-Zprofile broken in nightly? #50203

Activity

1. mssun

   changed the title [-]Does -Zprofile broken in nightly?[/-] [+]Is -Zprofile broken in nightly?[/+] on Apr 24, 2018
2. mssun

   changed the title [-]Is -Zprofile broken in nightly?[/-] [+]-Zprofile broken in nightly?[/+] on Apr 24, 2018
3. emilyalbini

   added

   T-compilerRelevant to the compiler team, which will review and decide on the PR/issue.Relevant to the compiler team, which will review and decide on the PR/issue.

   C-bugCategory: This is a bug.Category: This is a bug.

   on Apr 26, 2018
4. marco-c commented on Jun 8, 2018

   

   marco-c

   on Jun 8, 2018

   Contributor

   More actions

   It would be great if we could bisect this, to know what caused the regression.
5. mssun commented on Jun 8, 2018

   

   mssun

   on Jun 8, 2018

   ContributorAuthor

   More actions

   Thanks. It still breaks on the latest nightly.

   $ rustc -V
   rustc 1.28.0-nightly (4a9c58c6b 2018-06-05)
   $ cargo -V
   cargo 1.28.0-nightly (c3b09c968 2018-05-27)
   

   This feature is really important to get a proper code coverage profile.
6. marco-c commented on Jun 8, 2018

   

   marco-c

   on Jun 8, 2018

   Contributor

   More actions

   I'm trying to bisect manually.
   For now:
   2018-01-01 is broken.
   2017-12-15 is working.
7. marco-c commented on Jun 8, 2018

   

   marco-c

   on Jun 8, 2018

   Contributor

   More actions

   2017-12-24 is working. 2017-12-25 is broken.
   Here are the commits landed between those two days: https://github.com/rust-lang/rust/compare/master@%7B2017-12-24%7D...master@%7B2017-12-25%7D.
8. marco-c commented on Jun 8, 2018

   

   marco-c

   on Jun 8, 2018

   Contributor

   More actions

   I guess we should have added parsing of gcda as part of this test: https://github.com/rust-lang/rust/blob/master/src/test/run-make-fulldeps/profile/Makefile.
9. marco-c commented on Jun 12, 2018

   

   marco-c

   on Jun 12, 2018

   Contributor

   More actions

   @alexcrichton is the range from #50203 (comment) correct? I guess there's a precise way to tell what commit a given Nightly build correspons to?
10. kennytm commented on Jun 14, 2018

    

    kennytm

    on Jun 14, 2018

    Last edited by kennytm

    Member

    More actions

    C:\> rustc +nightly-2017-12-24 -vV
    rustc 1.24.0-nightly (169929308 2017-12-23)
    binary: rustc
    commit-hash: 16992930835ce3376a4aaed42307726e1fc78e45
    commit-date: 2017-12-23
    host: x86_64-pc-windows-msvc
    release: 1.24.0-nightly
    LLVM version: 4.0
    
    C:\> rustc +nightly-2017-12-25 -vV
    rustc 1.24.0-nightly (c284f8807 2017-12-24)
    binary: rustc
    commit-hash: c284f8807eb3a1d728242bb6a767b0306d6f6bd5
    commit-date: 2017-12-24
    host: x86_64-pc-windows-msvc
    release: 1.24.0-nightly
    LLVM version: 4.0
    

    The regression happens between 1699293...c284f88

    PRs in this range:

    • incr.comp.: Cache check_match and use ensure() for coherence-related queries. #46881 (incr.comp.: Cache check_match and use ensure() for coherence-related queries.)
    • Prevent unwinding past FFI boundaries #46833 (Prevent unwinding past FFI boundaries)
    • Add a feature gate for nested uses of impl Trait #46888 (Add a feature gate for nested uses of impl Trait)
    • Only mark unions as uninhabited if all of their fields are uninhabited #46859 (Only mark unions as uninhabited if all of their fields are uninhabited)
    • Const-eval array lengths in rustdoc. #46894 (Const-eval array lengths in rustdoc.)
    • fix debuginfo scoping of let-statements #46896 (fix debuginfo scoping of let-statements)
    • Update Cargo and its dependencies #46630 (Update Cargo and its dependencies)
    • Capture Command environment at spawn #46789 (Capture Command environment at spawn)

    Fast PR-by-PR bisection is impossible at this point since the artifacts have been deleted.
11. marco-c commented on Jun 18, 2018

    

    marco-c

    on Jun 18, 2018

    Last edited by marco-c

    Contributor

    More actions

    Thanks @kennytm. I've tested with 4910ed2 and it is working. So the regression is between 4910ed2 and c284f88.
12. kennytm

    added

    A-incr-compArea: Incremental compilationArea: Incremental compilation

    on Jun 18, 2018
13. kennytm commented on Jun 18, 2018

    

    kennytm

    on Jun 18, 2018

    Last edited by kennytm

    Member

    More actions

    It is caused by incremental compilation.

    ~:45$ rustc --crate-name hello_world src/main.rs --crate-type bin --emit=link -C debuginfo=2 -Zprofile -Copt-level=1 -Clink-dead-code -Ccodegen-units=1 -Zno-landing-pads -C metadata=e6641c514587dc22 -C extra-filename=-e6641c514587dc22 --out-dir ~/hello_world/target/debug/deps 
    
    ~:46$ ./target/debug/deps/hello_world-e6641c514587dc22
    Hello, world!
    
    ~:47$ rustc --crate-name hello_world src/main.rs --crate-type bin --emit=link -C debuginfo=2 -Zprofile -Copt-level=1 -Clink-dead-code -Ccodegen-units=1 -Zno-landing-pads -C metadata=e6641c514587dc22 -C extra-filename=-e6641c514587dc22 --out-dir ~/hello_world/target/debug/deps -C incremental=~/hello_world/target/debug/incremental
    
    ~:48$ ./target/debug/deps/hello_world-e6641c514587dc22
    Hello, world!
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous GCDA file: corrupt arc tag (0x00000001)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous GCDA file: corrupt arc tag (0x24246437)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous run count: corrupt object tag (0x24643775)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous GCDA file: corrupt arc tag (0x39383038)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous run count: corrupt object tag (0x32626265)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous GCDA file: corrupt arc tag (0x39383038)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous GCDA file: corrupt arc tag (0x00000000)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous run count: corrupt object tag (0x62333333)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous GCDA file: corrupt arc tag (0x38327472)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous GCDA file: corrupt arc tag (0x34737973)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous run count: corrupt object tag (0x78696e75)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous GCDA file: corrupt arc tag (0x246e6f69)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous run count: corrupt object tag (0x36245447)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous GCDA file: corrupt arc tag (0x38306266)
    profiling: ~/hello_world/target/debug/deps/hello_world-e6641c514587dc22.gcda: cannot merge previous run count: corrupt object tag (0x00004538)
    
14. kennytm

    added a commit that references this issue on Jun 18, 2018

    Disable incremental to workaround rust-lang/rust#50203.

    05d1c1e
15. marco-c commented on Jun 18, 2018

    

    marco-c

    on Jun 18, 2018

    Contributor

    More actions

    git bisect start
    # bad: [c284f8807eb3a1d728242bb6a767b0306d6f6bd5] Auto merge of #46789 - Diggsey:command-env-capture, r=dtolnay
    git bisect bad c284f8807eb3a1d728242bb6a767b0306d6f6bd5
    # good: [16992930835ce3376a4aaed42307726e1fc78e45] Auto merge of #46864 - estebank:closure-type-err-sp, r=nikomatsakis
    git bisect good 16992930835ce3376a4aaed42307726e1fc78e45
    # good: [16992930835ce3376a4aaed42307726e1fc78e45] Auto merge of #46864 - estebank:closure-type-err-sp, r=nikomatsakis
    git bisect good 16992930835ce3376a4aaed42307726e1fc78e45
    # good: [11a24d9c3940f60e527c571680d64e80e0889abe] Auto merge of #46888 - cramertj:nested-impl-trait-error, r=nikomatsakis
    git bisect good 11a24d9c3940f60e527c571680d64e80e0889abe
    # good: [304717bd86e42bc9b0c45ea5a6068e7ed9d13f2f] Auto merge of #46894 - detrumi:fix-const-eval-trait, r=eddyb
    git bisect good 304717bd86e42bc9b0c45ea5a6068e7ed9d13f2f
    

    Almost done.

    ❤️React with ❤️1Reacted by kennytm
16. marco-c commented on Jun 18, 2018

    

    marco-c

    on Jun 18, 2018

    Contributor

    More actions

    325d739 is the culprit.
17. marco-c commented on Jun 18, 2018

    

    marco-c

    on Jun 18, 2018

    Contributor

    More actions

    325d739 updated Cargo from 930f9d949b384fde9b0d7b9cc590515933f194c0 to e08f310188cc0d101550c21b3d88877cfba7e5e1, here's the list of commits that landed between the two: rust-lang/cargo@930f9d9...e08f310.
18. marco-c commented on Jun 18, 2018

    

    marco-c

    on Jun 18, 2018

    Contributor

    More actions

    So I guess it's simply rust-lang/cargo#4817.
19. kennytm commented on Jun 18, 2018

    

    kennytm

    on Jun 18, 2018

    Member

    More actions

    Well so we're back to square one 😞. Maybe inc. comp. is simply incompatible with -Zprofile.
20. marco-c commented on Jun 18, 2018

    

    marco-c

    on Jun 18, 2018

    Contributor

    More actions

    We could disable incremental compilation when profile is enabled, like for LTO.
21. marco-c commented on Jun 19, 2018

    

    marco-c

    on Jun 19, 2018

    Contributor

    More actions

    We could disable incremental compilation when profile is enabled, like for LTO.

    I've opened #51645 to do this.
22. bors

    added a commit that references this issue on Jul 2, 2018

    Auto merge of #51645 - marco-c:error_when_incremental_and_profile, r=…

    ...

    This issue will close once commit e75e782 is merged into the 'main' branch.

    e75e782
23. bors

    closed this as completedin #51645on Jul 2, 2018
24. wllenyj commented on Feb 16, 2022

    

    wllenyj

    on Feb 16, 2022

    Contributor

    More actions

    The same problem reappeared in 1.60.0-nightly (09cb29c 2022-02-15)

        ubuntu: info: default toolchain set to 'nightly-x86_64-unknown-linux-gnu'
        ubuntu:   nightly-x86_64-unknown-linux-gnu installed - rustc 1.60.0-nightly (09cb29c64 2022-02-15)
        ubuntu: 
        ubuntu: 
        ubuntu: Rust is installed now. Great!
    
    ...
    
    failures:
    
    ---- x86_64::mpspec::bindgen_test_layout_mpc_cpu stdout ----
    ---- x86_64::mpspec::bindgen_test_layout_mpc_cpu stderr ----
    profiling: /vagrant/target/debug/deps/db_boot-f38d7d4bfa105ed8.gcda: cannot merge previous GCDA file: corrupt arc tag (0x00000000)
    profiling: /vagrant/target/debug/deps/db_boot-f38d7d4bfa105ed8.gcda: cannot merge previous GCDA file: corrupt arc tag (0x00000001)
    profiling: /vagrant/target/debug/deps/db_boot-f38d7d4bfa105ed8.gcda: cannot merge previous GCDA file: corrupt arc tag (0x00000001)
    profiling: /vagrant/target/debug/deps/db_boot-f38d7d4bfa105ed8.gcda: cannot merge previous GCDA file: corrupt arc tag (0x75747375)
    profiling: /vagrant/target/debug/deps/db_boot-f38d7d4bfa105ed8.gcda: cannot merge previous GCDA file: corrupt arc tag (0x3638782d)
    profiling: /vagrant/target/debug/deps/db_boot-f38d7d4bfa105ed8.gcda: cannot merge previous GCDA file: corrupt arc tag (0x696c2f75)
    

    See https://github.com/wllenyj/dragonball-sandbox/runs/5212083346 for details.
25. workingjubilee

    added

    A-code-coverageArea: Source-based code coverage (-Cinstrument-coverage)Area: Source-based code coverage (-Cinstrument-coverage)

    on Nov 9, 2024