Chained constant expressions causes stack overflow #34997

polazarus · 2016-07-23T19:12:41Z

Rustc's constant expression evaluation overflows in stack while compiling this kind of code:

pub const CST_1: u32 = 0;
pub const CST_2: u32 = CST_1+1;
pub const CST_3: u32 = CST_2+1;
pub const CST_4: u32 = CST_3+1;
pub const CST_5: u32 = CST_4+1;
pub const CST_6: u32 = CST_5+1;
pub const CST_7: u32 = CST_6+1;
pub const CST_8: u32 = CST_7+1;
pub const CST_9: u32 = CST_8+1;
//...
pub const CST_350: u32 = CST_349+1;

On my machine (linux 64): ~350 constants on the chain is enough with rustc 1.10.0 (cfcb716 2016-10-03). What is quite troubling to me is that rustc seems to do the work for each constant and does not rely on previously computed results. One could think that Rust just always draws the short straw, but I tried random names and random order for my constants and got the same problem.

Also, on 1.12.0-nightly (936bfea 2016-07-20) and 1.11.0-beta.1 (8dc253b 2016-07-05), there is no stack overflow message just a painful silent abort (return code -1).

The text was updated successfully, but these errors were encountered:

oli-obk · 2016-07-25T08:46:06Z

@eddyb: is it worth fixing this in the current const evaluators? Miri already has a cache, and doesn't use recursion to solve such constant chains.

@eddyb

Replace all const evaluation with miri * error reporting in constants prints a stacktrace through all called const fns * Trivial constant propagation and folding in MIR (always active, irrelevant of the optimization level) * can now use floating constants in patterns (previously only floating point literals were allowed) * the future compat lint is still produced for both cases * can index into constant arrays during const eval (previously feature gated) * can create a constant union value with field `a` and read from field `b` * can dereference references into constants * can create references inside constants (`const X: &u32 = &22`) * Tuple struct constructors can be used in constants * regression in const eval errors spans (some of these need improvements in mir debug info) * can cast floats to ints and vice versa (in constants, and even nan/inf constants) * Mir dump prints false/true instead of 0u8/1u8 * `1i8 >> [8][0]` does not lint about exceeding bitshifts anymore. * Needs const propagation across projections * `foo[I]` produces a const eval lint if `foo: [T; N]` and `N < I` * Essentially all builtin panics produce lints if they can be statically proven to trigger at runtime. This is on a best effort basis, so there might be some complex cases that don't trigger. (The runtime panic stays there, irrelevant of whether the lint is produced or not) fixes #34997 (stack overflow with many constants) fixes #25574 (deref byte strings in patterns) fixes #27918 (broken mir ICE) fixes #46114 (ICE on struct constructors in patterns) fixes #37448 (`SomeStruct { foo } as SomeStruct`) fixes #43754 (`return` in const fn) fixes #41898 (tuple struct constructors) fixes #31364 (infinite recursion with const fn, fixed by miri's recursion limit) closes #29947 (const indexing stabilization) r? @eddyb

@eddyb

Replace all const evaluation with miri * error reporting in constants prints a stacktrace through all called const fns * Trivial constant propagation and folding in MIR (always active, irrelevant of the optimization level) * can now use floating constants in patterns (previously only floating point literals were allowed) * the future compat lint is still produced for both cases * can index into constant arrays during const eval (previously feature gated) * can create a constant union value with field `a` and read from field `b` * can dereference references into constants * can create references inside constants (`const X: &u32 = &22`) * Tuple struct constructors can be used in constants * regression in const eval errors spans (some of these need improvements in mir debug info) * can cast floats to ints and vice versa (in constants, and even nan/inf constants) * Mir dump prints false/true instead of 0u8/1u8 * `1i8 >> [8][0]` does not lint about exceeding bitshifts anymore. * Needs const propagation across projections * `foo[I]` produces a const eval lint if `foo: [T; N]` and `N < I` * Essentially all builtin panics produce lints if they can be statically proven to trigger at runtime. This is on a best effort basis, so there might be some complex cases that don't trigger. (The runtime panic stays there, irrelevant of whether the lint is produced or not) fixes #34997 (stack overflow with many constants) fixes #25574 (deref byte strings in patterns) fixes #27918 (broken mir ICE) fixes #46114 (ICE on struct constructors in patterns) fixes #37448 (`SomeStruct { foo } as SomeStruct`) fixes #43754 (`return` in const fn) fixes #41898 (tuple struct constructors) fixes #31364 (infinite recursion with const fn, fixed by miri's recursion limit) closes #29947 (const indexing stabilization) fixes #45044 (pattern matching repeat expressions) r? @eddyb

@eddyb

Replace all const evaluation with miri * error reporting in constants prints a stacktrace through all called const fns * Trivial constant propagation and folding in MIR (always active, irrelevant of the optimization level) * can now use floating constants in patterns (previously only floating point literals were allowed) * the future compat lint is still produced for both cases * can index into constant arrays during const eval (previously feature gated) * can create a constant union value with field `a` and read from field `b` * can dereference references into constants * can create references inside constants (`const X: &u32 = &22`) * Tuple struct constructors can be used in constants * regression in const eval errors spans (some of these need improvements in mir debug info) * can cast floats to ints and vice versa (in constants, and even nan/inf constants) * Mir dump prints false/true instead of 0u8/1u8 * `1i8 >> [8][0]` does not lint about exceeding bitshifts anymore. * Needs const propagation across projections * `foo[I]` produces a const eval lint if `foo: [T; N]` and `N < I` * Essentially all builtin panics produce lints if they can be statically proven to trigger at runtime. This is on a best effort basis, so there might be some complex cases that don't trigger. (The runtime panic stays there, irrelevant of whether the lint is produced or not) * can use `union`s to implement `transmute` for `Copy` types in constants without a feature gate. With all the greatness and nasal demons that come with this. fixes #34997 (stack overflow with many constants) fixes #25574 (deref byte strings in patterns) fixes #27918 (broken mir ICE) fixes #46114 (ICE on struct constructors in patterns) fixes #37448 (`SomeStruct { foo } as SomeStruct`) fixes #43754 (`return` in const fn) fixes #41898 (tuple struct constructors) fixes #31364 (infinite recursion with const fn, fixed by miri's recursion limit) closes #29947 (const indexing stabilization) fixes #45044 (pattern matching repeat expressions) fixes #47971 (ICE on const fn + references) r? @eddyb

@eddyb

Replace all const evaluation with miri * error reporting in constants prints a stacktrace through all called const fns * Trivial constant propagation and folding in MIR (always active, irrelevant of the optimization level) * can now use floating constants in patterns (previously only floating point literals were allowed) * the future compat lint is still produced for both cases * can index into constant arrays during const eval (previously feature gated) * can create a constant union value with field `a` and read from field `b` * can dereference references into constants * can create references inside constants (`const X: &u32 = &22`) * Tuple struct constructors can be used in constants * regression in const eval errors spans (some of these need improvements in mir debug info) * can cast floats to ints and vice versa (in constants, and even nan/inf constants) * Mir dump prints false/true instead of 0u8/1u8 * `1i8 >> [8][0]` does not lint about exceeding bitshifts anymore. * Needs const propagation across projections * `foo[I]` produces a const eval lint if `foo: [T; N]` and `N < I` * Essentially all builtin panics produce lints if they can be statically proven to trigger at runtime. This is on a best effort basis, so there might be some complex cases that don't trigger. (The runtime panic stays there, irrelevant of whether the lint is produced or not) * can use `union`s to implement `transmute` for `Copy` types in constants without a feature gate. With all the greatness and nasal demons that come with this. * can convert integers to `&'static T` in constants (useful for embedded) fixes #34997 (stack overflow with many constants) fixes #25574 (deref byte strings in patterns) fixes #27918 (broken mir ICE) fixes #46114 (ICE on struct constructors in patterns) fixes #37448 (`SomeStruct { foo } as SomeStruct`) fixes #43754 (`return` in const fn) fixes #41898 (tuple struct constructors) fixes #31364 (infinite recursion with const fn, fixed by miri's recursion limit) closes #29947 (const indexing stabilization) fixes #45044 (pattern matching repeat expressions) fixes #47971 (ICE on const fn + references) fixes #48081 (ICE on cyclic assoc const error) fixes #48746 (nonhelpful error message with unions) r? @eddyb even though 1k loc are added in tests, this PR reduces the loc in this repository by 700

@eddyb

Replace all const evaluation with miri * error reporting in constants prints a stacktrace through all called const fns * Trivial constant propagation and folding in MIR (always active, irrelevant of the optimization level) * can now use floating constants in patterns (previously only floating point literals were allowed) * the future compat lint is still produced for both cases * can index into constant arrays during const eval (previously feature gated) * can create a constant union value with field `a` and read from field `b` * can dereference references into constants * can create references inside constants (`const X: &u32 = &22`) * Tuple struct constructors can be used in constants * regression in const eval errors spans (some of these need improvements in mir debug info) * can cast floats to ints and vice versa (in constants, and even nan/inf constants) * Mir dump prints false/true instead of 0u8/1u8 * `1i8 >> [8][0]` does not lint about exceeding bitshifts anymore. * Needs const propagation across projections * `foo[I]` produces a const eval lint if `foo: [T; N]` and `N < I` * Essentially all builtin panics produce lints if they can be statically proven to trigger at runtime. This is on a best effort basis, so there might be some complex cases that don't trigger. (The runtime panic stays there, irrelevant of whether the lint is produced or not) * can use `union`s to implement `transmute` for `Copy` types in constants without a feature gate. With all the greatness and nasal demons that come with this. * can convert integers to `&'static T` in constants (useful for embedded) fixes #34997 (stack overflow with many constants) fixes #25574 (deref byte strings in patterns) fixes #27918 (broken mir ICE) fixes #46114 (ICE on struct constructors in patterns) fixes #37448 (`SomeStruct { foo } as SomeStruct`) fixes #43754 (`return` in const fn) fixes #41898 (tuple struct constructors) fixes #31364 (infinite recursion with const fn, fixed by miri's recursion limit) closes #29947 (const indexing stabilization) fixes #45044 (pattern matching repeat expressions) fixes #47971 (ICE on const fn + references) fixes #48081 (ICE on cyclic assoc const error) fixes #48746 (nonhelpful error message with unions) r? @eddyb even though 1k loc are added in tests, this PR reduces the loc in this repository by 700

polazarus changed the title ~~Relatively short chained constant expressions causes stack overflow~~ Chained constant expressions causes stack overflow Jul 23, 2016

steveklabnik added the A-compiler label Jul 25, 2016

steveklabnik added T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. and removed A-compiler labels Mar 24, 2017

Mark-Simulacrum added the A-const-fn label Jun 20, 2017

Mark-Simulacrum added the C-bug Category: This is a bug. label Jul 25, 2017

oli-obk mentioned this issue Jan 16, 2018

Replace all const evaluation with miri #46882

Merged

bors closed this as completed in #46882 Mar 8, 2018

RalfJung added the A-const-eval Area: Constant evaluation, covers all const contexts (static, const fn, ...) label Dec 1, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Chained constant expressions causes stack overflow #34997

Chained constant expressions causes stack overflow #34997

polazarus commented Jul 23, 2016

oli-obk commented Jul 25, 2016

Chained constant expressions causes stack overflow #34997

Chained constant expressions causes stack overflow #34997

Comments

polazarus commented Jul 23, 2016

oli-obk commented Jul 25, 2016