[Miri] Use a session variable instead of checking for an env var always #69888
Conversation
rust-highfive
added
the
S-waiting-on-review
src/librustc_session/session.rs
Outdated
| pub enum CtfeBacktrace { | ||
| /// Do nothing special, return the error as usual without a backtrace. | ||
| Disabled, | ||
| /// Capture a backtrace at the point the error is created and return it in the error. |
There was a problem hiding this comment.
| /// Capture a backtrace at the point the error is created and return it in the error. | |
| /// Capture a backtrace at the point the error is created and return it in the error | |
| /// (to be printed later if/when the error ever actually gets shown to the user). |
|
LGTM aside from the nit, but this is also touching global compiler infrastructure that I know little about... @oli-obk does this look reasonable? |
|
Yea this is OK, but at this point you could also just use a lazy static inside that function without affecting any code outside of it |
|
But then how is Miri going to change its value at run-time? |
|
You can lazily initialize from the env var. Since the initialization only ever happens once, just like in the current PR, you can't change it while it runs, but that was never a use case |
Yes it was, Miri does exactly that. |
|
That's OK, that happens before any error is ever thrown so the lazy in init hasn't happened yet |
|
The intend is for that code to run after MIR got generated and const-eval happened. It gets called in Otherwise there'd be no reason for this entire |
|
Right. I guess let's do this for now and reconsider if/when someone dislikes it |
In CTFE heavy code, checking the env var everytime is inefficient. We can do a lot better by using a `Session` variable instead.
wesleywiser
force-pushed
the
miri_exception_env_var_to_session_var
branch
from
ea85732
to
5357f83
Compare
|
@bors r=RalfJung |
|
📌 Commit 5357f83 has been approved by |
|
🌲 The tree is currently closed for pull requests below priority 1000, this pull request will be tested once the tree is reopened |
bors
added
S-waiting-on-bors
Rollup of 8 pull requests Successful merges: - #66472 (--show-coverage json) - #69603 (tidy: replace `make check` with `./x.py test` in documentation) - #69760 (Improve expression & attribute parsing) - #69828 (fix memory leak when vec::IntoIter panics during drop) - #69850 (panic_bounds_check: use caller_location, like PanicFnLangItem) - #69876 (Add long error explanation for E0739) - #69888 ([Miri] Use a session variable instead of checking for an env var always) - #69893 (librustc_codegen_llvm: Use slices instead of 0-terminated strings) Failed merges: r? @ghost
In CTFE heavy code, checking the env var everytime is inefficient. We
can do a lot better by using a
Sessionvariable instead.r? @RalfJung
Part of #69297