This is an archive of the discontinued LLVM Phabricator instance.

Differential D137275

[Driver][test] Fix test by creating empty archive instead of empty file
ClosedPublic

Authored by lamb-j on Nov 2 2022, 11:14 AM.

Details

Reviewers
yaxunl
Commits
rGf9de860ca4c4: [Driver][test] Fix test by creating empty archive instead of empty file
rG6b992bcce0c5: [Driver][test] Fix test by creating empty archive instead of empty file

Diff Detail

Unit TestsFailed

TimeTest
1,810 msx64 debian > MemProfiler-x86_64-linux-dynamic.TestCases::test_merge_mib.cpp

Event Timeline

lamb-j created this revision.Nov 2 2022, 11:14 AM
Herald added a project: Restricted Project. · View Herald TranscriptNov 2 2022, 11:14 AM
lamb-j requested review of this revision.Nov 2 2022, 11:14 AM
Herald added a project: Restricted Project. · View Herald TranscriptNov 2 2022, 11:14 AM
Herald added a subscriber: cfe-commits. · View Herald Transcript
Harbormaster completed remote builds in B195767: Diff 472704.Nov 2 2022, 12:36 PM
yaxunl accepted this revision.Nov 2 2022, 12:58 PM

LGTM. Thanks.

This revision is now accepted and ready to land.Nov 2 2022, 12:58 PM
This revision was landed with ongoing or failed builds.Dec 5 2022, 3:43 PM
Closed by commit rG6b992bcce0c5: [Driver][test] Fix test by creating empty archive instead of empty file (authored by lamb-j). · Explain Why
This revision was automatically updated to reflect the committed changes.
lamb-j added a commit: rG6b992bcce0c5: [Driver][test] Fix test by creating empty archive instead of empty file.
thakis added a subscriber: thakis.Dec 5 2022, 5:12 PM

This breaks check-clang: http://45.33.8.238/linux/93300/step_7.txt

Please take a look and revert for now if it takes a while to fix.

thakis added a comment.Dec 6 2022, 5:59 AM
In D137275#3972773, @thakis wrote:

This breaks check-clang: http://45.33.8.238/linux/93300/step_7.txt

Please take a look and revert for now if it takes a while to fix.

This ended up being a local config thing that I fixed in af95441ba7f3ca61c925409e58ee7e5486d84033

thakis added a comment.Dec 6 2022, 7:22 AM

…but even after that, things are still broken on macOS: http://45.33.8.238/macm1/50243/step_7.txt

ar: no archive members specified
usage:  ar -d [-TLsv] archive file ...
	ar -m [-TLsv] archive file ...
	ar -m [-abiTLsv] position archive file ...
	ar -p [-TLsv] archive [file ...]
	ar -q [-cTLsv] archive file ...
	ar -r [-cuTLsv] archive file ...
	ar -r [-abciuTLsv] position archive file ...
	ar -t [-TLsv] archive [file ...]
	ar -x [-ouTLsv] archive [file ...]
thakis added a comment.Dec 6 2022, 7:23 AM

Reverted this in 8e7994b9b513497589d4bf911f061378f517fb16 for now.

thakis added a reverting change: rGf2d878ba688a: Revert "[clang] Tweak test to tolerate clang being called "clang" instead of….Dec 6 2022, 7:23 AM
thakis added a reverting change: rG8e7994b9b513: Revert "[Driver][test] Fix test by creating empty archive instead of empty file".
yaxunl added inline comments.Dec 6 2022, 7:53 AM
clang/test/Driver/hip-link-bc-to-bc.hip
22

probably we should use llvm-ar instead of ar

lamb-j added a commit: rGf9de860ca4c4: [Driver][test] Fix test by creating empty archive instead of empty file.Dec 7 2022, 11:50 AM
lamb-j added a comment.Dec 7 2022, 11:53 AM

@thakis thanks for reverting that, sorry I didn't see your request sooner!

I've switched the patch to use "llvm-ar" instead of "ar", which should solve the issues

lamb-j marked an inline comment as done.Dec 7 2022, 11:54 AM

Revision Contents

PathSize
clang/
test/
Driver/
4 lines

Diff 472704

clang/test/Driver/hip-link-bc-to-bc.hip

// REQUIRES: x86-registered-target, amdgpu-registered-target // REQUIRES: x86-registered-target, amdgpu-registered-target
// See issue #58711
// XFAIL: *
// Check that clang unbundles the two bitcodes and links via llvm-link // Check that clang unbundles the two bitcodes and links via llvm-link
// RUN: rm -rf %t && mkdir %t // RUN: rm -rf %t && mkdir %t
// RUN: touch %t/bundle1.bc // RUN: touch %t/bundle1.bc
// RUN: touch %t/bundle2.bc // RUN: touch %t/bundle2.bc
// RUN: %clang -### --target=x86_64-unknown-linux-gnu --offload-arch=gfx906 --hip-link \ // RUN: %clang -### --target=x86_64-unknown-linux-gnu --offload-arch=gfx906 --hip-link \
// RUN: -emit-llvm -fgpu-rdc --cuda-device-only \ // RUN: -emit-llvm -fgpu-rdc --cuda-device-only \
// RUN: %t/bundle1.bc %t/bundle2.bc \ // RUN: %t/bundle1.bc %t/bundle2.bc \
// RUN: 2>&1 | FileCheck -check-prefix=BITCODE %s // RUN: 2>&1 | FileCheck -check-prefix=BITCODE %s
// BITCODE: "{{.*}}clang-offload-bundler" "-type=bc" "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx906" "-input={{.*}}bundle1.bc" "-output=[[B1HOST:.*\.bc]]" "-output=[[B1DEV1:.*\.bc]]" "-unbundle" "-allow-missing-bundles" // BITCODE: "{{.*}}clang-offload-bundler" "-type=bc" "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx906" "-input={{.*}}bundle1.bc" "-output=[[B1HOST:.*\.bc]]" "-output=[[B1DEV1:.*\.bc]]" "-unbundle" "-allow-missing-bundles"
// BITCODE: "{{.*}}clang-{{.*}}" "-o" "[[B1DEV2:.*bundle1-gfx906.bc]]" "-x" "ir" "[[B1DEV1]]" // BITCODE: "{{.*}}clang-{{.*}}" "-o" "[[B1DEV2:.*bundle1-gfx906.bc]]" "-x" "ir" "[[B1DEV1]]"
// BITCODE: "{{.*}}clang-offload-bundler" "-type=bc" "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx906" "-input={{.*}}bundle2.bc" "-output=[[B2HOST:.*\.bc]]" "-output=[[B2DEV1:.*\.bc]]" "-unbundle" "-allow-missing-bundles" // BITCODE: "{{.*}}clang-offload-bundler" "-type=bc" "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx906" "-input={{.*}}bundle2.bc" "-output=[[B2HOST:.*\.bc]]" "-output=[[B2DEV1:.*\.bc]]" "-unbundle" "-allow-missing-bundles"
// BITCODE: "{{.*}}clang-{{.*}}" "-o" "[[B2DEV2:.*bundle2-gfx906.bc]]" "-x" "ir" "[[B2DEV1]]" // BITCODE: "{{.*}}clang-{{.*}}" "-o" "[[B2DEV2:.*bundle2-gfx906.bc]]" "-x" "ir" "[[B2DEV1]]"
// BITCODE: "{{.*}}llvm-link" "-o" "bundle1-hip-amdgcn-amd-amdhsa-gfx906.bc" "[[B1DEV2]]" "[[B2DEV2]]" // BITCODE: "{{.*}}llvm-link" "-o" "bundle1-hip-amdgcn-amd-amdhsa-gfx906.bc" "[[B1DEV2]]" "[[B2DEV2]]"
// Check that clang unbundles the bitcode and archive and links via llvm-link // Check that clang unbundles the bitcode and archive and links via llvm-link
// RUN: touch %t/libhipbundle.a // RUN: ar rc %t/libhipbundle.a
yaxunlUnsubmitted
Done
ReplyInline Actions

probably we should use llvm-ar instead of ar

yaxunl: probably we should use llvm-ar instead of ar
// RUN: touch %t/bundle.bc // RUN: touch %t/bundle.bc
// RUN: %clang -### --target=x86_64-unknown-linux-gnu --offload-arch=gfx906 --hip-link \ // RUN: %clang -### --target=x86_64-unknown-linux-gnu --offload-arch=gfx906 --hip-link \
// RUN: -emit-llvm -fgpu-rdc --cuda-device-only \ // RUN: -emit-llvm -fgpu-rdc --cuda-device-only \
// RUN: %t/bundle.bc -L%t -lhipbundle \ // RUN: %t/bundle.bc -L%t -lhipbundle \
// RUN: 2>&1 | FileCheck -check-prefix=ARCHIVE %s // RUN: 2>&1 | FileCheck -check-prefix=ARCHIVE %s
// ARCHIVE: "{{.*}}clang-offload-bundler" "-type=bc" "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx906" "-input={{.*}}bundle.bc" "-output=[[HOST:.*\.bc]]" "-output=[[DEV1:.*\.bc]]" "-unbundle" "-allow-missing-bundles" // ARCHIVE: "{{.*}}clang-offload-bundler" "-type=bc" "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx906" "-input={{.*}}bundle.bc" "-output=[[HOST:.*\.bc]]" "-output=[[DEV1:.*\.bc]]" "-unbundle" "-allow-missing-bundles"
// ARCHIVE: "{{.*}}clang-{{.*}}" "-o" "[[DEV2:.*\.bc]]" "-x" "ir" "[[DEV1]]" // ARCHIVE: "{{.*}}clang-{{.*}}" "-o" "[[DEV2:.*\.bc]]" "-x" "ir" "[[DEV1]]"
// ARCHIVE: "{{.*}}clang-offload-bundler" "-unbundle" "-type=a" "-input={{.*}}libhipbundle.a" "-targets=hip-amdgcn-amd-amdhsa-gfx906" "-output=[[AR:.*\.a]]" "-allow-missing-bundles" "-hip-openmp-compatible" // ARCHIVE: "{{.*}}clang-offload-bundler" "-unbundle" "-type=a" "-input={{.*}}libhipbundle.a" "-targets=hip-amdgcn-amd-amdhsa-gfx906" "-output=[[AR:.*\.a]]" "-allow-missing-bundles" "-hip-openmp-compatible"
// ARCHIVE: "{{.*}}llvm-link" "-o" "bundle-hip-amdgcn-amd-amdhsa-gfx906.bc" "[[DEV2]]" "[[AR]]" // ARCHIVE: "{{.*}}llvm-link" "-o" "bundle-hip-amdgcn-amd-amdhsa-gfx906.bc" "[[DEV2]]" "[[AR]]"