drath [Thu, 21 Jun 2007 13:15:22 +0000 (13:15 +0000)]
- added support for Intel/Marvel PXA27x (XScale) targets
- added support for scans coming from or ending in Shift-DR or Shift-IR to bitbang code (required for XScale debugging)
- cleaned up errror handlers. only use when there's a catchable error
- fix segfault when etm was configured without a valid driver
ntfreak [Sat, 16 Jun 2007 14:45:55 +0000 (14:45 +0000)]
- added mingw elf patches from Vincent Palatin
- added str9x programming using flash controller tap (str9xpec), including option bytes and device lock/unlock
- inttypes.h now used for long long printf style declarations
drath [Thu, 14 Jun 2007 09:47:00 +0000 (09:47 +0000)]
- added manpage for OpenOCD (thanks to Uwe Hermann)
- fixed bug in ARM926EJ-S cache handling that caused cache linefills to be disabled after first debug entry
- added support for auto image type detection (thanks to Vincent Palatin)
- further work on ETM trace decoding (tested with a ETB interface using an ETM in normal 16-bit port mode, still experimental)
drath [Wed, 30 May 2007 15:47:18 +0000 (15:47 +0000)]
- reworked image handling to support multiple sections (tested with ihex file containing gaps)
This checkin is still experimental, not recommended for general use
drath [Tue, 29 May 2007 11:23:42 +0000 (11:23 +0000)]
- split fileio handling into fileio part and image handling
- reworked etm/etb into a generic etm part with trace capture drivers (currently only etb supported)
- added XScale debug handler binary to repository
- added Thumb disassembling (thanks to Vincent Palatin for this patch)
- added support for non-CFI compatible flashes to cfi driver (currently only SST39VFxxx devices supported)
This checkin is experimental, not suitable for general use
drath [Sat, 5 May 2007 11:02:03 +0000 (11:02 +0000)]
- further work on ETB decoding (not yet functional)
- fixed several uses of uninitialized bits (Thanks to Vincent Palatin for finding and fixing these bugs)
- explicitly disable monitor mode on ARM7/9 targets
- added "prepare_reset_halt()" to target_type_t, which allows reset_halt to be prepared before a reset is asserted, possibly preventing communication with the target
- arm7/9 devices now use a breakpoint at 0x0 or reset vector catching for debug out of reset
- correctly mask out bits that aren't part of a copied buffer
- fixed arm926ej-s CP15 register access handling
- correctly identify SYSCLK source in LPC3180 NAND flash controller driver
- disabled excessive debug output in jtag.h
- output all of EmbeddedICE version register in error message
- update OpenOCD's idea of the current core mode, and immediately change core mode, on CPSR changes
- added support for CFI cmdset 0002 (patch by Andrew Dyer, thanks a lot)
- enhanced CFI cmdset 0002 support to Atmel AT49 flashes (thanks to Joerg Krein for providing test hardware)
drath [Wed, 28 Mar 2007 16:31:55 +0000 (16:31 +0000)]
- merged XScale branch back into trunk
- fixed some compiler warnigns in amt_jtagaccel.c, bitbang.c, parport.c
- free working area and register stuff if str7x block write algorithm failed
- check PC after exiting a target algorithm in armv4_5.c
drath [Mon, 26 Mar 2007 21:47:26 +0000 (21:47 +0000)]
- fixed arm926 cp15 command bug (thanks to Vincent Palatin for this patch)
- fixed compiler warnings throughout the code (thanks to Vincent Palatin for this patch)
- added support for accessing ETB (embedded trace buffer) registers
drath [Thu, 15 Mar 2007 13:36:44 +0000 (13:36 +0000)]
- reworked file i/o. every fileaccess (target, flash, nand, in future configuration, too) should now go through the fileio subsystem
- added support for reading IHEX files (through fileio)
- load/dump_binary renamed to the more generic load/dump_image <file> <address> ['bin'|'ihex']
- added NAND framework (preliminary)
- added support for the LPC3180 SLC and MLC NAND controllers (preliminary)
- fix initialization for parport
- gw16012 fixes/cleanups
- added EmbeddedICE version 7 (preliminary, reported on two LPC23xx devices so far)
- added 'arm7_9 etm <target#>' configuration command to enable access to the ETM registers
drath [Wed, 31 Jan 2007 10:59:20 +0000 (10:59 +0000)]
- merged several changes from XScale
- complain about identify_chain scan with all bits one (jtag communication problem)
- add 0x80000 as a valid size for lpc2000_v2 flash banks (previously only the user accessible 0x7d000 were valid)
drath [Mon, 22 Jan 2007 14:47:00 +0000 (14:47 +0000)]
- fix incorrect parsing of whitespace in command.c (thanks to Magnus Lundin)
- fix infinite recursion in target_init_handler (thanks to jw and Magnus Lundin)
- fix CFI flash handlign with buswidth < 32bit (thanks to Daniele Orio for reporting this)
- add support for reading JTAG device id (currently only as debug output on startup)
- cleaned up handling of EmbeddedICE registers. Supported functionality and register size now determined by EmbeddedICE version number.
- small cleanups/fixes
drath [Tue, 9 Jan 2007 17:19:41 +0000 (17:19 +0000)]
- disabled FT2232 debug output (that code should be fairly stable)
- fixed bug in jtag.c where JTAG command type was wrong for PATHMOVE commands
- added lattice ISP cable (experimental)
- properly initialize parallel port (clear tristate bit)
- fixed jtag/Makefile.am that failed with some versions of autotools
- print error number if read() in gdbserver.c failed on Windows
- fixed handling of opcodes with bits 27-25 b011 (comment was correct, code had a bug)
- added support for AT91SAM7SE internal flash
drath [Wed, 22 Nov 2006 13:03:10 +0000 (13:03 +0000)]
- added a PLD (programmable logic device) subsystem for FPGA, CPLD etc. configuration
- added support for loading .bit files into Xilinx Virtex-II devices
- added support for the Gateworks GW16012 JTAG dongle
- merged CFI fixes from XScale branch
- a few minor fixes
drath [Tue, 7 Nov 2006 10:54:19 +0000 (10:54 +0000)]
- added support for the oocd-link (http://www.joernonline.de/dw/doku.php?id=en:projects:oocdlink)
- fixed breakpoint handling (this changes the target_t interface)
drath [Mon, 23 Oct 2006 14:02:34 +0000 (14:02 +0000)]
- several small fixes (thanks to Magnus Lundin and Spencer Oliver)
- added support for using Luminary Micro development board 'evb_lm3s811' as a JTAG interface (thanks to Magnus Lundin)
drath [Thu, 19 Oct 2006 10:07:24 +0000 (10:07 +0000)]
- fixed endianness handling in arm7_9_read_core_reg (thanks to Magnus Lundin for reporting this)
- correctly handle ft2232_device_desc and ft2232_serial when both are given at the same time
drath [Thu, 12 Oct 2006 16:20:47 +0000 (16:20 +0000)]
- allow FT2232 devices to be opened by serial number instead of device description ('ft2232_serial <serial>' command)
- redirect output from target event scripts (currently only reset) to the daemon output (INFO:)
- some minor fixes and enhancements
drath [Fri, 6 Oct 2006 15:46:25 +0000 (15:46 +0000)]
- added support for American Microsystem's M5960 (FT2232 based USB JTAG interface)
- added support for AT91SAM7A devices (thanks to Magnus Lundin)
- updated README with pointer to Amontec's JTAGkeyTiny
- str9x flash support (Thanks to Spencer Oliver)
- str75x flash support (Thanks to Spencer Oliver)
- correct reporting of T-Bit in CPSR (Thanks to John Hartman for reporting this)
- core-state (ARM/Thumb) can be switched by modifying CPSR
- fixed bug in gdb_server register handling
- register values > 32-bit should now be supported
- several minor fixes and enhancements
- allow writes to second flash bank on STR71x devices (cleaned up sector list building)
- cleaned up str7x flash erase and protect code (use two accesses to erase sectors in both banks)
- integrated patch from Magnus Lundin that fixes at91sam7 flash timing bugs and possible endianness problems (big endian hosts)
- correctly write trailing bytes on str7x flashes
- speed up str7x flash writing by offloading the algorithm to the target
drath [Thu, 31 Aug 2006 12:41:49 +0000 (12:41 +0000)]
- endianess fixes everywhere but in the flash code. flashing might still be broken on big-endian targets and/or hosts
- added access to ARM920T vector catch register (via generic register mechanism)
- don't disable linefills on ARM920T cores - this lead to lockups when accessing lines already contained in cache
- read content of ARM920T cache and tlb into file (arm920t read_flash/read_mmu commands)
- memory reading improved on ARM7/9, can be further accelerated with new "arm7_9 fast_memory_access enable" command (renamed from fast_writes)
- made in_handler independent from in field (makes the handler more flexible)
- added timeout to ft2232 when using D2XX library
- fixed STR7x protection bit handling on second bank (thanks to Bernard)
- added support for using the OpenOCD on AT91RM9200 systems (thanks to Anders Larsen)
- fixed AT91SAM7 flash handling when not running from 32kHz clock (thanks to Anders Larsen)
drath [Thu, 17 Aug 2006 14:53:15 +0000 (14:53 +0000)]
- renamed jtag_interface_t.support_statemove to jtag_interface_t.support_pathmove (it is used to indicate jtag_add_pathmove support)
- fixed small bug in str7x.c that printed an address instead of the target number in an error message
- added support for Olimex ARM-USB-OCD. The new ft2232 layout is called "olimex-jtag"
drath [Fri, 11 Aug 2006 17:49:41 +0000 (17:49 +0000)]
- added support for FreeBSD ppi (parallel port access similar to /dev/parport on linux) (thanks to Darius for this patch)
- unified 'ftd2xx' and 'ftdi2232' into a single interface 'ft2232'. The library used to access the hardware is choosen during configure with --enable-ft2232_ftd2xx or --enable-ft2232-libftdi.
drath [Sun, 6 Aug 2006 11:20:42 +0000 (11:20 +0000)]
- fixed a minor problem with the GDB server that could drop the first packet (non-fatal)
- fixed some small memory leaks (thanks to Spencer Oliver)
- verify chip- and buswidth of cfi flash configurations
- added support for ARM966E based systems (tested only with ST micro STR9, thanks to Spencer Oliver)
drath [Tue, 1 Aug 2006 09:45:22 +0000 (09:45 +0000)]
- allow building for MinGW using either -mno-cygwin or the MinGW gcc
- added GiveIO support to the amt_jtagaccel driver
- explicitly disable loopback mode for FT2232 devices
- changed configuration options n[st]rst_delay to jtag_n[st]rst_delay
- shutdown network services on exit
- added configurable delays after reset lines get deasserted. useful if reset circuitry keeps lines asserted for too long.
- additional debug output when opening the parallel port
- fixed counting of available arm7/9 watchpoint units
- 'flash write' now displays elapsed time
- Added support for native MinGW builds (thanks to Spencer Oliver and Michael Fischer) - you still need to install GiveIO (not part of OpenOCD)
- Added state-move support to ftd2xx and bitbang JTAG drivers (required for XScale, possibly useful for other targets, too)
- various fixes
drath [Sun, 25 Jun 2006 20:44:25 +0000 (20:44 +0000)]
- changed use of bzero (deprecated) to memset (thanks to Spen for pointing this out)
- changed fallback implementation of strndup to something that works on all systems (thanks to Spen for this patch)
drath [Sun, 25 Jun 2006 11:14:44 +0000 (11:14 +0000)]
- added "version" command (patch from John Hartman, thanks)
- fixed bug in telnet history handling (patch from John Hartman, thanks)
- OpenOCD version has been changed from SVN revision number to date/time
drath [Fri, 23 Jun 2006 07:54:01 +0000 (07:54 +0000)]
- added support for AT91SAM7A3 flash (patch from andre renaud, thanks)
- fix trunk build for mac os x (patch from Lauri Leukkunen, thanks)
- added check for host endianness, defines WORDS_BIGENDIAN on a big-endian host (e.g. mac os-x)
- fixed bug where endianness of memory accesses could be swapped on BE hosts
- added space for zero termination of ftd2xx_layout string (from Magnus Ludin, tahnks)
drath [Mon, 12 Jun 2006 16:49:49 +0000 (16:49 +0000)]
- the 'help' command now takes an optional argument to display help only on a certain command (thanks to Andrew Dyer for this enhancement)
- OpenOCD now includes the ability to diassemble instructions on its own (only ARM for now, Thumb might follow).
The command is "armv4_5 disassemble <address> <count> ['thumb']" (thumb is currently unsupported).
I've compared the produced disassembly against GDB/GNU Objdump output, and it seems to be correct, but there may still be some bugs left.