Fix feature gate for `#[link_args(..)]` attribute
Fix feature gate for `#[link_args(..)]` attribute so that it will fire regardless of context of attribute.
See also #29596 and #43106
Store all generic arguments for method calls in AST/HIR
The first part of https://github.com/rust-lang/rust/pull/42492.
Landed separately to start the process of merging libsyntax changes breaking rustfmt, which is not easy these days.
Fix stage 2 builds with a custom libdir.
When copying libstd for the stage 2 compiler, the builder ignores the
configured libdir/libdir_relative configuration parameters. This causes
the compiler to fail to find libstd, which cause any tools built with the
stage 2 compiler to fail.
To fix this, make the copy steps of rustbuild aware of the libdir_relative
parameter when the stage >= 2. Also update the dist target to be aware of
the new location of libstd.
Implemented `TypeFoldable` for `TypeError`s.
This is quite handy in some user code, for instance to pull out type errors from an inference context when `fresh_substs_for_item` has been used before.
Remove unused code from librustc_errors
While extracting librustc_errors into a [reusable library](https://github.com/kevinmehall/codemap-diagnostic), I noticed some obsolete code that the `dead_code` warning missed because it was marked `pub` but not used elsewhere.
Fold E0612, E0613 into E0609
As discussed in #42945, with PR 1506 tuple indices are no longer considered a separate case from normal field. This PR folds E06012 ("tuple index out of bounds") and E0613 ("type is not a tuple") into E0609 ("type does not have field with that name")
Resolves#42945
Fix Rustbuild linking on Illumos
Illumos (an OpenSolaris fork) expects to get several extra library references for some system functions used by Rust standard library. This commit adds required linker options to rustbuild, which is currently doesn't work on Illumos-based operating systems.
Implement O(1)-time Iterator::nth for Range*, and slim the Step trait
Fixes#43064.
Fixes part of #39975.
Fixes items 1 <s>and 3</s> of #42168.
CC #27741.
I think #42310 and #43012 should not have landed without the `nth` part of this PR, but oh well.
Skip the main thread's manual stack guard on Linux
Linux doesn't allocate the whole stack right away, and the kernel has its own stack-guard mechanism to fault when growing too close to an existing mapping. If we map our own guard, then the kernel starts enforcing a rather large gap above that, rendering much of the possible stack space useless.
Instead, we'll just note where we expect rlimit to start faulting, so our handler can report "stack overflow", and trust that the kernel's own stack guard will work.
Fixes#43052.
r? @alexcrichton
### Kernel compatibility:
Strictly speaking, Rust claims support for Linux kernels >= 2.6.18, and stack guards were only added to mainline in 2.6.36 for [CVE-2010-2240]. But since that vulnerability was so severe, the guards were backported to many stable branches, and Red Hat patched this all the way back to RHEL3's 2.4.21! I think it's reasonable for us to assume that any *supportable* kernel should have these stack guards.
At that time, the kernel only enforced one page of padding between the stack and other mappings, but thanks to [Stack Clash] that padding is now much larger, causing #43052. The kernel side of those fixes are in [CVE-2017-1000364], which Red Hat has backported to at least RHEL5's 2.6.18 so far.
[CVE-2010-2240]: https://access.redhat.com/security/cve/CVE-2010-2240
[CVE-2017-1000364]: https://access.redhat.com/security/cve/CVE-2017-1000364
[Stack Clash]: https://access.redhat.com/security/vulnerabilities/stackguard
When copying libstd for the stage 2 compiler, the builder ignores the
configured libdir/libdir_relative configuration parameters. This causes
the compiler to fail to find libstd, which cause any tools built with the
stage 2 compiler to fail.
To fix this, make the copy steps of rustbuild aware of the libdir_relative
parameter when the stage >= 2. Also update the dist target to be aware of
the new location of libstd.
Replace the global fulfillment cache with the evaluation cache
This uses the new "Chalk" ParamEnv refactoring to check "global" predicates in an empty environment, which should be correct because global predicates aren't affected by a consistent environment.
Fixes#39970.
Fixes#42796.
r? @nikomatsakis
This is just duplicating the logic from the old fulfillment cache, so
I'm not sure it is 100% correct, but it should not be more wrong than
the old logic.
Linux doesn't allocate the whole stack right away, and the kernel has
its own stack-guard mechanism to fault when growing too close to an
existing mapping. If we map our own guard, then the kernel starts
enforcing a rather large gap above that, rendering much of the possible
stack space useless.
Instead, we'll just note where we expect rlimit to start faulting, so
our handler can report "stack overflow", and trust that the kernel's own
stack guard will work.
Fixes#43052.
add a built-in MSP430 target
the MSP430 backend has been enabled for a while but no target was added to rustc
to encourage out of tree experimentation.
We believe the out of tree (custom) target has been iterated long enough and is
stable enough for inclusion in the compiler. Kudos to @pftbest and @awygle for
fixing several LLVM / codegen bugs this target had!
The target name chosen is a slight variation of the triple gcc uses, which is
simply `msp430-elf`. We picked `msp430-none-elf` to leave room for custom
targets that target some embedded OS running on MSP430 devices. (cf. the
custom `thumbv7m-tockos-eabi` target TockOS uses vs the built-in
`thumbv7m-none-eabi`).
There's one expected change in the specification of the proposed target: the
`asm_args` and `no_integrated_as` fields will change to their default values.
Once the LLVM backend gains the ability to directly produce MSP430 object files
we can stop depending on `msp430-elf-gcc` for producing object files; when that
occurs the `asm` related fields will change. This change won't break existing
user code.
r? @alexcrichton
cc @brson
