Near all pracc functions store $15 in DeSave and
restore it when exiting.
There is no need to save it, if mips32_pracc_read_regs()
save this register in Desave when entering debug mode.
mips32_pracc_write_regs() needs to update it when
exiting debug mode.
Other pracc functions must not modify DeSave.
The jump code in the fastdata transfer function needs also
some little modifications.
Remark:
Like in current code the user can read/modify $15
with the cp0 31 commands.
Change-Id: I5b7dfc1b6169da846f5d2dd3ad4209a9da2c3fad
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1565
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Currently the code is loaded calling mips32_pracc_write_mem_generic().
Cache synchronization is not performed.
If configured as write back cache there is no chance to execute the
handler. If configured as write through cache and the cache
lines written to are not cache resident (I-side cache miss) may work.
The patch makes possible to execute the handler in a cached active
memory segment (mainly from KSEG0), but nothing else. The data
is still loaded without performing cache synchronization, code loaded
may not be executable.
Performance may not be faster. At start, for example, the code resides in
main memory, not in cache, and the core must transfer code from
memory. We can really modify the code to force a wait for the first
transfer like we do with start and end addresses, making sure the code
is cache resident for the rest of the queued transfers.
This can also may happen if we execute code (greater than the I cache size)
and the handler code is evicted from the cache.
Code tested on ar7241.
Change-Id: Iffdb4dae108b872fef0e7bacc5ea99649cdc1630
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1564
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Currently the functions mips32_checksum_memory() and mips32_blank_check_memory()
load the code word by word.
The bug in cache code is a good reason for doing so.
If there is no other reason we can load the code as a buffer to save time.
mips_m4k_write_memory() expect a buffer in target endianness, this is done by
target_buffer_set_u32_array().
Cleaned up exit code.
Tested on ar7241 big endian and pic32mx little endian with verify_image.
Flash erase check only tested in pic32mx.
Change-Id: Ib63ed98732b2e23b058e7349a0a57934b7604905
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1562
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Follows the the same rules of optimization used by all pracc functions.
Solves some bugs in previous code and adds support for write through caches.
Change-Id: If88c6738ca8c8197f327f22b766120a24f71b567
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1557
Tested-by: jenkins
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
As exposed by arm-none-eabi build, fix the wrong modifiers.
Change-Id: Ia6ce7c5c1d40e95059525c3e5d81b752df2fea7c
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/2122
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Paging checking and disabling wasn't present for IO r/w,
so the commands were successful only when paging wasn't
enabled (e.g. EFI boot phase).
Change-Id: I41366c0fadff3ea1eb8a153291f20a46cd9ddec1
Signed-off-by: Ivan De Cesaris <ivan.de.cesaris@intel.com>
Reviewed-on: http://openocd.zylin.com/2118
Tested-by: jenkins
Reviewed-by: Peter Stuge <peter@stuge.se>
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
The TI TMS470 and TMS570 series of processors are BE-32 processors,
despite BE-32 not being supported by ARM in the Cortex-R4 core. TI
hacked in BE-32 support, which requires odd swizzling in OpenOCD to
make memory reads and writes function correctly. In particular,
without this change, OpenOCD word reads and writes had the bytes
reversed, and halfword and byte packed reads were reading garbage.
In my testing, this change fixes these problems.
Change-Id: I21dd30f4b9003f20fcc85f674ab833407bb61f74
Signed-off-by: Seth LaForge <sethml@google.com>
Reviewed-on: http://openocd.zylin.com/2064
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Since c6216201b2 gdb target description
generation support is enabled by default and it counts on checking
"feature" pointer in reg_list. Both mips32 and dsp563xx neither used
calloc nor explicitly set feature (as it was a newly introduced struct
field).
This patch changes all targets to use calloc for consistency.
Change-Id: I9eceadef8b04aacb108e24ae23cb51ca3009586f
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/2102
Tested-by: jenkins
Reviewed-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
According to the "Arm Arch Ref Manual ARMv7-a and ARMv7-R edition" the
CPSR encoding for Monitor mode is 0b10110 (22) not 0b11010 (26) as is
currently used.
Change-Id: I73373a0029a81abc92febf518b88bf0dd4dec1fa
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/2081
Reviewed-by: Jörg Wunsch <openocd@uriah.heep.sax.de>
Tested-by: jenkins
Reviewed-by: Younes REGAIEG <y.regaieg@gmail.com>
Reviewed-by: Tim Sander <tim@krieglstein.org>
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
If gdb was disconnected by the moment the target entered halted state,
update_halt_gdb would segfault.
Change-Id: I67477e9199c1df097be83a49e38602f975c083f5
Reported-by: Younes REGAIEG <younes.regaieg@imag.fr>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/2098
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This introduces a new global Tcl procedure that is run just after
init_targets and before init_boards.
Its default behaviour is to assign gdb-flash-erase-start and
gdb-flash-write-end to reasonable defaults.
The rationale for doing "reset init" before gdb erases and flashes
memory is that all flash drivers are written in assumption that they
can safely be used only after chip reset (plus chip-specific
configuration in the init handler if any). The need to use "reset
halt" after flashing is because a user expects running firmware after
loading to be the same as running it from power-on-reset.
Change-Id: I9ddc4047611904ca4ca779b73376d2739611948a
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/2062
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
COMMAND_PARSE_NUMBER may return, thus any memory allocated may not be
freed, simple reorder fixes the issue.
Change-Id: I0ce444a5b032f5c49b6d33a03a8c0b71cad49c8c
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/2065
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
GPL requires providing sources for any derived work. I do not see any
reason to not include the xscale stuff into release tarballs.
Wildcard matching is used because plain directory name matches
implicit rule for executables and xscale.c built is errorneously
attempted, and directory name with a slash duplicates a directory
(xscale/xscale) in dist.
Change-Id: Ie0266470dcb97be87a09ba2dda9b3957f7cbc2fa
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1911
Tested-by: jenkins
Reviewed-by: Robert Jarzmik <robert.jarzmik@free.fr>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Ignoring the value parameter in that function makes its code rather
misleading. Also the only caller of it, armv7m_restore_context already
does the whole "buf_get_u32" conversion business, so using
'value' also removes the waste of doing the conversion twice.
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Change-Id: I515979c314d9b59ee1065c55b5bb5747c7e93f01
Reviewed-on: http://openocd.zylin.com/2057
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The timer callback is started on target init, but it makes no sense to
poll until the target is fully setup.
Change-Id: I118201e125e39be3d0a920e3ef9a3f68a2035f39
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/2041
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The ST-LINK/V2 has limited internal buffering, such that trace data
can be missed if the target is generating data at a rate quicker than
the OpenOCD trace sampling. The issue of lost data is compounded since
individual TPIU packets may be split across individual STLINK_TRACE_EP
reads, and misleading results can occur if mid-packet loss occurs.
This patch increases the frequency of checking for pending trace data
with the aim of minimising such losses. Note: With the limited (I/O
and memory) bandwidth of the ST-LINK/V2 there cannot, however, be a
guarantee against trace data loss.
The timer callback is only added when enabling tracing, and is removed
when tracing is disabled.
Change-Id: Ibde9794b77793d3068f88cb5c1a26f9ceadcbd8a
Signed-off-by: James G. Smith <jsmith@ecoscentric.com>
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1661
Tested-by: jenkins
Make the APB-AB memory read routines handle endianess order
when running on big endian host. cortex_a8_read_apb_ab_memory
is also called by cortex_a8_write_apb_ab_memory and was breaking
both APB-AB read and write functions. Also fixed bug in write
function in calculating the offset of end of buffer data. The
change aslo fixes the read issues with all combinations of
aligned unaligned memory access found by 'test_mem_access' cmd.
Tested with target "test_mem_access 4000", also size 1-9,
'mdb/h/w' 'mwb/h/w' cmds and equivalent gdb 'x' 'set' cmds.
Change-Id: Ia927c60c4837617f5342a9beb6fdab1f061855fe
Signed-off-by: Kamal Dasu <kdasu.kdev@gmail.com>
Reviewed-on: http://openocd.zylin.com/1781
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Tested-by: jenkins
This makes the listing easier to read, imho. The tab indentation
technique causes the base address to precess as the parser proceeds
through the subtables, and can easily wrap.
Change-Id: Iea5e678255e6314a9d532e4b222a2572b5394390
Signed-off-by: Brad Riensche <brad.riensche@gmail.com>
Reviewed-on: http://openocd.zylin.com/1518
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Move the ROM table printing into a separate function to allow
recursive calls with nested tables. ROM tables can nest. The
printing is limited to 16 levels.
Update the types of tables printed. When an entry can't be read, print
a warning and continue.
Change-Id: Ib134edd9e987af2f5f606071521885b17af4d70f
Signed-off-by: Chris Johns <chrisj@rtems.org>
Reviewed-on: http://openocd.zylin.com/1427
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
v2:
- Split work into separate patches
The received packet will not be altered in any of the processing functions.
Some it can be made "const".
Change-Id: I7bb410224cf6daa74a6c494624176ccb9ae638ac
Signed-off-by: Christian Eggers <ceggers@gmx.de>
Reviewed-on: http://openocd.zylin.com/1919
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Perform proper ICache flush operations on memory writes. This should fix
inability to use software breakpoints for debugging with caches
enabled.
This patch is only compile-time tested.
Commit 1137eaedaf fixed the same issue
for arm920t. Among all the arm7_9_common targets only arm926ejs seems
to be broken in the same way.
Change-Id: I575306ac4319a69fc637b42f7c958f4595c5e81f
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1912
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Intel Quark X10xx SoC debug support added
Lakemont version 1 (LMT1) is the x86 core in Quark X10xx SoC
Generic x86 32-bit code is in x86_32_common.c/h
Change-Id: If2bf77275cd0277a82558cd9895b4c66155cf368
Signed-off-by: adrian.burns@intel.com
Reviewed-on: http://openocd.zylin.com/1829
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
On a target where AHB AP memory access is unavailable, care should be
taken to avoid treating addresses as virtual if the MMU was disabled
at the time the target was stopped.
Without this it's impossible to peek memory with Gdb when debugging
e.g. a bootloader because cortex_a8_read_memory() unconditionally
tried (and failed because of a sanity check in cortex_a8_mmu_modify)
to enable MMU.
Change-Id: Id7c63f4912920fb71a6104226ec6428d18c96a56
Reported-by: mbm@openwrt.org
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1787
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
According to the standard every operation returns at least an integer,
so PRIu8 format specifier is not suitable for these values as is.
This breaks build on OS X (x86_64-apple-darwin13.0.0) with "Apple LLVM
version 5.0 (clang-500.2.79) (based on LLVM 3.3svn)".
Fix by adding appropriate casts. In fact there's plenty of room (and
I'd say necessity) for factoring out common code in there, but it's
too invasive for a non-maintainer.
Change-Id: I7d2182eb1d2f86fa22c882fbbaa6cfadf1c3e8fc
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1878
Tested-by: jenkins
Reviewed-by: Xiaofan <xiaofanc@gmail.com>
Reviewed-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The argument 'force' enables a user to bypass the internal cache and read
a target register directly. However it is missing from the user guide.
Change-Id: I26f689eec20b38a0dc5294626b25df566b554446
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1897
Tested-by: jenkins
Commit ff5ec942d8 made this target
always use generic arm7_9 memory write routines for software
breakpoints which resulted in inability to debug and single-step
sources in Gdb when icache is active as generic routine doesn't
invalidate it. This should fix it (and is real-life tested against
Samsung S3C2442). I expect other arm7-9 targets to be affected as
well.
Change-Id: Id7980e370ae4db47ac6b1490321d81ffe85711c0
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1817
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
This commit adds support for passing the ARM4 target description to GDB
when enabling gdb_target_description, in order to expose all banked
registers.
Change-Id: Id618bc6226f00fe83397ea28888a84b64b09cafd
Signed-off-by: Luca BRUNO <lucab@debian.org>
Reviewed-on: http://openocd.zylin.com/1810
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The out only version of jtag_add_dr_scan smells like a bogus optimization
that complicates the minidriver API for questionable gain.
The function was only used by four old ARM targets. Rewrite the callers
to use the generic function and remove all implementations.
Change-Id: I13b643687ee8ed6bc9b6336e7096c34f40ea96af
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1801
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Review and modify to conform to C99 integer types format specifiers.
Use arm-none-eabi toolchain to build successfully.
Change-Id: If855072a8f88886809309155ac6d031dcfcbc4b2
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Signed-off-by: Hsiangkai <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1794
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This is used for the emulated DCC channel which is only maintained as long
as target->dbg_msg_enabled is set. Skip the saving and restoring if not
enabled to save one dap_run() per core register access.
Note that we could've probably queued all core register accesses in the
same transaction if the armv7 register framework hadn't required
synchronous register accesses.
Change-Id: I4fe6d713261ee5db42422203eb63035fdcc48891
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1848
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
It's far nicer to pass a size parameter than to split the calls to
separate wrappers which are combined to a single function anyway.
Change-Id: I716741ebf916f6f8e9358a31c8f4fe761107c82f
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1847
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This is based on work from:
https://github.com/TheShed/OpenOCD-CMSIS-DAP/tree/cmsis-dap
Main changes include moving over to using HIDAPI rather than libusb-1.0
and cleaning up to merge into master. Support for reset using srst has
also been added.
It has been tested on all the mbed boards as well as the Freedom board
from Freescale. These boards only implement SWD mode, however JTAG mode
has been tested with a Keil ULINK2 and a stm32 target - but requires a lot
more work.
Change-Id: I96d5ee1993bc9c0526219ab754c5aad3b55d812d
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Signed-off-by: Andrey Yurovsky <yurovsky@gmail.com>
Reviewed-on: http://openocd.zylin.com/1542
Tested-by: jenkins
This is not in the Intel hex file format specification but
some hex files may include comments (i.e. Altera USB-Blaster II
firmware) starting with '#'.
This patch makes image_ihex_buffer_complete_inner to skip
comment lines.
Change-Id: Id1f57d84d75da45e592f1c72b2b8c29193bc14e3
Signed-off-by: Franck Jullien <franck.jullien@gmail.com>
Reviewed-on: http://openocd.zylin.com/1842
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This add support to the Xilinx BSCAN_* virtual JTAG interface.
This is the Xilinx equivalent of the Altera sld_virtual_jtag interface,
it allows a user to connect to the debug unit through the main
FPGA JTAG connection.
Change-Id: Ia438e910650cff9cbc8f810b719fc1d5de5a8188
Signed-off-by: Sergio Chico <sergio.chico@gmail.com>
Reviewed-on: http://openocd.zylin.com/1806
Tested-by: jenkins
Reviewed-by: Franck Jullien <franck.jullien@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
If the flash is not ready (MDM_STAT_FREADY is 0) then
dap_syssec_kinetis_mdmap() would act as if the MDM_STAT_SYSSEC bit was
set and erase the flash. Wait until MDM_STAT_FREADY is set before
checking the MDM_STAT_SYSSEC bit.
Change-Id: I5c3352f625599016625ed9be8787033f49bfacea
Signed-off-by: Per Ekman <pekenator@gmail.com>
Reviewed-on: http://openocd.zylin.com/1762
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Attempt to use target_buffer_set_u32_array to convert to target endian
arrays rather reimplementing code.
This also removed cfi_fix_code_endian as its functionality is also
repeated.
Change-Id: I7c359dbe46ea791cd5f6fb18d8b0fb6895c599d3
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1783
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
if data size is bigger than transfer buffer, all portions are
transferred from/to the same target address - address advance
after successful transmission missed.
Change-Id: I79a6c388af197ac062d2807e397a2d7947400520
Signed-off-by: Sergey A. Borshch <sb-sf@users.sourceforge.net>
Signed-off-by: Mathias K <kesmtp@freenet.de>
Reviewed-on: http://openocd.zylin.com/1679
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Tested-by: jenkins
Found by grepping for pointer casts.
Also rewrite to reduce scope and allocate the few bytes needed on stack
instead of on heap.
Change-Id: Ia2a369fb612e807b981ee60ebcfd9c09c2fbdf4c
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1779
Tested-by: jenkins
Reviewed-by: Mathias Küster <kesmtp@freenet.de>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
When pointer casts are needed, cast directly to the correct type, instead
of going via void*.
Don't explicitly cast to void* if it would have been done implicitly.
Change-Id: I4093209200051c5eb62847d00a4b9c8567480068
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1669
Tested-by: jenkins
Reviewed-by: Mathias Küster <kesmtp@freenet.de>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Don't use const on pointers that hold heap allocated data, because that
means functions that free them must cast away the const.
Do use const on pointer parameters or fields that needn't be modified.
Remove pointer casts that are no longer needed after fixing the constness.
Change-Id: I5d206f5019982fd1950bc6d6d07b6062dc24e886
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1668
Tested-by: jenkins
Reviewed-by: Mathias Küster <kesmtp@freenet.de>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This eliminates most of the warnings reported when building for
arm-none-eabi (newlib).
Hsiangkai, there're many similar warnings left in your nds32 files, I
didn't have the nerve to clean them all, probably you could pick it
up.
Change-Id: Id3bbe2ed2e3f1396290e55bea4c45068165a4810
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1674
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
If the fpb_enabled is not set then as part of cortex_m3_set_breakpoint we
enable the fpb, however we do not signal the fpb as being enabled.
This issue only effects the hla target as the current cortex_m code enables
the fpb during cortex_m3_endreset_event.
Change-Id: I44d3fc65916c131b7a226869dd16aed5afb441b4
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1634
Tested-by: jenkins
Polling target does not change stste information
except if new state is TARGET_HALTED.
Connecting to the runing target result in target->state
not updated with retrieved value and remains "UNKNOWN"
until 'halt' command issued.
Change-Id: I803d6c0207f7f8699e648779d1df342c9ee7315a
Signed-off-by: Sergey A. Borshch <sb-sf@users.sourceforge.net>
Reviewed-on: http://openocd.zylin.com/1680
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Add a mandatory field in struct arm7_9_common for regular, non-optimized
memory writes. Together with the existing bulk_memory_write field, this
allows variants to select any combination of implementations for regular
and bulk writes, without risking infinite loops from accidentally using
bulk writes for implementing bulk writes.
ARM 7/9 targets may now select arm7_9_memory_write_opt as their
target.write_memory implementation, which will dispatch to
arm7_9_common.bulk_write_memory if possible, or fallback to
arm7_9_common.write_memory otherwise.
To avoid loops, bulk write implementations mustn't call any other
functions than arm7_9_write_memory_no_opt() to write memory; it will
unconditionally call arm7_9_common.write_memory. If they fail, they should
simply return error to allow the caller to fallback to regular writes.
Tested on a regular ARM7TDMI only.
Change-Id: Iae42a6e093e2df68c4823c927d757ae8f42ef388
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1685
Tested-by: jenkins
Reviewed-by: Sergey A. Borshch <sb-sf@users.sourceforge.net>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
We changed the actual target name quite a while ago.
This changes the actual target function names/defines to also match
this change.
Change-Id: I4f22fb107636db2279865b45350c9c776e608a75
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1626
Tested-by: jenkins
Create a single pair of relatively simple functions to handle all variants
of MEM-AP transfers. This replaces the many separate functions that
handled different access sizes and packed or non-packed transfers, which
were all implemented rather differently.
With this single implementation, performance should be more consistent,
regardless of transfer type.
Change-Id: I89960e437fc6ba68a389c074fab8eac91abcf844
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1658
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Accessing one byte of memory using a 16-bit access is not well defined.
The current implementation is forgiving and rounds up, but it should not
be relied upon.
Also, I suspect this code might fail if the byte order differs between
target and host, but I have no way of verifying it so I left it as it is.
Change-Id: I8d6a511151a194ed419f141703201f0632d84fc8
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1657
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The word count returned from max_tar_block_size() was compared with the
count of half-word/bytes in the u16 and u8 packed access functions,
causing an infinite loop if the access actually crossed the boundary.
Change max_tar_block_size() to return a byte count, and scale at the call
site.
Change-Id: I2fe9b5941eb485f3d8219cfdd29fb71e02006de4
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1649
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
For packed and/or unaligned accesses, the write functions reordered the
source buffer in place. Causing in the best case a segfault, in the worst
case silent data corruption.
Rewrite the data fetching to directly match the byte lane mapping
according to IHI0031C, without destroying the buffer.
Also slightly clean up variable usage and harmonize all the write
functions.
Change-Id: I9a01cfc5578653f9ec02043ff6b61a7a20f90d67
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1646
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Add support for OpenRISC target. This implementation
supports the adv_debug_sys debug unit core. The mohor
dbg_if is not supported. Support for mohor TAP core
and Altera Virtual JTAG core are also provided.
Change-Id: I3b1cfab1bbb28e497c4fca6ed1bd3a4362609b72
Signed-off-by: Franck Jullien <franck.jullien@gmail.com>
Reviewed-on: http://openocd.zylin.com/1547
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
It's been like this since the check was added, in 5aa93a5e.
Change-Id: Iaa0586e0cd1ce57ad92735dcc3e51108a133fe96
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1640
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Due to issues reported when using the jtag mode of the stlink (see Trac #61),
the functionality/checking has been moved to the driver.
This change also fixes unaligned 32bit memory read/write for the stlink.
From testing this change also brings a 3KiB/s speed increase, this is due
to the larger read/write packets.
Change-Id: I8234110e7e49a683f4dadd54c442ecdc3c47b320
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1632
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
This is the first part in moving the adapter specific api back into the
driver.
The next task is to also move the hla read/write buffer size handling.
Change-Id: I86a19144b50b6de18659bfcd05b3729b2cafc051
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1621
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Now that we support the target description format we do not need
these hacks anymore, we can now tell gdb what registers we support.
Change-Id: Ie774231d296420b35efcf708bc4435475c87ff5e
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1617
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Add support for the default gdb register description type "int" and "float".
When this is given to gdb it will use the bitsize to determine the reg size.
Change-Id: Iaeed594d1feece54044128eae1baff9858bdcae0
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1622
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Half-word and byte writes using mwh/mwb has been completely broken since
bf71e34cbf, because it dispatched all writes, regardless of access size,
to target_write_buffer(), which uses as large accesses as possible.
Partially revert the commit by changing back to the correct write method.
Change-Id: I60731fc576bf4a076a7da02bee7879e121c21d17
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1628
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Currently we try and use MEM-AP packed transfers as much as possible for
8/16bit transfers. However not all targets support packed transfers, so
check before using.
Change-Id: I66256007f25ccd0c583f23db5acf6d1aa8b5e57d
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1602
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
.soft_reset_halt is not necessary for nds32 target.
Remove the dependency.
Change-Id: Ic3b126d6c7eb995583a661b762627e736222fcaa
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1612
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
As GDB uses file-I/O protocol to access memory, use DMA to access
if no DCache. This commit improves the performance of Andes
Virtual Hosting.
Change-Id: I36bb2154b9f497fc4237625836cf8c7115330a60
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1580
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Modify handle_nds32_memory_access_command: do not use DCache
setting to block user's setting.
Change-Id: I2d33f893773e2a2e3e2f26edde5938ef5902609d
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1579
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
If multiple read watchpoints are hit at the same time, report
any one of these hit watchpoints.
Change-Id: I8d4439aa80e4b62bb7d48bbdc0450920e2008a2e
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1576
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Do not turn on/off polling as leave/enter debug mode.
Enable polling after gdb attached, and disable polling
after gdb detached.
Change-Id: Id64459b86f44937af7ea5ccfe2cd13e31732eecf
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1574
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
1. gprof uses 2-bytes as minimum bucket size.
2. As user wants to use gprof --sum to summarize multiple
profiling data files, the range MUST be the same.
Add new arguments to specify profiling range.
Change-Id: Ie7e6afa6a4d82250e2d194a0eed2b428c1479ea1
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1572
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Profiling could be target-specific. Add .profiling interface
to target_type.
Change-Id: Ic0eea9db742971db1350a474fbbb5ed24565922b
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1571
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
I do not know what is the reasonable number of buckets.
If there are enough buckets, the result will be accurate.
I propose increase the maximum number of buckets to 128K.
If the size of program text section is less than 256KB, every
two bytes will be occupied by one buckets.
(The minimum size of one buckets is 2 bytes in gprof implementation.)
Change-Id: If9147743cefdc36f40f21e6dc73b9b28f28c9e1e
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1608
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
1. high_pc should be (maximum sample + 1) (Refer to gprof source code)
2. bucket index should be
sample offset
--------------- x (number of bucket)
sample range
For example, if minimum sample is 0 and maximum sample is 5
and the number of bucket is 3.
a = sampled_address - 0
b = 3
c = 6
(a, b, c refer to source code variables)
sampled_address = 0, => a = 0, => bucket_index = 0
sampled_address = 1, => a = 1, => bucket_index = 0
sampled_address = 2, => a = 2, => bucket_index = 1
sampled_address = 3, => a = 3, => bucket_index = 1
sampled_address = 4, => a = 4, => bucket_index = 2
sampled_address = 5, => a = 5, => bucket_index = 2
Change-Id: Ia9fa0e4d9c7183e3e9d7ceaf73e63729f07aa2ce
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1607
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Change variable name 'length' to 'numBuckets'. It is more readable.
Change-Id: I913cba0746f887adf6da401a46cd5e9ea88d2c6d
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1606
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Rewrite the target_*_buffer_default to generate as large accesses as
possible while maintaining natural alignment.
These versions are easy to extend to generate 8-byte accesses to support
64-bit targets, although it requires some conformity from all target
implementations (i.e. they need to refuse unsupported access sizes with
some defined error code, so we can try again with a smaller one).
Change-Id: I00ddcbb1d2fd33f9f8b99cb448cc93505a2421fc
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1221
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The only remaining user is arm7_9 so remove it from the target API and add
it to struct arm7_9_common to support all its variants with minimal
changes. Many of the variants are likely not correct in the cache/mmu
handling when the bulk write is triggered. This patch does nothing to
change that, except for arm946e, where it was easier to do what might be
the right thing.
Change-Id: Ie73ac07507ff0936fefdb90760046cc8810ed182
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1220
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
We already set cpsr in armv7m_build_reg_cache, so lets use it for all other
accesses to this field.
Change-Id: I19b3b21ecf1571bbea12e1be664845e6544f6fa1
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1539
Tested-by: jenkins
Make sure the target support target requests before we enable any receivers.
Change-Id: I8ce42922eaff76fb5e7a114da716f2a6585a6ab5
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1536
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Let the default handler issue an unsupported warning rather than using
empty handler routines that may/may not issue a unsupported warning.
Change-Id: Iafe3e45146981a4cfae39771c3ab7370ac86da48
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1535
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-by: Hsiangkai Wang <hsiangkai@gmail.com>
If target does not support semi-hosting function, it has
no need to provide .get_gdb_fileio_info callback. OpenOCD
will use default function target_get_gdb_fileio_info_default.
The default function just return ERROR_FAIL and gdb_server
will treat every halted condition as normal halted and
return "Txx" to gdb.
Change-Id: I9ddb2be3a1145eae2ef5b712bdea89eb2e0fbc20
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1586
Tested-by: jenkins
Reviewed-by: Nemui Trinomius <nemuisan_kawausogasuki@live.jp>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
After the target was reset we can be sure it's not running any
algorithm.
This fixes the following failure scenario:
On my STM32F103 board after I start the firmware and then stop and try
to "load" in gdb (before doing mon reset halt), I get
Error: timeout waiting for algorithm, a target reset is recommended
However, target reset doesn't help as the flag is still there ("Error:
Target is already running an algorithm"), so I have no choice but to
restart the OpenOCD process.
I'm not sure yet what exactly prevents load from working after my
firmware is initialised, most probably some interrupt is firing and my
handler produces a fault due to garbled RAM.
Change-Id: Idd977f2780a64d84800e3abd412cffc1ab6801b0
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1512
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Enable reading the SWO trace output via STLinkv2 dongles that support
it.
This adds an optional initialization parameter "trace" with which the user
specifies a destination file where SWO trace output is appended as it comes in
as well as the trace module's source clock rate.
STLink will be configured for a 2MHz SWO data rate (STLink's highest
supported rate) if the source clock is > 2MHz, otherwise the source
clock is used as the data rate directly.
For example:
trace swo.log 168000000
If "trace" is specified with a usable file path, the stlink_usb driver will
attempt to configure and read SWO trace data as follows:
- on _run(), the target's TPI and TMI are configured and the STLinkv2 is told
to enable tracing. Only generic ARM TPI and TMI registers are
configured, any MCU-specific settings (ex: pin routing) are the
responsibility of the target firmware. The configuration applied is
based on the STLinkv2's capabilities (UART emulation).
- on _v2_get_status(), the trace data (if any) is fetched from the
STLink after the target status is checked and the target is found to
be running.
- on _halt(), the STLink is told to disable tracing.
When fetching trace data, the entire trace frame is written to the output file
and that data is flushed. An external tool may be used to parse the
trace data into a more human-readable format.
Tested on ARM Cortex M4F and M3 MCUs (STM32F407 and STM32L152).
Change-Id: Ic3983d46c82ba77010c23b0e18ce7b275d917f12
Signed-off-by: Andrey Yurovsky <yurovsky@gmail.com>
Reviewed-on: http://openocd.zylin.com/1524
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Currently, there is no way to notify gdb that program has exited.
Add new target_debug_reason called DBG_REASON_EXIT to notify gdb
the condition has occured. If the debug reason is DBG_REASON_EXIT,
gdb_server will send 'W' packet to tell gdb the process has exited.
Change-Id: I7a371da292716a3e6ac4cc2c31b009a651fe047a
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1242
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The File I/O remote protocol extension allows the target to use the
host's file system and console I/O to perform various system calls.
To use the function, targets need to prepare two callback functions:
* get_gdb_finish_info: to get file I/O parameters from target
* gdb_fileio_end: pass file I/O response to target
As target is halted, gdb_server will try to get file-I/O information
from target through target_get_gdb_fileio_info(). If the callback function
returns ERROR_OK, gdb_server will initiate a file-I/O request to gdb.
After gdb finishes system call, gdb will pass response of the system call
to target through target_gdb_fileio_end() and continue to run(continue or step).
To implement the function, I add a new data structure in struct target,
called struct gdb_fileio_info, to record file I/O name and parameters.
Details refer to GDB manual "File-I/O Remote Protocol Extension"
Change-Id: I7f4d45e7c9e967b6d898dc79ba01d86bc46315d3
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1102
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
In GDB remote serial protocol, the stop reply packet could contain more
detail stop reason. The currently defined stop reasons are listed below.
* watch
* rwatch
* awatch
* library
* replaylog
This commit adds stop reason, watch/rwatch/awatch, in stop reply packet for
just hit watchpoint. As manual indicates, at most one stop reason should be present.
The function needs target to implement new hook, hit_watchpoint. The hook will fill
the hit watchpoint in second parameter. The information will assist gdb to locate
the watchpoint. If no such information, gdb needs to scan all watchpoints by itself.
Refer to GDB Manual, D.3 Stop Reply Packets
Change-Id: I1f70a1a9cc772e88e641b6171f1a009629a43bd1
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1092
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Documentation says they should be given values in milliseconds,
DEFAULT_HALT_TIMEOUT matches that too.
Change-Id: Ic1a30fa90f75b412c43fe50ba187d01c3d0a5fba
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1504
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
When I do not have the JTAG adapter connected to the target, I often
end up always reading 1s from the chain. If the OpenOCD is configured
to connect to an ETM-equipped target (i.MX25 ARM9 in my case), this
results in writing garbage values in the etm reg_cache as the ETM bit
fields for the comparators, counters and outputs are wider than the
amount of entries in the corresponding arrays. This later results in a
segfault in the first etm_reg_lookup() call.
Change-Id: Ied81fdbf3a53a3dd749e2e5e97adf86c012df575
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1505
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
EJTAG 1.5, 2.0 and 2.5 have different breakpoint register addresses.
This patch add support of EJTAG 2.0, which is part some broadcom
SoCs.
This work was testet on Broadcom BCM7401.
Change-Id: I4b0ee23871fa9205f9001b7c9165e7b6ebe9ccbf
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Reviewed-on: http://openocd.zylin.com/1464
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Split function mips32_configure_break_unit
to mips32_configure_ibs and mips32_configure_dbs
to make code more readable.
This will probably make work easyer with differnet EJTAG versions.
Change-Id: I666f949fd7bc3656bdf75e7bcaadb164f15855dd
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Reviewed-on: http://openocd.zylin.com/1463
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
On some CPUs, like bcm7401 with EJTAG v2.0 we can have situation where
CPU do not stops execution. For example, all CP0 commands will have this issue.
In this case we should some hove recover our session. Currently
jump to 0xff20.0200 seems to be good option. If it brake some thing on
newer EJTAG, then check for EJTAG v2.0 should be added.
Change-Id: Icd8841f38a1a85e0f7682b6dc358af8dfaae0744
Signed-off-by: Oleksij Rempel <bug-track@fisher-privat.net>
Reviewed-on: http://openocd.zylin.com/1353
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This instruction we will need to make jump to 0xff20.0000
Change-Id: Ic723e683e8848492cd8e186e71fd668dbd1d97e6
Signed-off-by: Oleksij Rempel <bug-track@fisher-privat.net>
Reviewed-on: http://openocd.zylin.com/1338
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Some parts of code are version specific. It will be easier
if we extract ejtag version and store it some where.
Change-Id: I8f9addc42108cba5ae9d61b8ade8f9d7f02a0fb5
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Reviewed-on: http://openocd.zylin.com/1462
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
On slow targets we sometimes get false messages about timeouts due to poll
using 1sec rather than the default 5sec timeout.
Change-Id: Icc81c78e0ca86cebf9eeb2f5307cf7a82f1f4ee8
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1466
Tested-by: jenkins
Added missing breakpoint/watchpoint implementation to dsp563xx target.
Implementation is not yet complete, which means it does not leverage all
available features of the once debug interface.
This does NOT use the openocd breakpoint/watchpoint command because of
the "special" memory address spaces (X/Y/P/L) of the 56k DSP series.
Change-Id: I6840a3ff1e6fdebb38ab7758f164886aff773af6
Signed-off-by: Bernhard Kiesbauer <bernhard@kiesbauer.com>
Reviewed-on: http://openocd.zylin.com/1468
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
gcc (Gentoo Hardened 4.6.3 p1.13) produces a warning about the variable
assigned but not used. write() can sometimes write less than the specified
count so it's marked with warn_unused_result in the system headers and its
return value can't be ignored. The most correct solution would be to have a
loop writing the buffer until all bytes are written or an error is
returned but here it's impractical.
Change-Id: I75f7482e2b26fe0e6d70d34947518d3a8f0afe5c
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1490
Reviewed-by: Laszlo Papp <lpapp@kde.org>
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Double const is bogus and breaks the build on Clang 3.3.
Change-Id: I9f9394d17f66289ac74ae6dd48f3165483b72e9e
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1489
Reviewed-by: Laszlo Papp <lpapp@kde.org>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Tested-by: jenkins
On XScale architecture, to write debug control register DCSR
and activate JTAG debug (ie. to choose Halt Mode), the
enabling can only be done while the board is held in reset
state (ie. PXAxx #RST line held low).
The current implementation writes to the register before
asserting the SRST line. Swap the order to activate the SRST
line before writing to DCSR.
Change-Id: I914b9d53d39bdeb5fe4ee5e11068cafafe0da4d2
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Reviewed-on: http://openocd.zylin.com/1458
Tested-by: jenkins
Reviewed-by: Marek Vasut <marek.vasut@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Replace hexadecimal masks for vector catching with words
representing the caught exception, such as dabt for data
abort, etc ...
This way, the new xscale command is :
- xscale vector_catch
Reads back to the user the current vector catching status
- xscale vector_catch reset dabt pabt
Sets the caught vectors to data abort and prefetch abort
for example.
This is mostly taken from Cortex-M3 openocd code.
Change-Id: I66591d5796f0e07f0f31edc8d28722e1e48aa8c5
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Reviewed-on: http://openocd.zylin.com/1456
Tested-by: jenkins
Reviewed-by: Marek Vasut <marek.vasut@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This prevents segmentation fault that can occur on cortex_a targets if
gdb connection is established before it's halted and CPSR examined.
Change-Id: Id996200e0fd95440496509c1fecaabbdbf425e23
Tested-by: Henrik Nordstrom <hno@squid-cache.org>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1446
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
soft_reset_halt was only really intended for the older arm arch, eg. armv4/5.
Newer arch such as armv7m/mips do not need this as they have better alternatives.
Change-Id: Ifb360680b7e4bfa5365f3c79d82574bded952b45
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1442
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
mips_m4k_bulk_write_memory was only called from target_write_buffer as an
optimization when the word count was large enough.
Remove mips_m4k_bulk_write_memory from the target type, causing the default
implementation to call the regular mips_m4k_write_memory instead.
Perform the dispatch to bulk write in mips_m4k_write_memory, enabling the
optimization for target_write_memory() writes with size 4, in addition to
target_write_buffer() writes.
It also enables making the choice of bulk write vs regular write
specifically for the architecture and not relying on the generic target
code to make a sensible decision.
Change-Id: I295f21a67ceaa195802403f2518ea2e0a025c1c7
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1215
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Tested-by: jenkins
If mem_ap_sel_write_atomic_u32 fails then tmp_buff may not be released.
Detected by clang.
Change-Id: I3d5416bd47d0eea61f61ec02ac5e43c82f114108
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1433
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Change scope of the variable gdb_reg_list_idx from global to module.
Change-Id: Ib8273c0769c11c1d5a338e4711efbddb8e8a0243
Signed-off-by: Jim Norris <u17263@att.net>
Reviewed-on: http://openocd.zylin.com/1441
Tested-by: jenkins
Reviewed-by: Øyvind Harboe <oyvindharboe@gmail.com>
Add target code for Andes targets.
Change-Id: Ibf0e1b61b06127ca7d9ed502d98d7e2aeebbbe82
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1259
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Andes AICE uses USB to transfer packets between OpenOCD and AICE.
It uses high-level USB commands to control targets instead of using
JTAG signals. I define an interface as aice_port_api_s. It contains
all basic operations needed by target-dependent code.
Change-Id: I117bc4f938fab2732e44c509ea68b30172d6fdb9
Signed-off-by: Hsiangkai Wang <hsiangkai@gmail.com>
Reviewed-on: http://openocd.zylin.com/1256
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This file is used by all the Cortex-M family not just Cortex-M3.
Change-Id: Ie8680535b220c66bb8fcd862510407a46a73e8a0
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1429
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Native compilation on RaspberryPi with
gcc (Debian 4.6.3-1) 4.6.3
Target: arm-linux-gnueabi
ends with error:
cortex_a.c: In function 'cortex_a8_read_apb_ab_memory':
cortex_a.c:2063:40: error: cast increases required alignment of target type [-Werror=cast-align]
cc1: all warnings being treated as errors
Also check for malloc failure.
This patch is compile-tested only.
Change-Id: I580c505424d03ac3a565de54182db22277c52ac1
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/1369
Tested-by: jenkins
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This should cover all the cases when RESET_SRST_NO_GATING is set
without RESET_HAS_SRST. This might happen when RESET_SRST_NO_GATING is
automatically set by a target code (and not from tcl).
However, there're some other places (mips_m4k, arm7_9_common) where
adding RESET_SRST_PULLS_TRST would lead to trying to use SRST even if
it's not present. Currently it's impossible for the user to enable
that flag without enabling SRST.
Change-Id: Ib1c6f68feed0b8057d55afd5f260bb22ab332ced
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1405
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Commit d9ba56c295 did a bunch of
renaming of cortex_a8 to cortex_a, including the names in config
files. However that introduced a regression as the name in target_type
struct remained unchanged.
This adds the last missing bit: actual renaming of the target name as
understood by OpenOCD.
Also change the (hopefully) last instance of using it in the supplied
config files, namely from imx6.cfg.
Change-Id: Ib9289fc6d946630133ec6e36c20015ccb50acf61
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/1420
Tested-by: jenkins
Reviewed-by: Chris Johns <chrisj@rtems.org>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
There seems to be a few missing semihosting calls. I
am not sure why this one is actually missing, since it
seems simple enough to implement. It was tested using
an HTC HD7 connected to openocd through a "home brew"
ftdi 4232H board.
Change-Id: Ie17dc96c6d48227a3dc9ff1e21201a85498a10b1
Signed-off-by: Brandon Warhurst <roboknight@gmail.com>
Reviewed-on: http://openocd.zylin.com/1345
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Rename cortex_a8 target to use a more correct cortex_a name.
This also adds a deprecated_name var so that older scripts issue a warning
to update the target name.
cfg files have also been updated to the new target name.
Change-Id: I0eb1429c9281321efeb444b27a662a941a2ab67f
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1130
Tested-by: jenkins
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Rename cortex_m3 target to use a more correct cortex_m name.
This also adds a deprecated_name var so that older scripts issue a warning
to update the target name.
cfg files have also been updated to the new target name.
Change-Id: Ia8429f38e88da677249c5caa560c50f8ce56ea10
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1129
Tested-by: jenkins
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Sometimes the target may have breakpoint registers set from a previous
debug session, we can either sync them or as we have chosen here clear them.
Change-Id: I439a623ebbf010246a70e5596d04aa7d546da731
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1363
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Seems commit fc2abe63fd caused a regression
in that the arm reg cmd no longer worked. The issue was caused because we
changed the value of ARM_MODE_THREAD which was being checked in arm_init_arch_info.
Change-Id: Id571d4ab336d1b0e2b93363147af245d24b65ca5
Signed-off-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-on: http://openocd.zylin.com/1362
Tested-by: jenkins
Reviewed-by: Luca Bruno <lucab@debian.org>
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Now all the functions with only fetch accesses are modified.
The same delay between scans has been added to mips32_pracc_fastdata_xfer(), it should work
at the same scan rates as the other pracc functions, but it needs higher scan_delays
to work.
Change-Id: Ifb31d8ea6de9d22674385782913d221a2494dbbf
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1196
Tested-by: jenkins
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Functions mips32_pracc_read_mem(), mips32_cp0_read() and mips32_pracc_read_regs() are now modified.
mips32_cp0_read() is very similar to mips32_read_u32() with one store access.
mips32_pracc_read_regs() is the only function that can not be executed from only one queue.
Now this function is modified to use reg8, it saves all the registers but does not restore reg8.
To remedy this, mips_ejtag_config_step() is called after mips32_save_context() in
mips_m4k_debug_entry(). Function mips_ejtag_config_step() is modified to use reg8 and
restore it from ejtag info instead of using DeSave for save/restore.
Change-Id: Icc224f6d7e41abdec94199483401cb512cc0b450
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1195
Tested-by: jenkins
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Each pracc function defines a variable ctx of type struct pracc_queue_info.
To simplify the code tree auxiliary functions are defined: pracc_queue_init(), pracc_add() and
pracc_queue_free().
The second parameter in pracc_add() is the store address if the instruction is a store at dmseg,
otherwise it should be 0.
The code is executed by mips32_pracc_queue_exec(). If ejtag_info->mode is 0 mips32_pracc_exec()
is called and it should work like with current code.
To generate the delay between scans the number of clock ticks are calculated with the help of
jtag_get_speed_khz(). Due to delays in the execution of each single ftdi instruction the number of ticks
are higher as it should be, specially at higher scan rates.
mips32_pracc_read_u32() should now work with the new code.
Change-Id: I471590a4fc89b56af10bd46c48767b4c64de154f
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1194
Tested-by: jenkins
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
This patch and the following patches define another way of doing processor access without the need to read back
the pracc address as needed in current pracc code.
Current pracc code is executed linearly and unconditionally. The processor starts execution at 0xff200200
and the fetch address is ever incremented by 4, including the last instruction in the delay slot of the branch to start.
Most of the processor accesses are fetch and some are store accesses.
After a previous patch regarding the way of restoring registers (reg8 and reg9), there are no load processor accesses.
The pracc address for a store depends only on the store instruction given before.
m4k core has a 5 stage pipeline and the memory access is done in the 3rth stage. This means that the store access
will not arrive immediately after a store instruction, it appears after another instruction enters the pipeline.
For reference: MD00249 mips32 m4k manual.
A new struct pracc_queue_info is defined to help each function in generating the code. The field pracc_list holds in the
lower half the list of instructions and in the upper half the store addressess, if any. In this way the list can be used by
current code or by the new one to generate the sequence of pracc accesses.
For every pracc access only one scan to register "all" is used by calling the new function mips_ejtag_add_scan_96().
This function does not call jtag_execute_queue(), all the scans needed can be queued before calling for execution.
The pracc bit is not checked before execution, is checked after the queue has been executed.
Without calling the wait function the code works much faster, but the scan frequency must be limited. For pic32mx
with core clock at 4Mhz works up to 600Khz and with 8Mhz up to 1200. To increase the scan frequency a delay
between scans is added by calling jtag_add_cloks().
A time delay in nano seconds is stored in scan_delay, a new field in ejtag_info, and a handler is provided for it.
A mode field is added to ejtag_info to hold the working mode. If a time delay of 2ms (2000000 ns) or higher is set,
current code is executed, if lower, new code is executed.
Initial default values are set in function mips32_init_arch_info. A reset does not change this settings.
Change-Id: I266bdb386b24744435b6e29d8489a68c0c15ff65
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1193
Tested-by: jenkins
Reviewed-by: Freddie Chopin <freddie.chopin@gmail.com>
Add command to fix CSW_SPROT in register AP_CSW.
This solves dap apmem access in non secure access.
Change-Id: I7cfcb6434d75f5cfd4a2630a059901cdeea010ce
Signed-off-by: Michel JAOUEN <michel.jaouen@stericsson.com>
Reviewed-on: http://openocd.zylin.com/1276
Tested-by: jenkins
Reviewed-by: mike brown
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
After calling mips32_cp0_read() nothing has been queued, the call to jtag_exec_queue() is unnecessary.
Change-Id: Ie25438045a8e9b6b1b170df7b52609d45f284b5a
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1190
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
In current devel code there are 3 functions (related to m4k code) that need to restore register 8 from pracc stack:
mips32_pracc_read_u32()
mips32_cp0_read()
mips32_pracc_write_mem_generic()
And mips32_pracc_read_mem() needs to restore regs 8 and 9 from pracc stack.
Values in this registers should be the same as read by mips32_pracc_read_regs() when entering debug
mode and can be modified by mips32_pracc_write_regs() when leaving debug mode.
There is no need to read their values from the processor registers every time.
The fields reg8 and reg9 are added to struct mips_ejtag to store these register values
and the call to mips32_save_context() is shifted in mips_m4k_debug_entry() in order
to store them before any other function needs to restore these registers.
For the same reason in function mips_m4k_step() the call to mips_m4k_set_breakpoint(), if needed,
should be made after calling mips_m4k_debug_entry().
For single word write the number of pracc accesses are now 9 or 8, from 13 or 12 in current code,
single word read takes now 10 instead of 12.
This patch is really the first in a set of patches for an alternate m4k pracc code
much faster that current code. At least for me with pic32mx works fine.
Change-Id: Ibd9df5e8b9f78ce05a180949ba6a561c761b61d6
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Reviewed-on: http://openocd.zylin.com/1146
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
In this function after loading the handler code and the jump code there is a call
to wait_for_pracc_rw() to verify that a pracc access is pending.
Next the address is read to verify that the handler is running, the address should be at
fastdata area.
Next, another call is made to wait_for_pracc_rw(). This call is not needed, we now already
that a pracc access is pending.
Better we call this function before loading the end address to be sure it is loaded correctly.
Signed-off-by: Salvador Arroyo <sarroyofdez@yahoo.es>
Change-Id: If311450ea634786fc28cf1a8e18ed24ce5257d20
Reviewed-on: http://openocd.zylin.com/1142
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Salvador Arroyo
Spencer Oliver
Paul Fertser
Ivan De Cesaris
Seth LaForge
Tim Sander
Andreas Fritiofson
Andrey Smirnov
Oleksij Rempel
Kamal Dasu
Brad Riensche
Chris Johns
Christian Eggers
Antony Pavlov
Adrian Burns
Luca BRUNO
Hsiangkai Wang
Franck Jullien
Per Ekman
Sergio Chico
Dongxue Zhang
Mathias K
Sergey A. Borshch
Andrey Yurovsky
Oleksij Rempel
Bernhard Kiesbauer
Robert Jarzmik
Brent Roman
Jim Norris
Brandon Warhurst
Michel JAOUEN