Support for RISC-V RNMI registers

[kubeqz]

Do you plan to add support for "Smrnmi" RISC-V extension (https://docs.riscv.org/reference/isa/priv/rnmi.html) ?

I'm using latest release and when trying to read 'mnstatus' it just fails with

> reg mnstatus
register mnstatus not found in current target

[en-sc]

@kubeqz, this repository is stale, please, consider raising the issue in mainline OpenOCD (https://openocd.org/pages/bug-tracker.html).

In any case, supporting these registers in OpenOCD is two-step process:

1. Update src/target/riscv/encoding.h (auto-generated from https://github.com/riscv/riscv-opcodes.
2. [Optional] Add discovery mechanism. AFAIU, it can be similar to interrupts M-Mode CSRs:
   https://github.com/riscv-collab/riscv-openocd/blob/eb01c632a4bb1c07d2bddb008d6987c809f1c496/src/target/riscv/riscv_reg.c#L493C7-L500
   

   riscv-openocd/src/target/riscv/riscv-013_reg.c

   Lines 267 to 290 in eb01c63

   	static int examine_mtopi(struct target *target)
   	{
   	/* Assume the registers exist */
   	int res = assume_reg_exist(target, GDB_REGNO_MTOPI);
   	if (res != ERROR_OK)
   	return res;
   	res = assume_reg_exist(target, GDB_REGNO_MTOPEI);
   	if (res != ERROR_OK)
   	return res;
   	riscv_reg_t value;
   	if (riscv_reg_get(target, &value, GDB_REGNO_MTOPI) != ERROR_OK) {
   	res = riscv_reg_impl_set_exist(target, GDB_REGNO_MTOPI, false);
   	if (res != ERROR_OK)
   	return res;
   	return riscv_reg_impl_set_exist(target, GDB_REGNO_MTOPEI, false);
   	}
   	if (riscv_reg_get(target, &value, GDB_REGNO_MTOPEI) != ERROR_OK) {
   	LOG_TARGET_INFO(target, "S?aia detected without IMSIC");
   	return riscv_reg_impl_set_exist(target, GDB_REGNO_MTOPEI, false);
   	}
   	LOG_TARGET_INFO(target, "S?aia detected with IMSIC");
   	return ERROR_OK;
   	}

Please, consider submitting the patch to mainline OpenOCD as described in https://review.openocd.org/Documentation/user-upload.html

[TommyMurphyTM1234]

this repository is stale, please, consider raising the issue in mainline OpenOCD (https://openocd.org/pages/bug-tracker.html).

Please, consider submitting the patch to mainline OpenOCD as described in https://review.openocd.org/Documentation/user-upload.html

@en-sc, does that mean that further RISC-V development is now done upstream in the master OpenOCD project and not here in this repo?

If so then are there any plans to deprecate/archive this repo to avoid confusion?

[en-sc]

@en-sc, does that mean that further RISC-V development is now done upstream in the master OpenOCD project and not here in this repo?

Recently there was a sync with the mainline OpenOCD that covered most of the differences between the repos (see the updates in #979). IMHO, it's better to contribute RISC-V-related changes to mainline and sync the changes back into this repo if needed.

If so then are there any plans to deprecate/archive this repo to avoid confusion?

I can't do that, since my write permissions were taken away.
In any case, IMHO there are still things to do before archiving this repo:

• Finish the sync (e.g. the are still RTOS-related changes that are unsynced). At least these patches should be uploaded to mainline's Gerrit in order not to lose them.
• Comb the issues once more and transfer relevant ones into mainline's bug tracker.

[TommyMurphyTM1234]

Thanks for the explanation @en-sc - I wasn't aware of all that and it certainly clarifies matters.