summary refs log tree commit diff
path: root/src/ci/run.sh
AgeCommit message (Collapse)AuthorLines
2024-11-20ci: Disable full `debuginfo-level=2` in windows alt jobMarcoIeni-1/+1
2024-11-14Rollup merge of #132010 - cuviper:alt-full-debuginfo, r=Mark-SimulacrumGuillaume Gomez-1/+6
ci: Enable full `debuginfo-level=2` in `DEPLOY_ALT` It will be slower to build and produce larger artifacts, but hopefully it will help catch debuginfo regressions sooner, especially for problems that LLVM assertions would uncover. try-job: dist-x86_64-linux try-job: dist-x86_64-linux-alt
2024-11-12Auto merge of #132954 - matthiaskrgr:rollup-x3rww9h, r=matthiaskrgrbors-3/+1
Rollup of 7 pull requests Successful merges: - #131831 (extend the "if-unchanged" logic for compiler builds) - #132541 (Proper support for cross-crate recursive const stability checks) - #132657 (AIX: add run-make support) - #132901 (Warn about invalid `mir-enable-passes` pass names) - #132923 (Triagebot: Consolidate the T-compiler ad hoc assignment groups) - #132938 (Make precise capturing suggestion machine-applicable only if it has no APITs) - #132947 (clarify `must_produce_diag` ICE for debugging) r? `@ghost` `@rustbot` modify labels: rollup
2024-11-12Auto merge of #132282 - Noratrieb:it-is-the-end-of-serial, r=cjgillotbors-4/+1
Delete the `cfg(not(parallel))` serial compiler Since it's inception a long time ago, the parallel compiler and its cfgs have been a maintenance burden. This was a necessary evil the allow iteration while not degrading performance because of synchronization overhead. But this time is over. Thanks to the amazing work by the parallel working group (and the dyn sync crimes), the parallel compiler has now been fast enough to be shipped by default in nightly for quite a while now. Stable and beta have still been on the serial compiler, because they can't use `-Zthreads` anyways. But this is quite suboptimal: - the maintenance burden still sucks - we're not testing the serial compiler in nightly Because of these reasons, it's time to end it. The serial compiler has served us well in the years since it was split from the parallel one, but it's over now. Let the knight slay one head of the two-headed dragon! #113349 Note that the default is still 1 thread, as more than 1 thread is still fairly broken. cc `@onur-ozkan` to see if i did the bootstrap field removal correctly, `@SparrowLii` on the sync parts
2024-11-12Delete the `cfg(not(parallel))` serial compilerNoratrieb-4/+1
Since it's inception a long time ago, the parallel compiler and its cfgs have been a maintenance burden. This was a necessary evil the allow iteration while not degrading performance because of synchronization overhead. But this time is over. Thanks to the amazing work by the parallel working group (and the dyn sync crimes), the parallel compiler has now been fast enough to be shipped by default in nightly for quite a while now. Stable and beta have still been on the serial compiler, because they can't use `-Zthreads` anyways. But this is quite suboptimal: - the maintenance burden still sucks - we're not testing the serial compiler in nightly Because of these reasons, it's time to end it. The serial compiler has served us well in the years since it was split from the parallel one, but it's over now. Let the knight slay one head of the two-headed dragon!
2024-11-11ci: Enable full `debuginfo-level=2` in `DEPLOY_ALT`Josh Stone-1/+6
It will be slower to build and produce larger artifacts, but hopefully it will help catch debuginfo regressions sooner, especially for problems that LLVM assertions would uncover.
2024-11-11Revert "do not trust download-rustc=if-unchanged on CI for now"onur-ozkan-3/+1
This reverts commit b3c212103b826cde383093fab2f4237bb5736923.
2024-11-10do not trust download-rustc=if-unchanged on CI for nowRalf Jung-1/+3
2024-10-25remove `change-id` from CI scriptonur-ozkan-5/+0
Signed-off-by: onur-ozkan <work@onurozkan.dev>
2024-10-08Fix quotation marks around debug line in `src/ci/run.sh`Josh Stone-1/+1
Without this change, the markdown-style backticks are treated as a shell command substitution, which fails like so: /checkout/src/ci/run.sh: line 58: DISABLE_CI_RUSTC_IF_INCOMPATIBLE: command not found debug: configured.
2024-10-08force `download-rustc=if-unchanged` for x86_64-gnu-tools runneronur-ozkan-0/+1
Signed-off-by: onur-ozkan <work@onurozkan.dev>
2024-10-08fix `ci_rustc_if_unchanged_logic` testonur-ozkan-1/+2
Signed-off-by: onur-ozkan <work@onurozkan.dev>
2024-10-08handle CI rustc incompatible runnersonur-ozkan-0/+5
Signed-off-by: onur-ozkan <work@onurozkan.dev>
2024-10-08disable CI rustc when not using CI LLVMonur-ozkan-0/+2
Signed-off-by: onur-ozkan <work@onurozkan.dev>
2024-10-08use precompiled rustc for non-dist builders by defaultonur-ozkan-0/+4
Signed-off-by: onur-ozkan <work@onurozkan.dev>
2024-07-16Fix git safe-directory path for docker imagesEric Huss-1/+1
2024-06-28disable lld if external llvm is usedonur-ozkan-0/+4
Signed-off-by: onur-ozkan <work@onurozkan.dev>
2024-05-30compiletest: clarify COMPILETEST_NEEDS_ALL_LLVM_COMPONENTS errorRalf Jung-1/+1
also improve wording for an ignore reason
2024-05-29ci: Consolidate $IS_NOT_LATEST_LLVM into $EXTERNAL_LLVMJubilee Young-6/+2
We want to only demand that we check for all components we expect if we actually built the components we expect, which means we built the LLVM. Otherwise, it isn't worth checking.
2024-02-16Changes to CI related to mingw and MSYSMatt Harding-1/+1
2024-01-26Skip building cranelift for FuchsiaTyler Mandry-3/+5
This refactors run.sh to never override an explicit $CODEGEN_BACKENDS if set in the build.
2024-01-07Auto merge of #119556 - onur-ozkan:optimized-compiler-builtins, r=onur-ozkanbors-0/+9
Reland optimized-compiler-builtins config Copy of #102579 PR. From #102579: > No concerns on my side. Currently, Jyn isn't actively working on the project. I will close this PR; open another one to cherry-pick the commits, resolve conflicts, and then r+ it. > Fixes https://github.com/rust-lang/rust/issues/102560. Fixes https://github.com/rust-lang/rust/issues/101172. Helps with https://github.com/rust-lang/rust/issues/105065 (although there's some weirdness there - it's still broken when optimized-compiler-builtins is set to true). Fixes https://github.com/rust-lang/rust/issues/102560. Fixes https://github.com/rust-lang/rust/issues/101172. Helps with https://github.com/rust-lang/rust/issues/105065 r? ghost
2024-01-07add a new `optimized_compiler_builtins` optionjyn-0/+9
in particular, this makes the `c` feature for compiler-builtins an explicit opt-in, rather than silently detected by whether `llvm-project` is checked out on disk. exposing this is necessary because the `cc` crate doesn't support cross-compiling to MSVC, and we want people to be able to run `x check --target foo` regardless of whether they have a c toolchain available. this also uses the new option in CI, where we *do* want to optimize compiler_builtins. the new option is off by default for the `dev` channel and on otherwise.
2024-01-03Rollup merge of #119298 - onur-ozkan:silence-change-tracker-warning-for-ci, ↵León Orell Valerian Liehr-1/+6
r=Mark-Simulacrum suppress change-tracker warnings in CI containers Fixes #119296
2023-12-29suppress change-tracker warnings in containersonur-ozkan-1/+6
Signed-off-by: onur-ozkan <work@onurozkan.dev>
2023-12-28Remove `--enable-missing-tools` usage in CIJakub Beránek-4/+0
2023-11-30Fix cg_gcc CI runGuillaume Gomez-0/+8
2023-11-11merge `if-available` and `if-unchanged` for `download-ci-llvm`onur-ozkan-1/+1
Signed-off-by: onur-ozkan <work@onurozkan.dev>
2023-11-06Auto merge of #117435 - SparrowLii:nightly_parallel, r=oli-obk,davidtwcobors-2/+2
enable parallel rustc front end in nightly builds Refers to the [MCP](https://github.com/rust-lang/compiler-team/issues/681), this pr does: 1. Enable the parallel front end in nightly builds, and keep the default number of threads as 1. Then users can use the parallel rustc front end via -Z threads=n option. 2. Set it up to serial front end for beta/stable builds via bootstrap. 3. Switch over the alt builders from parallel rustc to serial, so we have artifacts without parallel to test against the artifacts with parallel. r? `@oli-obk` cc `@cjgillot` `@nnethercote` `@bjorn3` `@Kobzol`
2023-11-02Add comment explaining why the ENABLE_GCC_CODEGEN env variable is neededGuillaume Gomez-2/+4
2023-11-02Fix invalid enabling of gcc backend in `run.sh`Guillaume Gomez-2/+2
2023-11-02Rename `SKIP_CODEGEN_TESTS` into `ENABLE_GCC_CODEGEN`Guillaume Gomez-1/+1
2023-11-02Don't include GCC backend if SKIP_CODEGEN_TESTS is not enabledGuillaume Gomez-2/+7
2023-11-02Run rustc_codegen_gcc tests in the CIGuillaume Gomez-1/+1
2023-10-31enable parallel rustc in nightly buildsSparrowLii-2/+2
2023-10-27Explicitly mark which targets to distribute cg_clif for in CIbjorn3-2/+4
This avoids needlessly building cg_clif for other targets and makes it easier for the dist code to determine if it should distribute cg_clif as component.
2023-10-27Fix review commentsbjorn3-3/+2
2023-10-27Distribute cg_clif as a rustup componentbjorn3-1/+2
2023-10-05Enable new bors try branch to run on CIJakub Beránek-1/+2
2023-09-10Allow testing cg_clif using ./x.py testbjorn3-0/+3
2023-08-07Rollup merge of #114573 - Kobzol:ci-no-group-on-error, r=oli-obkMatthias Krüger-2/+14
CI: do not hide error logs in a group This PR avoids creating a GHA group at the very end of a CI workflow when some failure has happened. Before, when a failure has happened, its GHA group was not closed, however the clock drift check function would create a new group, which would actually close the group containing the error log, thus making errors hidden by default, which is not ideal. See discussion here: https://rust-lang.zulipchat.com/#narrow/stream/326414-t-infra.2Fbootstrap/topic/GHA.20groups.20being.20closed.20on.20failures r? bootstrap
2023-08-07Do not hide CI error logs in a group when a failure happensJakub Beránek-2/+14
2023-08-07Auto merge of #114206 - sethp:patch-1, r=Mark-Simulacrumbors-1/+1
fix(ci): Ensure idempotence of user creation Previously, re-running `run.sh` in the same container would fail at the useradd step, because the user already exists. Instead, change that step to "create if not exists" semantics to ease interactive debugging of CI failures. Split out from https://github.com/rust-lang/rust/pull/111891 per request by `@jackh726`
2023-08-02Add more context to `quit_if_file_exists` in `configure.py`Trevor Gross-0/+3
Currently, having a dirty `obj/` directory is sufficient to abort CI tests. This results in errors like the following: ``` ... == end clock drift check == sccache: Starting the server... configure: error: Existing 'config.toml' detected. == clock drift check == ... ``` This is subtle and doesn't give a good idea as to what causes the issue. With this patch, the error becomes more prominent and a resolution is suggested: ``` == end clock drift check == sccache: Starting the server... configure: ERROR: Existing 'config.toml' detected. Exiting Is objdir '/home/tmgross/projects/rust/obj' clean? == clock drift check == ```
2023-07-31fix: lookup user by name, not idsethp-1/+1
Previously, the `id -u $LOCAL_USER_ID` check would succeed, because there was already a user mapped to the $LOCAL_USER_ID with a different name. Looking up the entry in the database by name, however, correctly makes the `useradd` operation idempotent. An alternative fix would be to avoid duplicating an already-existing-ID, however that would require either probing for an available ID or identifying where LOCAL_USER_ID is being set and changing it to better match the ubuntu base image, neither of which is a one-line change.
2023-07-29fix(ci): Ensure idempotence of user creationsethp-1/+1
Previously, re-running `run.sh` in the same container would fail at the useradd step, because the user already exists. Instead, change that step to "create if not exists" semantics to ease interactive debugging of CI failures. Split out from https://github.com/rust-lang/rust/pull/111891 per request by @jackh726
2023-07-14Add even more GHA log groupsjyn-2/+6
This also adds a dynamic check that we don't emit nested groups, since GHA currently doesn't support them.
2023-06-19Print what is being executed on CIJakub Beránek-0/+1
This makes it easier to grep for executed commands in CI logs
2023-06-03disable truncation in CIjyn-0/+1
2023-04-25Auto merge of #110232 - Amanieu:old-llvm-components, r=petrochenkovbors-1/+5
Allow older LLVM versions to have missing components This check was introduced by #77280 to ensure that all tests that are filtered by LLVM component are actually tested in CI. However this causes issues for new targets (e.g. #101069) where support is only available on the latest LLVM version. This PR restricts the tests to only CI jobs that use the latest LLVM version.
generated by cgit-pink 1.4.1-3-g733a5 (git 2.37.1) at 2026-04-28 07:34:14 +0000