Page MenuHomeFreeBSD
Differential D5742

Remove old depend code and make FAST_DEPEND one true way.
ClosedPublic
Actions

Authored by bdrewery on Mar 25 2016, 9:56 PM.
Tags
None
Referenced Files
Unknown Object (File)
Sat, Apr 12, 5:56 PM
Unknown Object (File)
Sat, Apr 12, 4:06 AM
Unknown Object (File)
Mon, Apr 7, 3:14 AM
Unknown Object (File)
Thu, Mar 27, 4:17 PM
Unknown Object (File)
Mar 5 2025, 12:30 AM
Unknown Object (File)
Mar 1 2025, 9:47 PM
Unknown Object (File)
Feb 25 2025, 10:18 AM
Unknown Object (File)
Feb 25 2025, 2:12 AM
View All Files
Subscribers
wblock

Details

Reviewers
brooks
emaste
jhb
• hselasky
imp
Group Reviewers
manpages
Commits
rS297434: Remove the old depend (mkdep) code and make FAST_DEPEND the one true way.
Summary

All of the removed code is handled by the FAST_DEPEND code.

A suspicious looking removals are in bsd.prog.mk/bsd.lib.mk but they were
replaced by a OBJS_DEPEND_GUESS variable that bsd.dep.mk now handles
when there is no .depend.<foo.o> file.

I am unsure if I missed more in sys/conf that can come out.

Diff Detail

Repository
rS FreeBSD src repository - subversion
Lint
Lint Passed
Unit
No Test Coverage
Build Status
Buildable 3080
Build 3112: arc lint + arc unit

Event Timeline

bdrewery updated this revision to Diff 14614.Mar 25 2016, 9:56 PM
bdrewery retitled this revision from to Remove old depend code and make FAST_DEPEND one true way..
bdrewery updated this object.
bdrewery edited the test plan for this revision. (Show Details)
bdrewery added reviewers: imp, emaste, • hselasky, jhb.
bdrewery updated this revision to Diff 14616.Mar 25 2016, 9:59 PM

Fix a comment and remove more from sys/conf.

bdrewery added a reviewer: brooks.Mar 25 2016, 11:08 PM
emaste added inline comments.Mar 29 2016, 5:49 PM
Makefile.inc1
146–147

This is now redundant.

bdrewery added inline comments.Mar 29 2016, 5:49 PM
Makefile.inc1
146–147

Ah right, I intended to remove all of the 'make depend' in here too. Thanks.

bdrewery updated this revision to Diff 14720.Mar 29 2016, 6:01 PM

Remove leftover tree-walks and NO_KERNELDEPEND references. Also fixup build.7 documentation for new depend.

Herald added a reviewer: manpages. · View Herald TranscriptMar 29 2016, 6:01 PM
emaste added inline comments.Mar 29 2016, 6:07 PM
Makefile.inc1
625

This will break some of my finger memory :-)

• hselasky accepted this revision.Mar 29 2016, 6:11 PM
• hselasky edited edge metadata.

Looks OK with regard to the kern.post.mk mlx5 related changes.

This revision is now accepted and ready to land.Mar 29 2016, 6:11 PM
bdrewery updated this revision to Diff 14721.Mar 29 2016, 7:06 PM
bdrewery edited edge metadata.

Keep old stage numbers

This revision now requires review to proceed.Mar 29 2016, 7:06 PM
Herald edited edge metadata. · View Herald TranscriptMar 29 2016, 7:06 PM
emaste added inline comments.Mar 29 2016, 7:14 PM
Makefile.inc1
625

Although I do have searched build logs with /4\.4/, the comment was mostly in jest. Having a missing 4.3 is probably more confusing.

bdrewery updated this revision to Diff 14722.Mar 29 2016, 7:16 PM
bdrewery edited edge metadata.

Don't skip stage numbers

Herald edited edge metadata. · View Herald TranscriptMar 29 2016, 7:16 PM
emaste accepted this revision.Mar 29 2016, 7:19 PM
emaste edited edge metadata.

I think this is fine. This stuff is subtle and I'm a bit worried about edge cases, but I don't see a good way to find them other than committing.

This revision is now accepted and ready to land.Mar 29 2016, 7:19 PM
wblock added a subscriber: wblock.Mar 29 2016, 9:12 PM
wblock added inline comments.
Makefile.inc1
963

A word is missing (probably "on"), but "depending" is likely not the best choice of word here, either. The wording is so vague it is hard to suggest something clearer. Something about how TARGET affects which config files are used, at a guess.

share/man/man7/build.7
128

Quotes are not needed here. They are only necessary when single parameter will be mistaken to be multiples because of whitespace. No harm, though.

sys/conf/kern.post.mk
151

spelling error

emaste added inline comments.Mar 29 2016, 9:22 PM
Makefile.inc1
963

Indeed, but this predates @bdrewery's changes by a decade and a half or so. This should be improved in a separate commit.

sys/conf/kern.post.mk
151

As above, perhaps in a separate commit

brooks accepted this revision.Mar 30 2016, 5:54 PM
brooks edited edge metadata.

No obvious issues in the diff. I've been running our local tree with FAST_DEPEND as the default for quite a while now and haven't had any issues.

bdrewery updated this revision to Diff 14751.Mar 30 2016, 9:38 PM
bdrewery edited edge metadata.

A few more unneeded 'make depend' walks

This revision now requires review to proceed.Mar 30 2016, 9:38 PM
Herald edited edge metadata. · View Herald TranscriptMar 30 2016, 9:38 PM
Closed by commit rS297434: Remove the old depend (mkdep) code and make FAST_DEPEND the one true way. (authored by bdrewery). · Explain WhyMar 30 2016, 11:50 PM
This revision was automatically updated to reflect the committed changes.

Revision Contents
Changeset List

PathSize
2 lines
74 lines
2 lines
5 lines
gnu/
lib/
csu/
1 line
kerberos5/
2 lines
release/
picobsd/
build/
3 lines
secure/
2 lines
share/
man/
man5/
8 lines
man7/
23 lines
mk/
5 lines
57 lines
23 lines
7 lines
7 lines
1 line
sys/
conf/
7 lines
49 lines
23 lines
5 lines
modules/
pflog/
5 lines
pfsync/
5 lines
wtap/
5 lines
tools/
build/
options/
2 lines
usr.sbin/
ndiscvt/
5 lines

Diff 14751

View Options

Makefile

Show First 20 Lines • Show All 3,460 Lines • ▼ Show 20 Lines fork_context_gnu, /**< Called from GNU generated code, so must not invoke the
microtask internally. */ microtask internally. */
fork_context_intel, /**< Called from Intel generated code. */ fork_context_intel, /**< Called from Intel generated code. */
fork_context_last fork_context_last
}; };
extern int __kmp_fork_call(ident_t *loc, int gtid, extern int __kmp_fork_call(ident_t *loc, int gtid,
enum fork_context_e fork_context, kmp_int32 argc, enum fork_context_e fork_context, kmp_int32 argc,
microtask_t microtask, launch_t invoker, microtask_t microtask, launch_t invoker,
/* TODO: revert workaround for Intel(R) 64 tracker #96 */ /* TODO: revert workaround for Intel(R) 64 tracker #96 */
#if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64) && KMP_OS_LINUX #if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64)
emasteUnsubmitted
Not Done
Inline Actions

Came in from "aarch64 port sent by C. Bergstrom"
http://llvm.org/viewvc/llvm-project?view=revision&revision=225792
but this just added KMP_AARCH_64 to the existing ARM/X86_64 cases, KMP_OS_LINUX was already there

is there something FreeBSD-aarch64-specific here?

emaste: Came in from "aarch64 port sent by C. Bergstrom" http://llvm.org/viewvc/llvm-project?
val_packett.coolUnsubmitted
Done
Inline Actions

The va_list usage in the else cases (which treats it as if it is already a pointer) does not compile on FreeBSD/aarch64, it needs to be in the same case as Linux/(aarch64|amd64|arm), which treats it as opaque.

This is just another "upstream only considered Linux, we need to be included in the same way as Linux" case.

val_packett.cool: The `va_list` usage in the `else` cases (which treats it as if it is already a pointer) does…
dimAuthorUnsubmitted
Not Done
Inline Actions

This would typically not be acceptable for upstream. The condition should at least be rewritten as:

#if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64) && (KMP_OS_LINUX | KMP_OS_FREEBSD)

It should probably be defining something like KMP_HAVE_WEIRD_VA_LIST so the conditional expression does not have to be changed in many places.

dim: This would typically not be acceptable for upstream. The condition should at least be rewritten…
va_list *ap va_list *ap
#else #else
va_list ap va_list ap
#endif #endif
); );
extern void __kmp_join_call(ident_t *loc, int gtid extern void __kmp_join_call(ident_t *loc, int gtid
#if OMPT_SUPPORT #if OMPT_SUPPORT
▲ Show 20 Lines • Show All 440 Lines • Show Last 20 Lines
View Options

Makefile.inc1

Show First 20 Lines • Show All 3,460 Lines • ▼ Show 20 Lines fork_context_gnu, /**< Called from GNU generated code, so must not invoke the
microtask internally. */ microtask internally. */
fork_context_intel, /**< Called from Intel generated code. */ fork_context_intel, /**< Called from Intel generated code. */
fork_context_last fork_context_last
}; };
extern int __kmp_fork_call(ident_t *loc, int gtid, extern int __kmp_fork_call(ident_t *loc, int gtid,
enum fork_context_e fork_context, kmp_int32 argc, enum fork_context_e fork_context, kmp_int32 argc,
microtask_t microtask, launch_t invoker, microtask_t microtask, launch_t invoker,
/* TODO: revert workaround for Intel(R) 64 tracker #96 */ /* TODO: revert workaround for Intel(R) 64 tracker #96 */
#if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64) && KMP_OS_LINUX #if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64)
emasteUnsubmitted
Not Done
Inline Actions

Came in from "aarch64 port sent by C. Bergstrom"
http://llvm.org/viewvc/llvm-project?view=revision&revision=225792
but this just added KMP_AARCH_64 to the existing ARM/X86_64 cases, KMP_OS_LINUX was already there

is there something FreeBSD-aarch64-specific here?

emaste: Came in from "aarch64 port sent by C. Bergstrom" http://llvm.org/viewvc/llvm-project?
val_packett.coolUnsubmitted
Done
Inline Actions

The va_list usage in the else cases (which treats it as if it is already a pointer) does not compile on FreeBSD/aarch64, it needs to be in the same case as Linux/(aarch64|amd64|arm), which treats it as opaque.

This is just another "upstream only considered Linux, we need to be included in the same way as Linux" case.

val_packett.cool: The `va_list` usage in the `else` cases (which treats it as if it is already a pointer) does…
dimAuthorUnsubmitted
Not Done
Inline Actions

This would typically not be acceptable for upstream. The condition should at least be rewritten as:

#if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64) && (KMP_OS_LINUX | KMP_OS_FREEBSD)

It should probably be defining something like KMP_HAVE_WEIRD_VA_LIST so the conditional expression does not have to be changed in many places.

dim: This would typically not be acceptable for upstream. The condition should at least be rewritten…
va_list *ap va_list *ap
#else #else
va_list ap va_list ap
#endif #endif
); );
extern void __kmp_join_call(ident_t *loc, int gtid extern void __kmp_join_call(ident_t *loc, int gtid
#if OMPT_SUPPORT #if OMPT_SUPPORT
▲ Show 20 Lines • Show All 440 Lines • Show Last 20 Lines
View Options

Makefile.libcompat

Show First 20 Lines • Show All 3,460 Lines • ▼ Show 20 Lines fork_context_gnu, /**< Called from GNU generated code, so must not invoke the
microtask internally. */ microtask internally. */
fork_context_intel, /**< Called from Intel generated code. */ fork_context_intel, /**< Called from Intel generated code. */
fork_context_last fork_context_last
}; };
extern int __kmp_fork_call(ident_t *loc, int gtid, extern int __kmp_fork_call(ident_t *loc, int gtid,
enum fork_context_e fork_context, kmp_int32 argc, enum fork_context_e fork_context, kmp_int32 argc,
microtask_t microtask, launch_t invoker, microtask_t microtask, launch_t invoker,
/* TODO: revert workaround for Intel(R) 64 tracker #96 */ /* TODO: revert workaround for Intel(R) 64 tracker #96 */
#if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64) && KMP_OS_LINUX #if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64)
emasteUnsubmitted
Not Done
Inline Actions

Came in from "aarch64 port sent by C. Bergstrom"
http://llvm.org/viewvc/llvm-project?view=revision&revision=225792
but this just added KMP_AARCH_64 to the existing ARM/X86_64 cases, KMP_OS_LINUX was already there

is there something FreeBSD-aarch64-specific here?

emaste: Came in from "aarch64 port sent by C. Bergstrom" http://llvm.org/viewvc/llvm-project?
val_packett.coolUnsubmitted
Done
Inline Actions

The va_list usage in the else cases (which treats it as if it is already a pointer) does not compile on FreeBSD/aarch64, it needs to be in the same case as Linux/(aarch64|amd64|arm), which treats it as opaque.

This is just another "upstream only considered Linux, we need to be included in the same way as Linux" case.

val_packett.cool: The `va_list` usage in the `else` cases (which treats it as if it is already a pointer) does…
dimAuthorUnsubmitted
Not Done
Inline Actions

This would typically not be acceptable for upstream. The condition should at least be rewritten as:

#if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64) && (KMP_OS_LINUX | KMP_OS_FREEBSD)

It should probably be defining something like KMP_HAVE_WEIRD_VA_LIST so the conditional expression does not have to be changed in many places.

dim: This would typically not be acceptable for upstream. The condition should at least be rewritten…
va_list *ap va_list *ap
#else #else
va_list ap va_list ap
#endif #endif
); );
extern void __kmp_join_call(ident_t *loc, int gtid extern void __kmp_join_call(ident_t *loc, int gtid
#if OMPT_SUPPORT #if OMPT_SUPPORT
▲ Show 20 Lines • Show All 440 Lines • Show Last 20 Lines
View Options

UPDATING

Show First 20 Lines • Show All 3,460 Lines • ▼ Show 20 Lines fork_context_gnu, /**< Called from GNU generated code, so must not invoke the
microtask internally. */ microtask internally. */
fork_context_intel, /**< Called from Intel generated code. */ fork_context_intel, /**< Called from Intel generated code. */
fork_context_last fork_context_last
}; };
extern int __kmp_fork_call(ident_t *loc, int gtid, extern int __kmp_fork_call(ident_t *loc, int gtid,
enum fork_context_e fork_context, kmp_int32 argc, enum fork_context_e fork_context, kmp_int32 argc,
microtask_t microtask, launch_t invoker, microtask_t microtask, launch_t invoker,
/* TODO: revert workaround for Intel(R) 64 tracker #96 */ /* TODO: revert workaround for Intel(R) 64 tracker #96 */
#if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64) && KMP_OS_LINUX #if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64)
emasteUnsubmitted
Not Done
Inline Actions

Came in from "aarch64 port sent by C. Bergstrom"
http://llvm.org/viewvc/llvm-project?view=revision&revision=225792
but this just added KMP_AARCH_64 to the existing ARM/X86_64 cases, KMP_OS_LINUX was already there

is there something FreeBSD-aarch64-specific here?

emaste: Came in from "aarch64 port sent by C. Bergstrom" http://llvm.org/viewvc/llvm-project?
val_packett.coolUnsubmitted
Done
Inline Actions

The va_list usage in the else cases (which treats it as if it is already a pointer) does not compile on FreeBSD/aarch64, it needs to be in the same case as Linux/(aarch64|amd64|arm), which treats it as opaque.

This is just another "upstream only considered Linux, we need to be included in the same way as Linux" case.

val_packett.cool: The `va_list` usage in the `else` cases (which treats it as if it is already a pointer) does…
dimAuthorUnsubmitted
Not Done
Inline Actions

This would typically not be acceptable for upstream. The condition should at least be rewritten as:

#if (KMP_ARCH_ARM || KMP_ARCH_X86_64 || KMP_ARCH_AARCH64) && (KMP_OS_LINUX | KMP_OS_FREEBSD)

It should probably be defining something like KMP_HAVE_WEIRD_VA_LIST so the conditional expression does not have to be changed in many places.

dim: This would typically not be acceptable for upstream. The condition should at least be rewritten…
va_list *ap va_list *ap
#else #else
va_list ap va_list ap
#endif #endif
); );
extern void __kmp_join_call(ident_t *loc, int gtid extern void __kmp_join_call(ident_t *loc, int gtid
#if OMPT_SUPPORT #if OMPT_SUPPORT
▲ Show 20 Lines • Show All 440 Lines • Show Last 20 Lines
View Options

gnu/lib/csu/Makefile

Loading...
View Options

kerberos5/Makefile

Loading...
View Options

release/picobsd/build/Makefile.conf

Loading...
View Options

secure/Makefile

Loading...
View Options

share/man/man5/make.conf.5

Loading...
View Options

share/man/man7/build.7

Loading...
View Options

share/mk/bsd.clang-analyze.mk

Loading...
View Options

share/mk/bsd.dep.mk

Loading...
View Options

share/mk/bsd.lib.mk

Loading...
View Options

share/mk/bsd.opts.mk

Loading...
View Options

share/mk/bsd.prog.mk

Loading...
View Options

share/mk/local.meta.sys.mk

Loading...
View Options

sys/conf/kern.opts.mk

Loading...
View Options

sys/conf/kern.post.mk

Loading...
View Options

sys/conf/kern.pre.mk

Loading...
View Options

sys/conf/kmod.mk

Loading...
View Options

sys/modules/pflog/Makefile

Loading...
View Options

sys/modules/pfsync/Makefile

Loading...
View Options

sys/modules/wtap/Makefile

Loading...
View Options

tools/build/options/WITHOUT_FAST_DEPEND

Loading...
View Options

tools/build/options/WITH_FAST_DEPEND

Loading...
View Options

tools/tinder.sh

Loading...
View Options

usr.sbin/ndiscvt/ndisgen.sh

Loading...