trace: incorporate events #4456

nia-e · 2025-07-08T08:18:52Z

If tracing is enabled, use the info from the trace to actually update Miri's state. Pending a small change to rustc to expose some necessary interfaces to handle writes, this should be good!

nia-e · 2025-07-08T08:19:36Z

Ah, I forgot I rebased this on top of #4435. Just the last commit is relevant for now

src/shims/native_lib/mod.rs

RalfJung · 2025-07-08T20:18:58Z

What this PR definitely needs is some tests. :) Remember that tracing mode is off-by-default; none of the tests hit the tracing infra. So this needs some new native-mode tests that specifically check that we detect the ranges properly. For instance, sharing a 2-field uninit struct with C, having C initialize the first field, and then reading the 2nd field -- that should be UB.

nia-e · 2025-07-09T07:55:43Z

I actually wrote some! Just didn't know if they belong here. I'll add them, though

src/shims/native_lib/mod.rs

interpret/allocation: expose init + write_wildcards on a range Part of rust-lang/miri#4456, so that we can mark down when a foreign access to our memory happened. Should this also move `prepare_for_native_access()` itself into Miri, given that everything there can be implemented on Miri's side? r? `@RalfJung`

interpret/allocation: expose init + write_wildcards on a range Part of rust-lang/miri#4456, so that we can mark down when a foreign access to our memory happened. Should this also move `prepare_for_native_access()` itself into Miri, given that everything there can be implemented on Miri's side? r? ``@RalfJung``

interpret/allocation: expose init + write_wildcards on a range Part of rust-lang/miri#4456, so that we can mark down when a foreign access to our memory happened. Should this also move `prepare_for_native_access()` itself into Miri, given that everything there can be implemented on Miri's side? r? ```@RalfJung```

interpret/allocation: expose init + write_wildcards on a range Part of rust-lang/miri#4456, so that we can mark down when a foreign access to our memory happened. Should this also move `prepare_for_native_access()` itself into Miri, given that everything there can be implemented on Miri's side? r? ````@RalfJung````

interpret/allocation: expose init + write_wildcards on a range Part of rust-lang/miri#4456, so that we can mark down when a foreign access to our memory happened. Should this also move `prepare_for_native_access()` itself into Miri, given that everything there can be implemented on Miri's side? r? `````@RalfJung`````

Rollup merge of #143634 - nia-e:init-and-wildcards, r=RalfJung interpret/allocation: expose init + write_wildcards on a range Part of rust-lang/miri#4456, so that we can mark down when a foreign access to our memory happened. Should this also move `prepare_for_native_access()` itself into Miri, given that everything there can be implemented on Miri's side? r? `````@RalfJung`````

interpret/allocation: expose init + write_wildcards on a range Part of #4456, so that we can mark down when a foreign access to our memory happened. Should this also move `prepare_for_native_access()` itself into Miri, given that everything there can be implemented on Miri's side? r? `````@RalfJung`````

nia-e · 2025-07-14T07:48:32Z

I did readd logic (not much code) for whether an access spands multiple allocations since from what I understand this is theoretically allowed by LLVM at least and I'm a little scared it could happen in the wild; but I can replace that with just an assertion that accesses span a single allocation and ICE/error otherwise.

@rustbot ready

src/shims/native_lib/mod.rs

RalfJung

Okay that's really cool that this works, this is pure magic.^^

I still have a bunch of comments, of course. ;)

tests/native-lib/fail/trace_read.rs

tests/native-lib/fail/trace_read.stderr

tests/native-lib/fail/trace_write.rs

src/shims/native_lib/mod.rs

src/shims/native_lib/trace/parent.rs

nia-e · 2025-07-15T11:39:34Z

This should address review, lmk if there's anything more!

nia-e · 2025-07-15T12:00:08Z

nvm, seems like I messed something up, just a sec 😭

src/shims/native_lib/mod.rs

RalfJung · 2025-07-15T12:27:53Z

src/shims/native_lib/mod.rs

+                        let p_map = alloc.provenance();
+                        for idx in overlap {
+                            // If a provenance was read by the foreign code, expose it.
+                            if let Some(prov) = p_map.get(Size::from_bytes(idx), this) {


Please add a FIXME saying that we should really have a get_range on ProvenanceMap.

(Here's a next PR for you if you are interested. ;)

tests/native-lib/pass/ptr_read_access.rs

tests/ui.rs

RalfJung · 2025-07-15T12:31:33Z

@rustbot author

rustbot · 2025-07-15T12:31:37Z

Reminder, once the PR becomes ready for a review, use @rustbot ready.

nia-e · 2025-07-15T12:49:23Z

@rustbot ready

src/shims/native_lib/mod.rs

RalfJung · 2025-07-15T14:51:48Z

src/shims/native_lib/mod.rs

-                // The start of the overlap range will be the greater of the alloc base address and the
-                // lowest remaining address in the access' range; the end will be the lesser of the end
-                // of the allocation and the end of the access' range. Both are then shifted by `alloc_addr`
+                // The start of the overlap range will be `curr`; the end will be the lesser of the end of


Please make it clear that this range is in terms of offsets inside the allocation, i.e., relative to alloc_addr.

tests/ui.rs

RalfJung · 2025-07-15T14:53:41Z

This looks great, thanks! After resolving the last nits, please squash the commits using ./miri squash. Then write @rustbot ready after you force-pushed the squashed PR.

@rustbot author

tests allow accesses to span multiple allocs oopsie fixup rebase Apply suggestions from code review Co-authored-by: Ralf Jung <[email protected]> review apparently x86 has signed types for registers but x64 has unsigned... i give up doc for certain write more review last review

nia-e · 2025-07-15T21:42:41Z

@rustbot ready

nia-e force-pushed the trace-incorporate-events branch 2 times, most recently from acbf7f3 to a8694a4 Compare July 8, 2025 08:52

RalfJung added the S-blocked Status: blocked on something happening somewhere else label Jul 8, 2025

nia-e force-pushed the trace-incorporate-events branch 5 times, most recently from 4326ab5 to 180efe5 Compare July 8, 2025 09:30

nia-e commented Jul 8, 2025

View reviewed changes

src/shims/native_lib/mod.rs Outdated Show resolved Hide resolved

nia-e mentioned this pull request Jul 8, 2025

interpret/allocation: expose init + write_wildcards on a range rust-lang/rust#143634

Merged

nia-e commented Jul 9, 2025

View reviewed changes

src/shims/native_lib/mod.rs Outdated Show resolved Hide resolved

nia-e force-pushed the trace-incorporate-events branch 5 times, most recently from 6d91006 to 3d16495 Compare July 9, 2025 19:51

nia-e mentioned this pull request Jul 14, 2025

native-lib: allow passing structs as arguments #4466

Open

nia-e force-pushed the trace-incorporate-events branch from 5283a29 to 6454be8 Compare July 14, 2025 07:21

rustbot added S-waiting-on-review Status: Waiting for a review to complete and removed S-blocked Status: blocked on something happening somewhere else labels Jul 14, 2025

nia-e commented Jul 14, 2025

View reviewed changes

src/shims/native_lib/mod.rs Outdated Show resolved Hide resolved

RalfJung reviewed Jul 15, 2025

View reviewed changes

nia-e force-pushed the trace-incorporate-events branch from fc5760e to 94bf0d2 Compare July 15, 2025 11:42

nia-e force-pushed the trace-incorporate-events branch 2 times, most recently from b9e4399 to 59bc0a7 Compare July 15, 2025 12:30

RalfJung reviewed Jul 15, 2025

View reviewed changes

rustbot added S-waiting-on-author Status: Waiting for the PR author to address review comments and removed S-waiting-on-review Status: Waiting for a review to complete labels Jul 15, 2025

nia-e force-pushed the trace-incorporate-events branch from 8bfa229 to 8ed4303 Compare July 15, 2025 12:45

rustbot added S-waiting-on-review Status: Waiting for a review to complete and removed S-waiting-on-author Status: Waiting for the PR author to address review comments labels Jul 15, 2025

RalfJung reviewed Jul 15, 2025

View reviewed changes

src/shims/native_lib/mod.rs Outdated Show resolved Hide resolved

RalfJung reviewed Jul 15, 2025

View reviewed changes

tests/ui.rs Outdated Show resolved Hide resolved

rustbot added S-waiting-on-author Status: Waiting for the PR author to address review comments and removed S-waiting-on-review Status: Waiting for a review to complete labels Jul 15, 2025

nia-e force-pushed the trace-incorporate-events branch from 8ed4303 to 377c515 Compare July 15, 2025 21:42

rustbot added S-waiting-on-review Status: Waiting for a review to complete and removed S-waiting-on-author Status: Waiting for the PR author to address review comments labels Jul 15, 2025

trace: incorporate events #4456

Are you sure you want to change the base?

trace: incorporate events #4456

Conversation

nia-e commented Jul 8, 2025

Uh oh!

nia-e commented Jul 8, 2025

Uh oh!

Uh oh!

RalfJung commented Jul 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

nia-e commented Jul 9, 2025

Uh oh!

Uh oh!

nia-e commented Jul 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

RalfJung left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

nia-e commented Jul 15, 2025

Uh oh!

nia-e commented Jul 15, 2025

Uh oh!

Uh oh!

Uh oh!

RalfJung Jul 15, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

RalfJung commented Jul 15, 2025

Uh oh!

rustbot commented Jul 15, 2025

Uh oh!

nia-e commented Jul 15, 2025

Uh oh!

Uh oh!

RalfJung Jul 15, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

RalfJung commented Jul 15, 2025

Uh oh!

nia-e commented Jul 15, 2025

Uh oh!

Uh oh!

RalfJung commented Jul 8, 2025 •

edited

Loading

nia-e commented Jul 14, 2025 •

edited

Loading