On FreeBSD, sig_atomic_t is of type 'long' rather than 'int'. Remove
the assert() instead of replacing it with a more complex, size-agnostic
check that would handle both 'int' and 'long'. Since the signal number
comes from the operating system, an explicit range check is unnecessary.
Checkpatch-ignore: COMMIT_LOG_LONG_LINE
Build error on FreeBSD:
../src/server/server.c:616:39: error: result of comparison of constant 9223372036854775807 with expression of type 'int' is always true [-Werror,-Wtautological-constant-out-of-range-compare]
616 | assert(sig >= SIG_ATOMIC_MIN && sig <= SIG_ATOMIC_MAX);
| ~~~ ^ ~~~~~~~~~~~~~~
/usr/include/assert.h:55:21: note: expanded from macro 'assert'
55 | #define assert(e) ((e) ? (void)0 : __assert(__func__, __FILE__, \
| ^
../src/server/server.c:616:14: error: result of comparison of constant -9223372036854775808 with expression of type 'int' is always true [-Werror,-Wtautological-constant-out-of-range-compare]
616 | assert(sig >= SIG_ATOMIC_MIN && sig <= SIG_ATOMIC_MAX);
| ~~~ ^ ~~~~~~~~~~~~~~
/usr/include/assert.h:55:21: note: expanded from macro 'assert'
55 | #define assert(e) ((e) ? (void)0 : __assert(__func__, __FILE__, \
Change-Id: I6cf9242ab8206184bff9390ef0245843261656bd
Signed-off-by: Marc Schink <dev@zapb.de>
Reviewed-on: https://review.openocd.org/c/openocd/+/9477
Reviewed-by: Anatoly P <kupokupokupopo@gmail.com>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
Some clients probe or use SMP-related 'j' packets. If OpenOCD
received an unknown 'j' packet, gdb_read_smp_packet previously did
not send any reply. This may result cause the client to hang while
waiting for a response.
Send an empty reply packet for unsupported 'j' packets.
Change-Id: I84c5e8e99f946d41dd5c11163c3eeb8af2b6c90f
Signed-off-by: Cristian Dinca <hello@icmd.tech>
Reviewed-on: https://review.openocd.org/c/openocd/+/9461
Tested-by: jenkins
Reviewed-by: Evgeniy Naydanov <eugnay@gmail.com>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Add check for failed allocation.
Add warning for memory allocation fallback and change existing log
from user to more appropriate warning and debug.
Move allocation and check before changing flash information.
Change-Id: I5b2ab6bc12ea15a5d8f634ed00cf0a0bc7e5a517
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9384
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-by: Marc Schink <dev@zapb.de>
Before this fix, the "exit" command behaved differently
depending on where it was called from:
- "exit" in a telnet session: Session disconnected.
- "exit" in a Tcl session: Empty response sent but the
session remained connected.
- "exit" in a script (outside of Telnet or Tcl): OpenOCD
exited with code 0. Based on the help and documentation
"exit" was apparently not intended for these cases.
What's more, if "exit" is allowed in Tcl scripts,
user may confuse it with the Tcl's native "exit"
command that is not available in OpenOCD (it is
shadowed with this OpenOCD's "exit" command).
This fix makes the behavior of "exit" consistent:
- "exit" in a telnet session: Session disconnected
(no change).
- "exit" in a Tcl session: Session disconnected
(same as for telnet).
- "exit" in a script: Notify the user that "exit"
is deprecated outside of telnet/tcl but
still shut down OpenOCD (to preserve the original
behavior).
Update the documentation to make it very clear to users
when to use "exit" vs. "shutdown".
Change-Id: I790495330e1fa705b34097a1347fdc57aaa86de1
Signed-off-by: Jan Matyas <jan.matyas@codasip.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9380
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
Cortex-M debug AP could be on AHB3 or AHB5.
Finding AHB3 is fast as AP usually has a low address.
On the other hand, finding AP on the AHB5 equipped device is weird
without this patch. cortex_m_find_mem_ap()
first tries to read IDR from all possible 256 ADIv5 APs and checks
for AHB3 ID and if that fails, starts over and search for AHB5.
It takes long time (74 msec on fast USB HS based STLINK V3J15M7)
and logs lot of rubbish:
Debug: 168 117 target_examine_one(): [stm32u3x.cpu] Examination started
Debug: 169 117 target_call_event_callbacks(): event 19 (examine-start)
Debug: 170 117 arm_adi_v5.c:1201 dap_get_ap(): refcount AP#0x0 get 1
Debug: 171 118 arm_adi_v5.c:1226 dap_put_ap(): refcount AP#0x0 put 0
Debug: 172 118 arm_adi_v5.c:1201 dap_get_ap(): refcount AP#0x1 get 1
Debug: 173 118 arm_adi_v5.c:1226 dap_put_ap(): refcount AP#0x1 put 0
Debug: 174 118 arm_adi_v5.c:1201 dap_get_ap(): refcount AP#0x2 get 1
Debug: 175 119 arm_adi_v5.c:1226 dap_put_ap(): refcount AP#0x2 put 0
...
Debug: 188 123 arm_adi_v5.c:1201 dap_get_ap(): refcount AP#0x9 get 1
Debug: 189 123 stlink_usb_error_check(): STLINK_BAD_AP_ERROR
Debug: 190 123 arm_adi_v5.c:1226 dap_put_ap(): refcount AP#0x9 put 0
Debug: 191 123 arm_adi_v5.c:1201 dap_get_ap(): refcount AP#0xa get 1
Debug: 192 123 stlink_usb_error_check(): STLINK_BAD_AP_ERROR
Debug: 193 123 arm_adi_v5.c:1226 dap_put_ap(): refcount AP#0xa put 0
...
Debug: 926 190 arm_adi_v5.c:1201 dap_get_ap(): refcount AP#0xff get 1
Debug: 927 191 stlink_usb_error_check(): STLINK_BAD_AP_ERROR
Debug: 928 191 arm_adi_v5.c:1226 dap_put_ap(): refcount AP#0xff put 0
Debug: 929 191 arm_adi_v5.c:1154 dap_find_get_ap(): No MEM-AP AHB3 found
Debug: 930 191 arm_adi_v5.c:1201 dap_get_ap(): refcount AP#0x0 get 1
Debug: 931 191 arm_adi_v5.c:1144 dap_find_get_ap(): Found MEM-AP AHB5
at AP index: 0 (IDR=0x14770015)
Introduce dap_find_by_types_get_ap() to search for the array of AP IDs.
With the patch cortex_m_find_mem_ap() succeeds immediately:
Debug: 168 118 target_examine_one(): [stm32u3x.cpu] Examination started
Debug: 169 118 target_call_event_callbacks(): event 19 (examine-start)
Debug: 170 118 arm_adi_v5.c:1222 dap_get_ap(): refcount AP#0x0 get 1
Debug: 171 118 arm_adi_v5.c:1150 dap_find_by_types_get_ap(): Found MEM-AP
AHB5 at AP index: 0 (IDR=0x14770015)
Change-Id: Iabcfa1fd64a48febb0f759a213f15d69621ea5cf
Fixes: commit 0d47d85ff5 ("target/cortex_m: Add support for AHB5-AP")
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/9439
Tested-by: jenkins
Reviewed-by: Marc Schink <dev@zapb.de>
This is necessary for non-flash NVM which does not have the concept of
sectors, such as RRAM. Also small flash memory regions without sectors,
such as configuration areas, can now be exposed to GDB.
Tested with nRF54L15-DK, tested for regression with AT32F421C8T7.
Change-Id: Ifc96ddca7dcfcd4ab17683c0e59f1be912901941
Signed-off-by: Marc Schink <dev@zapb.de>
Reviewed-on: https://review.openocd.org/c/openocd/+/9290
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
About why the new log level LOG_LVL_DEBUG_USB has the same value a
LOG_LVL_DEBUG_MALLOC, see the mailing list discussion starting here:
Replacing --enable-verbose-usb-comms in configure.ac
https://sourceforge.net/p/openocd/mailman/message/59215751/
Other minor fixes included here which are probably
not worth submmitting in separate patches:
- In error message "level must be between -3 and 4", increase 4 to 5.
- LOG_DEBUG_IO was passing LOG_LVL_DEBUG instead of LOG_LVL_DEBUG_IO.
Change-Id: I71440bbabe4785338c0a27562cc76fa1b7d54bf5
Signed-off-by: R. Diez <rdiez-2006@rd10.de>
Reviewed-on: https://review.openocd.org/c/openocd/+/9432
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Similar as virtual flash but has write/erase suppressed.
'virtual' flash driver name check in flash_free_all_banks()
was replaced by a customized free_driver_priv()
Change-Id: I528760aad0ba55ebc57fc1fabfdfdf07c92cac94
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/5107
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
The only reason to define a read-only region is that gdb needs
a complete memory map to choose hard or soft breakpoints properly.
Change-Id: I9d05cb6b91f054ad5cc9333af6b14eb433dbdc99
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/5106
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
The functions in struct target_type::blank_check_memory() return
either an OpenOCD error or a positive value that indicates the
number of blocks checked.
To prevent the mix of error codes and returned values, return the
value through an additional parameter 'checked' and then return
ERROR_OK.
While there:
- change to unsigned int the parameter 'num_blocks';
- in armv7m_blank_check_memory() verify that the working area can
contain at least two 'algo_block'.
Change-Id: Ie22f5816819bc77ec611c3f251373d026ed9f784
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9386
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
1. Added comments like /* MSPM0G310x-Q1 */ to the parts table
based on the datasheet the parts came from.
2. Sorted the MSPM0G310x-Q1 device lines alphabetically: this is
how all other MSPM0G parts are organized.
Change-Id: I90d0f44289a538667c4ad0f23d44581f3c65d0c8
Signed-off-by: Mikhail Iakhiaev <iakhiaev@google.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9419
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-by: Nishanth Menon <nm@ti.com>
The 'dap info' command was not checking that the target was
an ARM before dereferencing the `arm` pointer. This would
cause a crash if the current target was (say) a mem_ap.
Add 'target_to_dap' function to safely get the dap
Change-Id: I0c765f915f2ef7b9a7d20c934e144559ca4e5f1c
Signed-off-by: Grant Ramsay <grant.ramsay@hotmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/8415
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
The incremental counter of logged lines is a signed int.
Change it to unsigned, since negative values has no sense in this
context.
Change-Id: Ia75f9ca038bba385c5f88fb9fa368faaf055f9a3
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9401
Reviewed-by: R. Diez <rdiez-2006@rd10.de>
Tested-by: jenkins
Like for Cortex-M7, also Cortex-M85 prevents detecting the cache
properties when the CPU is kept under reset.
Extend to Cortex-M85 the same deferred mechanism already in place
for Cortex-M7.
Change-Id: Id274bb6c0b46c568554eed9671ef690c34cf7cfa
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9397
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-by: Marc Schink <dev@zapb.de>
Previously, configure.ac only checked whether glibc was available,
but other C runtime libraries like Newlib have mallinfo too.
This is a first step to remove configuration option --enable-malloc-logging
and replace it with a debug level configurable at runtime.
Change-Id: If30fc98a84158459e222fddf08043f46d6fa4112
Signed-off-by: R. Diez <rdiez-2006@rd10.de>
Reviewed-on: https://review.openocd.org/c/openocd/+/9394
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
The function free_service() does not close the TCL socket when the
service is shutdown.
Add the necessarily code.
Change-Id: I58c5c4fee3111949fcc143c6226522c29d976493
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9376
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
During JSP service start, check that memory is properly allocated
and if add_service() fails release the allocated memory.
Change-Id: I7eb2ea8a10dbeccd27fd39485080552c5d0736d3
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9375
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
During GDB service start, check that memory is properly allocated
and if add_service() fails release the allocated memory.
While there, modify the code following the coding style.
Change-Id: Iebd1481a82f7391c110c5f6ad9878ba4abf052b3
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9374
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
The function add_service() can return on error when cannot create
the new service. In this case the caller cannot assume that the
parameter 'priv' has been taken in consideration and it should
take care of free() it.
To avoid a double free(), add_service() should not free() the
parameter 'priv' if it exits with error.
Replace the call to free_service() with a dedicated exit path on
error.
Change-Id: I340ec3ee46f471f31170c6717ed74fb632f0da20
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reported-by: Karl Palsson <karlp@tweak.au>
Fixes: 171454fffa ("server: fix a new double free()")
Fixes: 5ff384be08 ("semihosting: fix memory leak and double free")
Reviewed-on: https://review.openocd.org/c/openocd/+/9372
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Wrap the name field in braces when formatting the flash list Tcl structure
so that Tcl does not interpret backslash escape sequences.
This ensures Windows paths are preserved verbatim in the output and
can be reliably matched in Tcl scripts.
Fixes: 80fc9fabc6 ("flash: nor: rewrite command 'flash list' as COMMAND_HANDLER")
Change-Id: Ibd0a4b292828f084122e4a02732dc69ac0885f3c
Signed-off-by: HAOUES Ahmed <ahmed.haoues@st.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9360
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Move the configuration files into a dedicated vendor folder as required
by the developer guidelines.
Change-Id: I9ed39e32b6281a9cb8510914690f3f7751b795c8
Signed-off-by: Marc Schink <dev@zapb.de>
Reviewed-on: https://review.openocd.org/c/openocd/+/9271
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Use the new adapter I/O log level to more easily distinguish
adapter I/O messages from other debug messages.
Change-Id: Ib3aa2bf84871c038aef937452cfeb28de232d821
Signed-off-by: Marc Schink <dev@zapb.de>
Reviewed-on: https://review.openocd.org/c/openocd/+/9387
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
There are two allocated fields that are never freed,
contiguous_regs_desc and contiguous_regs_list.
Additionally, prevent memory leaks and invalid accesses
when xtregs command is called repeatedly.
Change-Id: Id6ab4a2565ddb19e1e9d3f1c3b822182b3a6fb9d
Signed-off-by: Samuel Obuch <samuel.obuch@espressif.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9113
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
Reviewed-by: Ian Thompson <ianst+cdns@cadence.com>
This commit prevents NULL dereference during malloc
failure on stm32x bank command.
Change-Id: Ib348a1863aee5f8ed9d1699d203cdc4c83557628
Signed-off-by: Igor Cudnik <igor@its.rel.pl>
Reviewed-on: https://review.openocd.org/c/openocd/+/9379
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Tested-by: jenkins
The command "flash erase_check" reported any error code from
flash driver erase_check method as "unknown" despite the error
is usually described in a log message from the flash driver layer.
E.g.:
Error: Target not halted
unknown error when checking erase state of flash bank #0 at 0x00000000
Reword the message not to mislead user.
Change-Id: Ica768fa2924c107fa36d0f654d2c3149aa608dc4
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/9331
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Apparently the last user of the hack in alloc_vprintf() is the
code in log_vprintf_lf().
Document it for future rework.
Change-Id: I6b69f8b7f3feb93e5daa9649c4122ec12e22d84f
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9317
Tested-by: jenkins
Commit 7f260f5009 ("helper/command: Handle Tcl return values
consistently") allows calling command_print with the first
parameter set to NULL.
This should not be allowed.
I cannot identify any part of OpenOCD that calls a command that in
turn calls command_print() with NULL command_invocation.
Adding
__attribute__((nonnull (1, 2)))
to the prototype of the functions does not trigger any positive
with GCC and scan-build.
Add an assert to detect such corner cases.
This change is kept small and self contained to allow reverting it
easily if a real use case is identified.
Change-Id: I8aa6e5c0c341e818c0327eaa0d2bd5b6304c93b8
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9091
Tested-by: jenkins
Avoid code duplication by merging command_print() and
command_print_sameline().
Detect the allocation error, keep track of it and let the command
return error.
Add a FIXME as the functions should always have 'cmd' properly
set. Should this be an assert()?
Change-Id: Iff704c42969a7ca9ea884520942adecd40bebbd6
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9090
Tested-by: jenkins
Rewrite the function command_print() without using any specific
API from jimtcl for string manipulation.
Change-Id: I1adddd493b43e30ead26e96da09a4ee8c0a41307
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9076
Reviewed-by: Evgeniy Naydanov <eugnay@gmail.com>
Tested-by: jenkins
The current code for cortex_a watchpoint sets the field DBGWCR:LSC
to '3', that corresponds to 'access' watchpoint.
Thus, any 'r' or 'w' watchpoint is considered to 'a'.
Convert the enum watchpoint_rw to the corresponding values for the
field DBGWCR:LSC.
Change-Id: Iccfddb3e34f3f26927983f3b00d9d5f81b06eb21
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/9291
Tested-by: jenkins
aarch64_resume() set the last_run_control_op on the current core
only and left last_run_control_op as is on the rest of the SMP group.
It caused semihosting call on other cores to stop because
last_run_control_op sticked on ARMV8_RUNCONTROL_STEP
Set last_run_control_op to ARMV8_RUNCONTROL_RESUME on all cores
in the SMP group.
Change-Id: I55a97bb1f7ea25bfc5937c3cc846532cdf390064
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/9247
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
Semihosting worked only on the first/gdb assigned core of a SMP group.
If a semihosting call was issued on another core, aarch64_update_halt_gdb()
emitted 'halted' event on core0 before semihosting decoding started.
Use target's smp_halt_event_postponed flag to keep events from emitting
until semihosting is decoded. If a semihosting call is confirmed,
clear flags and do not send 'halted' event for any core of SMP group.
Change-Id: Ie7eff7e493c2a4df3039f49fce1744d996050a59
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/9246
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
update_halt_gdb() called aarch64_poll() recursively with temporary
switching target's smp flag off to prevent deeper recursion.
This was not possible for gdb assigned target or hwthread failed
with "SMP node change, disconnect GDB from core/thread".
Therefore the aarch64_poll(gdb_target) resulted in the useless
recursion back to update_halt_gdb().
Introduce aarch64_poll_smp() with smp parameter to avoid
update_halt_gdb() recursion properly and without fiddling
with target's smp flags.
While on it, add 'aarch64_' prefix to update_halt_gdb() function.
Change-Id: I645166f50c106f4a6d4d35dc70ad49041d2442aa
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/9245
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
