Skip to content

Commit 3535ad0

Browse files
japaricjaparic
committed
add minutes for the past 3 meetings
1 parent 9d8a87a commit 3535ad0

File tree

4 files changed

+214
-0
lines changed

4 files changed

+214
-0
lines changed

‎minutes/2018-03-20.md

Lines changed: 44 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,44 @@
1+
Woo! Embedded devices Working Group! \o/
2+
3+
Coordination repository: https://github.com/rust-lang-nursery/embedded-wg
4+
Milestoned issues: https://github.com/rust-lang-nursery/embedded-wg/issues?q=is%3Aopen+is%3Aissue+milestone%3A%222018+edition%22
5+
Meetings: Tuesdays 9 PM CET (UTC+1) - Hangouts or IRC
6+
IRC: #rust-embedded
7+
8+
# Next meeting schedule
9+
10+
March 20 21:00 CET (UTC+1) - Hangouts
11+
12+
# Meeting agenda (& minutes)
13+
14+
- Requests priorities - https://paper.dropbox.com/doc/Domain-WG-requests-t93t94TQw20GDnlht0Hjp
15+
- assembly
16+
17+
- Define deliverables for existing milestone issues
18+
19+
- Issues nominated for the 2018 edition milestone
20+
- embedded-hal, what will be the deliverable?
21+
- The embedded Rust book
22+
- embedded Rust on stable
23+
- embedded testing
24+
25+
- Meeting time (is it CET or Europe/Paris?, only the horror of daylight savings is almost upon us)
26+
27+
- Next meeting is 21:00 Europe/Berlin :)
28+
29+
30+
- $YOUR_AGENDA_ITEM
31+
32+
# Rust All Hands
33+
34+
https://paper.dropbox.com/doc/Embedded-WG-Rust-All-Hands-2018-dtawBXADPn3XCJn6kGTcU
35+
36+
# Action items
37+
dvc94ch:
38+
update E310 drivers
39+
binary distribution riscv-rust
40+
E310/HX8K devboard
41+
42+
- milestone book
43+
- awesome embedded Rust -> community, discuss with Cargo team
44+
- docs testing: mock and Custom Test Frameworks

‎minutes/2018-03-27.irc.log

Lines changed: 118 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,118 @@
1+
2018-03-27 20:57:18 dvc hi
2+
2018-03-27 20:57:57 ~japaric dvc: o/
3+
2018-03-27 20:58:07 hannobraun Hello
4+
2018-03-27 21:01:35 ~japaric (let's wait a few minutes for the other members)
5+
2018-03-27 21:03:56 dvc japaric: thoughts on atomic operations with svd2rust?
6+
2018-03-27 21:04:23 dvc cortex-m doesn't have any atomic instructions right?
7+
2018-03-27 21:05:07 ~japaric dvc: it does; I don't remember the instruction name though
8+
2018-03-27 21:05:17 ~japaric you can do CAS loops on Cortex-M3+
9+
2018-03-27 21:05:18 dvc oh
10+
2018-03-27 21:05:30 ~japaric Cortex-M0 doesn't have the instruction to implement atomics
11+
2018-03-27 21:05:51 ~japaric so llvm produces a call to some compiler intrinsics for that target
12+
2018-03-27 21:06:30 hannobraun (and neither does Cortex-M0+, just to clarify)
13+
2018-03-27 21:08:11 ~japaric dvc: haven't read your svd2rust issue, btw so no comments right now
14+
2018-03-27 21:09:29 dvc on riscv we have amoor, amoand, amoxor instead of peripherals providing stateless registers for example for toggling a led in one instruction
15+
2018-03-27 21:11:00 dvc so it would make sense to use those in whatever svd2rust generates at least for the riscv target
16+
2018-03-27 21:11:17 ~japaric dvc: could you elaborate on what these instructions do?
17+
2018-03-27 21:11:30 dvc atomic or, and, xor
18+
2018-03-27 21:12:10 dvc so toggling a led is amoor gpio_address 0b1000 or something like that
19+
2018-03-27 21:12:16 dvc sorry xor
20+
2018-03-27 21:12:28 ~japaric this sounds like bitbanding on ARM
21+
2018-03-27 21:13:06 ~japaric that's not supported on all ARM devices; it's an optional component of the Cortex-M architecture
22+
2018-03-27 21:13:17 dvc ah ok, just looked up what it means
23+
2018-03-27 21:13:38 dvc it's not like mmap, but is supported on any memory address
24+
2018-03-27 21:14:12 ~japaric I see but it sounds like a single API could abstract what both amoor and bitbanding do
25+
2018-03-27 21:14:31 ~japaric like setting a single bit of a register
26+
2018-03-27 21:15:53 dvc well amoor also needs ordering flags
27+
2018-03-27 21:16:18 dvc not sure if it matters on single cores
28+
2018-03-27 21:17:36 dvc so everyone here? :)
29+
2018-03-27 21:18:05 ~japaric hannobraun: jamesmunns are right next to me :-)
30+
2018-03-27 21:18:20 dvc berlin?
31+
2018-03-27 21:18:34 ~japaric yep, we are in mozilla berlin office
32+
2018-03-27 21:18:36 jamesmunns I guess the rest can read the logs :)
33+
2018-03-27 21:18:51 jamesmunns Heres a summary of what we got up to today:
34+
2018-03-27 21:18:56 dvc :)
35+
2018-03-27 21:19:25 Douman some serious business you got here
36+
2018-03-27 21:19:46 jamesmunns 1. Getting Embedded Rust on Stable - there are three current blockers
37+
2018-03-27 21:20:19 jamesmunns a) compiler-builtins requires unstable
38+
2018-03-27 21:21:22 dvc is that a question? :)
39+
2018-03-27 21:21:25 jamesmunns We have two possibly solutions, first would be to teach the compiler about compiler builtins - though this requires introducing some new attributes
40+
2018-03-27 21:22:05 jamesmunns second would be to include compiler-builtins, and automatically insert it with the `#[no_std]` prelude
41+
2018-03-27 21:22:12 jamesmunns Check out https://github.com/rust-lang/rust/issues/49380 for more info
42+
2018-03-27 21:22:30 jamesmunns And we have support from the compiler team to get this merged for 2018
43+
2018-03-27 21:22:46 jamesmunns 2. Xargo requires nightly
44+
2018-03-27 21:23:23 jamesmunns the current plan is to ship a `rust-std` component for the thumb and msp430 components, basically shipping just `core` for those targets, so you can `rustup target add thumb*` etc.
45+
2018-03-27 21:23:42 jamesmunns Also Ok'd to get done for 2018, see https://github.com/rust-lang/rust/issues/49382 for more details
46+
2018-03-27 21:23:44 Douman JFYI Rust's PR #49316 seems to get approve so incremental linking gotta get fixed?
47+
2018-03-27 21:24:21 est31 jamesmunns: still no merging of xargo into cargo?
48+
2018-03-27 21:24:21 jamesmunns 3. panic-fmt is required
49+
2018-03-27 21:24:52 jamesmunns est31: that will come later (maybe after 2018 Era)
50+
2018-03-27 21:25:07 jamesmunns but shipping core for the embedded targets will definitely make it into 2018 era
51+
2018-03-27 21:25:18 est31 "the embedded targets"
52+
2018-03-27 21:25:21 est31 there are tons of them
53+
2018-03-27 21:25:31 ~japaric (jamesmunns is making a summary of what we talked about with the other Rust teams today at the all hands event)
54+
2018-03-27 21:25:41 est31 ah right
55+
2018-03-27 21:25:43 ~japaric est31: there's like 5 of them supported by rustc
56+
2018-03-27 21:25:45 * est31 stops arguing
57+
2018-03-27 21:25:57 ~japaric those are the ones that will get a rust-std component
58+
2018-03-27 21:25:58 est31 <jamesmunns> 3. panic-fmt is required
59+
2018-03-27 21:26:08 est31 ther was an RFC about it by japaric right?
60+
2018-03-27 21:26:30 jamesmunns re: panic-fmt, upstream compiler team will fix for 2018 era, with "something that works", and they are aware about the code bloat issue, but that may or may not make it for the 2018 deadline
61+
2018-03-27 21:27:01 jamesmunns Outside of those top-prio issues, there are some additional things we worked on today
62+
2018-03-27 21:27:19 dvc code bloat when impl panic-fmt?
63+
2018-03-27 21:27:47 dvc I thought that had something to do with #[derive(Debug)]
64+
2018-03-27 21:28:04 dvc anyway continue - not that important
65+
2018-03-27 21:28:21 jamesmunns `asm!()` is unlikely to stabilize for 2018. Our short term mitigation is to add a `core::asm::$arch` module, with common intrinsics like bkpt, wfi, etc.
66+
2018-03-27 21:28:36 est31 dvc: iirc panic-fmt means that in many cases, the entire fmt machinery is shipped
67+
2018-03-27 21:29:04 jamesmunns No upstream issue for this yet, but we will make an RFC, and collect what kind of intrinsics should be included. These intrinsics will basically by a safe interface on unstable inline assembly
68+
2018-03-27 21:29:05 ~japaric est31: yes, we are likely to implement and stabilize rfc #2070
69+
2018-03-27 21:29:33 ~japaric we can hopefully fix the binary size problem (panic strings kept in the binary even when panic_fmt = loop {}) by leveraging pure MIR rlibs
70+
2018-03-27 21:29:39 jamesmunns const-fn will move unto FCP soon, so we may see that for 2018 era
71+
2018-03-27 21:30:08 est31 why is const fn specifically important for embedded?
72+
2018-03-27 21:30:14 jamesmunns last low prio issue was getting `#[used]` attribute stabilized, which everyone seemed happy with moving to FCP
73+
2018-03-27 21:30:26 jamesmunns (almost done, will start answering questions soon)
74+
2018-03-27 21:30:32 jamesmunns Other misc things:
75+
2018-03-27 21:31:13 jamesmunns We will start shipping more of `llvm-binutils`, probably behind some facade to prevent adding to the user's path, something like cargo-objcopy, etc.
76+
2018-03-27 21:31:20 ~japaric est31: it's a P-med unstable feature; it's not being stabilize because embedded needs but because it has been in the compiler for a while and we are unlikely to change the design (e.g. drop the const qualifier)
77+
2018-03-27 21:31:38 ~japaric a P-med (medium priority) unstable feature for embedded**
78+
2018-03-27 21:32:05 ~japaric const-fn is needed to initialize a static variable that contains a Mutex, for example
79+
2018-03-27 21:32:21 ~japaric it's not only needed in embedded / bare metal, but also comes up in OS dev
80+
2018-03-27 21:32:24 jamesmunns We also had a good talk about the embedded rust book, and decided to initially target it at developers who already know rust, and already know embedded, and point to other resources for learning core concepts. This is committing to ship that much of the embedded rust book in 2018
81+
2018-03-27 21:32:44 jamesmunns After that ships, we can start expanding to cover more possible people interested in Rust.
82+
2018-03-27 21:33:01 est31 japaric: are there things you do in embedded programming where you need const fn?
83+
2018-03-27 21:33:13 est31 i was just wondering why you are tracking it
84+
2018-03-27 21:33:22 jamesmunns That was all the important stuff discussed today, We'll start answering questions now :)
85+
2018-03-27 21:33:28 dvc the book is coming along. I perused it today but didn't get to reading it thoroughly
86+
2018-03-27 21:33:58 dvc I'm very happy with the target audience =P
87+
2018-03-27 21:34:11 ~japaric est31: we are tracking it because it came up high in this survey https://github.com/rust-lang-nursery/embedded-wg/issues/31
88+
2018-03-27 21:34:31 jamesmunns dvc are you talking about the "embeddednomicon"?
89+
2018-03-27 21:34:37 dvc yes
90+
2018-03-27 21:35:01 est31 japaric: I see :)
91+
2018-03-27 21:35:26 jamesmunns We want that to handle low level impl, and also ship a "stable embedded rust book", which covers some more of the high level "how to write your first embedded application using existing board/chip support crates, embedded-hal traits, and existing drivers"
92+
2018-03-27 21:37:11 ~japaric dvc: the embedonomicon != the embedded Rust book. the nomicon is targeted towards advanced users that want to look into low level impl details; the embedded Rust book is / will be the getting started guide
93+
2018-03-27 21:37:24 dvc ok
94+
2018-03-27 21:39:04 dvc I'm reimplementing riscv-rt currently so more info on how to handle interrupts would be good
95+
2018-03-27 21:39:33 dvc like how do you share a peripheral accross the isr and main properly
96+
2018-03-27 21:40:33 ~japaric dvc: you can use Mutex<RefCell<Option<T>>> ... or implement rtfm
97+
2018-03-27 21:41:38 dvc still need to be able to share one timer for all tasks
98+
2018-03-27 21:42:11 ~japaric dvc: rtfm doesn't make use of timers at all. At least not right now
99+
2018-03-27 21:42:24 ~japaric it may optionally require a timer when timing semantics are added
100+
2018-03-27 21:42:42 dvc I thought each task is wired to some interrupt
101+
2018-03-27 21:42:52 ~japaric it is
102+
2018-03-27 21:42:58 Douman you need to enable timers
103+
2018-03-27 21:43:06 Douman for them to be actually running afaik
104+
2018-03-27 21:43:24 ~japaric timers? is that a riscv specific thing?
105+
2018-03-27 21:43:39 dvc no
106+
2018-03-27 21:43:46 ~japaric on Cortex-M you have to enable the interrupt in the NVIC peripheral
107+
2018-03-27 21:44:17 dvc on riscv there is a CLINT which means Core Local Interrupt
108+
2018-03-27 21:44:31 ~japaric but don't need to enable timers to make interrupts work
109+
2018-03-27 21:44:34 dvc so each core has a MachineTimer interrupt handled by the clint
110+
2018-03-27 21:44:46 dvc and there is a PLIC for external interrupts
111+
2018-03-27 21:44:58 dvc which is like NVIC
112+
2018-03-27 21:45:19 dvc CLINT is like SYSTICK I think
113+
2018-03-27 21:47:56 dvc I'll give rtfm another go after releasing e310x-hal
114+
2018-03-27 21:50:06 dvc so end of transmission?
115+
2018-03-27 21:51:07 ~japaric yeah, unless someone wants to discuss something else
116+
2018-03-27 21:51:12 ~japaric we only had the summary in the agenda
117+
2018-03-27 21:53:13 dvc till next week then
118+
2018-03-27 21:53:17 ~japaric dvc: o/

‎minutes/2018-03-27.md

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,8 @@
1+
# Meeting agenda
2+
3+
- what happened today at Rust all hands?
4+
- https://paper.dropbox.com/doc/Berlin-All-Hands-Tuesday-recap-SwYY8zYb2IkvJFJswTNRv
5+
6+
# Rust All Hands
7+
8+
https://paper.dropbox.com/doc/Embedded-WG-Rust-All-Hands-2018-dtawBXADPn3XCJn6kGTcU

‎minutes/2018-04-03.md

Lines changed: 44 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,44 @@
1+
# Meeting agenda (& minutes)
2+
3+
- Triage milestone items
4+
- embedded Rust on stable
5+
- panic-fmt lang item
6+
- implement RFC 2070
7+
- @nagisa will mentor
8+
- fix binary size issue later using pure MIR rlibs (not yet implemented)
9+
- Xargo
10+
- we'll provide rust-std components for the embedded targets
11+
- made PR for Cortex-M
12+
- https://github.com/rust-lang/rust/pull/49563
13+
- TODO msp430 (tier 2) -- compiler errors shouldn't block PR from landing
14+
- compiler-builtins
15+
- PR to add extern crate compiler_builtins to #![no_std] expansion
16+
- https://github.com/rust-lang/rust/pull/49503
17+
- const-fn
18+
- proposed for stabilization
19+
- https://github.com/rust-lang/rust/issues/24111#issuecomment-376605402
20+
- #[used]
21+
- RFC for stabilization
22+
- https://github.com/rust-lang/rfcs/pull/2386
23+
24+
- tests in rust-lang/rust CI
25+
- Cortex-M will be tier 1
26+
- tier 1: rust-std component, has tests and test errors block PRs from landing
27+
- others will be tier 2/3 depending on the state of the llvm backend
28+
- tier 2: like tier 1 but tests don't block PRs from landing, rust-std component may be missing sometimes
29+
- tier 3: no tests or rust-std component
30+
- adding tests will be the Increasing Rust Reach project
31+
- http://reach.rust-lang.org/
32+
33+
- llvm-binutils
34+
- discussion issue https://github.com/rust-lang/rust/issues/49584
35+
36+
- cargo-size, et al.
37+
- first step: just directly expose the underlying tool (has to search for it in the sysroot first)
38+
39+
- website
40+
- made a wireframe during Rust All Hands
41+
- https://imgur.com/a/enu4V
42+
43+
- awesome-embedded-rust list https://github.com/rust-embedded/awesome-embedded-rust
44+
- TODO add link to README

0 commit comments

Comments
 (0)