Skip to content

Add support for no_split_stack attribute to LLVM. Use it on main #1226

New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

#

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

Jump to bottom
Closed
brson opened this issue Nov 28, 2011 · 10 comments
Closed

Add support for no_split_stack attribute to LLVM. Use it on main #1226

brson opened this issue Nov 28, 2011 · 10 comments
Labels
A-LLVM Area: Code generation parts specific to LLVM. Both correctness bugs and optimization-related issues. I-slow Issue: Problems and improvements with respect to performance of generated code.

Comments

@brson
Copy link
Contributor

brson commented Nov 28, 2011

Currently our __morestack method has to do an ugly check to detect whether it is being called from main. gcc supports a no_split_stack attribute that can be used to suppress the split-stack prologue. We should add it to LLVM and use it.
@brson
Copy link
Contributor Author

brson commented Nov 28, 2011

C shim functions run on the C stack so they need to not have the prologue either

@brson brson mentioned this issue Feb 16, 2012
Closed
@brson
Copy link
Contributor Author

brson commented Mar 6, 2012

Ditto for crust functions.

@brson
Copy link
Contributor Author

brson commented Mar 6, 2012

There was some noise on llvm-dev about this recently but I didn't notice if anything became of it.

@ghost ghost assigned brson Apr 12, 2012
@bblum
Copy link
Contributor

bblum commented Jun 11, 2012

After the changed C stack switching design, it will be essential that C shims have the prologue. Same for crust functions.

My changes to i386 __morestack make the "ugly check" less ugly, but the check could still stand to go away entirely.

@bblum
Copy link
Contributor

bblum commented Jun 21, 2012

C stack switching was dropped; nothing was done about the ugly check.

This was referenced Jan 14, 2013
Closed
Closed
@brson
Copy link
Contributor Author

brson commented Jan 14, 2013

@amuxtux I'm not clear on the obstacles to implementing __morestack on ARM but I opened an issue about it: #4489. If you intend to disable stack growth on ARM then the easiest way is probably to completely turn off the split stack prologue on ARM (the code for which is here).

@brson brson mentioned this issue May 3, 2013
Closed
30 tasks
@emberian
Copy link
Member

Triage bump; still relevant, esp with recent split-stack discussions.

@brson brson mentioned this issue Jul 28, 2013
Closed
4 tasks
@brson
Copy link
Contributor Author

brson commented Jul 28, 2013

This is required for the new scheduler in order to implement split stacks.

@alexcrichton alexcrichton mentioned this issue Aug 6, 2013
Closed
alexcrichton added a commit to alexcrichton/rust that referenced this issue Aug 20, 2013
@alexcrichton
Upgrade llvm to current HEAD
d1e4815 
* This has one workaround patch (everything's testing just fine...)
* I reworked the fixedstacksegment attribute to be specified with a string
  rather than using a keyword and an integer and modifying the parser
* I added a "no-split-stack" attribute along the same lines as the
  "fixedstacksegment" attribute for rust-lang#1226
bors added a commit that referenced this issue Aug 20, 2013
@bors
auto merge of #8328 : alexcrichton/rust/llvm-head, r=brson
a8c3fe4 
The first commit message is pretty good, but whomever reviews this should probably also at least glance at the changes I made in LLVM. I basically reorganized our pending patch queue to be a bit more organized and clearer in what needs to go where. After this, our queue would be:

* Add the `no-split-stack` attribute
* Add the `fixedstacksegment` attribute
* Add split-stacks for arm android
* Add split-stacks for arm linux
* Add split stacks for mips

Then there's a patch which I added to get rust to build at all on LLVM-head, and I'm not quite sure why it's there, but nothing seems to be crashing for now! (famous last words).

Otherwise, I just updated code to reflect the changes I made in LLVM with the only major change being the advent of the new `no_split_stack` attribute. This is work towards #1226, but someone more familiar with the code should probably actually assign the attribute to the appropriate functions.

Also as a bonus, I've verified that this closes #5774
@alexcrichton
Copy link
Member

This attribute has now been implemented in LLVM, and it's available for use via the no_split_stack attribute. Currently there are no users of it, however.

@catamorphism
Copy link
Contributor

Done, closing

bjorn3 added a commit to bjorn3/rust that referenced this issue May 15, 2022
@bjorn3
Fix assert_assignable for array types
74b9232 
Fixes rust-lang#1226
Kobzol pushed a commit to Kobzol/rust that referenced this issue Dec 30, 2024
@Lee-Janggun
Update links to rustc_mir in Two Phase Borrows (rust-lang#1226)
e38c344 
rust-lang#80522 split the `rustc_mir` crates into 5 crates, effectively invalidating all the direct links to `rustc_mir` in the docs.

I found this while looking at the Two Phase Borrows doc, which is why I am giving out this PR to fix this.
bors pushed a commit to rust-lang-ci/rust that referenced this issue Jan 2, 2025
@Lee-Janggun
Update links to rustc_mir in Two Phase Borrows (rust-lang#1226)
c3c390f 
rust-lang#80522 split the `rustc_mir` crates into 5 crates, effectively invalidating all the direct links to `rustc_mir` in the docs.

I found this while looking at the Two Phase Borrows doc, which is why I am giving out this PR to fix this.
# for free to join this conversation on GitHub. Already have an account? # to comment
Assignees
No one assigned
Labels
A-LLVM Area: Code generation parts specific to LLVM. Both correctness bugs and optimization-related issues. I-slow Issue: Problems and improvements with respect to performance of generated code.
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
5 participants
@alexcrichton @brson @catamorphism @emberian @bblum