Makefile: fix dependency for $(UNIT_TEST_DIR)/clar/clar.o #1810
Conversation
|
/submit |
|
Submitted as pull.1810.git.1728323033680.gitgitgadget@gmail.com To fetch this version into To fetch this version to local tag |
|
On the Git mailing list, Junio C Hamano wrote (reply to this): "Philippe Blain via GitGitGadget" <gitgitgadget@gmail.com> writes:
> From: Philippe Blain <levraiphilippeblain@gmail.com>
>
> The clar source file '$(UNIT_TEST_DIR)/clar/clar.c' includes the
> generated 'clar.suite', but this dependency is not taken into account by
> our Makefile, so that it is possible for a parallel build to fail if
> Make tries to build 'clar.o' before 'clar.suite' is generated.
>
> Correctly specify the dependency.
>
> Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
> ---
> Makefile: fix dependency for $(UNIT_TEST_DIR)/clar/clar.o
>
> Hi Patrick,
>
> I tried building v2.47.0 and stumbled onto this small issue. It
> reproduces for me from a fresh clone on my old 2009 Mac with make -j -l
> 2.5, it's a little curious that no one ran into this yet.
I suspect that nobody tells make to build clar.o (and nothing else).
Instead, the t/unit-tests/bin/unit-tests target is what is typically
built, which is part of $(CLAR_TEST_PROG) that has clar.suite as one
of its dependencies.
$ make
$ rm -f t/unit-tests/clar.suite t/unit-tests/clar/clar.o
$ make -j1 t/unit-tests/bin/unit-tests
GEN t/unit-tests/clar.suite
CC t/unit-tests/clar/clar.o
LINK t/unit-tests/bin/unit-tests
What is possible to happen from the broken dependencies is when I
did not remove clar.o in the above experiment. We may rebuild
clar.suite and then link clar.o that is outdated without realizing.
> I found it best to declare the dependency directly on clar.o, instead of
> adjusting the dependencies of CLAR_TEST_OBJS on the line above, since it
> is really only this clar.c that includes clar.suite
>
> Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1810%2Fphil-blain%2Fclar-build-dep-fix-v1
> Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1810/phil-blain/clar-build-dep-fix-v1
> Pull-Request: https://github.com/gitgitgadget/git/pull/1810
>
> Makefile | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/Makefile b/Makefile
> index 2dde1fd2b8b..b615de74811 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -3911,6 +3911,7 @@ $(UNIT_TEST_DIR)/clar-decls.h: $(patsubst %,$(UNIT_TEST_DIR)/%.c,$(CLAR_TEST_SUI
> $(UNIT_TEST_DIR)/clar.suite: $(UNIT_TEST_DIR)/clar-decls.h
> $(QUIET_GEN)awk -f $(UNIT_TEST_DIR)/clar-generate.awk $< >$(UNIT_TEST_DIR)/clar.suite
> $(CLAR_TEST_OBJS): $(UNIT_TEST_DIR)/clar-decls.h
> +$(UNIT_TEST_DIR)/clar/clar.o: $(UNIT_TEST_DIR)/clar.suite
> $(CLAR_TEST_OBJS): EXTRA_CPPFLAGS = -I$(UNIT_TEST_DIR)
> $(CLAR_TEST_PROG): $(UNIT_TEST_DIR)/clar.suite $(CLAR_TEST_OBJS) $(GITLIBS) GIT-LDFLAGS
> $(call mkdir_p_parent_template)
>
> base-commit: 777489f9e09c8d0dd6b12f9d90de6376330577a2 |
|
On the Git mailing list, Patrick Steinhardt wrote (reply to this): On Mon, Oct 07, 2024 at 05:53:41PM -0700, Junio C Hamano wrote:
> "Philippe Blain via GitGitGadget" <gitgitgadget@gmail.com> writes:
>
> > From: Philippe Blain <levraiphilippeblain@gmail.com>
> >
> > The clar source file '$(UNIT_TEST_DIR)/clar/clar.c' includes the
> > generated 'clar.suite', but this dependency is not taken into account by
> > our Makefile, so that it is possible for a parallel build to fail if
> > Make tries to build 'clar.o' before 'clar.suite' is generated.
> >
> > Correctly specify the dependency.
> >
> > Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
> > ---
> > Makefile: fix dependency for $(UNIT_TEST_DIR)/clar/clar.o
> >
> > Hi Patrick,
> >
> > I tried building v2.47.0 and stumbled onto this small issue. It
> > reproduces for me from a fresh clone on my old 2009 Mac with make -j -l
> > 2.5, it's a little curious that no one ran into this yet.
>
> I suspect that nobody tells make to build clar.o (and nothing else).
>
> Instead, the t/unit-tests/bin/unit-tests target is what is typically
> built, which is part of $(CLAR_TEST_PROG) that has clar.suite as one
> of its dependencies.
>
> $ make
> $ rm -f t/unit-tests/clar.suite t/unit-tests/clar/clar.o
> $ make -j1 t/unit-tests/bin/unit-tests
> GEN t/unit-tests/clar.suite
> CC t/unit-tests/clar/clar.o
> LINK t/unit-tests/bin/unit-tests
>
> What is possible to happen from the broken dependencies is when I
> did not remove clar.o in the above experiment. We may rebuild
> clar.suite and then link clar.o that is outdated without realizing.
Makes sense. In any case, the fix looks good to me, thanks!
Patrick |
|
This patch series was integrated into seen via git@0ba0f9d. |
|
This branch is now known as |
|
This patch series was integrated into seen via git@9decb09. |
|
This patch series was integrated into seen via git@fac0be3. |
|
On the Git mailing list, Philippe Blain wrote (reply to this): Hi,
I had a closer look at how the header dependencies are handled in the Makefile and I think a cleaner and more idiomatic way to fix the problem would be to add clar.suite to GENERATED_H.
I’ll try to send a v2 tomorrow so maybe hold on to merging it to next for the moment, Junio.
Thanks
Philippe.
> Le 7 oct. 2024 à 23:58, Patrick Steinhardt <ps@pks.im> a écrit :
>
> On Mon, Oct 07, 2024 at 05:53:41PM -0700, Junio C Hamano wrote:
>> "Philippe Blain via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>
>>> From: Philippe Blain <levraiphilippeblain@gmail.com>
>>>
>>> The clar source file '$(UNIT_TEST_DIR)/clar/clar.c' includes the
>>> generated 'clar.suite', but this dependency is not taken into account by
>>> our Makefile, so that it is possible for a parallel build to fail if
>>> Make tries to build 'clar.o' before 'clar.suite' is generated.
>>>
>>> Correctly specify the dependency.
>>>
>>> Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
>>> ---
>>> Makefile: fix dependency for $(UNIT_TEST_DIR)/clar/clar.o
>>>
>>> Hi Patrick,
>>>
>>> I tried building v2.47.0 and stumbled onto this small issue. It
>>> reproduces for me from a fresh clone on my old 2009 Mac with make -j -l
>>> 2.5, it's a little curious that no one ran into this yet.
>>
>> I suspect that nobody tells make to build clar.o (and nothing else).
>>
>> Instead, the t/unit-tests/bin/unit-tests target is what is typically
>> built, which is part of $(CLAR_TEST_PROG) that has clar.suite as one
>> of its dependencies.
>>
>> $ make
>> $ rm -f t/unit-tests/clar.suite t/unit-tests/clar/clar.o
>> $ make -j1 t/unit-tests/bin/unit-tests
>> GEN t/unit-tests/clar.suite
>> CC t/unit-tests/clar/clar.o
>> LINK t/unit-tests/bin/unit-tests
>>
>> What is possible to happen from the broken dependencies is when I
>> did not remove clar.o in the above experiment. We may rebuild
>> clar.suite and then link clar.o that is outdated without realizing.
>
> Makes sense. In any case, the fix looks good to me, thanks!
>
> Patrick |
|
User |
|
On the Git mailing list, Junio C Hamano wrote (reply to this): Philippe Blain <levraiphilippeblain@gmail.com> writes:
> I had a closer look at how the header dependencies are handled in
> the Makefile and I think a cleaner and more idiomatic way to fix
> the problem would be to add clar.suite to GENERATED_H.
If it is a header file (e.g. we could run "make hdr-check" on it to
make sure it is self-contained, if we wanted to check it), that
sounds like a very appropriate thing to do.
> I’ll try to send a v2 tomorrow so maybe hold on to merging it to
> next for the moment, Junio.
Thanks. |
|
This patch series was integrated into seen via git@4a9f863. |
|
There was a status update in the "Cooking" section about the branch Build fix. Expecting a reroll. cf. <C05B01E0-5007-4FB9-94CD-CBE74E79F9B7@gmail.com> source: <pull.1810.git.1728323033680.gitgitgadget@gmail.com> |
The clar source file '$(UNIT_TEST_DIR)/clar/clar.c' includes the generated 'clar.suite', but this dependency is not taken into account by our Makefile, so that it is possible for a parallel build to fail if Make tries to build 'clar.o' before 'clar.suite' is generated. Correctly specify the dependency. Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
phil-blain
force-pushed
the
clar-build-dep-fix
branch
from
42ae65c
to
f20c8a0
Compare
|
/submit |
|
Submitted as pull.1810.v2.git.1728667787227.gitgitgadget@gmail.com To fetch this version into To fetch this version to local tag |
|
This patch series was integrated into seen via git@712f36a. |
|
This patch series was integrated into seen via git@6f8ee0e. |
|
There was a status update in the "Cooking" section about the branch Build fix. Will merge to 'next'. source: <pull.1810.v2.git.1728667787227.gitgitgadget@gmail.com> |
|
On the Git mailing list, Patrick Steinhardt wrote (reply to this): On Fri, Oct 11, 2024 at 05:29:47PM +0000, Philippe Blain via GitGitGadget wrote:
> From: Philippe Blain <levraiphilippeblain@gmail.com>
>
> The clar source file '$(UNIT_TEST_DIR)/clar/clar.c' includes the
> generated 'clar.suite', but this dependency is not taken into account by
> our Makefile, so that it is possible for a parallel build to fail if
> Make tries to build 'clar.o' before 'clar.suite' is generated.
>
> Correctly specify the dependency.
>
> Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
Thanks, this version looks good to me.
Patrick |
|
This patch series was integrated into seen via git@e4d0535. |
|
This patch series was integrated into seen via git@bba4f9d. |
|
On the Git mailing list, Taylor Blau wrote (reply to this): On Mon, Oct 14, 2024 at 01:59:25PM +0200, Patrick Steinhardt wrote:
> On Fri, Oct 11, 2024 at 05:29:47PM +0000, Philippe Blain via GitGitGadget wrote:
> > From: Philippe Blain <levraiphilippeblain@gmail.com>
> >
> > The clar source file '$(UNIT_TEST_DIR)/clar/clar.c' includes the
> > generated 'clar.suite', but this dependency is not taken into account by
> > our Makefile, so that it is possible for a parallel build to fail if
> > Make tries to build 'clar.o' before 'clar.suite' is generated.
> >
> > Correctly specify the dependency.
> >
> > Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
>
> Thanks, this version looks good to me.
Thanks, both. Junio had already marked this one to merge to 'next', so I
will go ahead and merge it down in the next integration cycle (which
should be tomorrow, assuming I remember how to do it correctly).
Thanks,
Taylor |
|
User |
|
This patch series was integrated into seen via git@9029bc8. |
|
This patch series was integrated into seen via git@8a17d45. |
|
This patch series was integrated into next via git@ae6e80b. |
|
This patch series was integrated into seen via git@36bbd9f. |
|
This patch series was integrated into seen via git@8a8efad. |
|
This patch series was integrated into seen via git@d954ba7. |
|
Any chance you could use this to fix the cmake buildsystem as well in contrib/buildsystems? The claro changes broke it... when using cmake and a builddir... |
|
FYI the claro changes in 2.47.0 causing an out of tree/builddir based cmake build to break can be fixed with this: --- a/contrib/buildsystems/CMakeLists.txt 2024-10-23 14:51:53.474906768 -0400
+++ b/contrib/buildsystems/CMakeLists.txt 2024-10-23 14:52:24.263046592 -0400
@@ -1036,8 +1036,8 @@ string(APPEND clar_suites
"};\n"
"static const size_t _clar_suite_count = ${clar_suites_count};\n"
"static const size_t _clar_callback_count = ${clar_cbs_count};\n")
-file(WRITE "${CMAKE_BINARY_DIR}/t/unit-tests/clar-decls.h" "${clar_decls}")
-file(WRITE "${CMAKE_BINARY_DIR}/t/unit-tests/clar.suite" "${clar_decls}" "${clar_cbs}" "${clar_suites}")
+file(WRITE "${CMAKE_SOURCE_DIR}/t/unit-tests/clar-decls.h" "${clar_decls}")
+file(WRITE "${CMAKE_SOURCE_DIR}/t/unit-tests/clar.suite" "${clar_decls}" "${clar_cbs}" "${clar_suites}")
list(TRANSFORM clar_test_SUITES PREPEND "${CMAKE_SOURCE_DIR}/t/unit-tests/")
list(TRANSFORM clar_test_SUITES APPEND ".c") |
|
This patch series was integrated into seen via git@49c22e3. |
@satmandu Thank you for reporting a bug. However, this is not the place the Git projects expects bug reports: please send it instead to the Git mailing list. See https://git-scm.com/community for more details. |
|
In fact, what you report is already fixed by this patch, from what I understand: https://lore.kernel.org/git/bb005979e7eb335b0178094251b5c37682d7d47b.1729506329.git.ps@pks.im/ |
Thank you! I look forward to that being in an upcoming release! This was our prior bug report, which did get a response. I should have responded to that directly. https://lore.kernel.org/git/CAGjHeYfyH+cOMYYYHnFR+Vu9T+RbmzO1SpB_-kbmBSf1DitJhA@mail.gmail.com/ |
|
This patch series was integrated into seen via git@dca32b8. |
|
This patch series was integrated into master via git@dca32b8. |
|
This patch series was integrated into next via git@dca32b8. |
|
Closed via dca32b8. |
v2:
Compared to v1, I just moved the new line one line up
to keep the dependencies of
$(CLAR_TEST_OBJS)together.Regarding what I wrote in [1], I took a closer look and
actually clar.suite is already added to GENERATED_H.
The issue really is that for generated headers, the dependency
must be specified manually. This is the case for other generated
headers (command-list.h, etc). So my initial approach is correct.
[1] https://lore.kernel.org/git/C05B01E0-5007-4FB9-94CD-CBE74E79F9B7@gmail.com/
v1:
Hi Patrick,
I tried building v2.47.0 and stumbled onto this small issue. It reproduces for me from a fresh clone on my old 2009 Mac with
make -j -l 2.5, it's a little curious that no one ran into this yet.I found it best to declare the dependency directly on
clar.o, instead of adjusting the dependencies ofCLAR_TEST_OBJSon the line above, since it is really only thisclar.cthat includesclar.suiteCC: Patrick Steinhardt ps@pks.im
cc: Philippe Blain levraiphilippeblain@gmail.com
cc: Taylor Blau me@ttaylorr.com