interpret: make read-pointer-as-bytes a CTFE-only error with extra information #101101

RalfJung · 2022-08-27T21:50:55Z

Next step in the reaction to #99923. Also teaches Miri to implicitly strip provenance in more situations when transmuting pointers to integers, which fixes rust-lang/miri#2456.

Pointer-to-int transmutation during CTFE now produces a message like this:

   = help: this code performed an operation that depends on the underlying bytes representing a pointer
   = help: the absolute address of a pointer is not known at compile-time, so such operations are not supported

r? @oli-obk

rustbot · 2022-08-27T21:50:58Z

Some changes occurred to the CTFE / Miri engine

cc @rust-lang/miri

Some changes occurred to the CTFE / Miri engine

cc @rust-lang/miri

Some changes occurred in compiler/rustc_codegen_gcc

cc @antoyo

Some changes occurred in compiler/rustc_codegen_cranelift

cc @bjorn3

and show some extra information when it happens in CTFE

saethlin · 2022-08-28T16:43:17Z

compiler/rustc_middle/src/mir/interpret/allocation.rs

+            // We are *not* reading a pointer.
+            // If we can just ignore provenance, do exactly that.
+            if Prov::OFFSET_IS_ADDR {


The fact that OFFSET_IS_ADDR also indicates whether we support plain reads from bytes with provenance seems like a logical jump. Even the docs for this const only say this:

If `false`, ptr-to-int casts are not supported. The offset *must* be relative in that case.

which is identifying ptr-to-int casts which someone familiar with Miri may be sure is different from a ptr-to-int transmute. So I think this jump from OFFSET_IS_ADDR to "provenance can be discarded" isn't (clearly) documented anywhere that's easy to find, and it would be nice if these branches on the const were more clear.

Can we have something like Prov::can_discard_provenance() or Prov::permits_ptr_int_transmute() then branch on that in these places? Such a function would also be a good place to explain the connection.

Yeah this is really not about casts any more, it is about transmutation. I have updated the docs.

There is a 1:1 correlation between "we can just ignore provenance and get a meaningful result" and "the offset is a pointer is the absolute address": the reason we can strip provenance is that the pointer offset is the absolute address, so revealing the pointer bytes to the interpreted program is meaningful.

I have updated the comment, I hope it is more clear now.

oli-obk · 2022-08-29T10:43:59Z

@bors r+

took some brain mangling to grok, even with the comments, but I don't think it can be improved, it's just a complex operation

bors · 2022-08-29T10:44:00Z

📌 Commit c46e803 has been approved by oli-obk

It is now in the queue for this repository.

…oli-obk interpret: make read-pointer-as-bytes a CTFE-only error with extra information Next step in the reaction to rust-lang#99923. Also teaches Miri to implicitly strip provenance in more situations when transmuting pointers to integers, which fixes rust-lang/miri#2456. Pointer-to-int transmutation during CTFE now produces a message like this: ``` = help: this code performed an operation that depends on the underlying bytes representing a pointer = help: the absolute address of a pointer is not known at compile-time, so such operations are not supported ``` r? `@oli-obk`

Rollup of 9 pull requests Successful merges: - rust-lang#95376 (Add `vec::Drain{,Filter}::keep_rest`) - rust-lang#100092 (Fall back when relating two opaques by substs in MIR typeck) - rust-lang#101019 (Suggest returning closure as `impl Fn`) - rust-lang#101022 (Erase late bound regions before comparing types in `suggest_dereferences`) - rust-lang#101101 (interpret: make read-pointer-as-bytes a CTFE-only error with extra information) - rust-lang#101123 (Remove `register_attr` feature) - rust-lang#101175 (Don't --bless in pre-push hook) - rust-lang#101176 (rustdoc: remove unused CSS selectors for `.table-display`) - rust-lang#101180 (Add another MaybeUninit array test with const) Failed merges: r? `@ghost` `@rustbot` modify labels: rollup

Adjust for supporting more implicit ptr-to-int transmutation This is the Miri side of rust-lang/rust#101101. Fixes #2456.

…=pnkfelix beta-backport of provenance-related CTFE changes This is all part of dealing with rust-lang#99923. The first commit backports the effects of rust-lang#101101. `@pnkfelix` asked for this and it turned out to be easy, so I think this is uncontroversial. The second commit effectively repeats rust-lang#99965, which un-does the effects of rust-lang#97684 and therefore means rust-lang#99923 does not apply to the beta branch. I honestly don't think we should do this; the sentiment in rust-lang#99923 was that we should go ahead with the change but improve diagnostics. But `@pnkfelix` seemed to request such a change so I figured I would offer the option. I'll be on vacation soon, so if you all decide to take the first commit only, then someone please just force-push to this branch and remove the 2nd commit.

Fixes model-checking#1615 Relevant changes to rustc: - rust-lang/rust#101483: Change to intrinsics. - rust-lang/rust#94075: Change to niche opt. - rust-lang/rust#101101: Method rename.

Fixes model-checking#1615 Relevant changes to rustc: - rust-lang/rust#101483: Change to intrinsics. - rust-lang/rust#94075: Change to niche opt. - rust-lang/rust#101101: Method rename. Co-authored-by: Adrian Palacios <73246657+adpaco-aws@users.noreply.github.com>

Fixes #1615 Relevant changes to rustc: - rust-lang/rust#101483: Change to intrinsics. - rust-lang/rust#94075: Change to niche opt. - rust-lang/rust#101101: Method rename. Co-authored-by: Adrian Palacios <73246657+adpaco-aws@users.noreply.github.com>

…oli-obk interpret: make read-pointer-as-bytes a CTFE-only error with extra information Next step in the reaction to rust-lang#99923. Also teaches Miri to implicitly strip provenance in more situations when transmuting pointers to integers, which fixes rust-lang/miri#2456. Pointer-to-int transmutation during CTFE now produces a message like this: ``` = help: this code performed an operation that depends on the underlying bytes representing a pointer = help: the absolute address of a pointer is not known at compile-time, so such operations are not supported ``` r? ``@oli-obk``

interpret: rename relocation → provenance

e63a625

rustbot added the T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. label Aug 27, 2022

rust-highfive assigned oli-obk Aug 27, 2022

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Aug 27, 2022

This was referenced Aug 27, 2022

Adjust for supporting more implicit ptr-to-int transmutation rust-lang/miri#2516

Merged

Regression in consteval: error[E0080]: could not evaluate static initializer (unable to turn pointer into raw bytes) #99923

Closed

RalfJung force-pushed the read-pointer-as-bytes branch 2 times, most recently from f3c61d1 to 24cf766 Compare August 27, 2022 22:27

RalfJung added 2 commits August 27, 2022 18:37

interpret: make read-pointer-as-bytes *always* work in Miri

2e17247

and show some extra information when it happens in CTFE

adjust tests

cf2c65b

RalfJung force-pushed the read-pointer-as-bytes branch from 24cf766 to cf2c65b Compare August 27, 2022 22:38

validation should only catch UB errors

1a1220c

RalfJung force-pushed the read-pointer-as-bytes branch from 7465f7d to 1a1220c Compare August 28, 2022 15:49

saethlin reviewed Aug 28, 2022

View reviewed changes

improve OFFSET_IS_ADDR docs

c46e803

RalfJung force-pushed the read-pointer-as-bytes branch from 02091ba to c46e803 Compare August 28, 2022 17:05

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Aug 29, 2022

RalfJung changed the title ~~interpret: make read-pointer-as-pointers a CTFE-only error with extra information~~ interpret: make read-pointer-as-bytes a CTFE-only error with extra information Aug 29, 2022

matthiaskrgr mentioned this pull request Aug 30, 2022

Rollup of 8 pull requests #101182

Closed

Dylan-DPC mentioned this pull request Aug 30, 2022

Rollup of 9 pull requests #101183

Merged

bors merged commit 81f3841 into rust-lang:master Aug 30, 2022

rustbot added this to the 1.65.0 milestone Aug 30, 2022

RalfJung deleted the read-pointer-as-bytes branch August 31, 2022 10:45

bors added a commit to rust-lang/miri that referenced this pull request Aug 31, 2022

Auto merge of #2516 - RalfJung:read-pointer-as-bytes, r=RalfJung

872bea4

Adjust for supporting more implicit ptr-to-int transmutation This is the Miri side of rust-lang/rust#101101. Fixes #2456.

bors added a commit to rust-lang/miri that referenced this pull request Aug 31, 2022

Auto merge of #2516 - RalfJung:read-pointer-as-bytes, r=RalfJung

0da5f76

Adjust for supporting more implicit ptr-to-int transmutation This is the Miri side of rust-lang/rust#101101. Fixes #2456.

RalfJung mentioned this pull request Sep 2, 2022

beta-backport of provenance-related CTFE changes #101320

Merged

celinval mentioned this pull request Sep 30, 2022

Upgrade toolchain to nightly-2022-09-13 model-checking/kani#1737

Merged

4 tasks

rust-timer mentioned this pull request Oct 24, 2022

Rollup of 7 pull requests #103471

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

interpret: make read-pointer-as-bytes a CTFE-only error with extra information #101101

interpret: make read-pointer-as-bytes a CTFE-only error with extra information #101101

RalfJung commented Aug 27, 2022 •

edited

Loading

rustbot commented Aug 27, 2022

saethlin Aug 28, 2022 •

edited

Loading

RalfJung Aug 28, 2022

oli-obk commented Aug 29, 2022

bors commented Aug 29, 2022

interpret: make read-pointer-as-bytes a CTFE-only error with extra information #101101

interpret: make read-pointer-as-bytes a CTFE-only error with extra information #101101

Conversation

RalfJung commented Aug 27, 2022 • edited Loading

rustbot commented Aug 27, 2022

saethlin Aug 28, 2022 • edited Loading

Choose a reason for hiding this comment

RalfJung Aug 28, 2022

Choose a reason for hiding this comment

oli-obk commented Aug 29, 2022

bors commented Aug 29, 2022

RalfJung commented Aug 27, 2022 •

edited

Loading

saethlin Aug 28, 2022 •

edited

Loading