Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

calloc() memory per register. #658

Merged
merged 1 commit into from Nov 2, 2021
Merged

calloc() memory per register. #658

merged 1 commit into from Nov 2, 2021

Conversation

timsifive
Copy link
Collaborator

This replaces a static array with 8 bytes per register. When there are
vector registers larger than 8 bytes, they would end up clobbering each
other's values. I can't believe I didn't catch this earlier.

Change-Id: I9df4eaf05617a2c8df3140fff9fe53f61ab2b261
Signed-off-by: Tim Newsome tim@sifive.com

This replaces a static array with 8 bytes per register. When there are
vector registers larger than 8 bytes, they would end up clobbering each
other's values. I can't believe I didn't catch this earlier.

Change-Id: I9df4eaf05617a2c8df3140fff9fe53f61ab2b261
Signed-off-by: Tim Newsome <tim@sifive.com>
@timsifive timsifive merged commit 0ac6930 into riscv Nov 2, 2021
@timsifive timsifive deleted the big_regs branch November 2, 2021 16:55
ntfreak pushed a commit to openocd-org/openocd that referenced this pull request Dec 24, 2021
This replaces a static array with 8 bytes per register. When there are
vector registers larger than 8 bytes, they would end up clobbering each
other's values. I can't believe I didn't catch this earlier.

See riscv-collab/riscv-openocd#658

Change-Id: I9df4eaf05617a2c8df3140fff9fe53f61ab2b261
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6775
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
timsifive added a commit that referenced this pull request Dec 29, 2021
* jimtcl: update to version 0.81 (2021-11-28)

Change-Id: I374348047331b88e35b6584a5c65d8a73240b8a1
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6757
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>

* tcl: stm32mp13x: add target and board config files

The stm32mp13x has one core Cortex-A7.

The board Discovery Kit includes an on-board STLink-V3 with SWD
connection.
The webpage of the board is not active yet.

Change-Id: I8836b26612a160ead79766955ebefaf3d21a329c
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6675
Tested-by: jenkins
Reviewed-by: Tarek BOCHKATI <tarek.bouchkati@gmail.com>

* doc: mention SWD multidrop

Change-Id: I361dc351c407b67613eadef4d29e27ed0e00bd05
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6744
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins

* flash/nor/kinetis_ke: add .help fields for tcl commands

Add help texts from similar driver kinetis.c

While on it fix one existing help: the flash is obviously not NAND

Change-Id: Ibd295105586b008aaabf2fb4e4a75bf551266e38
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6727
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* flash/nor/kinetis_ke: remove 'kinetis mdm test_securing' cmd

The command might be a leftover from development of the driver.
There is no documentation what it does.

Change-Id: Iaa5aa1ac51638bd6acce172a5dd03846a165dc27
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6728
Tested-by: jenkins
Reviewed-by: Ivan-Artekit <ivan@artekit.eu>

* doc: document 'atsamv gpnvm' command

Change-Id: I922c75e2d074447558f85bfda5421aa24a850be1
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6729
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* flash/nor/at91samd: remove 'at91samd info' command

The command is a stub only, does nothing.

Change-Id: Ib3b8c2122a9f6f2e179bee34ac56d0adf367bfcc
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6730
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* doc: fix list of supported nRF52 devices

Change-Id: I64d6d851dd9333689556b7062b1ff38d41b5ad32
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6731
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* doc: document 'cmsis-dap cmd' command

Change-Id: I90b7f020ee2bf078fdf22e3d21423f333ae061ac
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6732
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* doc/openocd.texi: Document add_help_text and add_usage_text

Document add_help_text and add_usage_text under Genral Commands ->
Misc Commands.  The text are taken from command's help strings.

Change-Id: I2bbbee8e5faa8d9654227c304e0528f5144275e8
Signed-off-by: Yasushi SHOJI <yashi@spacecubics.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6734
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* doc/openocd.texi: Document usage command

Document usage command under General Commands -> Server Commands. The
text is taken from the command's help string.

Change-Id: I957cb46cca98181c3c5e676228c5c103e47bb655
Signed-off-by: Yasushi SHOJI <yashi@spacecubics.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6735
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* flash/nor/fespi: algorithm, large address, errors

* Move more smarts into the target algorithm code, and rewrite that in C
  so it's easier to understand/maintain.
* Support >24-bit addresses.
* Check for errors.

Change-Id: I3b1a143589fe6defafb8f95820aa682acc9646e7
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6679
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>

* target,flash: allow target_free_working_area on NULL area pointer

Standard C library free() allows NULL pointer as a parameter.

Change target_free_working_area() to conform this convention.

Remove NULL pointer tests before target_free_working_area() calls.

While on it add missing setting pointer to NULL after target_free_working_area().

Change-Id: I7c692ab04a9933398ba5bc614723ad0bdecb87b3
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6712
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* gdb_server: added and improved several debug prints

Added and improved several prints related to the GDB connection
and various error states that may occur in relation to this
connection.

Change-Id: I233246190b613cc925b783561cfa3aa5267360fd
Signed-off-by: Jan Matyas <matyas@codasip.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6288
Reviewed-by: Tim Newsome <tim@sifive.com>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins

* cortex_m: remove last references to debugport_init()

The function debugport_init() has never existed in OpenOCD code,
but few comments erroneously references it in place of the
existing function ahbap_debugport_init().

Commit 00dbc18 ("arm_adi_v5: Split ahbap_debugport_init")
splits the function ahbap_debugport_init() in dap_dp_init() and
mem_ap_init(), but did not removed all the incorrect comments
about debugport_init(). Few of such comments has been removed in
later patches.

Remove the last comment that references debugport_init().

Change-Id: Ibd1f125475386e5653340fedf706903a0ee15897
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6694
Tested-by: jenkins

* adi_v5_swd: add jtag-to-swd through dormant

ARM IHI 0031A does not support SWJ-DP, so no switch between JTAG
and SWD is considered.
ARM IHI 0031B is not publicly available and it's reported as
"Confidential Beta" in the history list in following doc versions.
From ARM IHI 0031C the direct switch between JTAG and SWD is
already deprecated in favor of passing through dormant mode. With
no access to IHI 0031B we haven't info if any device strictly
requires the direct switch.

OpenOCD implements only the deprecated direct switch, so changing
it could cause regression on devices that do not implement dormant
mode.
Plus, not all the adapters support dormant mode.

Nevertheless there are already target devices that only allow
entering in SWD by passing through dormant.

Let the code try both method, alternating one tentative with the
deprecated legacy direct switch, then another tentative passing
through dormant, and repeat till timeout.
This would work on any device that don't support dormant, on new
devices that require switch through dormant and will work with
adapters that don't support dormant.

Change-Id: Ib8619635277d497872079a33fa4e38be9beb84a0
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6695
Tested-by: jenkins

* doc: list internal commands called by init

Change-Id: I5534c94a983906533a1b38c30ec3eb982e62dee7
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6760
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* doc: document noinit command

Change-Id: I915c7d3231935fd44407b5fa65d86d4cb93e8db5
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6761
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* target/target: Check checksum_memory before call

Make sure checksum_memory is present.  Otherwise it'll segfault.

Change-Id: If31123323bd8a03282da43505c9604fde735ad0e
Signed-off-by: Yasushi SHOJI <yashi@spacecubics.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6758
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* doc/openocd.texi: Document command mode command

Document "command mode" command Server Configuration -> Configuration
Stage.  The text is taken from command's help string.

In addition to the help string, this commit explain the words
mismatch, the doc uses "stage" but the command and source code uses
"mode" to describe the same thing.

Change-Id: I1d5fc8c64a3a0b07ea8430bf016bcbd54e52da1c
Signed-off-by: Yasushi SHOJI <yashi@spacecubics.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6736
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* doc/openocd.texi: Document find and ocd_find command

Document both find and ocd_find command under Config File Guidelines
-> Interface Config Files.  find command is used in the previous
section as well but the previous section is more about using OpenOCD.

The section added is "aimed at any user who needs to write a config
file".

Change-Id: I698207fe58d564c615a15b50756aa5547a5f40b7
Signed-off-by: Yasushi SHOJI <yashi@spacecubics.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6737
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* doc: Update doc for commands "riscv expose_csrs" and "riscv expose_custom"

These commands were extended/improved in the last drop
of RISC-V target updates. Update also the documentation
to properly describe how the commands should be used now.

Change-Id: I9e2ba6adbe1a4c032b96f5f8ff2d4791fa4c2527
Signed-off-by: Jan Matyas <matyas@codasip.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6738
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-by: Tim Newsome <tim@sifive.com>

* doc: Updated RISC-V memory-related cmds in documentation

- "riscv set_prefer_sba" - removed from the doc, superseded
  by the latter
- "riscv set_mem_access" - new command, new entry added
  into the doc

This change only addresses the documentation. The corresponding
implementation of the RISC-V commands is already merged.

Change-Id: I3c07672cde94324407cf667504dba5402f63a543
Signed-off-by: Jan Matyas <matyas@codasip.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6743
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins

* Add Microchip SAME51 Curiosity Nano board

This board is (software wise) similar to the existing SAME54 Xplained
Pro board, with just a slightly different MCU. (Hardware wise, it
features a vastly different form factor.)

Signed-off-by: Jörg Wunsch <openocd@uriah.heep.sax.de>
Change-Id: I5e5435d49d333fb74471994ee84de59ed983153c
Reviewed-on: https://review.openocd.org/c/openocd/+/6747
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-by: Axel W.
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>

* linuxgpiod: Allow using multiple GPIO chips.

Allow passing optional gpiochip number before gpio number.
If no optional chip number is passed, the one from the 'gpiochip'
configuration directive is used.

Change-Id: I16933d81581d9af4d1600c5f9fdbc832ef3fda94
Signed-off-by: Laszlo Sitzer <dlsitzer@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6742
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* target/arm_jtag.h: fix wrong comparison in arm_jtag_set_instr

Change [1] introduced a regression that results in comparison
in arm_jtag_set_instr() to be always true if the length of the
IR register is not 8 bit. The value on the left side
of the != operator contains only tap->ir_length number of
bits while value on the right is full 8-bit instruction code.

This forces OpenOCD to update the JTAG IR register on each
transaction even if the instruction in the JTAG IR register
is correct. This causes noticeable performance degradation,
especially with slow JTAG adapters.

[1] https://review.openocd.org/c/openocd/+/6285

time ./src/openocd -s tcl/ -f interface/cmsis-dap.cfg \
  -c "transport select jtag" -f target/psoc6.cfg \
  -c "init; load_image data.bin 0x08000000; exit"

Without this change:
real    0m4,863s
user    0m0,074s
sys     0m0,128s

With this change:
real    0m3,083s
user    0m0,038s
sys     0m0,098s

Signed-off-by: Bohdan Tymkiv <bohdan200@gmail.com>
Change-Id: Iaded83a04ecc7e65f18256afae582267ccc1fc59
Reviewed-on: https://review.openocd.org/c/openocd/+/6762
Tested-by: jenkins
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* flash/nor/spi: add micron MT25QU01G

1Gbit SPI flash on VCU118 Rev. 2.0

Contributed to riscv-openocd in #487

Signed-off-by: Tim Newsome <tim@sifive.com>
Change-Id: I93447dd970d9901a671567fe8ab9e407432f8db9
Reviewed-on: https://review.openocd.org/c/openocd/+/6764
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins

* jtag/drivers/vsllink: fix memory leak

Close libusb context in vsllink_quit()

Change-Id: I85da8d7228b1b2b033a32b2f9ae9ed0726546b55
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6766
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* target/cortex_m: minor refactoring in cortex_m_store_core_reg_u32()

Unlike cortex_m_load_core_reg_u32() storing core register uses
the same code pattern around DHCSR read as offered by the convenience
helper cortex_m_read_dhcsr_atomic_sticky().

Use the helper.

Change-Id: Ia947204944a8b549f3c2be7fb2f717aad18970c4
SeeAlso: 65d7629 (cortex_m: poll S_REGRDY on register r/w)
SeeAlso: 0dcf95c (target/cortex_m: cumulate DHCSR sticky bits)
Signed-off-by: Tomas Vanek <vanekt@fbl.cz>
Reviewed-on: https://review.openocd.org/c/openocd/+/6767
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* openocd: add keep_alive during command sleep

The command sleep holds the host CPU until it completes.
Send keep_alive to GDB, so it will not timeout.

Change-Id: I92e9c5fc871b4e6a7695cdc449ca9fb3c1f1d9ec
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6770
Tested-by: jenkins
Reviewed-by: Jan Matyas <matyas@codasip.com>

* gdb_server: fix a comment in gdb_new_connection()

On 2008-03-05, before git's age, commit 6d95014 adds a
comment about unobserved ACK supposedly sent by GDB at connection.

The ACK is sent since GDB 3.95 (1999-05-04), but a bug introduced
in GDB 6.5 (2006-06-21) and fixed in GDB 7.0 (2009-10-06) makes
GDB sending the query for "supported packets" before sending the
ACK. Due to the bug, the author of the commit failed to see the
ACK.

Change-Id: I574a8013e7d159d1c71087af83b7c2ce92be86bd
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6769
Tested-by: jenkins

* semihosting: use macro COMMAND_HANDLER

We have the macro
	#define COMMAND_HANDLER(name) \
		static __COMMAND_HANDLER(name)
Use it!

Change-Id: I0e5385cb54197c743348f0d2ce215c93b8e396a4
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6786
Tested-by: jenkins

* target/riscv: calloc() memory per register.

This replaces a static array with 8 bytes per register. When there are
vector registers larger than 8 bytes, they would end up clobbering each
other's values. I can't believe I didn't catch this earlier.

See #658

Change-Id: I9df4eaf05617a2c8df3140fff9fe53f61ab2b261
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6775
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>

* contrib/udev: Add missing J-Link USB PID

See libjaylink/contrib/99-libjaylink.rules

Change-Id: I36733e18eb4bc11828c5fa563c687d28be9e34bd
Signed-off-by: Marc Schink <dev@zapb.de>
Reviewed-on: https://review.openocd.org/c/openocd/+/6778
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

* Document how vector registers are exposed to gdb.

See #570

Change-Id: Ie7cdef3717e107a9df0b48316cfbc547dea9a7fd
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6776
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>

Co-authored-by: Antonio Borneo <borneo.antonio@gmail.com>
Co-authored-by: Tomas Vanek <vanekt@fbl.cz>
Co-authored-by: Yasushi SHOJI <yashi@spacecubics.com>
Co-authored-by: Jan Matyas <matyas@codasip.com>
Co-authored-by: Joerg Wunsch <openocd@uriah.heep.sax.de>
Co-authored-by: Laszlo Sitzer <dlsitzer@gmail.com>
Co-authored-by: Bohdan Tymkiv <bohdan200@gmail.com>
Co-authored-by: Jaehoon Park <52432979+jaehoon-sifive@users.noreply.github.com>
Co-authored-by: Marc Schink <dev@zapb.de>
dnltz pushed a commit to aesc-silicon/elements-openocd that referenced this pull request Jan 5, 2022
This replaces a static array with 8 bytes per register. When there are
vector registers larger than 8 bytes, they would end up clobbering each
other's values. I can't believe I didn't catch this earlier.

See riscv-collab/riscv-openocd#658

Change-Id: I9df4eaf05617a2c8df3140fff9fe53f61ab2b261
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6775
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Shunichi-K pushed a commit to sonydevworld/spresense-openocd that referenced this pull request Mar 8, 2023
This replaces a static array with 8 bytes per register. When there are
vector registers larger than 8 bytes, they would end up clobbering each
other's values. I can't believe I didn't catch this earlier.

See riscv-collab/riscv-openocd#658

Change-Id: I9df4eaf05617a2c8df3140fff9fe53f61ab2b261
Signed-off-by: Tim Newsome <tim@sifive.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6775
Tested-by: jenkins
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-by: Tomas Vanek <vanekt@fbl.cz>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants