diff options
author | Paul Fertser <fercerpav@gmail.com> | 2013-07-29 17:22:07 +0400 |
---|---|---|
committer | Spencer Oliver <spen@spen-soft.co.uk> | 2013-08-07 21:02:51 +0000 |
commit | fccc55225a7f874ee9d795286e9d92dc051b5b9d (patch) | |
tree | 17397d8f71d554e3b2e64d3cfc47b60dd2746144 /contrib | |
parent | d998ea40f3a323cfbc7b80d9c9d5057fbc76c6a9 (diff) | |
download | riscv-openocd-fccc55225a7f874ee9d795286e9d92dc051b5b9d.zip riscv-openocd-fccc55225a7f874ee9d795286e9d92dc051b5b9d.tar.gz riscv-openocd-fccc55225a7f874ee9d795286e9d92dc051b5b9d.tar.bz2 |
mdr32fx: support for Milandr's MDR32Fx internal flash memory
This adds example config and flash driver for russian Cortex-M3
microcontroller model.
Run-time tested on MDR32F9Q2I evaluation board; the flash driver
should be compatible with MDR32F2x (Cortex-M0) too but I lack hardware
to test.
There're no status bits at all, the datasheets specifies some delays
for flash operations instead. All being in <100us range, they're hard
to violate with JTAG, I hope. There're also no flash identification
registers so the flash size and type has to be hardcoded into the
config.
The flashing is considerably complicated because the flash is split
into pages, and each page consists of 4 interleaved non-consecutive
"sectors" (on MDR32F9 only, MDR32F2 is single-sectored), so the
fastest way is to latch the page and sector address and then write
only the part that should go into the current page and current sector.
Performance testing results with adapter_khz 1000 and the chip running
on its default HSI 8MHz oscillator:
When working area is specified, a target helper algorithm is used:
wrote 131072 bytes from file testfile.bin in 3.698427s (34.609 KiB/s)
This can theoretically be sped up by ~1.4 times if the helper
algorithm is fed some kind of "loader instructions stream" to allow
sector-by-sector writing.
Pure JTAG implementation (when target memory area is not available)
flashes all the 128k memory in 49.5s.
Flashing "info" memory region is also implemented, but due to the
overlapping memory addresses (resulting in incorrect memory map
calculations for GDB) it can't be used at the same time, so OpenOCD
needs to be started this way: -c "set IMEMORY true" -f
target/mdr32f9q2i.cfg
It also can't be read/verified because it's not memory-mapped anywhere
ever, and OpenOCD NOR framework doesn't really allow to provide a
custom handler that would be used when verifying.
Change-Id: I80c0632da686d49856fdbf9e05d908846dd44316
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1532
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Diffstat (limited to 'contrib')
-rw-r--r-- | contrib/loaders/flash/mdr32fx.S | 125 |
1 files changed, 125 insertions, 0 deletions
diff --git a/contrib/loaders/flash/mdr32fx.S b/contrib/loaders/flash/mdr32fx.S new file mode 100644 index 0000000..73f4b6f --- /dev/null +++ b/contrib/loaders/flash/mdr32fx.S @@ -0,0 +1,125 @@ +/*************************************************************************** + * Copyright (C) 2011 by Andreas Fritiofson * + * andreas.fritiofson@gmail.com * + * * + * Copyright (C) 2013 by Paul Fertser * + * fercerpav@gmail.com * + * * + * 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., * + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. * + ***************************************************************************/ + + .text + .syntax unified + .cpu cortex-m0 + .thumb + .thumb_func + .global write + + /* Params: + * r0 - flash base (in), status (out) + * r1 - count (32bit) + * r2 - workarea start + * r3 - workarea end + * r4 - target address + * Clobbered: + * r5 - rp + * r6 - wp, tmp + * r7 - current FLASH_CMD + */ + +#define FLASH_CMD 0x00 +#define FLASH_ADR 0x04 +#define FLASH_DI 0x08 + +#define FLASH_NVSTR (1 << 13) +#define FLASH_PROG (1 << 12) +#define FLASH_MAS1 (1 << 11) +#define FLASH_ERASE (1 << 10) +#define FLASH_SE (1 << 8) +#define FLASH_YE (1 << 7) +#define FLASH_XE (1 << 6) + + ldr r7, [r0, #FLASH_CMD] +wait_fifo: + ldr r6, [r2, #0] /* read wp */ + cmp r6, #0 /* abort if wp == 0 */ + beq exit + ldr r5, [r2, #4] /* read rp */ + cmp r5, r6 /* wait until rp != wp */ + beq wait_fifo + + ldr r6, [r5] /* "*target_address++ = *rp++" */ + str r4, [r0, #FLASH_ADR] + str r6, [r0, #FLASH_DI] + + ldr r6, =(FLASH_XE | FLASH_PROG) + orrs r7, r7, r6 + str r7, [r0, #FLASH_CMD] + # wait 5us + movs r6, #5 + bl delay + ldr r6, =#FLASH_NVSTR + orrs r7, r7, r6 + str r7, [r0, #FLASH_CMD] + # wait 10us + movs r6, #13 + bl delay + movs r6, #FLASH_YE + orrs r7, r7, r6 + str r7, [r0, #FLASH_CMD] + # wait 40us + movs r6, #61 + bl delay + movs r6, #FLASH_YE + bics r7, r7, r6 + str r7, [r0, #FLASH_CMD] + ldr r6, =#FLASH_PROG + bics r7, r7, r6 + str r7, [r0, #FLASH_CMD] + # wait 5us + movs r6, #5 + bl delay + ldr r6, =#(FLASH_XE | FLASH_NVSTR) + bics r7, r7, r6 + str r7, [r0, #FLASH_CMD] + + adds r5, #4 + adds r4, #4 + + cmp r5, r3 /* wrap rp at end of buffer */ + bcc no_wrap + mov r5, r2 + adds r5, #8 +no_wrap: + str r5, [r2, #4] /* store rp */ + subs r1, r1, #1 /* decrement word count */ + cmp r1, #0 + beq exit /* loop if not done */ + b wait_fifo +exit: + mov r0, r6 /* return status in r0 */ + bkpt #0 + + /* r6 - in + * for r6 == 1 it'll take: + * 1 (prepare operand) + 4 (bl) + 2 (subs+cmp) + 1 (bne) + 3 (b) -> + * 11 tacts == 1.4us with 8MHz + * every extra iteration will take 5 tacts == 0.6us */ +delay: + subs r6, r6, #1 + cmp r6, #0 + bne delay + bx lr |