Teach rustc about the Xtensa VaListImpl #127565
Conversation
rustbot
added
S-waiting-on-review
This comment has been minimized.
This comment has been minimized.
|
r? @davidtwco |
|
cc @workingjubilee who has been doing a lot of varargs work recently |
There was a problem hiding this comment.
I haven't reviewed the codegen changes in detail as things will go much more smoothly if I have the bits of information I am asking for details on. If you don't have all this information, that's fine, and don't go off for a year digging for answers or anything. I just want as much of it that is on-hand and easily referenced.
|
@rustbot author |
rustbot
added
S-waiting-on-author
|
Gentle ping for a review whenever you have time some spare cycles :) |
There was a problem hiding this comment.
At each step, I prefer to confirm the code implements the spec rather than read the code and try to discern what spec it implements. There's only so many ways to implement a spec, but there's an unbounded set of specs a given fragment could implement. So, this could use a few more comments.
Perhaps a less overloaded naming scheme would suffice? But when we're already checking in about 100 lines of code, we can afford a bit of extra padding to smooth things for the next reader.
|
@MabezDev |
MabezDev
force-pushed
the
xtensa-vaargs
branch
from
bcc9c0e to
c711221
Compare
|
Thank you for the review @workingjubilee (and sorry for the delay), I think I've addressed most of your initial concerns. I'm a little unsure on how to improve the stack ndx check, so any ideas would be appreciated. |
alex-semenyuk
added
S-waiting-on-review
|
Don't worry too much if we don't figure out anything to change. I expect nonspecific responses... including no response... to review comments that don't say anything particularly specific! In this case, it was mostly to signal what I am going to give further attention later when the code was structured to make it easier to reach that point, without feeling as-distracted by the smaller details of the code preceding. In a sense it was an invitation for you to have a sudden flash of insight if you had one to spare... it's completely understandable if you spent that on something else, like the meaning of love, attaining nirvana, or the best wine-and-cheese pairing for an appetizer. Otherwise it was mostly a reminder to myself. |
MabezDev
force-pushed
the
xtensa-vaargs
branch
from
c711221 to
710b544
Compare
rustbot
removed
the
S-waiting-on-review
|
Thank you for the review! You were correct in your understanding of the LLVM codegen API, and I have added the relevant comments above the blocks in Rust code. This was a nice suggestion as it makes it a lot more readable. |
MabezDev
force-pushed
the
xtensa-vaargs
branch
from
ec0e3dd to
aa5c9bc
Compare
This comment has been minimized.
This comment has been minimized.
MabezDev
force-pushed
the
xtensa-vaargs
branch
from
aa5c9bc to
f6f4ddb
Compare
|
Thanks for bearing with me, and sorry for the miscommunication. I believe I've addressed your comments properly now, let me know if there is anything else you'd like me to adjust. |
|
r=me with the typo settled. :3 @bors delegate+ |
MabezDev
force-pushed
the
xtensa-vaargs
branch
from
31d9868 to
a572055
Compare
…vaarg for xtensa. LLVM does not include an implementation of the va_arg instruction for Xtensa. From what I understand, this is a conscious decision and instead language frontends are encouraged to implement it themselves. The rationale seems to be that loading values correctly requires language and ABI-specific knowledge that LLVM lacks. This is true of most architectures, and rustc already provides implementation for a number of them. This commit extends the support to include Xtensa. See https://lists.llvm.org/pipermail/llvm-dev/2017-August/116337.html for some discussion on the topic. Unfortunately there does not seem to be a reference document for the semantics of the va_list and va_arg on Xtensa. The most reliable source is the GCC implementation, which this commit tries to follow. Clang also provides its own compatible implementation. This was tested for all the types that rustc allows in variadics. Co-authored-by: Brian Tarricone <brian@tarricone.org> Co-authored-by: Jonathan Bastien-Filiatrault <joe@x2a.org> Co-authored-by: Paul Lietar <paul@lietar.net>
MabezDev
force-pushed
the
xtensa-vaargs
branch
from
a572055 to
059f627
Compare
|
@bors r=workingjubilee |
|
I think I broke bors, let me try again 😅. @bors r=workingjubilee |
|
Sometimes things just get out of sync @bors r=workingjubilee |
bors
added
S-waiting-on-bors
…ilee Teach rustc about the Xtensa VaListImpl Following on from the target Xtensa target PRs (rust-lang#125141, rust-lang#126380), this PR teaches rustc about the structure of the VA list on the Xtensa arch, as well as adding the required lowering to be able to actually use it.
Rollup of 6 pull requests Successful merges: - rust-lang#127565 (Teach rustc about the Xtensa VaListImpl) - rust-lang#128004 (codegen `#[naked]` functions using global asm) - rust-lang#133256 (CI: use free runners for i686-gnu jobs) - rust-lang#133472 (Run TLS destructors for wasm32-wasip1-threads) - rust-lang#133632 (CI: split x86_64-msvc job) - rust-lang#133827 (CI: rfl: move job forward to Linux v6.13-rc1) r? `@ghost` `@rustbot` modify labels: rollup
…ilee Teach rustc about the Xtensa VaListImpl Following on from the target Xtensa target PRs (rust-lang#125141, rust-lang#126380), this PR teaches rustc about the structure of the VA list on the Xtensa arch, as well as adding the required lowering to be able to actually use it.
Rollup of 6 pull requests Successful merges: - rust-lang#127565 (Teach rustc about the Xtensa VaListImpl) - rust-lang#133844 (clarify simd_relaxed_fma non-determinism) - rust-lang#133867 (Fix "std" support status of some tier 3 targets) - rust-lang#133882 (Improve comments for the default backtrace printer) - rust-lang#133888 (Improve bootstrap job objects) - rust-lang#133898 (skip `setup::Hook` on non-git sources) r? `@ghost` `@rustbot` modify labels: rollup
Rollup of 6 pull requests Successful merges: - rust-lang#127565 (Teach rustc about the Xtensa VaListImpl) - rust-lang#133844 (clarify simd_relaxed_fma non-determinism) - rust-lang#133867 (Fix "std" support status of some tier 3 targets) - rust-lang#133882 (Improve comments for the default backtrace printer) - rust-lang#133888 (Improve bootstrap job objects) - rust-lang#133898 (skip `setup::Hook` on non-git sources) r? `@ghost` `@rustbot` modify labels: rollup
Rollup merge of rust-lang#127565 - esp-rs:xtensa-vaargs, r=workingjubilee Teach rustc about the Xtensa VaListImpl Following on from the target Xtensa target PRs (rust-lang#125141, rust-lang#126380), this PR teaches rustc about the structure of the VA list on the Xtensa arch, as well as adding the required lowering to be able to actually use it.
Following on from the target Xtensa target PRs (#125141, #126380), this PR teaches rustc about the structure of the VA list on the Xtensa arch, as well as adding the required lowering to be able to actually use it.