this patch contains several changes to run control and state
handling together with gdb:
- graceful handling of target/gdb desync on resume, step and halt
- a default gdb-attach event executing the "halt" command, to meet gdb
expectation of target state when it attaches
- call target_poll() after Ctrl-C command from gdb
- call target_poll() after resume and step through a vCont packet
- fix log message forwarding on vCont stepping, also move an aarch64
log message from INFO to DEBUG level to prevent messing up the gdb
console during source-line stepping
- fix oversight in vCont support that messes up breakpoint handling
during stepping
Change-Id: Ic79db7c2b798a35283ff752e9b12475486a1f31a
Fixes: d301d8b42f
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
Reviewed-on: http://openocd.zylin.com/4432
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
target_start_algorithm does not download the algorithm. It only starts
it. It expects someone else to have already written the algorithm code
into the proper location before calling it.
Change-Id: I5e04406eed0ebb1c23e550dbf8d9f1204c432603
Signed-off-by: Christopher Head <chead@zaber.com>
Reviewed-on: http://openocd.zylin.com/4435
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
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
if armv7a_l1_d_cache_clean_inval_all will fail, error value is never
forwarded. So make sure we do it from now.
Change-Id: I02acfaa938ec09f58df77191d13d8f4bb3308720
Signed-off-by: Oleksij Rempel <linux@rempel-privat.de>
Reviewed-on: http://openocd.zylin.com/4384
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Images build improperly (by simply concatenating separate images) were accepted,
but anything after the first end-of-file record *silently* ignored. Now emit warning
for intel and motorola images upon non-whitespace after first end-of-file record but
continue reading anyway.
ST ships some images broken that way in their CubeMX packages ...
Change-Id: I0c5d08fa90070fed11fb805c5f0dc39817048176
Signed-off-by: Andreas Bolsch <hyphen0break@gmail.com>
Reviewed-on: http://openocd.zylin.com/4281
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
... by disabling all triggers, single stepping, enabling them, and then
resuming as usual. Without this change, you'd just be stuck on an
address trigger and would have to manually disable it.
Change-Id: I5834984671baa6b64f72e533c4aa94555c64617e
Fixed abstract register access for registers that aren't XLEN wide.
Avoided excessive errors cases where we attempted to execute a fence but
failed.
Don't mark all the CSRs as caller-save. gdb was saving/restoring
dscratch, which broke function calls as a side effect. dscratch is
accessible for people who really know what they're doing, but gdb should
never quietly access it. The same is probably true for other CSRs.
Change-Id: I7bcdbbcb7e3c22ad92cbc205bf537c1fe548b160
It's not tested because spike never reports any busy errors since every
access happens instantaneously.
Change-Id: If43ea233a99f98cd419701dc98f0f4a62aa866eb
scan-build reported a couple of problems with code in aarch64.c,
this patch cleans them up. No functional changes.
Change-Id: Ie210237ddc840a8bbcd535f86a3a5faf473132f2
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
Reviewed-on: http://openocd.zylin.com/4346
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Some of these changes actually fix broken comparisons which could
occasionally fail. Others just clean up the code and make it more clear.
Change-Id: I6c398bdc45fa0d2716f48a74822457d1351f81a5
Signed-off-by: Christopher Head <chead@zaber.com>
Reviewed-on: http://openocd.zylin.com/4380
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Command
mdw 0 0x40000000
triggers Segmentation fault on an arm.
Size parameter is a nonsence that may happen e.g. if you
mistype mdw instead of mww.
Add checking for calloc() NULL return in mdb/h/w.
Use calloc() instead of malloc() as multiplication
count * sizeof(uint32_t) overflows for size >= 0x40000000.
Change-Id: I968c944d863d1173ef932a7077d526fccb9381ae
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4349
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Don't use atomic dap operations when not necessary. Also remove
loop trying to set core power request, didn't find a platform
where it actually worked and it's slowing examination down.
Change-Id: I44e5c2f289f951b8f4579f08a841172404a52053
Signed-off-by: Matthias Welwarsky <matthias.welwarsky@sysgo.com>
Reviewed-on: http://openocd.zylin.com/4143
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
added missing instructions from latest available ejtag specification
(MD00047 v6.1 at time of writting) for trace control, fast data channel
and pcsample.
Change-Id: I30293951611265ffc2bd896f9d3ca6b310e5cac6
Signed-off-by: Jiri Kastner <cz172638@gmail.com>
Reviewed-on: http://openocd.zylin.com/3950
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
This duration vs sample count is _significantly_ closer to the truth
than simply declaring the value to be 100Hz.
Change-Id: Ie8d8bdf1959e1aa7cead0631cd2c86afe77d1efc
Signed-off-by: Karl Palsson <karlp@tweak.net.au>
Reviewed-on: http://openocd.zylin.com/4221
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Leverages the existing work that added profiling via DWT_PCSR.
hla_target doesn't have direct access to the mem_ap for doing a bulk
repeated read, but simply reading the DWT_PCSR register repeatedly is
still ~2 order of magnitude faster than halt/resume.
Change-Id: Ibe451aa95143694398370fdad6939cfb6191d56f
Signed-off-by: Karl Palsson <karlp@tweak.net.au>
Reviewed-on: http://openocd.zylin.com/4220
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Use DWT_PCSR if present (reads nonzero); otherwise do halt-and-sample pc.
Signed-off-by: Simon Schubert <2@0x2c.org>
Change-Id: Id2dc4665e5008cc497a2e6e6493522d038d5af42
Reviewed-on: http://openocd.zylin.com/4211
Tested-by: jenkins
Reviewed-by: Karl Palsson <karlp@tweak.net.au>
Reviewed-by: Christopher Head <chead@zaber.com>
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
We're using an I/O port reset by default. The only board currently
supported (Galileo) doesn't have SRST routed on the JTAG connector.
When using 'reset halt', we must rely on Reset Break because our
adapters don't have support for PREQ#/PRDY# signals.
Tested with Intel Galileo GEN2.
Change-Id: Ia406e31c156f8001717d5b6a08bd03f71de790d3
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Reviewed-on: http://openocd.zylin.com/4016
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Just some misc cleanups without any functional changes. It's just
easier to read.
Change-Id: I66bee31f297bcbdb8cc4446b774d7714fbaa7c83
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Reviewed-on: http://openocd.zylin.com/4015
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
No functional changes, cleanup only
Change-Id: I53c422be16d0a4ff157745d31362f6483093e5eb
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Reviewed-on: http://openocd.zylin.com/4014
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
When testing with Intel Galileo GEN2 I have noticed a few iterations
where probemode entry took a little longer. At most I had to read
tapstatus twice. This patch uses a loop of up to 100 iterations to
wait for tapstatus to update with PM entry results.
Change-Id: I1c87d7dde715255b3fd2460d299b9493218533fc
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Reviewed-on: http://openocd.zylin.com/4013
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
These symbols are only used within this C source file. They don't need
to be exposed to the outside.
Change-Id: Idb04550ecca7f12c3fdc8c6447eeeb871961add3
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Reviewed-on: http://openocd.zylin.com/4012
Tested-by: jenkins
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Reported by thesourcerer on IRC.
Change-Id: I36d4d95de176a19fdd0bf80d8b419be6ca637e98
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Reviewed-on: http://openocd.zylin.com/3893
Tested-by: jenkins
Functions mem_ap_read() and mem_ap_write() incremented address even
if addrinc=false. I overlooked this fact and moved mem_ap_setup_tar()
set wrong addresses in no-incr mode.
Fixed by preventing address increment in no-incr mode.
Change-Id: I512e12a6a64e30cf6bc5bf77e3d57d35cc33e058
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Suggested-by: Matthias Welwarsky <matthias@welwarsky.de>
Reviewed-on: http://openocd.zylin.com/4326
Tested-by: jenkins
Reviewed-by: Christopher Head <chead@zaber.com>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
MEM-AP access through banked data registers MEM_AP_REG_BD0..3
does not increment TAR regardless of the current autoincrement mode.
mem_ap_read_u32() and mem_ap_write_u32() can keep the current
autoincrement mode instead of switching autoincrement off.
Change-Id: Ib7ec688d3e04f1da678363cd2819ce90e8910e58
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4163
Tested-by: jenkins
Reviewed-by: Andreas Bolsch <hyphen0break@gmail.com>
Reviewed-by: Christopher Head <chead@zaber.com>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Problem: If the same memory location is accessed alternatively
by MEM-AP banked data registers without autoincrement and by standard
autoincremented read/write, TAR register is not updated correctly.
How to replicate: On a Cortex-M issue
mdw 0xe000edf0
multiple times. When poll is on (poll reads the same memory location)
only the first read is correct.
0xe000edf0: 01000000
0xe000edf0: 00000000
0xe000edf0: 20002640
0xe000edf0: 01000000
0xe000edf0: 00000000
0xe000edf0: 00000000
No problems with poll off.
0xe000edf0: 01000000
0xe000edf0: 01000000
0xe000edf0: 01000000
mem_ap_setup_tar() writes to MEM_AP_REG_TAR if requested TAR value
changed or CSW_ADDRINC_... is currently active.
However if an autoincremented access has been issued and autoinc
switched off in CSW afterwards, TAR does not get updated.
The change introduces mem_ap_update_tar_cache() which is called
after queuing of any access to MEM_AP_REG_DRW. It simulates
TAR increment to keep tar_value in sync with MEM_AP.
Crossing tar autoincrement block boundary invalidates cached value.
mem_ap_write() and mem_ap_read() do not check tar autoincrement
block boundary, mem_ap_setup_tar() is called before each transfer instead.
dap_invalidate_cache() is introduced to ensure invalidation
of all cached values during dap_dp_init() and swd_connect()
Change-Id: I815c2283d2989cffd6ea9a4100ce2f29dc3fb7b4
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4162
Tested-by: jenkins
Reviewed-by: Christopher Head <chead@zaber.com>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
This avoids trying to read memory from the wrong hart, if the current
hart was changed by an earlier call (eg. to poll()).
Change-Id: I73da1e01c8d01d68f01ac7fdd6c548380a70cfd3
The existing code only used Memory Access mode to read memory,
which uses 32 bit operations only.
Rework the code to check the alignment/size of the read/write operation,
and use the Memory Access mode to read aligned 32 bit memory.
When using unaligned access, or 8 or 16 bit reads, use LDR{BHW} and STR{BHW}
instead.
The exception handling is still the same as it was before (meaning it breaks
when things go wrong), but I can now read an 8 bit register correctly.
Change-Id: I739a5ee825c0226ed4a89c32895cc2a047b8dc15
Signed-off-by: Bas Vermeulen <bas@daedalean.ai>
Reviewed-on: http://openocd.zylin.com/4301
Tested-by: jenkins
Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de>
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
As part of this I improved the memory read/write fatal error handling a
bit. Now at least we try to leave autoexec turned off, and will even
restore the temp registers if the situation isn't too hosed for that.
Partly addresses Issue #142
Change-Id: I79fe3f862f11c6d20441f39162423357e73a40c1
This lets users tell OpenOCD which non-standard CSRs exist on their
target, that will also be accessible and whose existence will be
communicated to gdb.
Change-Id: I56163a9fcb84ad7ebe815ae74fbd9fcc208f5a9d
(It's really only 2 bits, but something wonky happens between gdb and
OpenOCD if I make it that size.)
Change-Id: I562a65cb0ebe5aa0edcc54c251d0fea0e26f9cb1
Events reset-halt-pre, reset-halt-post, reset-wait-pre and
reset-wait-post are not used anywhere.
Change-Id: I9a0f94875b102d9b08f6c2fd9d73a9f05f8e8e79
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: http://openocd.zylin.com/4285
Tested-by: jenkins
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
New STM8 target based mostly on mips4k. Target communication
through STLINK/SWIM. No flash driver yet but it is still possible
to program flash through load_image command. The usual target debug
methods are implemented.
Change-Id: I7216f231d3ac7c70cae20f1cd8463c2ed864a329
Signed-off-by: Ake Rehnman <ake.rehnman@gmail.com>
Reviewed-on: http://openocd.zylin.com/3953
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-by: Paul Fertser <fercerpav@gmail.com>
Change-Id: Ieea1b0dec88818e9e8d5c8c5d54aa8959556d77b
Signed-off-by: Marc Schink <openocd-dev@marcschink.de>
Reviewed-on: http://openocd.zylin.com/4275
Tested-by: jenkins
Reviewed-by: Christopher Head <chead@zaber.com>
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Because there is no instruction that moves just half of a 64-bit FPR
to/from a GPR, we need to use scratch memory for this operation. This
code can theoretically use:
1. DMI_DATA, if it is memory mapped in the target.
2. DMI_PROGBUF, if it is writable in the target.
3. A user-configured address.
I have only tested this code very lightly. One reason is that gdb thinks
that on RV32 harts every register is 32 bits wide. Another is that this
is mostly proof-of-concept to satisfy the small program buffer code
review, which I don't want to drag out forever.
Existing tests don't realize that floating support was broken with
RV32D, and don't realize that it still doesn't work because of the gdb
problem mentioned above.
This change improves Issue #110 but there's more work to be done.
Change-Id: I99b8a36e5fea26f1d9e16e36cf99adc7be26b944
The dhcsr_save variable was used to save the value of
cortex_m->dcb_dhcsr so it could be restored later. However, all writes
in between the save and the restore use mem_ap_write_atomic_u32, not
cortex_m_write_debug_halt_mask, which means cortex_m->dcb_dhcsr isn’t
changed anyway. Delete the unnecessary local.
Change-Id: I064a3134e21398e1ecfc9f1fa7efd7b020b52341
Signed-off-by: Christopher Head <chead@zaber.com>
Reviewed-on: http://openocd.zylin.com/4240
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
In cortex_m_assert_reset, in two locations, DHCSR is written directly
using mem_ap_write_u32. This means that the cached version,
target_to_cm(target)->dcb_dhcsr, is not updated when these writes are
performed, so subsequent writes to DHCSR that use
cortex_m_write_debug_halt_mask will change those bits back to their old
values which, unless modified in that particular invocation, come from
the cache. This causes an actual, observable bug on an STM32F7 in which
running “reset run” immediately after “program” can in some cases result
in execution proceeding with C_MASKINTS set (it is cleared on line 1021
but is then set immediately afterward in cortex_m_clear_halt), causing
failure of the application. Replace these mem_ap_write_u32 calls with
cortex_m_write_debug_halt_mask calls to do the same jobs.
Change-Id: Id35ca7f6057c2df2ba9cd67c53a73b50816d0b71
Signed-off-by: Christopher Head <chead@zaber.com>
Reviewed-on: http://openocd.zylin.com/4239
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
GCC7 with -Wall warns about possible string truncation with
snprint()-type functions with "directive output may be truncated writing
1 byte into a region of size between 0 and 9
[-Werror=format-truncation=]" + "note: ‘snprintf’ output between 5 and
14 bytes into a destination of size 12" (or similar). Fix this by
increasing sizes of buffers.
See https://gcc.gnu.org/gcc-7/changes.html
Change-Id: Ib848f2a56dd658783534158947ae1be7c0e99d45
Signed-off-by: Freddie Chopin <freddie.chopin@gmail.com>
Reviewed-on: http://openocd.zylin.com/4175
Reviewed-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Tested-by: jenkins
Reviewed-by: Andreas Bolsch <hyphen0break@gmail.com>
GCC7 with -Wextra warns about switch-case blocks which fallthrough with
"this statement may fall through [-Werror=implicit-fallthrough=]". This
can be fixed by adding "special" comments: "/* fallthrough */".
See https://gcc.gnu.org/gcc-7/changes.html
Change-Id: Iba0be791dbdd86984489b2d9a0592bb59828da1e
Signed-off-by: Freddie Chopin <freddie.chopin@gmail.com>
Reviewed-on: http://openocd.zylin.com/4174
Tested-by: jenkins
Add "arm semihosting_cmdline [argv0 argv1 ...]" for setting the
command line arguments for the debuggee.
[andreas.fritiofson@gmail.com]: Dynamic allocation, empty default
Change-Id: I831ddd161d602f251940e29608a154e9590fdee1
Signed-off-by: Christian Groessler <chris@groessler.org>
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/3106
Tested-by: jenkins
They're never used, so just drop them.
Change-Id: Ie137deed3e7258f9d6af7e0cb508e73df0f53ee0
Signed-off-by: Karl Palsson <karlp@tweak.net.au>
Reviewed-on: http://openocd.zylin.com/4131
Tested-by: jenkins
Reviewed-by: Spencer Oliver <spen@spen-soft.co.uk>
Previously it might read an address multiple times if an abstract
command took longer to execute than expected.
The new implementations reads from the target how far it has gotten
along reading memory, and resumes from there if cmderr=busy.
This ended up being a bigger change than I envisioned, but in the end it
deleted more lines than it added, so I'm happy. :-)
The downloaded program now post-increments, and there's no longer an
attempt to read the current address from the target. This made it easier
to fix the problem where at the start of the loop the current address
was already read (in regular entry) or has not yet been read (when the
first round through the loop encountered busy more than once, or busy
was encountered at least once later on).
The interesting new code concerns ignore_prev_addr and
this_is_last_read.
Additionally, I tweaked some debug output, and optimized
riscv_batch_run() when the batch is empty.
The actual implementation of triggers didn't change between those two
versions, so there's no need to duplicate the code.
In the process, I also fixed a minor multicore bug where tselect didn't
always get written on all harts.
This is the only way the spec guarantees that GPRs are accessible, and
depending on the implementation this might be the only way that CSRs are
accessible.
Also changed the debug code that parses out DMI fields to be simpler to
maintain (albeit a little slower).
riscv013_execute_debug_buffer() now automatically clears cmderr if the
command fails. That feels like the right behavior. (It does return the
error to its caller.)
When first connecting to a target, have the debugger disable any
hardware triggers that are set by a previously connected debugger.
The 0.11 code already did this, but 0.13 did not.
To achieve this I decided to share the code to enumerate triggers
between 0.11 and 0.13, which required me to implement get_register() and
set_register() for 0.11, which made the whole change a lot larger than
you might have guessed.
Hopefully this sets us up to in the future share the code to set/remove
triggers as well.
They were implemented, and people want to keep using them.
Also make OpenOCD tolerate cores that have $misa at 0xf10 instead of the
current address of 0x301.
Actually return an error when we fail to read a CSR.
Tweak cache_set32() debug output.
Commit 47b8cf842 changed the fixed type of the value argument to snprint
but didn't change the format string to match for sizes != 64 bit.
Change-Id: I908b06f49ab69d04224282949190a0de883048e0
Signed-off-by: Andreas Fritiofson <andreas.fritiofson@gmail.com>
Reviewed-on: http://openocd.zylin.com/4167
Tested-by: jenkins
Reviewed-by: Stian Skjelstad <stian@nixia.no>
Reviewed-by: Philipp Guehring <pg@futureware.at>
Now that we're supporting non-RTOS multi-hart mode there's some more
assertions that you're running on the right hart. Those assertions
aren't sane very early in examine, so I avoid them.
Rather than having a bunch of "if rtos" stuff, I now just check "if
hart_enabled". This makes some code paths cleaner, all of which were
buggy in the non-RTOS multi-hart mode.
Tomas Vanek
Christopher Head
Tim Newsome
Girts Folkmanis
Ake Rehnman
Peter Mamonov
Matthias Welwarsky
Oleksij Rempel
Kaspar Schleiser
Andreas Bolsch
Megan Wachs
Gleb Gagarin
Paul Fertser
Jiri Kastner
Karl Palsson
Simon Schubert
Felipe Balbi
Bas Vermeulen
Marc Schink
Jonas Norling
Freddie Chopin
Andreas Fritiofson
Fabio Utzig
Liviu Ionescu
Palmer Dabbelt
Dmitry Ryzhov