Previously the code made the assumption (which is valid for conventional
RTOSs) that special registers (e.g. CSRs) are the same across threads.
26/45 tests pass.
Change-Id: Ibb3398790d7354a995d506772375d869f608f1f0
This patch covers the fact that cortex_m could be resumed externally by
Cross Trigger Interface or by direct write to DHSCR ...
To reproduce:
- halt the target
- then run the core through DHCSR (mww 0xe000edf0 0xa05f0001)
=> this resumes the core, but target state in OpenOCD remains HALTED.
Change-Id: Ifa1ae18645bfeb863acc78a039bbf04873fd78fe
Signed-off-by: Tarek BOCHKATI <tarek.bouchkati@gmail.com>
Reviewed-on: http://openocd.zylin.com/4817
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
cortex_a_debug_entry and update_halt_gdb are called in two consecutive
conditions which are complementary, so externalizing the common code
makes the conditions' body lighter
With the removal of LOG_DEBUG(" ") since it does not look too informative
Change-Id: I0c54e413619576bb3af164f2dcf256c5a862c5fd
Signed-off-by: Tarek BOCHKATI <tarek.bouchkati@gmail.com>
Reviewed-on: http://openocd.zylin.com/4832
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Tested-by: jenkins
If malloc fails in __command_name, the following strcpy will
segfault, thus preventing __command_name to return.
The actual calls to command_name() implement the correct check
for the NULL pointer, but propagate error -ENOMEM, that is not
an error value coherent within OpenOCD. Plus, in one case it
overwrites an already detected error.
Check the pointer returned by malloc and, in case of failure,
issue an error message and return the NULL pointer.
Let the caller of command_name() to keep the already detected
error or to return ERROR_FAIL in case of end of memory.
Change-Id: I151a24569409777dd5bc09a3daf5dba2b8e2829b
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4838
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
The notation "(tck tms tdi tdo)* " is incorrect, because it means the
quadruple of gpio can be repeated on the command-line.
The correct syntax of the command requires instead to provide either
all the four gpio numbers (in order to set the values) or to pass an
empty command-line (to dump the values previously set).
Change the .usage field to "[tck tms tdi tdo]".
Change similarly the corresponding .usage field for SWD command.
Add the .usage field for the commands that individually set each gpio.
Change-Id: If5b3c618097b71dfe7fcf988fb3c1499ae03a6d5
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4833
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
I don't understand how it was ever possible to change the registers on a
thread that's not the current active one when a halt happened. Really
instead of the RTOS tracking what the currently selected thread is, it
would make more sense to have gdb_server do that and simply pass it
along in every call to the RTOS layer.
Now MulticoreRegTest passes.
Change-Id: I399b9b2b05a147aa6b41463714ed3a39534b1fc8
Valgrind and Clang Static Analyzer have no complaints about this change.
Change-Id: I7757615ec52448372bdc57729cdf97c7016d97e8
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: http://openocd.zylin.com/4656
Tested-by: jenkins
Reviewed-by: Philipp Guehring <pg@futureware.at>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Commands to stlink devices are typically comprised of multiple
transactions with each transaction completing before moving to the next.
This change allows for multiple USB transactions to be issued at once
followed by a check that all transactions completed successfully. This
improves performance on some machines where there is a large turn-around
time between USB transfers such as is seen on some virtual machines.
This change is only supported when compiled with libusb1 as libusb1
supports and asynchronous interface.
Multi-transaction queueing introduced in this change paves the way for
improving speed of other transactions in the future such as memory and
register reads where multiple USB transactions in succession are
required to complete a command. Multiple USB transactions can be
submitted at once using jtag_libusb_bulk_transfer_n function.
Change-Id: I924e049217a789ef445b14e00aa1983576970fbf
Signed-off-by: Austin Phillips <austin_phillips@hotmail.com>
Reviewed-on: http://openocd.zylin.com/4484
Tested-by: jenkins
Reviewed-by: Andreas Bolsch <hyphen0break@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
When few adapters of the same type are in use, the serial string is
the way to select the right one.
Currently a serial string that does not match any of the connected
adapters will just fail the open, without specific information to
track the issue.
Add a specific message to highlight that the open failure is caused
by a serial mismatch.
Change-Id: I5cb77f1045cc746e532d395b2e5ced40a23ab638
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4701
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Where possible, keep the same style for returning error.
Change-Id: I3a04220c0b9f129a36e9fe83038b7c19dd57fe61
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4699
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Commit d1b7437633 fixes libusb1 to
return OpenOCD error values instead of negative errors in Linux
kernel's style.
The same fix should be applied to libusb0 too.
Fix return value of libusb0 to uniform it to OpenOCD style.
Change-Id: I68478c29c91c6be720074f58c432fe51477e03ed
Fixes: d1b7437633 ("libusb: return oocd error values")
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4698
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
GDB can be built for multi-architecture through the command
./configure --enable-targets=all && make
Such multi-architecture GDB requires the target's architecture to
be selected either manually by the user through the GDB command
"set architecture" or automatically by the target description sent
by the remote target (i.e. OpenOCD).
Commit e65acd889c ("gdb_server: add
support for architecture element") already provides the required
infrastructure to support multi-architecture gdb.
aarch64-linux-gnu-gdb 8.2 uses "aarch64" as default architecture,
but also supports the value "aarch64:ilp32" and all the values
supported by arm-none-eabi-gdb.
These values can be displayed on arm gdb prompt by typing
"set architecture " followed by a TAB for autocompletion.
Set the gdb architecture value for aarch64 target to "aarch64".
Change-Id: I63e9769f47d8e73f048eb84fa73e082dd1c8e52c
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4755
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
GDB can be built for multi-architecture through the command
./configure --enable-targets=all && make
Such multi-architecture GDB requires the target's architecture to
be selected either manually by the user through the GDB command
"set architecture" or automatically by the target description sent
by the remote target (i.e. OpenOCD).
Commit e65acd889c ("gdb_server: add
support for architecture element") already provides the required
infrastructure to support multi-architecture gdb.
arm-none-eabi-gdb 8.2 uses "arm" as default architecture, but also
supports the following values: "arm_any", "armv2", "armv2a",
"armv3", "armv3m", "armv4", "armv4t", "armv5", "armv5t", "armv5te",
"armv5tej", "armv6", "armv6k", "armv6kz", "armv6-m", "armv6s-m",
"armv6t2", "armv7", "armv7e-m", "armv8-a", "armv8-m.base",
"armv8-m.main", "armv8-r", "ep9312", "iwmmxt", "iwmmxt2", "xscale".
These values can be displayed on arm gdb prompt by typing
"set architecture " followed by a TAB for autocompletion.
Set the gdb architecture value for all arm targets to "arm".
Change-Id: I176cb89878606e1febd546ce26543b3e7849500a
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4754
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
GDB can be built for multi-architecture through the command
./configure --enable-targets=all && make
Such multi-architecture GDB requires the target's architecture to
be selected either manually by the user through the GDB command
"set architecture" or automatically by the target description sent
by the remote target (i.e. OpenOCD).
Commit e65acd889c ("gdb_server: add
support for architecture element") already provides the required
infrastructure to support multi-architecture gdb.
The gdb patches for stm8 are still not merged in the official
repository and are temporarily hosted in
https://stm8-binutils-gdb.sourceforge.io/
The latest patch set
stm8-binutils-gdb-sources-2018-03-04.tar.gz
define only one possible value ("stm8") for this architecture; it
can be displayed typing "set architecture " followed by a TAB for
autocompletion in gdb for stm8.
Set the gdb architecture value for stm8 to "stm8".
Change-Id: I643ceba662de46cecf061d1dc672b9178a077f1b
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4753
Tested-by: jenkins
Reviewed-by: Ake Rehnman <ake.rehnman@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
xPSR.T sets the processor to Thumb mode when set to 1. ARMv7-M only
supports execution of Thumb instructions, so it must always be set to 1.
If xPSR.T is set to 0 on armv7m, a usage fault is generated when a
instruction execution is attempted.
On armv7m, issuing a reset causes the vector table to be examined. PC
and xPSR.T are loaded from the vector table at byte offset 4. xPSR.T is
taken from the least significant bit this value, PC from the remaining
bits. This occurs even with `reset halt`, as the reset itself causes
this load to occur without the execution of any instructions.
As a result of this, following a reset with a "bad" value programmed in
the vector table, openocd would be unable to run algorithms on the
target, as running them would immediately result in a usage fault due to
xPSR.T being unset (0).
Allow algorithms to run regardless of the content of the vector table by
explicitly setting xPSR so that xPSR.T=1 prior to executing an
algorithm. One can think of this as openocd more closely emulating a
reset or branch instruction in executing it's algorithms.
Ticket: https://sourceforge.net/p/openocd/tickets/203/
Signed-off-by: Cody P Schafer <openocd@codyps.com>
Change-Id: I4dc3427ab195d06c3fd780ea768027fefccc4c28
Reviewed-on: http://openocd.zylin.com/4658
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The Agama family of devices (CC26x2/CC13x2) required an
additional bit to be set when adding the core's TAP into
the scan chain. The cancel reset bit 0x10000 tells the
ICEPick to take the bus out of reset so that the other
bits will take effect. This bit is a NOP on other devices
and ICEPicks, so the change shouldn't adversely affect
other devices.
Change-Id: I9245eef0936ea7eea28ae84ab5e8ce05fa63af40
Signed-off-by: Edward Fewell <efewell@ti.com>
Reviewed-on: http://openocd.zylin.com/4789
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The XDS110 stand-alone version has the ability to supply
voltage to the target board via it's AUX FUNCTIONS port.
Added command to enable setting the voltage on the XDS110
stand-alone.
Change-Id: I2f21c4a3d15ed99e649f3a83973c5e724c4bfeb6
Signed-off-by: Edward Fewell <efewell@ti.com>
Reviewed-on: http://openocd.zylin.com/4793
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The fact that one needs to always push contributions to a single
remote reference (refs/for/master) might seem odd to people unfamiliar
with Gerrit. GitHub, for instance, hosts personal repositories where
developers typically create topic branches for each contribution and
use a proprietary mecanism to request a review (the "pull request").
More generally, one normally does not expect to be able to push
non-fast-forwarding stuff to a remote branch.
This commit adds a clarifying note to the patch guidelines.
Change-Id: Ia750b815b82b18e92b6109c07f451000dcbecf9b
Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-on: http://openocd.zylin.com/4806
Tested-by: jenkins
Reviewed-by: Christopher Head <chead@zaber.com>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
refs/publish/master is deprecated and gives a warning in newer Gerrit.
Replace with refs/for/master.
Change-Id: I56871cc6e80c014ba81f4458230cd67dc318ecb3
Suggested-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-on: http://openocd.zylin.com/4810
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
That function might change from NULL reg_list but then return failure.
In that case reg_list shouldn't be freed.
Change-Id: I5380630c871d056fb52e25bda16836e346bd74b2
These devices differ from LPC8xx devices in that they have a different
IAP entry point, but everything else is the same. Using Tcl to pass
different IAP entry point.
no new Clang analyser warnings and no new build sanitizers issues.
Change-Id: I2d654dd250f416e74262c0228cad8713a283402f
Signed-off-by: Rod Boyce <developer@teamboyce.co.uk>
Reviewed-on: http://openocd.zylin.com/4684
Reviewed-by: Jean-Christian de Rivaz <jcamdr70@gmail.com>
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Currently it is impossible to flash ELF with correct offsets. The reason
is a bogus offset calculation extracted from base.
Since any other spi drivers do not care about base, do the same for
ath79 as well.
Change-Id: I9e46e01c9e7a709c2d07da9203c634f302603afd
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Reviewed-on: http://openocd.zylin.com/4821
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Longest erase all FLASH for nRF5 series is 295.3 ms for nRF52832.
Timeout period now is set to 340 ms (295.3 + 15%)
Change-Id: Iae00ed7b634f111b9798db11e35e4e066d4aaa95
Signed-off-by: Jānis Skujenieks <janis.skujenieks@gmail.com>
Reviewed-on: http://openocd.zylin.com/4822
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
This patch adds "hwthread", a pseudo rtos that represents cpu cores
in an SMP system as threads to gdb. This allows to debug SMP
system kernels in a more sensible manner and removes the current
atrocities of switching gdb manually between CPU cores to update
the context.
Change-Id: Ib781c6c34097689d21d9e02011e4d74a4a742379
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
Source: http://openocd.zylin.com/#/c/3999
this permits the full control of CTI from config files
Change-Id: Ia27ac8e12e08ec72da05f26dcbd81d24fa1a0f6f
Signed-off-by: Tarek BOCHKATI <tarek.bouchkati@gmail.com>
Reviewed-on: http://openocd.zylin.com/4815
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
w600 is a wifi soc from winner micro(www.winnermicro.com).
Change-Id: Ib8ccd6e52baefca6547fb97d29db75db0ee73948
Signed-off-by: Simon Qian <versaloon@simonqian.com>
Reviewed-on: http://openocd.zylin.com/4801
Tested-by: jenkins
Reviewed-by: yichen <wdyichen@wdyichen.cn>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
target configure -work-area-xxx calls target_free_all_working_areas()
and sets the desired new parameter. Without this change the working area
does not get reallocated if it has been allocated before.
target_free_all_working_areas() results in work area containing one block
marked as free.
Completely free working area in target_free_all_working_areas()
Change-Id: I79c681082f32f2a96a2b40eb3b8751e427549693
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4797
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
The block of code moved without any changes
Change-Id: I70b82dc3315dcc3f34de0537b362bee230007d02
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4796
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Remove option 'srst' which is not recognized from on-line help and texi.
Check parameter and return syntax error if wrong option is entered.
Change-Id: I87daa423a9f53193a0b015080594820b933628f5
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4795
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Cortex-M0, M0+ and M1 do not support VECTRESET bit in AIRCR.
Without this change the 'reset' command silently fails if VECTRESET
is requested.
Detect these cores, show warning if VECTRESET is about to use
and use SYSRESETREQ instead.
Change-Id: Ief174373e3ef0e6b287c57911c0aca4dfa8209f2
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4794
Tested-by: jenkins
Reviewed-by: Jean-Christian de Rivaz <jcamdr70@gmail.com>
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
This commit adds TARGET_HALTED check in armv8_get_core_reg32()
and armv8_set_core_reg32() to void a crash issue when gdb connects
but fails to halt the ARM core. Similar logic can be found in
armv8_get_core_reg() and armv8_set_core_reg().
Below is the call stack information of this case when gdb connects.
(gdb) bt
regnum=regnum@entry=0, dpm=0x990110) at src/target/armv8_dpm.c:657
r=0x9c7240, regnum=0, mode=<optimized out>) at src/target/armv8_dpm.c:974
at src/target/armv8.c:1487
packet=0x8ec8e0 <gdb_packet_buffer.9962> "g",
packet_size=<optimized out>, connection=<optimized out>)
at src/server/gdb_server.c:1200
at src/server/gdb_server.c:3180
command_context=command_context@entry=0x935010)
at src/server/server.c:566
...
Change-Id: I159837b533f110998184f910a0abe48409bd58f1
Signed-off-by: Liming Sun <lsun@mellanox.com>
Reviewed-on: http://openocd.zylin.com/4758
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Arm architecture reference manual DDI0406C reports at page 2024 in
table C3-1 the processor behaviour on debug events depending on
the debug-mode (none, monitor or halt), mode selected through the
bits MDBGen and HDBGen in DSCR register.
The halt request is served independently from the debug-mode. Thus
it's useless to enable the halt debug-mode in cortex_a_halt() by
setting the bit HDBGen (macro DSCR_HALT_DBG_MODE).
On the other side, halting for a breakpoint, a watchpoint or a
vector catch requires being in halt debug-mode.
Today HDBGen is set only in cortex_a_halt(), so we are forced to
halt the core at least once before it can be halted for hitting a
breakpoint/watchpoint/vector-catch. This is annoying since there
is no need to halt the target to set a HW breakpoint.
Move in cortex_a_init_debug_access() the selection of the halt
debug-mode, so the mode is set during examine.
To prevent a misconfigured hardware breakpoint/watchpoint/vector
catch to halt the target when OpenOCD has already quit, return to
debug-mode none at OpenOCD exit.
Change-Id: I68a1c51de3572ca1b89e90caf7eb20374268e926
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4783
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Tim Newsome
Tarek BOCHKATI
Severin Junker
Antonio Borneo
Tomas Vanek
Darius Rad
Austin Phillips
Cody P Schafer
Edward Fewell
Jerome Forissier
Rod Boyce
Oleksij Rempel
Jānis Skujenieks
Matthias Welwarsky
Tarek BOUCHKATI
Jean-Christian de Rivaz
Simon Qian
Liming Sun