This involves halting the target, which might have unintended side
effects, but when the debugger is connected software breakpoints must
trap to the debugger. Anything else is a terrible user experience.
Change-Id: I1f7bb610eeeb054cc3042dc6bcfc16589ce12a31
Signed-off-by: Tim Newsome <tim@sifive.com>
Intended as a place where we can interact with the target without too
much concern about preserving state and doing exactly the right thing
while poll() is going on.
Change-Id: Ic9bd441caae85901a131fd45e742599803df89b5
Signed-off-by: Tim Newsome <tim@sifive.com>
Specifically, call into the RISC-V version when target becomes halted,
running, or unavailable.
I'll be using unavailable shortly.
Change-Id: I9ffffdccbf22e053fe6390d656b362bf9ab9559a
Signed-off-by: Tim Newsome <tim@sifive.com>
Will be used later when we want to do a quick halt/resume.
Change-Id: Ib80166234c4c277b7d9ce26b7566ac0f93017e64
Signed-off-by: Tim Newsome <tim@sifive.com>
* Only set ebreak bits that might be supported based on misa.
* Don't write dcsr if its value wouldn't change.
Change-Id: I7087af0b0df0fbdbf994373b5c887b9b389df872
Signed-off-by: Tim Newsome <tim@sifive.com>
Make it callable earlier, handle `supported` being NULL, and make enum
names more clear.
Change-Id: If4d286b54ccfc01eb5de5a57eb18f748c920e979
Signed-off-by: Tim Newsome <tim@sifive.com>
The riscv013_on_halt function was being called but its implementation was
empty, providing no additional functionality. Removed the function declaration,
calls to it, and its implementation since it is not required.
Change-Id: I425ea890deadeec945f0a47af247f3f99172e801
Signed-off-by: Tim Newsome <tim@sifive.com>
When halted we don't need to read all 3 instructions before deciding the
sequence doesn't match.
Change-Id: I9f8345960ce27e859265af901a368166a70b9fde
Signed-off-by: Tim Newsome <tim@sifive.com>
These functions used to exist but don't anymore. (Pointed out in #863)
Change-Id: Iac6b5edd320bdff7628a788861e332f956dcd93d
Signed-off-by: Tim Newsome <tim@sifive.com>
Previously, progbuf execution did not flush or invalidate the register cache which could lead to incorrect behavior. This patch fixes it as well as refactors few sore points in the code related to it.
Change-Id: I353b931ca70a1828d4a9cc512aead00441730875
Signed-off-by: Marek Vrbka <marek.vrbka@codasip.com>
Otherwise we may end up modifying DCSR of a different hart than
intended.
Change-Id: I39bde21a1444623ed150f2b3d504b9318b9d6191
Signed-off-by: Tim Newsome <tim@sifive.com>
Register access on running target should fail if mstatus needs to be
modified.
Change-Id: Iec8e8d514ef2f5ca42606a5534cce55aaaa99180
Signed-off-by: Evgeniy Naydanov <evgeniy.naydanov@syntacore.com>
This way if you connect to a running target, before it's hit a breakpoint,
then when it does hit the breakpoint OpenOCD will catch it.
Change-Id: I6f1e5f169fa385f46759015786e664693c3872e4
Signed-off-by: Tim Newsome <tim@sifive.com>
Poll failure just means poll failed. It's safer to assume the target is
still running, because then if it is running and subsequently halts we can
relay this to gdb correctly. We can't do the other way around, because once
gdb thinks the target has halted, it can't deal with it spontaneously
running.
Change-Id: Idb56137f1d6baa9afc1b0e55e4a48f407b8ebe83
Signed-off-by: Tim Newsome <tim@sifive.com>
When a DM was powered down, we end up in examine() again, and clearly if
the DM was powered down we need to invalidate that cache.
Change-Id: I5eb6a289939f313e06c09cac22245db083026aa3
Signed-off-by: Tim Newsome <tim@sifive.com>
The error state is sticky, so this has to be done to recover.
Change-Id: I589f3cdab0f2351fd25f89951830cbc16c39bd93
Signed-off-by: Tim Newsome <tim@sifive.com>
This commit introduces a new function, which can be used to reduce number
of register accesses.
Change-Id: I125809726eb7797b11121175c3ad66bedb66dd0d
Signed-off-by: Evgeniy Naydanov <evgeniy.naydanov@syntacore.com>
Since writing a register can make some GPRs dirty (e.g. S0, S1), registers
should be flushed in reverse order, so GPRs are flushed last.
Change-Id: Ice352a4df4ae064619c0f9905db634a7b57e4711
Signed-off-by: Evgeniy Naydanov <evgeniy.naydanov@syntacore.com>
In case in the future I have the same idea of optimizing progbuf writes
again.
Change-Id: Ie383487691cceeff75e2c22f4c85fc1fe4873937
Signed-off-by: Tim Newsome <tim@sifive.com>
Small cleanup in target.c to get rid of few upstream differences:
- removed a pointless check for `reg->exists` - already checked
few lines above
- unify one log message with what's in upstream
Change-Id: I3fd761157382670611fa90de84e2dfc90192f473
Signed-off-by: Jan Matyas <jan.matyas@codasip.com>
If hart can't change pc (e.g. it is running), resume command should
fail.
Signed-off-by: Evgeniy Naydanov <evgeniy.naydanov@syntacore.com>
Change-Id: I14627366d574d806ea16262b7d305d8161f8bcc2
Since the deletion of `-rtos hwthread`, there is no need to treat harts
with `-rtos` specified differently on reset.
Signed-off-by: Evgeniy Naydanov <evgeniy.naydanov@syntacore.com>
Change-Id: I88a9129936b5172bb7479dfa1255e29ff460c054
No other attempt is made at doing anything hypervisor-specific. Are
other things necessary?
Change-Id: Ib65f114888840cf0878f9bfe028c9a42b436aa3f
Signed-off-by: Tim Newsome <tim@sifive.com>
Icount triggers don't have a maskmax field at all. This is a cut and
paste error.
Change-Id: I001b3d41bf683599706dba713f7be475e8dd1668
Signed-off-by: Tim Newsome <tim@sifive.com>
When permission bits R, W, and X in PTE all three are zero,
the PTE is a pointter to the next level of the page table;
otherwise, it is a leaf PTE. Here PTE_W is missed.
Change-Id: I82a4cc4e64280f0fcad75b20e51b617520aff29b
Signed-off-by: panciyan <panciyan@eswincomputing.com>
Untested, because I don't have a target that implements this.
Change-Id: Iff82c124e7caf8e8960a9da62d8e727afb2c6b8a
Signed-off-by: Tim Newsome <tim@sifive.com>
It is possible for triggers of the same type to support different match
field values, so it is needed to try all the triggers, not just the
first one.
Fixes issue #788.
Signed-off-by: Evgeniy Naydanov evgeniy.naydanov@syntacore.com
Change-Id: I4c9fbc98bae7259377456d9ad8e770232724a592
We should avoid using x16~x31 register in program buffer because there
are no such general purpose registers in RVE(Embedded) extension.
For targets that support rvE, when the parameter increment=0
and count>1 of the read_memory_progbuf function, openocd will cause
an error due to the use of the s2 register.
For example:
{Command} {riscv repeat_read} count address [size=4]
Change-Id: I8b74dcc15cd00a400f2f1354c577a82132394435
Signed-off-by: Hang Xu <xuhang@eswincomputing.com>
* target/riscv: hide_csrs configuration option
This option allows users to mark certain CSRs as hidden so they could be
expluded from *reg* output and target.xml
Change-Id: Iddf8456cd3901f572f8590329ebba5229974d24a
* Update doc/openocd.texi
Co-authored-by: Jan Matyas <50193733+JanMatCodasip@users.noreply.github.com>
Signed-off-by: Anatoly Parshintsev <114445139+aap-sc@users.noreply.github.com>
* Update src/target/riscv/riscv.c
Co-authored-by: Jan Matyas <50193733+JanMatCodasip@users.noreply.github.com>
Signed-off-by: Anatoly Parshintsev <114445139+aap-sc@users.noreply.github.com>
---------
Signed-off-by: Anatoly Parshintsev <114445139+aap-sc@users.noreply.github.com>
Co-authored-by: Jan Matyas <50193733+JanMatCodasip@users.noreply.github.com>
* Add command "exec_progbuf"
Command "exec_progbuf" allows to execute a user-specified sequence
of instructions using the program buffer.
Change-Id: If3b9614129d0b6fcbc33fade29d3d60b35e52f98
Signed-off-by: Jan Matyas <jan.matyas@codasip.com>
* Updated the doc:
- Minor reword and reorder of the sentences.
- Added information about C-instructions in progbuf.
- Fixed a typo (per the review).
- Added examples.
Change-Id: I88c9a3ff3c6b60614be7eafd3a6f21be722a77b7
Signed-off-by: Jan Matyas <jan.matyas@codasip.com>
* Cosmetic changes
Change-Id: I7135c9f435f640e189c7d7922a2702814dfd595f
Signed-off-by: Jan Matyas <jan.matyas@codasip.com>
---------
Signed-off-by: Jan Matyas <jan.matyas@codasip.com>
Co-authored-by: Jan Matyas <jan.matyas@codasip.com>
Also refactor shared code for clearing itrigger/etrigger/icount.
Change-Id: Iac2e756332c89d2ed43435391e3c097abc825255
Signed-off-by: Tim Newsome <tim@sifive.com>
Because riscv_program_exec() tries to add an instruction every time
through.
This would cause an error accessing vector registers where VL > 14(?).
Change-Id: Ie676ca8c9be786b46aa2a4b4028ac8b27f7a4b40
Signed-off-by: Tim Newsome <tim@sifive.com>
This should make vector accesses work on 64-bit harts that implement
Zve32*. There doesn't appear to be any way to easily determine what vsew
values are allowed, so try and notice the failure.
Change-Id: Ide0722d0d67da402a4fbe88163830094e46beb84
Signed-off-by: Tim Newsome <tim@sifive.com>
OpenOCD currently uses improper "fence" instruction:
"FENCE" opcode with empty predecessor and successor sets.
Such instruction has no effect and is reserved for future use
as a HINT instruction (RISC-V Unprivileged ISA spec V20191213,
section 2.9).
This patch fixes it by using the proper "fence rw,rw"
instruction.
Change-Id: Ia2a66059009153efef27279410850ddfd73dae38
Signed-off-by: Jan Matyas <jan.matyas@codasip.com>
This generalizes the little endian CRC32 function used in the OR1K
target and moves it to a common helper, so that other places do not need
to reinvent the wheel. It is directly used in the OR1K target.
Change-Id: I0e55340281a5bfd80669bb1994f3a96fecc1248a
Signed-off-by: Marian Buschsieweke <marian.buschsieweke@ovgu.de>
Reviewed-on: https://review.openocd.org/c/openocd/+/7415
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
The org.gnu.gdb.arm.m-system GDB feature defines the name in lowercase
letters.[1] Not adhering to the definition can cause issues with tools
interacting with the GDB which expect the correct casing.
[1]: https://sourceware.org/gdb/onlinedocs/gdb/ARM-Features.html
Change-Id: I0b6584a78f86b053947d79686baad5dac3ec4a00
Signed-off-by: Boris-Chengbiao Zhou <bobo1239@web.de>
Reviewed-on: https://review.openocd.org/c/openocd/+/7292
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Cortex-M support for SMP multicore targets.
This SMP implementation unlike older ones does not act
on the first halted target found. It polls targets
until a SMP group is finished and stores eventual
'halted' events instead of emitting them. As soon as polling
of a group is done, poll proceeds with SMP related tasks.
This approach improves detection of a reason why debug
stopped - a correct reason is detected for all targets,
not only for the first found.
Drawback: SMP target group should be defined in the same
order as the targets were defined.
Obsolete gdb 'J' packet/smp_gdb command core switching is not implemented,
use with rtos hwthread.
Only one core is resumed if debug_execution is requested.
Some ideas taken from Graham Sanderson's [4936]
and src/target/aarch64.c
Added error checking of armv7m_restore_context().
Change-Id: I60f5b79e74b624dc2b5835ff10e38ac2ccb23792
Link: [4936]: target/cortex_m: Add smp support for Cortex M | https://review.openocd.org/c/openocd/+/4936
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/7239
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
When pointer authentication is enabled, some upper bits of the link
register (LR[63:VA_SIZE]) are used to store a signature. Therefore, GDB
need to remove the signature to get backtraces.
GDB has support of pointer authentication. When pointer authenticaion is
enabled, GDB requests 8-bytes mask to the target to remove the
signature. mask[63:VA_SIZE] should be all set and mask[VA_SIZE-1:0]
should be all cleared. GDB removes the signature by addr&~mask or
addr|mask.
I added a feature to provide the mask for pointer authentication.
Signed-off-by: Koudai Iwahori <koudai@google.com>
Change-Id: I56fbbf9cc23619b6536ecd326f350c8bf137f322
Reviewed-on: https://review.openocd.org/c/openocd/+/7248
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
In any case flag will be set as examined.
Signed-off-by: Erhan Kurubas <erhan.kurubas@espressif.com>
Change-Id: I5177ee656f726a807269e2f4725223f50e49e855
Reviewed-on: https://review.openocd.org/c/openocd/+/7231
Tested-by: jenkins
Reviewed-by: Ian Thompson <ianst@cadence.com>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
The target nds32 and its companion adapter aice have not received
any real improvement since 2013.
It has been hard to keep them aligned during the evolution of
OpenOCD code, with no way for maintainers to really check if they
are still working.
No real documentation is present for them in OpenOCD.
The nds32 code triggers ~50 errors/warnings with scan-build.
The arch nds32 has been dropped from Linux kernel v5.18-rc1.
For all the reasons above, this code has been deprecated with
commit 2e5df83de7 ("nds32: deprecate it, together with aice
adapter driver") and tagged to be dropped before v0.13.0.
Let it r.i.p. in OpenOCD git history.
While there, drop from checkpatch list the camelcase symbols that
where only used in this code.
Change-Id: Ide52a217f2228e9da2f1cc5036c48f3536f26952
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/7382
Tested-by: jenkins
This patch fixes two corner-cases of finding RTT control block.
The first one is when there was a partial match (even single byte) at
the end of loaded buffer (uint8_t buf[1024]), but this was not part of
full match. In that case `cb_offset` was not updated correctly and the
returned `*address` was lower by the legth of the partial match. In case
of searched 'SEGGER RTT' (the default control block ID) string, it was
enough to match `buf[1023] == 'S'`, which is quite likely to happen, and
the `*address` was offset by 1 (e.g. it was 0x20000fff instead of
0x20010000).
Updating (or even maintaining) `cb_offset` is not needed, as start
address of control block can be calculated based on memory address that
was loaded into `uint8_t buf[1024]`, the offset within this buffer and
the length of expected string.
The second issue is when control block is prepended with a byte that
matches first ID character, e.g. there is `SEGGER RTT` control block ID
is prepended by another `S`, making memory contents be `SSEGGER RTT`. In
that case there was no match found.
Fix that issue by making sure that tested byte is always compared with
first byte of expected control block ID.
While at it, change names of local variables to better describe their
meaning.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Change-Id: I12aa6e202bf12bedcbb888ab595751a2a2518a24
Reviewed-on: https://review.openocd.org/c/openocd/+/7429
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
This command is supposed to be a start at a compliance test for system
bus access. It doesn't pass against spike because it doesn't handle all
cases where the interface might be busy. It's not documented. As far as
I know nobody uses it.
So delete 400 lines of code instead of trying to fix it.
Change-Id: Ib94f2acb95a48f7c07d4f44206ff7373b03857f3
Signed-off-by: Tim Newsome <tim@sifive.com>
Signed-off-by: Tim Newsome <tim@sifive.com>
Tim Newsome
Marek Vrbka
Evgeniy Naydanov
Jan Matyas
Parshintsev Anatoly
Mark Zhuang
panciyan
Hang Xu
Anatoly Parshintsev
Jan Matyas
Marian Buschsieweke
Boris-Chengbiao Zhou
Antonio Borneo
Tomas Vanek
Koudai Iwahori
Erhan Kurubas
Marcin Niestroj