The memory allocated to hold the watchpoints is not freed at
OpenOCD exit.
Free the watchpoint memory at OpenOCD exit.
Change-Id: I518c9ce0dc901cde2913d752e3154734f878b854
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/6210
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Use byte address select for 1 and 2 bytes length.
Use normal mode for 4 bytes length.
Change-Id: I28d182f25145d0635de64d0361d456f1ad96640e
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/6197
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Decrement the available watchpoints only when succeed setting it.
Initialize the available watchpoint with the correct value.
Change-Id: I0f93b347300b8ebedbcd9e718d4ba32b26cf6846
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/6196
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
The current implementation of OpenOCD does not support watchpoints for
cortex_a architecture. So, I replaced the add_watchpoint and
remove_watchpoint with the specific implementation for the
cortex a and using the breakpoint implementation and the arm
documentation [1] as reference. In particular, I have made the
following changes:
* added the following functions
- cortex_a_add_watchpoint
This wrapper function check whether there are any watchpoint
available on the target device by calling cortex_a_set_watchpoint.
- cortex_a_set_watchpoint
This function is responsible for settings the watchpoint register
pair. In particular, it sets the WVR and the WCR registers with
the cortex_a_dap command.
- cortex_a_remove_watchpoint
This wrapper function the selected watchpoint on the target device
by calling cortex_a_unset_watchpoint.
- cortex_a_unset_watchpoint
This function sets both the WVR and the WCR registers to zero, thus
unsetting the watchpoint.
[1]
http://infocenter.arm.com/help/topic/com.arm.doc.ddi0464f/BCGDHIEJ.html
Change-Id: I86611dab474cb84836662af572b17636dc68e282
Signed-off-by: Chengyu Zheng <chengyu.zheng@polimi.it>
Reviewed-on: http://openocd.zylin.com/3913
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Reviewed-by: Marc Schink <dev@zapb.de>
Tested-by: jenkins
Some target name were marked as deprecated in release v0.7.0 and
v0.8.0, almost 7 years ago, and replaced with more 'actual' names.
We can reasonably expect that in these 7 years any user of OpenOCD
has already migrated to v0.8.0 or to some newer release, thus has
already updated any local/personal script to get rid of the
deprecated message.
Drop the target names already deprecated in v0.8.0.
Change-Id: I7c7491496db1b302b4eb1e9fc6090b58d4acf05a
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/6086
Reviewed-by: Tarek BOCHKATI <tarek.bouchkati@gmail.com>
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
The target code for assert reset on cortex_a has been patched on
commit b0698501b0 ("cortex_a: fix cortex_a_assert_reset() if
srst_gates_jtag") then in cdba6ba0ad ("cortex_a: fix reset for
SWD transport") to workaround the mismatch between jtag and swd
implementations. See discussion for the second patch at
http://openocd.zylin.com/3641/
While all of these mismatches should hopefully be cleaned by the
reset framework rework, an extension of the workaround of the
second patch is required for dapdirect transports, either
dapdirect_swd and dapdirect_jtag.
Extend the existing workaround to all non-jtag transports.
Change-Id: Ia6a9d43bab524cbb3de4c37ce24c45f25187353d
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5979
Tested-by: jenkins
Functions and variables that are not used outside the file should
be declared as static.
Change-Id: I9f97571a528f0cb3c3c26f873577ab16fdec3cdc
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5895
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Modify the format strings to properly handle uint32_t data types.
While there, fix prototype mismatch between header and C file of
the function armv7a_l1_d_cache_inval_virt().
Change-Id: I434bd241fa5c38e0c15d22cda2295097050067f5
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5818
Tested-by: jenkins
There is no method to free the register cache, allocated in
armv4_5, so we get a memory leak.
Issue identified by valgrind.
Implement the method arm_free_reg_cache() and call it in cortex_a
deinit and to exit for error during arm_dpm_setup().
Tested on dual cortex-A stm32mp15x.
This change is inspired from similar fix in commit b01b5fe13a
("armv7m: Fix memory leak in register caching.").
The same allocation is also used by target types "arm7tdmi",
"arm9tdmi", "arm11" and "xscale" but they all lack the deinit
method and I do not have relevant HW to test the fix. For such
reasons they are not addressed in this patch.
Change-Id: I4da1e1f12e36ec245d1f3b11a4eafcbd9a1d2e25
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5693
Tested-by: jenkins
The tcl script src/target/startup.tcl has already the proper
centralized support to wait for all targets to halt after the
command "reset halt". The extra wait in cortex_a_deassert_reset()
is not required.
This extra wait is also an issue for multi-core support, because
waiting for one core to halt can delay the halt request to the
other cores.
Replace the indirect call to cortex_a_halt(), that embeds the wait
for halt, with a low-level halt sequence.
The on-going work on the reset framework is compatible with this
change; in fact it keeps in startup.tcl the wait for targets to
halt, even if current code proposal for cortex_a simply removes
the function cortex_a_deassert_reset().
Change-Id: Ic661c3791a29ba7d520e31f85a61f939a646feb5
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5472
Tested-by: jenkins
Reviewed-by: Tarek BOCHKATI <tarek.bouchkati@gmail.com>
Hypervisor mode is present only if the optional virtualization
extensions are available. Moreover, virtualization extensions
require that also security extensions are implemented.
Add the required infrastructure for the shadowed registers in
hypervisor mode.
Make monitor shadowed registers visible in hypervisor mode too.
Make hypervisor shadowed registers visible in hypervisor mode
only.
Check during cortex_a examine if virtualization extensions are
present and then conditionally enable the visibility of both
hypervisor and monitor modes shadowed registers.
Change-Id: I81dbb1ee8baf4c9f1a2226b77c10c8a2a7b34871
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5261
Tested-by: jenkins
Accordingly to ARM DDI 0406C at B1.5, the security extensions for
armv7a are optional extensions and can be detected by reading
ID_PFR1.
The monitor mode is part of the security extensions and the shadow
registers "sp_mon", "lr_mon" and "spsr_mon" are only present with
the security extensions.
Read the register ID_PFR1 during cortex_a examine, determine if
security extension is present and then conditionally enable the
visibility of the monitor mode shadow registers.
Change-Id: Ib4834698659046566f6dc5cd35b44de122dc02e5
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5259
Tested-by: jenkins
The fields core_type and core_mode use the same enum arm_mode
but encode different information, making the code less immediate
to read.
Use a different enum arm_core_type for the field core_type.
The code behavior is not changed.
Change-Id: I60f2095ea6801dfe22f6da81ec295ca71ef90466
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5258
Tested-by: jenkins
A common problem with target configurations appears to be broken
debug base address configuration. ARM DDI0406C.d specifies in App. D,
1.4.1, that bit 31 of the debug base address serves as identification
of an external debugger, as opposed to an internal access to memory
mapped debug registers by the CPU. External accesses are treated
as privileged and require no debug authentification via the lock
access register.
Sometimes the base address of a debug component is wrong even
in the targets' ROM table. In this case, the correct base address
must be specified using the -dbgbase argument when creating the
target.
This patch adds a warning when bit 31 of the debug base address
is not set, as a hint to the user.
Change-Id: I9c41d85a138123c657ef655e3436a2aa39249dcc
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
Reviewed-on: http://openocd.zylin.com/5105
Tested-by: jenkins
Reviewed-by: Tommy Vestermark <tov@vestermark.dk>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
In preparation for supporting the ARM MCRR and MRRC commands which will
require using two 32-bit registers to read/write a 64-bit internal
register, extract the common logic to read/write from/to a register
to/from DCC and make that parameterized such that we can do this through
not just r0.
Change-Id: Iadb73f5cde8cf5961b5a18ddd198bf39d791e610
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-on: http://openocd.zylin.com/5227
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Armv7a is able to read and write memory at un-aligned address, but
only when bit SCTLR.A (Alignment check enable) is zero and the
address belongs to a memory space with attribute "Normal" (see [1]
chapter A3.2.1 "Unaligned data access"). In all the other cases
the memory access will trigger an alignment fault data abort
exception.
Memory attributes are explained in [1] chapter A3.5 "Memory types
and attributes and the memory order model".
Disabling the MMU cause a change in memory attribute, as explained
in [1] chapter B3.2 "The effects of disabling MMUs on VMSA
behavior".
This can cause several issues. e.g. a SW breakpoint on un-aligned
4-byte Thumb instruction, set when MMU is on, can be impossible to
remove when MMU turns off.
While is possible to check all the possible conditions before an
un-aligned memory access, it's clearly more maintainable to skip
such complexity and only perform aligned accesses.
Check the alignment and eventually modify the data size before
calling the functions cortex_a_{read,write}_cpu_memory_slow().
Change the comment in the two functions above to comply with the
new behaviour.
[1] ARM DDI 0406C.d - "ARM Architecture Reference Manual, ARMv7-A
and ARMv7-R edition"
Change-Id: I57b4c11e7fa7e78aaaaee4406a5734b48db740ae
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5138
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Replace jtag specific API jtag_add_reset() with transport
independent API adapter_{de}assert_reset().
Change-Id: I1b917a4c1205115c4e0315373d81a9305e931258
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4944
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
In preparation for adding super section decoding, do not restrict
armv7a_mmu_translate_va_pa() to 32-bit virtual addresses since ARMv7-A
processors with VMSA extensions (including LPAE) can issue wider
physical addresses. Update casting to uint32_t where necessary.
Change-Id: Id1c3d0d5ac324cbdc334259d9ea75fe4981671a1
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-on: http://openocd.zylin.com/5211
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
To prepare for handling TCL return values consistently, all calls
to command_print/command_print_sameline should switch to CMD as
first parameter.
Change prototype of command_print() and command_print_sameline()
to pass CMD instead of CMD_CTX.
Since the first parameter is currently not used, the change can be
done though scripts without manual coding.
This patch is created using the command:
sed -i PATTERN $(find src/ doc/ -type f)
with all the following patters:
's/\(command_print(cmd\)->ctx,/\1,/'
's/\(command_print(CMD\)_CTX,/\1,/'
's/\(command_print(struct command_\)context \*context,/\1invocation *cmd,/'
's/\(command_print_sameline(cmd\)->ctx,/\1,/'
's/\(command_print_sameline(CMD\)_CTX,/\1,/'
's/\(command_print_sameline(struct command_\)context \*context,/\1invocation *cmd,/'
This change is inspired by http://openocd.zylin.com/1815 from Paul
Fertser but is now done through scripting.
Change-Id: I3386d8f96cdc477e7a2308dd18269de3bed04385
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/5081
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
To prepare for handling TCL return values consistently, all calls
to command_print/command_print_sameline should be ready to switch
to CMD as first parameter.
Change prototype of armv7a_handle_cache_info_command() to pass CMD
instead of CMD_CTX.
This change was part of http://openocd.zylin.com/1815 from Paul
Fertser and has been extracted and rebased to simplify the review.
Change-Id: I27f507869cb41c74722a759e69dfd45658fcd4be
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5069
Tested-by: jenkins
We have the function to wait for bits in dscr. Use it whenever
possible.
When the bit to test is DSCR_INSTR_COMP, use instead the wrapper
function cortex_a_wait_instrcmpl().
Change-Id: I5c54c239a00f489712af448eb97752210b4b38b8
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5113
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
In function cortex_a_wait_dscr_bits() the last read on dscr gets
ignored in case of timeout, even if it finally provides the value
that would trigger a successful return.
Check the returned value before testing the timeout.
Also, print a message on failure reading dscr.
Change-Id: I261ac1545113db39374833a55be911a4da71d893
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5112
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Depending on adapter speed, the function cortex_a_poll() can take
time to execute and can complete successfully when the timeout is
already expired. Checking the timeout at function return causes a
timeout error while that could be avoided.
Check the target status at cortex_a_poll() return, before checking
for the timeout.
Change-Id: I4c1581f6e718298c566df7b1359255e16e3955d5
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/5111
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
The code for this command is currently replicated in cortex_a and
in mips_m4k and is going to be added again for cortex_m.
Plus, it is going to be obsoleted by the HW thread pseudo RTOS.
Consolidate in smp.c a single instance of the command. This will
simplify both obsoleting it and adding it to cortex_m, whatever
change comes first.
Change-Id: I03cd857e21fa3f7202fdcee36bcbd5aae30a609d
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4991
Tested-by: jenkins
Reviewed-by: Tarek BOCHKATI <tarek.bouchkati@gmail.com>
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Seams over-engineered having two separate commands to turn SMP
on/off. Plus it is missing the possibility to dump the current
status of SMP and would be weird adding an additional command
for it. Moreover, such commands are replicated in few targets so
it would make sense centralizing them.
- Deprecate the commands "smp_on" and "smp_off".
- Add a new command "smp" that accepts optional parameters
"[on|off]" and prints the SMP status when run without
parameters. This replaces the two commands above.
- Put the deprecated and the new command handlers in smp.c
- Update the documentation, except for mips_m4k, since it is not
available yet.
- Promote the macro foreach_smp_target to global context and use
it where possible.
Change-Id: Ia72841c1a3bd6edd4db4cc809046322f498617e6
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4615
Tested-by: jenkins
Reviewed-by: Graham Sanderson <graham.sanderson@gmail.com>
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
If the function cortex_a_target_create() or the function
cortex_r4_target_create() exit for an error, the value of pointer
cortex_a get lost, causing a memory leak.
Move the allocation of cortex_a after the check on the parameters.
While there, add checks on the value returned by calloc().
Issue highlighted by clang 7.0.0.
Change-Id: Ic6eeb9c3b39922d016f6d0319eada1a97a6681f0
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4925
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
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
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>
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>
The functions cortex_a_dap_{read,write}_coreregister_u32() are
duplicate of the functions dpm_{read,write}_reg().
Remove both duplicated functions in cortex_a.c while export only
dpm_read_reg(), since dpm_write_reg() is currently not used.
Rename dpm_read_reg() as arm_dpm_read_reg() to keep uniform the
naming style.
Change-Id: I501bc99dc402039e630c47917a086a0bb382782c
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4747
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Tested-by: jenkins
The name of the function dpm_modeswitch() does not follow the
common style of the other function names in the same file.
Rename it as arm_dpm_modeswitch().
Change-Id: Idebf3c7bbddcd9b3c7b44f8d0dea1e5f7549b0eb
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4756
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
The armv7m debug port provides a direct access to the CPU memory
bus, allowing the debugger to bypass the CPU for every memory
operation.
The armv7a debug port doesn't offer the same feature, mainly
because CPU caches and MMU makes the direct memory access more
tricky. Nevertheless most SoC with armv7a provide direct memory
access through an AHB bus available on another DAP access port,
different from the debug port.
The original port of cortex_a in OpenOCD was inspired from the
working cortex_m code, and provided optional memory access
through the AHB, if present.
The code for AHB access is problematic and partially buggy due
to incomplete management of cache coherency and missing check of
page boundary during virtual address operations.
With the commit 5d458cf727
("target/mem_ap: generic mem-ap target") we have a clean support
for memory access through system buses connected to DAP AP, which
obsoletes the buggy memory AP hack in cortex_a.
Remove any code that uses the memory AP accesses in cortex_a.
Change-Id: I7cd1f94885e5817448058953e043d8da90dea3cc
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4748
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
The variable fast_reg_read is always zero, causing some code to
never be executed.
Such code try to read the target registers by dumping them in
memory and then reading back the memory through the debugger. But
it is broken due to lack of cache and MMU management.
This code also uses the broken memory_ap access that is going to
be removed soon.
Remove all the code that depends on fast_reg_read not zero.
Add a missing check on arm_dpm_read_current_registers() return.
Keep the unused function cortex_a_dap_write_coreregister_u32()
to balance the used "read" version.
Change-Id: If2ff28a8c49eb0a87dc85207f5431978efd158db
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4746
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Commit c8926d1457 introduces the
context and hybrid breakpoint types beside existing SW and HW
types. The new field "asid" is non-zero only for these new types.
The commit above did not properly initialize "asid" to 0 for a
temporarily HW breakpoint used during single step. This causes
cortex_a_unset_breakpoint() to identify this breakpoint as of
type "hybrid".
Identified through valgrind error:
Conditional jump or move depends on uninitialised value(s)
Actually valgrind triggers a total of 10 messages about use of
uninitialized variables, but they are all caused by the first
conditional jump bases on "asid != 0".
Fixed by initializing "asid" to 0 in cortex_a_step().
Fixes: c8926d1457 ("cortex_a hybrid & context breakpoints")
Change-Id: Ib674d8457d1e02e10332fd9d73554d13719ef93d
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4613
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
The periodic poll scans all the targets in the same order they
have been declared in the configuration file.
When one target in a SMP node halts, the transition is detected
in the following poll and this triggers a halt request to all the
other cores of the SMP node.
The targets that will be polled afterwards will be identified as
"halted", but the targets already scanned will remain as
"running" until the next periodic poll.
This creates a race condition with GDB; GDB sets the breakpoints
when runs the target and removes them as soon as the target is
halted. When it receives the halt event, it starts removing the
breakpoints and fails on the targets that are still reported as
"running".
Fixed by polling all the targets in the SMP node before informing
GDB about the halt event.
This implementation is almost copy/paste from the one in aarch64.
Change-Id: Id2bd99f1e56b014e48e9e34ccb891b4219c518f8
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4622
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
There is no reason to restrict the command "cortex_a dacrfixup"
to the EXEC phase only.
Change the command mode to ANY so the command can be used in
the initialization phase too.
Change-Id: I498cc6b2dbdc48b3b2dd5f0445519a51857b295f
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4623
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Commit ad6c71e151 introduced the
variable "mmu_enabled" whose pointer is passed to cortex_a_mmu()
that initialises it.
This initialization is not visible to the compiler that issue
a compile error.
The same situation is common across the same file and the usual
workaround is to initialize it to zero; thus the same fix i
applied here.
Ticket: https://sourceforge.net/p/openocd/tickets/197/
Fixes: commit ad6c71e151 ("cortex_a: fix virt2phys when mmu is disabled")
Change-Id: I77dec41acdf4c715b45ae37b72e36719d96d9283
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4619
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
When the MMU is not enabled on debug state entry, virt2phys cannot
perform a translation since it is unknown whether a valid MMU
configuration existed before. In this case, return the virtual
address as physical address.
Change-Id: I6f85a7a5dbc200be1a4b5badf10a1a717f1c79c0
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
Reviewed-on: http://openocd.zylin.com/4480
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Commit bfc5c764df avoids reading
ttbcr and ttb0/1 at every virt2phys translation by caching them,
and it updates the cached values in armv7a_arch_state().
But the purpose of any (*arch_state)() method, thus including
armv7a_arch_state(), is to only print out and inform the user
about some architecture specific status.
Moreover, to reduce the verbosity during a GDB session, the
method (*arch_state)() is not executed anymore at debug state
entry (check use of target->verbose_halt_msg in src/openocd.c),
thus the state of translation table gets out-of-sync triggering
Error: Address translation failure
or even using a wrong address in the memory R/W operation.
In addition, the commit above breaks the case of armv7r by
calling armv7a_read_ttbcr() unconditionally.
Fixed by moving in cortex_a_post_debug_entry() the call to
armv7a_read_ttbcr() on armv7a case only.
Remove the call to armv7a_read_ttbcr() in armv7a_identify_cache()
since it is (conditionally) called only in the same procedure
cortex_a_post_debug_entry().
Fixes: bfc5c764df ("armv7a: cache ttbcr and ttb0/1 on debug
state entry")
Change-Id: Ifc20eca190111832e339a01b7f85d28c1547c8ba
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: http://openocd.zylin.com/4601
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
- add 'dap create' command to create dap instances
- move all dap subcmmand into the dap instance commands
- keep 'dap info' for convenience
- change all armv7 and armv8 targets to take a dap
instance instead of a jtag chain position
- restructure tap/dap/target relations, jtag tap no
longer references the dap, daps are now independently
created and initialized.
- clean up swd connect
- re-initialize DAP also on JTAG errors (e.g. after reset,
power cycle)
- update documentation
- update target files
Change-Id: I322cf3969b5407c25d1d3962f9d9b9bc1df067d9
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
Reviewed-on: http://openocd.zylin.com/4468
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
and move declaration of all transport_is_xxx() functions to transport.h
Change-Id: Ib229115b5017507b49655bc43b517ab6fb32f7a6
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4469
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
This patch adds support in openOCD to read/write Arm vector/floating
point registers. This is compatible with Arm vfp v3 target xml in GDB.
Please refer to binutils-gdb/gdb/features/arm/arm-vfpv3.xml
Change-Id: Id4dd1bddef51c558f1a86300c1a876d159463f18
Signed-off-by: Omair Javaid <omair.javaid@linaro.org>
Reviewed-on: http://openocd.zylin.com/4421
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Tested-by: jenkins
When debugging Thumb-2 code, Gdb will at times send a
breakpoint packet 'Z0,<addr>,3', the number 3 denoting that
the instruction to break on is 32 bits long. Handle this by
replacing it with two consecutive 16bit Thumb BKPTs and make
sure to save and restore the full, original 32bit
instruction.
Note that this fix is only applicable if you debug a bare-metal program
(like the linux kernel) with the 'wrong' gdb, e.g. use an
"arm-linux" gdb instead of an "arm-eabi" gdb. But since most people
may not know about the subtle differences between gdb configurations
regarding thumb2 breakpoints it's still valid.
Change-Id: Ib93025faf35b11f0dba747a8c1fc36fd09a4c0f8
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
Reviewed-on: http://openocd.zylin.com/4241
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Tested-by: jenkins
Make DSCR_RUN_MODE() usable for armv8 and arm7 debug
Change-Id: Ib3ba3000d5b6aa03e590f3ca4969e677474eb12c
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
