
+       /* scrub any old ECC state */
+       target_read_u32(target, info->aemif + NANDERRVAL, &ecc4);
+
+       target_read_u32(target, fcr_addr, &fcr);
+       fcr &= ~(0x03 << 4);
+       fcr |= (1 << 12) | (info->chipsel << 4);
+
+       do {
+               uint32_t raw_ecc[4], *p;
+               int i;
+
+               /* start 4bit ecc on csX */
+               target_write_u32(target, fcr_addr, fcr);
+
+               /* write 512 bytes */
+               davinci_write_block_data(nand, data, 512);
+               data += 512;
+               data_size -= 512;
+
+               /* read the ecc, then save it into 10 bytes in the oob */
+               for (i = 0; i < 4; i++) {
+                       target_read_u32(target, ecc4_addr + 4 * i, &raw_ecc[i]);
+                       raw_ecc[i] &= 0x03ff03ff;
+               }
+               for (i = 0, p = raw_ecc; i < 2; i++, p += 2) {
+                       oob[*l++] =   p[0]        & 0xff;
+                       oob[*l++] = ((p[0] >>  8) & 0x03) | ((p[0] >> 14) & 0xfc);
+                       oob[*l++] = ((p[0] >> 22) & 0x0f) | ((p[1] <<  4) & 0xf0);
+                       oob[*l++] = ((p[1] >>  4) & 0x3f) | ((p[1] >> 10) & 0xc0);
+                       oob[*l++] =  (p[1] >> 18) & 0xff;
+               }
+
+       } while (data_size);
+
+       /* write OOB into spare area */
+       return davinci_writepage_tail(nand, oob, oob_size);
+}
+
+/*
+ * "Infix" OOB ... like Linux ECC_HW_SYNDROME.  Avoided because it trashes
+ * manufacturer bad block markers, except on small page chips.  Once you
+ * write to a page using this scheme, you need specialized code to update
+ * it (code which ignores now-invalid bad block markers).
+ *
+ * This is needed *only* to support older firmware.  Older ROM Boot Loaders
+ * need it to read their second stage loader (UBL) into SRAM, but from then
+ * on the whole system can use the cleaner non-infix layouts.  Systems with
+ * older second stage loaders (ABL/U-Boot, etc) or other system software
+ * (MVL 4.x/5.x kernels, filesystems, etc) may need it more generally.
+ */
+static int davinci_write_page_ecc4infix(struct nand_device *nand, uint32_t page,
+               uint8_t *data, uint32_t data_size, uint8_t *oob, uint32_t oob_size)
+{
+       struct davinci_nand *info = nand->controller_priv;
+       struct target *target = info->target;
+       const uint32_t fcr_addr = info->aemif + NANDFCR;
+       const uint32_t ecc4_addr = info->aemif + NAND4BITECC;
+       uint32_t fcr, ecc4;
+
+       davinci_write_pagecmd(nand, NAND_CMD_SEQIN, page);
+
+       /* scrub any old ECC state */
+       target_read_u32(target, info->aemif + NANDERRVAL, &ecc4);
+
+       target_read_u32(target, fcr_addr, &fcr);
+       fcr &= ~(0x03 << 4);
+       fcr |= (1 << 12) | (info->chipsel << 4);
+
+       do {
+               uint32_t raw_ecc[4], *p;
+               uint8_t *l;
+               int i;
+
+               /* start 4bit ecc on csX */
+               target_write_u32(target, fcr_addr, fcr);
+
+               /* write 512 bytes */
+               davinci_write_block_data(nand, data, 512);
+               data += 512;
+               data_size -= 512;
+
+               /* read the ecc */
+               for (i = 0; i < 4; i++) {
+                       target_read_u32(target, ecc4_addr + 4 * i, &raw_ecc[i]);
+                       raw_ecc[i] &= 0x03ff03ff;
+               }
+
+               /* skip 6 bytes of prepad, then pack 10 packed ecc bytes */
+               for (i = 0, l = oob + 6, p = raw_ecc; i < 2; i++, p += 2) {
+                       *l++ =   p[0]        & 0xff;
+                       *l++ = ((p[0] >>  8) & 0x03) | ((p[0] >> 14) & 0xfc);
+                       *l++ = ((p[0] >> 22) & 0x0f) | ((p[1] <<  4) & 0xf0);
+                       *l++ = ((p[1] >>  4) & 0x3f) | ((p[1] >> 10) & 0xc0);
+                       *l++ =  (p[1] >> 18) & 0xff;
+               }
+
+               /* write this "out-of-band" data -- infix */
+               davinci_write_block_data(nand, oob, 16);
+               oob += 16;
+               oob_size -= 16;
+
+       } while (data_size);
+
+       /* the last data and OOB writes included the spare area */
+       return davinci_writepage_tail(nand, NULL, 0);
+}
+
+static int davinci_read_page_ecc4infix(struct nand_device *nand, uint32_t page,
+               uint8_t *data, uint32_t data_size, uint8_t *oob, uint32_t oob_size)
+{
+       davinci_write_pagecmd(nand, NAND_CMD_READ0, page);
+
+       /* large page devices need a start command */
+       if (nand->page_size > 512)
+               davinci_command(nand, NAND_CMD_READSTART);
+
+       if (!davinci_nand_ready(nand, 100))
+               return ERROR_NAND_OPERATION_TIMEOUT;
+
+       /* NOTE:  not bothering to compute and use ECC data for now */
+
+       do {
+               /* write 512 bytes */
+               davinci_read_block_data(nand, data, 512);
+               data += 512;
+               data_size -= 512;
+
+               /* read this "out-of-band" data -- infix */
+               davinci_read_block_data(nand, oob, 16);
+               oob += 16;
+               oob_size -= 16;
+       } while (data_size);
+
+       return ERROR_OK;
+}
+
+NAND_DEVICE_COMMAND_HANDLER(davinci_nand_device_command)
+{
+       struct davinci_nand *info;
+       struct target *target;
+       unsigned long chip, aemif;
+       enum ecc eccmode;
+       int chipsel;
+
+       /* arguments:
+        *  - "davinci"
+        *  - target
+        *  - nand chip address
+        *  - ecc mode
+        *  - aemif address
+        * Plus someday, optionally, ALE and CLE masks.
+        */
+       if (CMD_ARGC < 5) {
+               LOG_ERROR("parameters: %s target "
+                               "chip_addr hwecc_mode aemif_addr",
+                               CMD_ARGV[0]);
+               goto fail;
+       }
+
+       target = get_target(CMD_ARGV[1]);
+       if (!target) {
+               LOG_ERROR("invalid target %s", CMD_ARGV[1]);
+               goto fail;
+       }
+
+       COMMAND_PARSE_NUMBER(ulong, CMD_ARGV[2], chip);
+       if (chip == 0) {
+               LOG_ERROR("Invalid NAND chip address %s", CMD_ARGV[2]);
+               goto fail;
+       }
+
+       if (strcmp(CMD_ARGV[3], "hwecc1") == 0)
+               eccmode = HWECC1;
+       else if (strcmp(CMD_ARGV[3], "hwecc4") == 0)
+               eccmode = HWECC4;
+       else if (strcmp(CMD_ARGV[3], "hwecc4_infix") == 0)
+               eccmode = HWECC4_INFIX;
+       else {
+               LOG_ERROR("Invalid ecc mode %s", CMD_ARGV[3]);
+               goto fail;
+       }
+
+       COMMAND_PARSE_NUMBER(ulong, CMD_ARGV[4], aemif);
+       if (aemif == 0) {
+               LOG_ERROR("Invalid AEMIF controller address %s", CMD_ARGV[4]);
+               goto fail;
+       }
+
+       /* REVISIT what we'd *like* to do is look up valid ranges using
+        * target-specific declarations, and not even need to pass the
+        * AEMIF controller address.
+        */
+       if (aemif == 0x01e00000                 /* dm6446, dm357 */
+                       || aemif == 0x01e10000  /* dm335, dm355 */
+                       || aemif == 0x01d10000  /* dm365 */
+                       ) {
+               if (chip < 0x02000000 || chip >= 0x0a000000) {
+                       LOG_ERROR("NAND address %08lx out of range?", chip);
+                       goto fail;
+               }
+               chipsel = (chip - 0x02000000) >> 25;
+       } else {
+               LOG_ERROR("unrecognized AEMIF controller address %08lx", aemif);
+               goto fail;
+       }
+
+       info = calloc(1, sizeof *info);
+       if (info == NULL)
+               goto fail;
+
+       info->target = target;
+       info->eccmode = eccmode;
+       info->chipsel = chipsel;
+       info->aemif = aemif;
+       info->data = chip;
+       info->cmd = chip | 0x10;
+       info->addr = chip | 0x08;
+
+       nand->controller_priv = info;
+
+       info->io.target = target;
+       info->io.data = info->data;
+
+       /* NOTE:  for now we don't do any error correction on read.
+        * Nothing else in OpenOCD currently corrects read errors,
+        * and in any case it's *writing* that we care most about.
+        */
+       info->read_page = nand_read_page_raw;
+
+       switch (eccmode) {
+       case HWECC1:
+               /* ECC_HW, 1-bit corrections, 3 bytes ECC per 512 data bytes */
+               info->write_page = davinci_write_page_ecc1;
+               break;
+       case HWECC4:
+               /* ECC_HW, 4-bit corrections, 10 bytes ECC per 512 data bytes */
+               info->write_page = davinci_write_page_ecc4;
+               break;
+       case HWECC4_INFIX:
+               /* Same 4-bit ECC HW, with problematic page/ecc layout */
+               info->read_page = davinci_read_page_ecc4infix;
+               info->write_page = davinci_write_page_ecc4infix;
+               break;
+       }
+
+       return ERROR_OK;
+
+fail:
+       return ERROR_NAND_OPERATION_FAILED;
+}
+
+struct nand_flash_controller davinci_nand_controller = {
+       .name                   = "davinci",
+       .nand_device_command    = davinci_nand_device_command,
+       .init                   = davinci_init,
+       .reset                  = davinci_reset,
+       .command                = davinci_command,
+       .address                = davinci_address,
+       .write_data             = davinci_write_data,
+       .read_data              = davinci_read_data,
+       .write_page             = davinci_write_page,
+       .read_page              = davinci_read_page,
+       .write_block_data       = davinci_write_block_data,
+       .read_block_data        = davinci_read_block_data,
+       .nand_ready             = davinci_nand_ready,
+};

Linking to existing account procedure

If you already have an account and want to add another login method you MUST first sign in with your existing account and then change URL to read https://review.openocd.org/login/?link to get to this page again but this time it'll work for linking. Thank you.

SSH host keys fingerprints

1024 SHA256:YKx8b7u5ZWdcbp7/4AeXNaqElP49m6QrwfXaqQGJAOk gerrit-code-review@openocd.zylin.com (DSA)
384 SHA256:jHIbSQa4REvwCFG4cq5LBlBLxmxSqelQPem/EXIrxjk gerrit-code-review@openocd.org (ECDSA)
521 SHA256:UAOPYkU9Fjtcao0Ul/Rrlnj/OsQvt+pgdYSZ4jOYdgs gerrit-code-review@openocd.org (ECDSA)
256 SHA256:A13M5QlnozFOvTllybRZH6vm7iSt0XLxbA48yfc2yfY gerrit-code-review@openocd.org (ECDSA)
256 SHA256:spYMBqEYoAOtK7yZBrcwE8ZpYt6b68Cfh9yEVetvbXg gerrit-code-review@openocd.org (ED25519)
+--[ED25519 256]--+
|=..              |
|+o..   .         |
|*.o   . .        |
|+B . . .         |
|Bo. = o S        |
|Oo.+ + =         |
|oB=.* = . o      |
| =+=.+   + E     |
|. .=o   . o      |
+----[SHA256]-----+
2048 SHA256:0Onrb7/PHjpo6iVZ7xQX2riKN83FJ3KGU0TvI0TaFG4 gerrit-code-review@openocd.zylin.com (RSA)