Activity since 2025-02-01

• esp-rs/esp-hal repo issues and pull requests

  • Allow `set_power_saving` even for coex #3081 [bjoernQ]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • jessebraham: Agree with @bugadani regarding the migration guide, but otherwise LGTM

    • bugadani

  • Enable async ADC for C2 and H2 #3082 [bugadani]

    They just work. Verified by running the example, and tying GPIO4 to GND (~2000 count) and 3V3 (4095 count). Also ran...

    • davoclavo: LGTM. Good catch on disabling interrupts only when all async adcs are released 👍🏼

    • bjoernQ: maybe I'm missing something here but why only for ESP32-C3?

      • bugadani: I think that's the only chip with 2 ADCs, for the rest we don't need to count to 1

      • bjoernQ: the only RISCV - true Mind adding a comment since it's (at least to me) not obvious

      • bugadani: I think I'd rather add an adc1 metadata symbol instead

      • bjoernQ: that's for sure much better 👍

    • jessebraham

    • bjoernQ: I guess this should be Cpu::all

    • bjoernQ: Thanks

  • Does esp-hal-procmacros need to be stable also? #3078 [MabezDev] RFC package:esp-hal-procmacros 1.0-blocker

    As part of not having dependencies <1.0 in the public API. I assumed this was a yes and was going to open a separate...

    • bjoernQ: Yes looks like this for me

    • bugadani: We probably want to make it clear that we don't intend to stabilize procmacros and directly...

  • ci: Test Xtensa action #3087 [SergioGasquez]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

  • DMA channel priority configuration #3089 [bugadani] peripheral:dma

    Somewhere in the refactors this has gone away - #3088 removed the last vestiges. We should figure out how we can...

  • DmaChannel::set_priority doesn't work anymore #3085 [Dominaezzz] bug peripheral:dma

    Bug description Since #2545 landed, it's no longer possible to set the priority on DMA channel, due to the lazy init...

  • Remove `DmaChannel::set_priority` #3088 [bugadani]

    Closes #3085 The intended way is to configure either the dma-enabled peripheral, or the transfer itself to run at a...

    • jessebraham

    • bjoernQ: The intended way is to configure either the dma-enabled peripheral, or the transfer itself to...

    • bjoernQ: Definitely fixes the linked bug

    • MabezDev: As @bjoernQ mentioned, please open an issue about dma prio. LGTM.

    • bugadani: #3089

  • Use self-hosted mac mini for the small, auxiliary tasks #3072 [bugadani]

    Total CPU time goes down by 13 or so minutes (https://github.com/esp-rs/esp-hal/actions/runs/13074331617/usage,...

    • SergioGasquez: LGTM, but left 2 questions

      • SergioGasquez: It was not worth to use the macos self hosted runner here? Also, we can already use...

        • bugadani: We can certainly try an ARM runner, but the difference will be seconds. This workflow isn't a...

        • bugadani: Well that didn't work out :) I can try fiddling with non-cross builds but maybe not in this one.

        • SergioGasquez: Thanks for the tests!

    • jessebraham

    • JurajSadel: Thanks!

  • Create an issue when nightly-ci fails #3080 [JurajSadel] skip-changelog

    closes #3068 I've tested this with another repo, works fine there, hopefully I haven't missed anything. I'm...

    • bjoernQ: I imagined the created issue could contain a bit more useful information about the problems or...

    • JurajSadel: I've changed the approach a bit here - instead of an action (I couldn't linked the failed job in...

    • bjoernQ: I've changed the approach a bit here - instead of an action (I couldn't linked the failed job in...

    • JurajSadel: I've changed the approach a bit here - instead of an action (I couldn't linked the failed job in...

    • bjoernQ: LGTM

    • MabezDev: Thank you!

  • Issue creation when nightly runs fail #3068 [MabezDev] CI

    It's hard to see when this goes wrong, and we should do a better job of keeping on top of it.

  • Migrate PARL_IO driver to newer DMA API #3033 [Dominaezzz]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • liebman: On error this eats self and the BUF - it should return a tuple right?

      • Dominaezzz: Ahh you're right. I'll update that

    • liebman: Looking at the failed tests, maybe one of the higher bits is connected to bit1 which would...

    • Dominaezzz: It does transition after number 32. The parl io is in 4 bit mode so each pin is responsible for...

    • Dominaezzz: With this commit, the weirdness goes away locally, I can use any pin just fine now.

    • liebman: that makes sense, you don't want your read sampling to be on the same edge that data is changing.

    • bugadani: Thanks

    • bjoernQ: seems like reformatting went bad

    • bjoernQ: seems like reformatting went bad

    • bjoernQ: Thanks

  • Allow switching between AP, STA etc at runtime #2224 [timovelten] package:esp-wifi

    I would like to initialize the Wifi interface in ApSta mode and later disable the access point (and switch to Sta...

    • ivmarkov: @bjoernQ I think in terms of flexibility, your API perhaps goes a tad too far? What I don't like...

    • bjoernQ: Thanks for taking time to look into this and for the feedback! Seems like this goes into a...

  • Migrate AES driver to DMA move API #3084 [Dominaezzz]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • bjoernQ: LGTM

    • bugadani: Thank you

  • `dma_tx_buffer` no longer returns an error #3086 [Dominaezzz]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • bugadani: IMHO we shouldn't try to fix this macro, but remove it completely.

    • Dominaezzz: Yeah probably, are you thinking users should use the dma_buffers macro and create DmaTxBuf...

    • bjoernQ: IMHO we shouldn't try to fix this macro, but remove it completely. I agree - the macros served...

    • Dominaezzz: Can't tell what the replacement is but looks like this pr isn't the desired direction.

  • I2C master: Read data does not match actually present on bus. #3034 [katyo] bug peripheral:i2c beta-blocker

    Bug description The data in read buffer does not match data which slave writes to the bus. For example it reads 0x83...

    • MabezDev: @katyo could you try the linked PR when you get the chance?

  • A time/rate solution #2923 [MabezDev] api beta-blocker

    We currently use fugit, which has served us well, and works nicely overall. I do see a few issues with it...

  • Newtype fugit Rate, Instant and Duration #3083 [bugadani] skip-changelog

    Closes #2923 skip-changelog because of embassy/wifi updates.

    • MabezDev

      • MabezDev: We can implement it on our own Rate struct now

      • MabezDev: I wonder if we're going with our own types we should also make Instant::now a thing instead of...

        • bugadani: Instant is meaningless without the HAL, you can't create an Instant anyway. That the underlying...

        • MabezDev: So I'm not sure if we want to move these out of the hal, or what we would gain by doing that I...

        • bugadani: We can always add an embassy_time_04 feature (if #3063 goes that way) to esp-hal.

    • bjoernQ: We probably could expand this comment a bit now

    • bjoernQ: Thanks! We could consider a From<core::time::Duration> impl but not necessarily in this PR (it's...

    • MabezDev: This LGTM, thanks for taking care of the impl @bugadani! The only thing left to do now is...

      • MabezDev: So as per the parent issue, let's bikeshed this a bit. Imo this still looks the cleanest, and...

        • bjoernQ: +1 for mhz

        • bugadani: mghz is just confusing, mega_hz or mega_hertz look sloppy as I think the prefixes are supposed...

        • MabezDev: I think the guideline suggestion has resolved the concerns here for me, I suppose I should have...

    • Ben-PH: fugit forces the exclusive choice of u32 or u64. this creates frustrations when wanting to...

    • MabezDev: fugit forces the exclusive choice of u32 or u64. this creates frustrations when wanting to...

    • bugadani: Awhile ago, I ran into this issue, and it's a pain. What's your use case?

    • Ben-PH: Use case was ages ago, so I forget the details, but the key issues is that there were...

    • JurajSadel: Thanks!

    • MabezDev: Noted. What do you plan to do for the cases where the types chosen for the const generics cause...

    • MabezDev

  • esp32c6 really bad WiFi range compared to C program #3075 [jpdisco] bug package:esp-wifi

    Bug description When I make program with C in arduino editor for seeed lab xiao esp32c6 wifi works very well...

    • ivmarkov: Could it be related to the problem described in #3081 ? @jpdisco In other words, a quick...

    • jpdisco: I have already set powersave to none, it only reduced latency not range problems. When I have...

  • ESP32-S2: Fix creation of .rwdata_dummy section #3096 [bjoernQ]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • jessebraham: Thanks!

    • bugadani

  • PCNT clock may be deactivated after migration to 0.23 #3094 [DamienEspitallier] bug

    Bug description 0.23 has changed the clock enable algorithm to use GenericPeripheralGuard. But it does not work when...

    • Dominaezzz: AHH, interesting use case. Each object basically needs a guard field in them

  • Fix BLE controller test #3097 [bugadani]

    On the H2, the connector sometimes returns two packets, which the previous test did not expect. This PR breaks up...

    • bjoernQ: Thanks

    • JurajSadel: Thanks!

  • Don't run CI on main #3066 [MabezDev] CI

    We already do all the merge checks in the merge queue.

    • jessebraham: I don't think this is an issue anymore, it seems GitHub have made some changes on their end. The...

    • MabezDev: It's nice to not have to do anything to close an issue every once in while :D

  • Add xtal frequency configuration #3054 [bugadani]

    This PR adds ESP_HAL_CONFIG_XTAL_FREQUENCY - although only 32 and C2 have meaningful configurability. THe point of...

    • MabezDev: Are we going to run into issues if someone wants to set this config but also support multiple...

      • bjoernQ: I guess #3020 will support this scenario better

    • bjoernQ: shouldn't this be unreachable!?

      • bugadani: it can't be, const eval just ends up panicking if you configure anything else

      • bjoernQ: ah somehow missed const

    • bjoernQ: LGTM

    • jessebraham

    • JurajSadel

  • HCI: don't take a critical section for each byte read #3098 [bugadani] skip-changelog

    • MabezDev

    • bjoernQ: Thanks

  • Fix waiting for short periods #3093 [bugadani]

    This PR (built on top of #3083) shortens wait times in tests (which trigger this issue), and sets the interrupt...

    • bugadani: Okay I'll have to investigate more, whether we should use level_int_en instead of the int_ena...

    • Dominaezzz: Clearing the interrupt here breaks cancellation I think, you ideally want to be able to...

      • Dominaezzz: Ah nvm, I don't think the cancellation semantics where ever decided on the timers.

      • bugadani: Yeah this future isn't cancellation-safe. Neither does it have to be, either.

      • bugadani: Still, I removed this clear in the last commit, where I also rework the async wait a bit.

    • bjoernQ: maybe we want to keep and extend the original comment (// always use level interrupt) - i.e. why...

    • bjoernQ: Thanks

    • MabezDev: LGTM, thanks!

  • `Pull` doesn't support pull up and pull down at the same time #3095 [MabezDev] peripheral:gpio 1.0 non-breaking

    Unsure if we should even support this but it's a real mode. The context I found for this is...

    • bugadani: we can add this to flex, I guess, but otherwise I wouldn't try to confuse anyone in the mainline...

    • MabezDev: Agreed.

  • Radio clock refactor #3100 [bugadani] skip-changelog

    This PR removes a trait in favour of a driver struct, so that the RADIO_CLK peripheral no longer implements logic....

    • bugadani: How on Earth did this turn into +800 -1200 lines 😶

    • bugadani: I wanted to merge RadioPeripherals into Peripheral so that we don't special-case them so much,...

    • MabezDev: LGTM, I would like a bit of testing around the phy clock change, just so we don't break esp-wifi with it

      • MabezDev: I'd be interested to see if this breaks things, I'm not sure we ever got to a point where we...

        • bugadani: 32 and S2 have this: pub(crate) unsafe extern "C" fn phy_common_clock_disable() { ...

    • bjoernQ: I did some testing and everything seems to work (also the changes are mostly "mechanical")

  • esp32s3 with wifi boot loop, `BROWNOUT_RST` #3102 [kaspar030] bug status:needs-attention

    Bug description I have multiple esp32-s3 devkits that seem to reboot with BROWNOUT_RST. I'm pretty sure they worked...

    • bugadani: Brownout reset means something's up with your devkits' power supply.

    • kaspar030: Brownout reset means something's up with your devkits' power supply. I fell for the USB hub not...

  • I2C slave driver #1909 [okhsunrog] peripheral:i2c

    Are there any plans for I2C slave driver? I only found embassy example for I2C slave for rp2040...

    • hcabel: I agree, +1 for i2c slave driver

  • 2Mbps UART crashes on initialization #3103 [TravisThomp] bug

    Bug description Setting the baud rate to 2Mbps causes the change_baud function to overflow during...

    • MabezDev: Could you try #3104?

    • TravisThomp: No longer getting an overflow error, Thank you :)

  • Allow more baud rates #3104 [MabezDev]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • MabezDev: Cool, I didn't completely break things, time to try and fix the real issue :D.

    • bugadani: 👍

    • bugadani: Could we just divide twice? I'm not a fan of u64 maths

    • bugadani: TYVM! Let's wait for feedback, but otherwise looks good to me.

    • jessebraham: Nitpick, I know it was already present but probably we don't need to keep this commented-out line?

      • bjoernQ: or maybe at some point (not here) we should do it - there are quite a few such delays in...

      • MabezDev: Thanks, I reworded the comment to something more useful instead of just commented out code.

  • More validation and `# Errors` entries #3074 [playfulFence]

    Introduces more validation for SPI and UART, allows setting baudrate tolerance for UART config, adds more # Errors...

    • MabezDev: Not sure if you're looking for review feedback now, but I guess you're gunna get it :D. In...

      • MabezDev: Shouldn't we specify which variants can be returned?

      • MabezDev: I think we should probably mark this as unstable for now, but Closest seems like a good default to me.

        • MabezDev: and also do the same for the other tolerance related type in the other drivers.

    • bjoernQ: this comment is not accurate for the H2 - probably there is also no need to mention the limits...

    • bjoernQ: no need for the leading space

    • bjoernQ: not to be resolved in this PR but I think in some cases passing zero-length buffers are no-ops...

    • bjoernQ: we should tell the real limit (which is different for H2) - we already do it in other places

    • bjoernQ: maybe it's worth to deduplicate the code block (we have it three times)

    • jessebraham: Suggested change ...

    • jessebraham: There are a number of identical lines in each branch of the cfg_if! here, probably could use...

    • bjoernQ: LGTM

    • SergioGasquez: Some formatting nitpicks

      • SergioGasquez: Suggested change /// # Errors ...

      • SergioGasquez: Suggested change /// # Errors. ...

      • SergioGasquez: Suggested change /// #...

      • SergioGasquez: Suggested change /// #...

      • SergioGasquez: Suggested change /// # Errors ...

      • SergioGasquez: Suggested change /// # Errors ...

      • SergioGasquez: Suggested change /// # Errors. ...

      • SergioGasquez: We should use a commit instead of master for the link

        • jessebraham: Yeah linking to specific lines without a commit hash is just a recipe for future confusion

        • playfulFence: Agreed

    • SergioGasquez

    • MabezDev: Thank you! This is looking good, most of my comments are related to the wording or layout, the...

      • MabezDev: Can we move this check to the validate function? It's fine to stay as a panic, but we should try...

      • MabezDev: I think this could be simplified a bit if you wrote it more...

        • MabezDev: Oh and please remove the period in the # Errors heading.

      • MabezDev: You might be able to collapse these two branches once you rebase, see the logic here:...

        • playfulFence: Good catch, but I think in this specific case it will actually only add new lines because extra...

        • MabezDev: yeah, ignore this one I think, it's cleaner to dupe here

      • MabezDev: I think these can be i32 and u32?

        • playfulFence: Can they? What if baudrate or actual_baud is greater than i32::MAX? Wouldn't it overflow?

        • MabezDev: i32::MAX is over 2 billion, our max supported baud is 5 million, even if a new chip comes out...

        • bugadani: 2GHz UART wen

      • MabezDev: I forget if we can mark enum variants as unstable (cc: @bugadani) but as this is related to the...

        • bugadani: It's impossible to mark enum variants as unstable. All we can do is cfg them away and return an...

        • MabezDev: Gotcha so we can do what we do for spi dma error:...

      • MabezDev: Similar to my other comment, I'd avoid writing the conditions in long sentences when an error...

      • MabezDev: I can't comment on the diff, but could you add some more tests to...

    • MabezDev: LGTM, thanks!

  • Peripheral drivers should validate configuration before applying it to hardware #2839 [bugadani] bug beta-blocker

    apply_config is fallible, which means two things: It can fail The user can ignore the failure and keep using the...

  • Calibration calc overflow #3090 [yuri-ncs]

    Submission Checklist 📝 I have updated existing examples or added new ones (if applicable). I have used cargo...

    • bjoernQ: nitpick but probably we should revert the Cargo.toml reformat

    • yuri-ncs: Sorry, just trying to solve the problem i was having and its also my first cotribution in a OS...

    • bjoernQ: Thanks for working on this! One thing I noticed is that the adc_calibaration qa-test is async...

    • yuri-ncs: i removed the async now. I was using Timer to delay and in the las minute changed for delay.

    • MabezDev: HIL test gets run with every PR, qa test only gets looked at around release time or some...

    • yuri-ncs: The test conversion is feasible, but regarding the reference issue, I need to study more about...

    • MabezDev: The test conversion is feasible, but regarding the reference issue, I need to study more about...

  • Remove note about fixed cargo bug #3105 [bugadani] skip-changelog

    rust-lang/cargo#10358 has been fixed, we probably no longer want to call it out. The fix will be part of 1.86, so we...

    • MabezDev: Should we also remove the work around we did in esp-config?

    • bjoernQ: I was too slow .... yes should we remove it: esp-hal/esp-config/src/generate.rs ...

  • Update PACs to use newly defined radio bits #3101 [bugadani] skip-changelog

    This PR rewrites the regi2c_write functions of C6 and H2, by using the I2C_ANA_MST peripheral defined in the...

    • MabezDev: LGTM, thanks! I merged esp-rs/esp-pacs#316 not sure if you need to change the pacs commit or not.

    • bjoernQ: Did some testing and seems to work fine

  • Specifying `TransferDirection` when creating `Preparation`s in `Dma[Tx|Rx]Buffer`s feels redundant. #3109 [Dominaezzz] peripheral:dma

    This is just to continue the conversation from here where we were discussing how to best to deal with the redundancy...

  • ESP32S2: `dma_mem2mem` support #2313 [SergioGasquez] chip:esp32s2

    Issue created from #1524

    • Dominaezzz: esp-rs/esp-pacs#312

    • Dominaezzz: esp-rs/esp-pacs#318

  • fix: ethernet frame header is 14 not 18 #3025 [Easyoakland] skip-changelog

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • bjoernQ: I think we just size our internal buffer larger than needed and we don't even need to account...

  • refactor esp-wifi to ease using external task scheduler #3106 [kaspar030] skip-changelog

    Submission Checklist 📝 I have updated existing examples or added new ones (if applicable). I have used cargo...

    • kaspar030: this function was identical for riscv and xtensa. the comments are a mix. I don't know which...

      • bugadani: Nice, this change will probably make it simpler to just throw all this out in favour of...

    • bjoernQ: Very nice! Lots of things I wanted to do for quite some time - from my side this looks already mergeable

    • bugadani: Hmm don't we have #[cfg(xtensa)] and #[cfg(riscv)] in esp-wifi?

      • bjoernQ: ah we use esp-metadata - so yes

      • kaspar030: I [pushed a...

    • bugadani: Thanks!

    • kaspar030: Should I rebase this to main?

    • bugadani: No need, it will merge cleanly, just needs a +1 from someone else, ideally @bjoernQ :)

    • jessebraham: Changes LGTM, but will let Bjoern have the final say before merging. Thanks!

    • bjoernQ: Thanks!

    • kaspar030: Thanks for the quick review!

  • Format PanicInfo with defmt #3112 [DBLouis] skip-changelog

    Remove use of Display2Format to format PanicInfo. Format has been implemented on...

    • jessebraham

    • bjoernQ

  • Cannot install esp-println #3114 [maelp] bug

    I get this error with v0.13 error[E0412]: cannot find type `PhantomData` in this scope -->...

    • MabezDev: Closed via #2979, we still haven't released esp-println yet though. Could you use the git...

    • maelp: Thanks, so you mean using a git hash?

    • maelp: I did, with the latest commit, but now I have this, any idea? error: mach-o section specifier...

    • jessebraham: Given the mention of mach-o in the error, looks like you are building for macOS and not targeting Xtensa.

    • maelp: Interesting, because I'm doing cargo +esp build, not sure how to configure? Here's my...

    • maelp: The error begins this way if it helps Compiling xtensa-lx-rt v0.18.0...

    • jessebraham: #[build] The build section is commented out, so the target is not being applied most likely.

    • maelp: wooops you're right, now I have an error in my code so it compiles it::MillisDurationU64; | ...

    • bugadani: Try to pass esp_hal::time::Duration and not fugit's type. Recent change.

    • maelp: Thank you indeed this works... now just have to figure out why my esp32-s3 device is not...

    • maelp: Weirdly esptool.py doesn't seem to detect the device, althouh I know I was able to flash it...

    • maelp: Hmmm it seems it was a driver issue with the board, installing this fixes it...

  • How to use EEPROM？ #1045 [Song-aff]

    • DBLouis: For persistence you can use https://github.com/esp-rs/esp-storage to write and read the flash....

    • bjoernQ: esp-idf uses it but no-std doesn't

  • Don't enable the interrupt when waiti isn't used #3116 [bugadani] skip-changelog

    This probably isn't the cause of our HIL issues, but without low_power_wait the interrupt isn't bound to a handler...

    • bjoernQ: Makes sense

    • jessebraham

  • DO NOT MERGE: polishing experiments #3117 [MabezDev] skip-changelog

    I suggest bulding and viewing the docs locally, then reviewing this commit by commit. The intention is not to merge...

  • Unstabilize CPU-related functionality, merge modules #3099 [bugadani] skip-changelog

    Multi-core operations, as well as CPU reset are no longer stable. CPU-related operations are now contained in the...

    • MabezDev: Sorry to bikeshed, I feel like this would be better placed in a system or soc module. Which also...

      • MabezDev: I suppose this (and the ROM function) should have a return type of -> !?

    • bugadani: Good shout, although that system module is equally pointless IMHO Radio clocks belong to maybe...

    • MabezDev: Good shout, although that system module is equally pointless IMHO Radio clocks belong to maybe...

    • bjoernQ: seems it's only used in rtc_cntl - should we move it there?

    • bugadani: I moved everything into system, and removed the EnumIter thing that was a perma-stable-public...

    • bjoernQ: LGTM

    • MabezDev: LGTM, just one question that I either can't tell from the diff or has some other effect I'm not...

      • MabezDev: Does this need to be stable? iirc it's all internal?

        • bugadani: It's hidden, so does it matter?

        • bugadani: Just checked - Peripheral needs to be public, because Info in e.g. UART is public, and it needs...

        • MabezDev: Oops, missed the doc hidden :D, thanks!

  • Make the main example work with the stable HAL #3113 [bugadani] skip-changelog

    • MabezDev: Thanks!

  • Add auto-writeback support to DMA buffers #3107 [Dominaezzz]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • bugadani: Thanks

    • bjoernQ: LGTM

  • Fix auto writeback on CryptoDMA #3108 [Dominaezzz]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • bugadani

    • bjoernQ: LGTM

  • Use correct Crypto DMA interrupt #3111 [Dominaezzz] skip-changelog

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • jessebraham

    • bjoernQ

  • DMA move API tracking issue #2269 [bugadani] peripheral:dma

    Basics #1716 #1719 #1856 #1976 Peripheral support SPI master SPI slave I8080 #2191 #2268 Camera #2242 AES...

  • Implement application description for ESP-IDF image format #3058 [SergioGasquez] linker scripts 1.0-blocker

    New versions of ESP-IDF bootloader expect to have application description, see...

    • ivmarkov: Adding the 1.0 blocker as we need to solve this update the bootloaders in espflash for v4?...

    • ivmarkov: There's also this new guy named mmu_page_size which appears in ESP-IDF bootloaders >= 5.4. I...

    • ivmarkov: Related: esp-rs/espflash#741

  • Rethink WiFi API #3027 [bjoernQ]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • bjoernQ: I don't think adding an example is interesting enough but here is the wifi_dhcp example modified...

    • MabezDev: I don't think adding an example is interesting enough but here is the wifi_dhcp example modified...

    • bjoernQ: What happens when a user wants to do AP_STA mode? We have the...

    • Dominaezzz: so we could have some intermediate "device" that can do some limited packet parsing to route the...

    • MabezDev: Maybe I was a bit too ambitious with the routing point, but I think its still worth trying to...

    • bjoernQ: Maybe I was a bit too ambitious with the routing point, but I think its still worth trying to...

    • bugadani: We should probably add a FIXME to remove this after the next rustc release, the c_char...

    • Dominaezzz: When I was reading the examples I thought this was a getter. I propose this should be called...

      • bugadani: If we want any sort of stability in esp-wifi, this would need to be create_smoltcp_<version>_interface.

    • Dominaezzz: My only note here with the API is this would've been nice as a one liner like so. let (mut...

      • bugadani: I agree that we are probably integrating smoltcp a bit more tightly than ideal. Since it's at...

      • bjoernQ: I decided to remove that completely - for now users can do it themselves and mid-term/long-term...

    • bugadani: Well this PR is borderline unreviewable on the github diff page, but here are a few comments.

      • bugadani: This can be a fetch_or.

        • Dominaezzz: Or a swap

        • bjoernQ: using fetch_or but honestly that sniffer and esp-now should become (very) special interfaces...

      • bugadani: Meaningless nitpick of the day, but this cna be so much nicer if you collected these into a...

        • Dominaezzz: I'll 2nd that and suggest using map and fold as well.

    • AnthonyGrondin: Side note; Now that Generics have been removed, can this now...

      • bugadani: At that point this can also turn into something like: #[embassy_executor::task(pool_size =...

      • MabezDev: I'm not sure if I'm meant to be playing a game of spot the difference here, but aren't they the same? :D

      • bugadani: Apparently I'm just blind and I just saw the diff, but I actually made a syntax error so there...

      • MabezDev: but I actually made a syntax error so there is a difference 🙈 Aww, that means I lost :(

      • Dominaezzz: What's the difference then? 😅

      • MabezDev: Missing closing bracket on task pool (yes it still took me another good 5 mins to find it 🤣)

  • Add linker scripts (and any additional required code) to support `mcuboot` #2479 [jessebraham] linker scripts 1.0 non-breaking

    Build environment for the bootloader can be found here: https://github.com/jessebraham/espressif-mcuboot-docker

  • esp-wifi: support using an external scheduler #3115 [kaspar030]

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

    • bjoernQ: I guess this one deserves a changelog entry This is using "extern" functions, which...

    • bugadani: You could go the embassy-time-driver route, add a Scheduler trait and a macro that ties the...

    • kaspar030: You could go the embassy-time-driver route, add a Scheduler trait and a macro that ties the...

    • bugadani: That's probably a bit more complicated than what I have in mind :D

    • kaspar030: That's probably a bit more complicated than what I have in mind :D That's two functions, here...

    • kaspar030: That's two functions, here we have a whopping seven 🙂. But I guess, manual labor it is for...

    • kaspar030: This needs to be public so preempt::Scheduler is available.

      • bugadani: At this point this should probably be always on & public, and our internal implementation can...

      • kaspar030: Implemented in c4024321. (It does feel a bit odd to go through the opaque raw pointers within...

      • bugadani: A bit, but maybe this will keep us honest (i.e. we probably won't break the third party...

      • MabezDev: We probably (read: I want to :D) move the scheduler bits out of here anyways, so I think keeping...

    • kaspar030: Updated Ariel diff is here, it looks cleaner indeed.

    • kaspar030: CI complained before that there were docs on a private item. I set this to ignore, as I don't...

      • bugadani: This should be an inner doc comment (//!), right now rustdoc thinks you documented the c_void import

    • MabezDev: Would be nice to remove this from the trait some how, but I don't think we can because the...

      • kaspar030: I feel the same. Maybe we could just keep it out, and document that a certain tick should be set...

      • bugadani: I feel the same. Maybe we could just keep it out, and document that a certain tick should be set...

      • kaspar030: ok, that would probably require splitting the init function: // old esp_wifi::init(timer, rng,...

      • MabezDev: We can probably make esp-wifi take the scheduler as an init argument, that way we don't have to...

      • kaspar030: Whilst the scheduler here is still built in, we can just call the setup function somewhere in...

      • MabezDev: Sorry I was referencing @bugadani with You still need something to signal that the scheduler...

    • MabezDev: This turned out really nice! Thanks for working on this! LGTM with CI green.

    • kaspar030: Linting esp-wifi fails now b/c timer is unused. Does linting run w/o default-features? I'm...

    • bugadani: Looks like that. You can add the feature a few lines above.

    • kaspar030: Looks like that. You can add the feature a few lines above. thanks!

    • kaspar030: Those HIL tests seem to only fail on riscv. Hm. It seems the hil tests also run without default...

    • kaspar030: It seems the hil tests also run without default features:...

    • bugadani: Huh, those look like legit failures. panicked at...

    • MabezDev: I think we need to setup the timer before enabling the scheduler, that should fix the errors?

      • kaspar030: That does indeed seem to fix the error. I don't understand why, and it seem dangerous. The timer...

  • Expose CopyDmaChannel #3118 [Dominaezzz] skip-changelog

    Thank you for your contribution! We appreciate the time and effort you've put into this pull request. To help us...

created at 2025-02-08 01:39 by dinghy 0.13.1.