about summary refs log tree commit diff
path: root/src/librustc_codegen_ssa/back
AgeCommit message (Collapse)AuthorLines
2020-05-10Auto merge of #72020 - alexcrichton:fix-incremental-linker-plugin-lto, r=oli-obkbors-30/+39
Fix disagreeement about CGU reuse and LTO This commit fixes an issue where the codegen backend's selection of LTO disagreed with what the codegen later thought was being done. Discovered in #72006 we have a longstanding issue where if `-Clinker-plugin-lto` in optimized mode is compiled incrementally it will always panic on the second compilation. The underlying issue turned out to be that the production of the original artifact determined that LTO should not be done (because it's being postponed to the linker) but the CGU reuse selection thought that LTO was done so it was trying to load pre-LTO artifacts which were never generated. The fix here is to ensure that the logic when generating code which determines what kind of LTO is being done is shared amongst the CGU reuse decision and the backend actually doing LTO. This means that they'll both be in agreement about whether the previous compilation did indeed produce incremental pre-LTO artifacts. Closes #72006
2020-05-09Fix disagreeement about CGU reuse and LTOAlex Crichton-30/+39
This commit fixes an issue where the codegen backend's selection of LTO disagreed with what the codegen later thought was being done. Discovered in #72006 we have a longstanding issue where if `-Clinker-plugin-lto` in optimized mode is compiled incrementally it will always panic on the second compilation. The underlying issue turned out to be that the production of the original artifact determined that LTO should not be done (because it's being postponed to the linker) but the CGU reuse selection thought that LTO was done so it was trying to load pre-LTO artifacts which were never generated. The fix here is to ensure that the logic when generating code which determines what kind of LTO is being done is shared amongst the CGU reuse decision and the backend actually doing LTO. This means that they'll both be in agreement about whether the previous compilation did indeed produce incremental pre-LTO artifacts. Closes #72006
2020-05-08Simplify bitcode embedding - either None or FullTom Karpiniec-16/+6
2020-05-07Provide configurable LLVM cmdline section via target specTom Karpiniec-0/+2
The App Store performs certain sanity checks on bitcode, including that an acceptable set of command line arguments was used when compiling a given module. For Rust code to be distributed on the app store with bitcode rustc must pretend to have the same command line arguments.
2020-05-07Force embed-bitcode on non-simulator iOS/tvOS targetsTom Karpiniec-0/+2
At this time Apple recommends Bitcode be included for iOS apps, and requires it for tvOS. It is unlikely that a developer would want to disable bitcode when building for these targets, yet by default it will not be generated. This presents a papercut for developers on those platforms. Introduces a new TargetOption boolean key for specific triples to indicate that bitcode should be generated, even if cargo attempts to optimise with -Cembed-bitcode=no.
2020-05-04Auto merge of #71754 - alexcrichton:no-bitcode-in-cache, r=nnethercotebors-21/+6
Don't copy bytecode files into the incr. comp. cache. It's no longer necessary now that bitcode is embedded into object files. This change meant that `WorkProductFileKind::Bytecode` is no longer necessary, which means that type is no longer necessary, which allowed several places in the code to become simpler. This commit was written by @nnethercote in https://github.com/rust-lang/rust/pull/70458 but that didn't land. In the meantime though we managed to land it in https://github.com/rust-lang/rust/pull/71528 and that doesn't seem to be causing too many fires, so I'm re-sending this patch!
2020-05-02cleanup: `config::CrateType` -> `CrateType`Vadim Petrochenkov-52/+49
2020-05-02Auto merge of #71716 - alexcrichton:bitcode-follow-up, r=nnethercotebors-2/+2
Rename `bitcode-in-rlib` option to `embed-bitcode` This commit finishes work first pioneered in #70458 and started in #71528. The `-C bitcode-in-rlib` option, which has not yet reached stable, is renamed to `-C embed-bitcode` since that more accurately reflects what it does now anyway. Various tests and such are updated along the way as well. This'll also need to be backported to the beta channel to ensure we don't accidentally stabilize `-Cbitcode-in-rlib` as well.
2020-05-01Rename `bitcode-in-rlib` option to `embed-bitcode`Alex Crichton-2/+2
This commit finishes work first pioneered in #70458 and started in #71528. The `-C bitcode-in-rlib` option, which has not yet reached stable, is renamed to `-C embed-bitcode` since that more accurately reflects what it does now anyway. Various tests and such are updated along the way as well. This'll also need to be backported to the beta channel to ensure we don't accidentally stabilize `-Cbitcode-in-rlib` as well.
2020-05-01Don't copy bytecode files into the incr. comp. cache.Nicholas Nethercote-21/+6
It's no longer necessary now that bitcode is embedded into object files. This change meant that `WorkProductFileKind::Bytecode` is no longer necessary, which means that type is no longer necessary, which allowed several places in the code to become simpler.
2020-05-01Auto merge of #71623 - petrochenkov:localink, r=estebankbors-1/+16
Disable localization for all linkers We previously disabled non-English output from `link.exe` due to encoding issues (#35785). In https://github.com/rust-lang/rust/pull/70740 it was pointed out that it also prevents correct inspection of the linker output, which we have to do occasionally. So this PR disables localization for all linkers.
2020-05-01Auto merge of #70674 - cjgillot:query-arena-all, r=matthewjasperbors-8/+5
Have the per-query caches store the results on arenas This PR leverages the cache for each query to serve as storage area for the query results. It introduces a new cache `ArenaCache`, which moves the result to an arena, and only stores the reference in the hash map. This allows to remove a sizeable part of the usage of the global `TyCtxt` arena. I only migrated queries that already used arenas before.
2020-04-30Auto merge of #70175 - Amanieu:remove_nlp, r=pnkfelixbors-2/+2
Remove -Z no-landing-pads flag Since #67502, `-Z no-landing-pads` will cause all attempted unwinds to abort since we don't generate a `try` / `catch`. This previously worked because `__rust_try` was located in libpanic_unwind which is always compiled with `-C panic=unwind`, but `__rust_try` is now directly inline into the crate that uses `catch_unwind`. As such, `-Z no-landing-pads` is now mostly useless and people should use `-C panic=abort` instead.
2020-04-29Auto merge of #71528 - alexcrichton:no-more-bitcode, r=nnethercotebors-49/+5
Store LLVM bitcode in object files, not compressed This commit is an attempted resurrection of #70458 where LLVM bitcode emitted by rustc into rlibs is stored into object file sections rather than in a separate file. The main rationale for doing this is that when rustc emits bitcode it will no longer use a custom compression scheme which makes it both easier to interoperate with existing tools and also cuts down on compile time since this compression isn't happening. The blocker for this in #70458 turned out to be that native linkers didn't handle the new sections well, causing the sections to either trigger bugs in the linker or actually end up in the final linked artifact. This commit attempts to address these issues by ensuring that native linkers ignore the new sections by inserting custom flags with module-level inline assembly. Note that this does not currently change the API of the compiler at all. The pre-existing `-C bitcode-in-rlib` flag is co-opted to indicate whether the bitcode should be present in the object file or not. Finally, note that an important consequence of this commit, which is also one of its primary purposes, is to enable rustc's `-Clto` bitcode loading to load rlibs produced with `-Clinker-plugin-lto`. The goal here is that when you're building with LTO Cargo will tell rustc to skip codegen of all intermediate crates and only generate LLVM IR. Today rustc will generate both object code and LLVM IR, but the object code is later simply thrown away, wastefully.
2020-04-29Store LLVM bitcode in object files, not compressedAlex Crichton-49/+5
This commit is an attempted resurrection of #70458 where LLVM bitcode emitted by rustc into rlibs is stored into object file sections rather than in a separate file. The main rationale for doing this is that when rustc emits bitcode it will no longer use a custom compression scheme which makes it both easier to interoperate with existing tools and also cuts down on compile time since this compression isn't happening. The blocker for this in #70458 turned out to be that native linkers didn't handle the new sections well, causing the sections to either trigger bugs in the linker or actually end up in the final linked artifact. This commit attempts to address these issues by ensuring that native linkers ignore the new sections by inserting custom flags with module-level inline assembly. Note that this does not currently change the API of the compiler at all. The pre-existing `-C bitcode-in-rlib` flag is co-opted to indicate whether the bitcode should be present in the object file or not. Finally, note that an important consequence of this commit, which is also one of its primary purposes, is to enable rustc's `-Clto` bitcode loading to load rlibs produced with `-Clinker-plugin-lto`. The goal here is that when you're building with LTO Cargo will tell rustc to skip codegen of all intermediate crates and only generate LLVM IR. Today rustc will generate both object code and LLVM IR, but the object code is later simply thrown away, wastefully.
2020-04-29Remove Session::no_landing_pads()Amanieu d'Antras-2/+2
2020-04-28Use the query system to allocate.Camille GILLOT-8/+5
2020-04-27Disable localization for all linkersVadim Petrochenkov-1/+16
2020-04-26codegen_llvm: Simplify logic for relaxing PIC into PIEVadim Petrochenkov-1/+1
2020-04-26rustc_target: Stop using "string typing" for relocation modelsVadim Petrochenkov-11/+2
Introduce `enum RelocModel` instead.
2020-04-25Rollup merge of #71544 - cuviper:filter_map_next, r=Mark-SimulacrumDylan DPC-5/+1
Replace filter_map().next() calls with find_map() These are semantically the same, but `find_map()` is more concise.
2020-04-25Rollup merge of #71364 - Amanieu:zprofile_compiler_builtins, r=cramertjDylan DPC-5/+18
Ignore -Zprofile when building compiler_builtins #70846 made the `compiler_builtins` crate ignore the default codegen-units setting and instead always split each function into a different codegen unit. This unfortunately breaks `-Zprofile` which requires a single codegen unit per crate (see #71283). You can notice this when building with `cargo -Zbuild-std` and `RUSTFLAGS` containing `-Zprofile`. This PR works around this issue by just ignoring `-Zprofile` for the `compiler-builtins` crate.
2020-04-24Replace filter_map().next() calls with find_map()Josh Stone-5/+1
These are semantically the same, but `find_map()` is more concise.
2020-04-23Modify `as_local_hir_id` to return a bare `HirId`marmeladema-1/+1
2020-04-23Modify `as_local_hir_id` to accept a `LocalDefId` instead of a `DefId`marmeladema-2/+2
2020-04-23librustc_middle: return LocalDefId instead of DefId in local_def_idmarmeladema-5/+5
2020-04-22Add a new option `-Cbitcode-in-rlib`.Nicholas Nethercote-1/+2
It defaults to true, but Cargo will set this to false whenever it can to reduce compile times.
2020-04-20Ignore -Zprofile when building compiler_builtinsAmanieu d'Antras-5/+18
2020-04-20Auto merge of #70729 - nnethercote:a-big-options-clean-up, r=petrochenkovbors-19/+6
A big options clean-up Lots of improvements here. r? @Centril
2020-04-19Dogfood more or_patterns in the compilerJosh Stone-3/+5
2020-04-19Replace uses of `parse_opt_*` with `parse_*` where possible.Nicholas Nethercote-19/+6
This lets us specify the default at the options declaration point, instead of using `.unwrap(default)` or `None | Some(default)` at some use point far away. It also makes the code more concise.
2020-04-16Rollup merge of #71145 - pfmooney:illumos-triple, r=nagisaDylan DPC-1/+1
Add illumos triple This fixes rust-lang/rust#55553 and adds support for `illumos` as a `target_os` on `x86_64`. In addition to the compile spec and libstd additions, several library dependencies have been bumped in order to permit working builds of cargo and rustup for the new target. Work originally started by @jasonbking, with subsequent additions by @pfmooney and @jclulow.
2020-04-14Add illumos triplePatrick Mooney-1/+1
Co-Authored-By: Jason King <jason.brian.king@gmail.com> Co-Authored-By: Joshua M. Clulow <jmc@oxide.computer>
2020-04-13linker: Pass `/NODEFAULTLIB` in a more regular wayVadim Petrochenkov-16/+3
2020-04-13rustc_target: Make sure lld-link is treated as link.exe by defaultVadim Petrochenkov-2/+4
The differences if they are discovered will need to be explicitly documented
2020-04-11Auto merge of #71031 - Dylan-DPC:rollup-zr8hh86, r=Dylan-DPCbors-162/+140
Rollup of 5 pull requests Successful merges: - #70644 (Clean up `ModuleConfig` initialization) - #70937 (Fix staticlib name for *-pc-windows-gnu targets) - #70996 (Add or_insert_with_key to Entry of HashMap/BTreeMap) - #71020 (Store UNICODE_VERSION as a tuple) - #71021 (Use write!-style syntax for MIR assert terminator) Failed merges: r? @ghost
2020-04-11Rollup merge of #70644 - nnethercote:clean-up-ModuleConfig-init, ↵Dylan DPC-162/+140
r=Mark-Simulacrum Clean up `ModuleConfig` initialization Because it's currently a mess. r? @Mark-Simulacrum
2020-04-11Auto merge of #70161 - cjgillot:query-arena, r=nikomatsakisbors-4/+3
Allocate some query results on an arena This avoids a cloning few `Lrc` and `Vec`s in the queries.
2020-04-09Rollup merge of #70868 - petrochenkov:linkorder, r=nagisa,mati865Mazdak Farrokhzad-261/+413
rustc_codegen_ssa: Refactor construction of linker arguments And add comments. This PR doesn't reorder any linker arguments and therefore shouldn't contain any observable changes. The next goal here is to - Factor out order-independent linker arguments in the compiler code and in target specifications and pass them together. Such arguments generally apply to the whole linking session or the produced linking result rather to individual object files or libraries. - Figure out where exactly among the remaining order-dependent arguments we should place customization points like `-C link-args` and `-Z pre-link-args`. - Possibly provide command line opt-outs for options that are currently passed unconditionally (like CRT objects or arguments defined by the target spec). - Document and stabilize the customization points that are not yet stable (https://github.com/rust-lang/rust/pull/70505).
2020-04-07linker: Some minor code cleanupVadim Petrochenkov-58/+38
2020-04-07Use assoc integer constants in librustc_*Linus Färnstrand-1/+1
2020-04-07linker: Factor out more parts of `linker_with_args` and add some docsVadim Petrochenkov-121/+183
2020-04-07linker: Add more markup and comments to code producing linker argumentsVadim Petrochenkov-23/+61
2020-04-07linker: Factor out addition of pre-, post- and late link argsVadim Petrochenkov-35/+82
2020-04-07linker: Factor out linking of pre- and post-link objectsVadim Petrochenkov-23/+37
2020-04-07linker: Combine argument building into a single functionVadim Petrochenkov-115/+118
2020-04-07linker: Make argument building interface in `trait Linker` richerVadim Petrochenkov-24/+32
by redirecting everything to `Command`
2020-04-05Remove Arcs in queries.Camille GILLOT-4/+3
2020-04-04Do not lose or reorder user-provided linker argumentsVadim Petrochenkov-16/+8
2020-04-01Rename `modules_config` as `regular_config`.Nicholas Nethercote-4/+4
That way it matches `ModuleKind::Regular`.
generated by cgit-pink 1.4.1-3-g733a5 (git 2.37.1) at 2026-02-25 11:17:06 +0000