implement Debug for readable fields of registers

[liebman]

contributes towards #48

Notes:

• I've only run this with esp (esp32/esp32s3) pacs
• I'd love some feedback on where I've implemented this and if there is a cleaner way

[burrbull]

I've missed that you targeting esp32. Do I understand right that you don't have other way to debug except this?
IDE extensions like probe-rs or Cortex-m-Debug for VSCode?

[liebman]

Cortex debug can work easily for newer esp32 family chips (esp32s3, esp32c3, etc) that have builtin usb - for those without builtin usb its only available if the JTAG pins are available.

One specific use case here is that many of the peripherals (for me this is RTC_CTRL power management) are not well documented. And since there is a std hal for esp32 that is built on esp-idf one thing I can do is dump peripherals registers on the std hal to help with implementations of the no-std hal by being able to compare what registers are set differently. After manually building this for esp32 I was starting on esp32s3 and realized this would be much better placed in svd2rust and could be automatically generated - thats when I came across #48.

[liebman]

What I really want is to be able to dump all registers from the peripheral level as registers (and fields) differ a fair amount between esp32 chip types. (but at least having the registers handled automatically is a huge help.

[burrbull]

What I really want is to be able to dump all registers from the peripheral level as registers (and fields) differ a fair amount between esp32 chip types. (but at least having the registers handled automatically is a huge help.

If you want to read a range of registers, you should also decide what to do with registers which have read side effects (see readAction).

[burrbull]

Don't do merge master. Use rebase instread. We want to have linear commit history.

[liebman]

I'm already skipping registers that have read side effects. Currently not even generating the Debug implementation but maybe I should and just print the register name and no value or something short indicating it was not read.

[liebman]

Don't do merge master. Use rebase instread. We want to have linear commit history.

How can I fix this?

[liebman]

I've added debug at the RegisterBlock level so one can print all registers belonging to a peripheral with:

    info!("SHA {:#?}", unsafe { &*esp::SHA::ptr() });

Write only registers are printed with a note that they are read only and registers that have read actions are also printed with a message. All readable registers have their fields printed in struct format and if there are no fields then the raw register value is printed. Register clusters "should" also be handled.

[burrbull]

Looks good for me.
@Emilgardis ?

[burrbull]

@liebman Could you how some usage example output?

[Emilgardis]

this is good as is, just some improvements on naming and suggestion for the future

some examples would be good also to add, and a mention in the documentation wouldn't hurt

[Emilgardis]

I'd like it if this field took a enum instead, which would make it possible to use defmt instead in the future.

Ofcourse, this can be added later with a breaking change

[liebman]

that or add another bool option for impl_defmt allowing both to be specified.

[liebman]

@burrbull Here is a (truncated) sample run on an esp32s3 using println!("RTC_CNT {:#?}", unsafe { &*esp32s3::RTC_CNTL::ptr() });

RTC_CNT RegisterBlock {
    options0: OPTIONS0 {
        sw_stall_appcpu_c0: 0,
        sw_stall_procpu_c0: 0,
        bb_i2c_force_pd: false,
        bb_i2c_force_pu: false,
        bbpll_i2c_force_pd: false,
        bbpll_i2c_force_pu: false,
        bbpll_force_pd: false,
        bbpll_force_pu: false,
        xtl_force_pd: false,
        xtl_force_pu: false,
        xtl_en_wait: 2,
        xtl_force_iso: false,
        pll_force_iso: false,
        analog_force_iso: false,
        xtl_force_noiso: true,
        pll_force_noiso: true,
        analog_force_noiso: true,
        dg_wrap_force_rst: false,
        dg_wrap_force_norst: false,
    },
    slp_timer0: SLP_TIMER0 {
        slp_val_lo: 0,
    },
    slp_timer1: SLP_TIMER1 {
        slp_val_hi: 0,
    },
    time_update: TIME_UPDATE {
        timer_sys_stall: false,
        timer_xtl_off: false,
        timer_sys_rst: false,
    },
    time_low0: TIME_LOW0 {
        timer_value0_low: 46911,
    },
    time_high0: TIME_HIGH0 {
        timer_value0_high: 0,
    },
    state0: STATE0 {
        apb2rtc_bridge_sel: false,
        sdio_active_ind: false,
        slp_wakeup: true,
        slp_reject: false,
        sleep_en: false,
    },
<snip>

[liebman]

@Emilgardis I like the naming suggestions and will try to update examples and documentation.

[Emilgardis]

awesome! thank you :)

I think we can merge this, I'll leave the final r+ for @burrbull

[Emilgardis]

Might be good to squash some of the commits btw!

[liebman]

squashed

[Emilgardis]

bors r=emilgardis,burrbull

thank you!

[bors]

Build succeeded!

The publicly hosted instance of bors-ng is deprecated and will go away soon.

If you want to self-host your own instance, instructions are here.
For more help, visit the forum.

If you want to switch to GitHub's built-in merge queue, visit their help page.

• CI
• Rustfmt

[liebman]

Is there a date planned for the next release?