Merge up to commit '1293ddd65713d6551775b67169387622ada477c1' from upstream #816
Conversation
Signed-off-by: Erhan Kurubas <erhan.kurubas@espressif.com> Change-Id: I21506526e3ebd9c3a70a25ba60bf83aee431feb6 Reviewed-on: https://review.openocd.org/c/openocd/+/7016 Tested-by: jenkins Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
There is an accsess to wrong index, when arm semihosting_basedir command not used or basedir set to empty string. Signed-off-by: Erhan Kurubas <erhan.kurubas@espressif.com> Change-Id: I3afa049d74b30496f5c03ba4ef67431784f81bdc Fixes: ce5027a ("semihosting: add semihosting_basedir command") Reviewed-on: https://review.openocd.org/c/openocd/+/7005 Tested-by: jenkins Reviewed-by: Tarek BOCHKATI <tarek.bouchkati@gmail.com> Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
While an ADIv5 DAP can only have 256 AP, ADIv6 can provide till 2**40 (1,099,511,627,776) AP per DAP. The actual trivial code implementation for ADIv5 (that uses an array of 256 ap in the struct adiv5_dap) cannot be extended as-is to handle ADIv6. The simple array of 256 AP can be reused as a dynamic storage for ADIv6 ap: - the ADIv5 AP number is replaced by the ADIv6 base address; - the index of the array (equal to ADIv5 AP number) has no link to any ADIv6 property; - the ADIv6 base_address has to be searched in the array of AP. The 256 elements in the AP array should be enough for any device available today. In future it can be easily increased, if needed. To efficiently use the 256 elements in the AP array, the code should associate one element of the array to an ADIv6 AP (through the AP base address), then cancel the association when the AP is not anymore needed. This is important to avoid saturating the AP array while exploring the device through 'dap apreg' commands. Add a reference counter in the struct adiv5_ap to track how many times the struct has been associated with the same base address. Introduce the function dap_get_ap() to associate and return the struct, and dap_put_ap() to release the struct. For the moment the code covers ADIv5 only, so the association is through the index. Use the two functions above and dap_find_get_ap() throughout the code. Check the return value of dap_get_ap(). It is always not NULL in the current ADIv5-only implementation, but can be NULL for ADIv6 when there are no more available AP in the array. Instrument dap_queue_ap_read() and dap_queue_ap_write() to log an error message if the AP has reference counter zero, meaning that the AP has not been 'get' yet. This helps identifying AP used without get/put, e.g. code missed by this patch, or merged later. Instrument dap_cleanup_all() to log an error message if an AP has reference counter not zero at openocd exit, meaning that the AP has not been 'put' yet. Change-Id: I98316eb42b9f3d9c9bbbb6c73b1091b53f629092 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6455 Reviewed-by: Daniel Goehring <dgoehrin@os.amperecomputing.com> Tested-by: jenkins
Add flags to 'dap create' command and set the field adi_version in struct adiv5_dap. Actually only ADIv5 is functional. Other patches are needed to get ADIv6 working. Split from change https://review.openocd.org/6077/ Change-Id: I63d3902f99a7f139c15ee4e07c19eae9ed4534b9 Signed-off-by: Kevin Burke <kevinb@os.amperecomputing.com> Signed-off-by: Daniel Goehring <dgoehrin@os.amperecomputing.com> Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6456 Tested-by: jenkins
ADIv5 MEM-AP registers are a subset of ADIv6 MEM-AP registers and are located at different offset. To prepare for introducing ADIv6, add 'struct adiv5_dap *' as argument to ADIv5 registers macro. Check the ADI version and use the proper address. Both adapter drivers rshim and stlink are ADIv5 only, so let them use the ADIv5 macros only. Split from change https://review.openocd.org/6077/ Change-Id: Ib861ddcdab74637b2082cc9f2612dea0007d77b1 Signed-off-by: Kevin Burke <kevinb@os.amperecomputing.com> Signed-off-by: Daniel Goehring <dgoehrin@os.amperecomputing.com> Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6457 Tested-by: jenkins
Required for parsing ADIv6 ROM tables. Split from change https://review.openocd.org/6077/ Change-Id: I849543b7b4a4455b10bd9fc7da38a37849d71700 Signed-off-by: Kevin Burke <kevinb@os.amperecomputing.com> Signed-off-by: Daniel Goehring <dgoehrin@os.amperecomputing.com> Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6458 Tested-by: jenkins
By accessing invalid AP in JTAG mode, it's possible to trigger the error: JTAG-DP STICKY ERROR After that the sticky error is never cleared and the whole DAP gets not anymore accessible. Clean-up the sticky error once detected. Change-Id: I8b07263b30f9e46645f0c29084b8f1626e241f45 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6430 Tested-by: jenkins
swd and dap-direct are not implemented yet Split from change https://review.openocd.org/6077/ Change-Id: I6d73d8adf6a6090001c5d4771325fb1d63c45e3c Signed-off-by: Kevin Burke <kevinb@os.amperecomputing.com> Signed-off-by: Daniel Goehring <dgoehrin@os.amperecomputing.com> Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6459 Tested-by: jenkins
ADIv5 reports: Accessing AP registers or debug resources in connected device through an AP can be subjected to other variable response delays in the system. A debugger that can adapt to these delays and avoid wasting WAIT scans operates more efficiently and provides higher maximum data throughput. The existing code in OpenOCD uses extra tck only for accessing resources through an AP. Extend the use of extra tck also for accessing an AP register. Split from change https://review.openocd.org/6077/ Change-Id: I2082362e098d09f4ba0668e01f5196afc965c8f3 Signed-off-by: Kevin Burke <kevinb@os.amperecomputing.com> Signed-off-by: Daniel Goehring <dgoehrin@os.amperecomputing.com> Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6460 Tested-by: jenkins
During enter in SWD read DP_DPIDR without selecting the register bank through DP_SELECT_DPBANK. Handle the different format of DP_SELECT register. Change-Id: Iea1b8eb6ec94177e16a430d5885595a38e833eeb Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6697 Tested-by: jenkins
ADIv5 DAP can only have 256 AP, while ADIv6 can provide till 2**40 (1,099,511,627,776) AP per DAP. Reuse the field ap_num in struct adiv5_ap, currently used on ADIv5 to hold the ADIv5 AP number (apsel), to contain the ADIv6 AP base address. Convert struct adiv5_ap->ap_num to 64 bit and initialize it to DP_APSEL_INVALID for unused AP. Restrict dap_find_get_ap() to ADIv5 only. To be enhanced. On ADIv6, let dap_get_ap() return an already allocated AP, or allocate and return an unused AP. Add function is_ap_num_valid() and use it. Change-Id: Ib2fe8c7ec0d08393cd91c29fdac5d632dfc1e438 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6461 Reviewed-by: Daniel Goehring <dgoehrin@os.amperecomputing.com> Tested-by: jenkins
On ADIv6 the system root ROM table is found by reading the DAP DP registers BASEPTR0 and BASEPTR1. Add option 'root' to the commands 'dap info' to let it retrieve the system root ROM table's AP from DAP DP, then use such AP for following dump. Change-Id: I1789457a005faa3870c5d14f763378d2f6a5f095 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6462 Tested-by: jenkins
ADIv6 adds AP that only contain a ROM table in the AP itself, that can point to other AP containing either another AP level ROM table or a MEM-AP to be parsed as usual. To handle recursive AP access, reorganize the code to: - pass the depth==0 from the command 'dap info'; - print the AP number as first line, adding proper indentation on depth>0; - align the following print with proper indentation. Change-Id: I5b811810c807fc51b307bd60f67817d9de2aa095 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6466 Tested-by: jenkins
ADIv6 adds AP that only contain a ROM table in the AP itself, that can point to other AP containing either another AP level ROM table or a MEM-AP to be parsed as usual. Add support for parsing AP level ROM tables. Change-Id: Ic25863b16463b8a6adc3b15e26db7fdca858d6df Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6467 Tested-by: jenkins
Configuration file can specify, as target's debug AP, an AP that contains a ROM table that points, in turn, to other APs. Current code in cortex_a and aarch64 is not able to handle a return from dap_lookup_cs_component() that points to another AP. While it could be interesting to specify 'root' as target's debug AP, drop any found value if it's not in the starting AP. Change-Id: Id206e4fa7a29e9402c8e2393026817b410bbb8bd Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6826 Tested-by: jenkins
Arm "CoreSight System-on-Chip SoC-600" specification describes a bridge "Access Port v1 adapter" aimed to "connect a legacy Access Port (AP) ... into an CoreSight Architecture v3 system". A ROM table can be located in the "legacy" part of the system, on the legacy AP behind the APv1 adapter. For the purpose of scanning the ROM tables, consider an ADIv6 SoC-600 APv1 adapter as an ADIv5 AP. Change-Id: I97d42fb77013c1251fb68d0caa4274086bf38a70 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/6827 Tested-by: jenkins
Add Ampere Altra ("Quicksilver") and Ampere Altra Max ("Mystique")
target/board configuration files.
The target configuration file supports silicon and emulation.
The board configuration files support 1 and 2 socket platforms.
Tested on Ampere emulation and silicon
Change-Id: I036c798a50624e30ab51ccd2895b6f60c40be096
Signed-off-by: Daniel Goehring <dgoehrin@os.amperecomputing.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/5591
Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Tested-by: jenkins
ESP32 is a dual core Xtensa SoC Not full featured yet. Some of the missing functionality: -Semihosting -Flash breakpoints -Flash loader -Apptrace -FreeRTOS Signed-off-by: Erhan Kurubas <erhan.kurubas@espressif.com> Change-Id: I76fb184aa38ab9f4e30290c038b5ff8850060750 Reviewed-on: https://review.openocd.org/c/openocd/+/6989 Tested-by: jenkins Reviewed-by: Ian Thompson <ianst@cadence.com> Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
ESP32-S3 is a dual core Xtensa SoC Not full featured yet. Some of the missing functionality: -Semihosting -Flash breakpoints -Flash loader -Apptrace -FreeRTOS Signed-off-by: Erhan Kurubas <erhan.kurubas@espressif.com> Change-Id: I44e17088030c96a9be9809f6579a4f16dbfc5794 Reviewed-on: https://review.openocd.org/c/openocd/+/6990 Tested-by: jenkins Reviewed-by: Ian Thompson <ianst@cadence.com> Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Signed-off-by: Erhan Kurubas <erhan.kurubas@espressif.com> Change-Id: Id685408281478cec0e7e886dbedb3b8972c7b652 Reviewed-on: https://review.openocd.org/c/openocd/+/7020 Tested-by: jenkins Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-by: Ian Thompson <ianst@cadence.com>
Custom user syscalls can be handled with target events in the TCL scripts. This patch gives another opportunity to handle custom syscalls in the c files. Besides that some utility functions are also exported for the custom handlers. Signed-off-by: Erhan Kurubas <erhan.kurubas@espressif.com> Change-Id: Ice13d527540a0de0b2a8abda912ae3dcff3834b7 Reviewed-on: https://review.openocd.org/c/openocd/+/6889 Tested-by: jenkins Reviewed-by: Ian Thompson <ianst@cadence.com> Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
In order to facilitate debugging multiple cores, specify the coreid and the hwthread rtos in the imx8m target configuration. Change-Id: Ibd871517a160ceca15002fb10e27cb793f14d086 Signed-off-by: Alvin Šipraga <alsi@bang-olufsen.dk> Reviewed-on: https://review.openocd.org/c/openocd/+/7019 Tested-by: jenkins Reviewed-by: Antonio Borneo <borneo.antonio@gmail.com>
Commit 6c01516 ("aarch64: add support for "reset halt"") introduces the register setting to halt at reset vector, but: - does not consider the case 'srst_pulls_trst' that makes useless setting the registers as they will be erased by the pulled trst; - does not clean sticky errors in case of 'srst_gates_jtag'. Avoid any register initialization on 'srst_pulls_trst' and move the cleaning of sticky errors in the common block. Change-Id: I6f839f06f7b091e234ede31ec18096e51f017bcd Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Fixes: 6c01516 ("aarch64: add support for "reset halt"") Reviewed-on: https://review.openocd.org/c/openocd/+/7034 Tested-by: jenkins Reviewed-by: Christian Hoff <christian.hoff@advantest.com>
Commit 20adf85 ("linuxgpiod: add SWDIO buffer") introduces an additional gpio for SWDIO direction, but does not release it at driver's exit. Release the gpio at exit. Change-Id: If7ea31f79ffed04af585864e49bcf1f35e118bdd Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/7032 Tested-by: jenkins Reviewed-by: Steve Marple <stevemarple@googlemail.com>
We already have a helper to release the gpio. Extend it to also release its corresponding gpio chip. As side effect, remove comparison with NULL. Change-Id: I47cd446edfaead662d63c3330f25a791b747e882 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/7033 Tested-by: jenkins Reviewed-by: Steve Marple <stevemarple@googlemail.com>
Commit b9526f1 ("semihosting: permit redirection of semihosting I/O to TCP") introduces a new comparison with NULL. Remove it. Change-Id: Ice4333c50d16f7592f0ff86b1640217fa42e34f6 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Fixes: b9526f1 ("semihosting: permit redirection of semihosting I/O to TCP") Reviewed-on: https://review.openocd.org/c/openocd/+/7031 Tested-by: jenkins
Change-Id: I83d2477e8bc837aeac69bd5d08fdd923fd00a37c Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/7023 Tested-by: jenkins
Remove trailing '.' While there, add newline to file's last line. Change-Id: I3a727e406b572d051b28e17688c24627e55520c4 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/7024 Tested-by: jenkins
OpenOCD project is switching to SPDX tags. Replace the few FSF boilerplate in tcl folder. Change-Id: I15b146eb77cc491ed7355178f684f3e76fc763b4 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/7025 Tested-by: jenkins Reviewed-by: Tarek BOCHKATI <tarek.bouchkati@gmail.com>
The SPDX tag is aimed at machine handling and it's thus expected to be placed in the first line. Change-Id: I3992856eeb28b333c38d010ef286e22471ede263 Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com> Reviewed-on: https://review.openocd.org/c/openocd/+/7026 Tested-by: jenkins
Hi @timsifive - thanks for updating the upstream merge. Apologies if it's a dumb question, but what exactly needs to be reviewed? I presume not all ~ 200 commits that are part of this PR, but, rather, something more specific? |
@TommyMurphyTM1234 This is not a dumb question. I honestly don't really know. My experience has been that when people review code, even these crazy merges, sometimes they find bugs. So I stick with the process. :-) @JanMatCodasip I can certainly wait a few days for your review. |
|
Hi @timsifive - since my first question above wasn't dumb after all, I'll try again. :-D
Is this because upstream does not actually enforce checkpatch passing before accepting patches and (often?) accepts patches even if they fail the check? |
I have never dug into why upstream code does not satisfy checkpatch. All I know is that we're getting errors that would have been obvious to anyone using the tool, in code that was not modified in this branch. This is not a new thing. |
|
Several checkpatch rules were added just before 0.12.0-rc1 tag. Now errors coming because the new checkpatch runs against the old codes. e.g. I see a lot of xtensa related errors that were fine during the review stage. |
Thanks @erhankur - that's interesting and could explain some (or all?) of the discrepancies alright. |
|
Instead of disabling the checkpatch for the next PRs, I'd suggest getting the commit count in the branch and running git diff against just PRs commit. Something like; |
That is better than the current fixed count of 20, but does not actually solve the problem of checkpatch failing because of existing code. (I just tried the commands manually, and checkpatch still complained about all kinds of old code.) A few representative errors: And so on... There are 1694 of them. |
|
@timsifive Edit: I take this back, this was my oversight. Both these files were changed in the trunk ( |
@timsifive and others - do you think it's of any use to flag the issue on the OpenOCD development mailing list or via an OpenOCD SourceForge issue and ask if there might be any willingness to align historical patches or the codebase as a whole with the latest checkpatch requirements? If there is then I am happy to raise the issue there. And maybe to help upstream if necessary. |
|
I have quick-reviewed this merge visually and it looks all right. I noticed just a detail: In |
Change-Id: I7537c122d581ec1848a1e7902874506e0bbb6e31 Signed-off-by: Tim Newsome <tim@sifive.com>
|
Thanks, @JanMatCodasip. I moved the flash device entry. |
This is probably an improvement, and some people there seem to really enjoy this kind of work. (Look at e.g. replacing |
Hi @timsifive - sorry, I don't understand your reply. Are you expressing positivity or negativitity towards what I suggested? If the former then I will raise the issue upstream if you/others think it might be of any use in making future downstream <-> upstream merges easier. But if you/others think it's of no real use then I won't bother. |
|
I was a bit on the fence about it, because I thought it would only affect things when merging down. But now it's affecting every other PR (until we get enough changes merged that this is far in history again). So yes, this would be helpful. |
Thanks @timsifive - ok, I will raise the issue upstream via a SourceForge issue or on the openocd-devel mailing list to see what others there think. I wouldn't expect any quick action given that obvious rebuttals are (a) it could cause instability by changing lots of historical patches and already committed code and (b) why are you doing development in a fork and not upstream in the master project. But let's see what happens.... |
I've posted here about this so let's see what people have to say about it. |
Hi @timsifive As mentioned by @AntonioBorneo here: would adding (via some script/automation?) |
* rtos/FreeRTOS: pxCurrentTCB should be used for judgment. The current TCB is stored in pxCurrentTCB, which is somehow RISC-V-specific, should not be overwritten from upstream (#816). * fix the code style check. Signed-off-by: Chao Du <duchao@eswincomputing.com> Change-Id: I9ffa8947f0cb9e93c7d96866882a5a1e8e69afad * revert some over-changes in last commit. Change-Id: Ie88bd75b59190503db11ee4538281bd13b554e50 Signed-off-by: Chao Du <duchao@eswincomputing.com> --------- Signed-off-by: Chao Du <duchao@eswincomputing.com>
This includes https://sourceforge.net/p/openocd/mailman/message/37710818/, which should fix #814.
No regressions against spike, HiFive Unleashed, and HiFive1.