[abseil_cpp_internal] Fix static linking patch commands #21961
Conversation
rpoyner-tri
added
priority: low
status: single reviewer ok
There was a problem hiding this comment.
Manual test technique:
$ ls bazel-bin/_solib_k8/*abseil*
That command will find some shared library files in a built master workspace. After this patch (and possibly a "bazel clean"), it should not find any shared library files.
Reviewable status: needs platform reviewer assigned, needs at least one assigned reviewer (waiting on @rpoyner-tri)
rpoyner-tri
force-pushed
the
abseil-static
branch
from
0eb6ec0
to
49b99d1
Compare
There was a problem hiding this comment.
Reviewable status: 1 unresolved discussion, needs platform reviewer assigned, needs at least one assigned reviewer (waiting on @rpoyner-tri)
a discussion (no related file):
After this patch, many mold versions should successfully link drake. I've tried 1.03 (apt install mold on jammy), and 2.34 (latest built from source).
There was a problem hiding this comment.
+@jwnimmer-tri for review.
Reviewed 1 of 1 files at r2, all commit messages.
Reviewable status: 2 unresolved discussions, LGTM missing from assignee jwnimmer-tri(platform)
tools/workspace/abseil_cpp_internal/repository.bzl line 17 at r2 (raw file):
], patch_cmds = [ # Back-fill a linkopts key missing in the upstream build system.
nit I think edit this would be substantially clearer as a patch file instead of sed. Also, everywhere else in abseil sets copts = ABSL_DEFAULT_COPTS and linkopts = ABSL_DEFAULT_LINKOPTS, so I imagine that's how our patch file should spell it also?
It also seems like a the kind of possible oversight that upstream would appreciate knowing about, i.e., we should propose that patch to upstream -- although given abseil/abseil-cpp#1003 (comment) maybe they would decline.
There was a problem hiding this comment.
Reviewable status: 1 unresolved discussion, LGTM missing from assignee jwnimmer-tri(platform)
tools/workspace/abseil_cpp_internal/repository.bzl line 17 at r2 (raw file):
Previously, jwnimmer-tri (Jeremy Nimmer) wrote…
nit I think edit this would be substantially clearer as a patch file instead of sed. Also, everywhere else in abseil sets
copts = ABSL_DEFAULT_COPTSandlinkopts = ABSL_DEFAULT_LINKOPTS, so I imagine that's how our patch file should spell it also?It also seems like a the kind of possible oversight that upstream would appreciate knowing about, i.e., we should propose that patch to upstream -- although given abseil/abseil-cpp#1003 (comment) maybe they would decline.
Am I guaranteed that commands are applied after patch files?
There was a problem hiding this comment.
Reviewable status: 2 unresolved discussions, LGTM missing from assignee jwnimmer-tri(platform)
tools/workspace/abseil_cpp_internal/repository.bzl line 17 at r2 (raw file):
Previously, rpoyner-tri (Rick Poyner (rico)) wrote…
Am I guaranteed that commands are applied after patch files?
When I skimmed the patching implementation, it seemed like "yes" to me. If it doesn't work, we can revisit.
rpoyner-tri
force-pushed
the
abseil-static
branch
from
49b99d1
to
bcae7d8
Compare
There was a problem hiding this comment.
Reviewable status: 1 unresolved discussion, LGTM missing from assignee jwnimmer-tri(platform)
tools/workspace/abseil_cpp_internal/repository.bzl line 17 at r2 (raw file):
Previously, jwnimmer-tri (Jeremy Nimmer) wrote…
When I skimmed the patching implementation, it seemed like "yes" to me. If it doesn't work, we can revisit.
That file doesn't import ABSL_DEFAULT_LINKOPTS, and I don't need it, so I left that out of the patch.
There was a problem hiding this comment.
Reviewed 2 of 2 files at r3, all commit messages.
Reviewable status: 2 unresolved discussions
tools/workspace/abseil_cpp_internal/patches/civil_time_linkopts.patch line 3 at r3 (raw file):
This patch could be upstreamed.
nit This comment leaves the point murky. This specific patch should not be upstreamed. What could be upstreamed is a patch that imported both of the ..._DEFAULT_... constants and using them.
Amend the patch commands to convert all of the cc_library() rules in the upstream's bazel files to use "linkstatic = 1". This prevents generation of "accidental" shared libraries, which in turn were causing link failures with more modern linkers (.i.e., ld.mold).
rpoyner-tri
force-pushed
the
abseil-static
branch
from
bcae7d8
to
9fe33aa
Compare
There was a problem hiding this comment.
Reviewable status: 1 unresolved discussion
tools/workspace/abseil_cpp_internal/patches/civil_time_linkopts.patch line 3 at r3 (raw file):
Previously, jwnimmer-tri (Jeremy Nimmer) wrote…
This patch could be upstreamed.
nit This comment leaves the point murky. This specific patch should not be upstreamed. What could be upstreamed is a patch that imported both of the
..._DEFAULT_...constants and using them.
Sure. I'm still not certain I know why this particular build file is different from all the others.
What I really think is that keying on arguments is the wrong seam. We should probably be reimplementing cc_library(). I had hopes for the documented cc_static_library(), but alas, it seems to have been abandoned. https://bazel.build/reference/be/c-cpp#cc_static_library
There was a problem hiding this comment.
Reviewed 1 of 1 files at r4, all commit messages.
Reviewable status: 1 unresolved discussion
tools/workspace/abseil_cpp_internal/patches/civil_time_linkopts.patch line 3 at r3 (raw file):
Previously, rpoyner-tri (Rick Poyner (rico)) wrote…
Sure. I'm still not certain I know why this particular build file is different from all the others.
What I really think is that keying on arguments is the wrong seam. We should probably be reimplementing cc_library(). I had hopes for the documented
cc_static_library(), but alas, it seems to have been abandoned. https://bazel.build/reference/be/c-cpp#cc_static_library
This BUILD file is different because the code is vendored (as noted here: abseil/abseil-cpp#1003 (comment)).
Longer term, we could submit an upstream ABSL_DEFAULT_LINKSTATIC to match how they handle default copts and linkopts. Their architecture is to use named constants to configure cc_library defaults. Using a wrapper cc_library reimplementation would go against how they've chosen to manage their project.
There was a problem hiding this comment.
Reviewable status: 1 unresolved discussion
tools/workspace/abseil_cpp_internal/patches/civil_time_linkopts.patch line 3 at r3 (raw file):
Previously, jwnimmer-tri (Jeremy Nimmer) wrote…
This BUILD file is different because the code is vendored (as noted here: abseil/abseil-cpp#1003 (comment)).
Longer term, we could submit an upstream
ABSL_DEFAULT_LINKSTATICto match how they handle default copts and linkopts. Their architecture is to use named constants to configure cc_library defaults. Using a wrapper cc_library reimplementation would go against how they've chosen to manage their project.
Oh, I'm not talking about upstreaming a cc_library hook...
Amend the patch commands to convert all of the cc_library() rules in the upstream's bazel files to use "linkstatic = 1". This prevents generation of "accidental" shared libraries, which in turn were causing link failures with more modern linkers (.i.e., ld.mold).
This change is