Running `make libfoo.ll` or `make libfoo.bc` within a library directory
will now give us an LLVM IR version of the library, and `make foo.full.ll`
or `make foo.full.bc` will give us an IR version of a binary.
As part of this change, we add an LLVM_LINK variable to sys.mk that can be
specified/overridden using an external toolchain.
Reviewed by: bdrewery, brooks
Approved by: rwatson (mentor)
Sponsored by: DARPA, AFRL
Differential Revision: https://reviews.freebsd.org/D8388
Refactor make suffix rules into separate files (one for POSIX and one not),
and rationalise the rules so that bsd.lib.mk can contain only those rules
that are library-specific (.c.po and .c.pico).
This can be accomplished by adding ${STATIC_CFLAGS} to the .c.o rule
unconditionally. STATIC_CFLAGS are only defined for use by sys.mk rules in
lib/libpam/Makefile.inc (see r227797), so it should be safe to include
them unconditionally in sys.mk's .c.o rule (tested by make universe and a
ports exp-run).
Reviewed by: bdrewery, sjg
Approved by: rwatson (mentor)
Sponsored by: DARPA, AFRL
Differential Revision: https://reviews.freebsd.org/D6805
build can break when different source files create the same target
files (case-insensitivity speaking). This is the case for object
files compiled with -fpic and shared libraries. The former uses
an extension of ".So", and the latter an extension ".so". Rename
shared object files from *.So to *.pico to match what NetBSD does.
See also r305855
MFC after: 1 month
Sponsored by: Bracket Computing
Differential Revision: https://reviews.freebsd.org/D7906
This is important to allow a Makefile to override OBJS_DEPEND_GUESS for
handling in META_MODE when its depend files are missing.
Approved by: re (implicit)
Sponsored by: EMC / Isilon Storage Division
A simplified example of the library targets with WITH_DEBUG_FILES is:
libgeom.so.5: libgeom.so.5.full
cp libgeom.so.5.full libgeom.so.5
libgeom.so.5.full:
ln -s libgeom.so.5 libgeom.so
cc -o libgeom.so.5.full *.o
Before, or without, WITH_DEBUG_FILES it is:
libgeom.so.5:
ln -s libgeom.so.5 libgeom.so
cc -o libgeom.so.5 *.o
The problem is that bmake considers the link source for the libgeom.so
link in the libgeom.so.5.full target as being a dependency for
libgeom.so.5.full. That resolves to libgeom.so.5. Thus a cyclic
dependency is created. The result of this is that if libgeom.so.5 is
created with a newer timestamp than libgeom.so.5.full, then
libgeom.so.5.full will be rebuilt on the next build. This causes a
chain reaction of everything in the build relinking, or hitting the
problem itself.
Moving the link creation to the target that actually creates
libgeom.so.5 fixes the problem. The simplest fix here is to just
duplicate the logic.
Submitted by: sjg
Approved by: re (implicit)
This breaks cross-building with WITH_META_MODE since it will rebuild
'build-tools' during the 'everything' phase.
A more proper fix is coming to bmake to implicitly require .META unless
.NOMETA (and other restrictions) are in place.
Adding .META to targets-to-build will ensure that they will rebuild if there
is no .meta file.
Adding it to all SUFFIXES and objects ensures that at least objects will
rebuild if there is no .meta file.
This will be reverted if bmake's behavior changes to rebuild on missing .meta
files.
Sponsored by: EMC / Isilon Storage Division
We want to avoid .text relocations in shared objects. libcrypto was the
only consumer and it is now fixed (as of r299389). Remove the now-unused
support for turning off the linker warning.
Reviewed by: kib
Differential Revision: https://reviews.freebsd.org/D6323
This overrides the cross-compiler's default sysroot to use the WORLD32's
sysroot for building the lib32 libraries. Previously the cross-compiler
would default the sysroot to the 64bit WORLDTMP and -B/-L/-isystem flags
were used to build using the lib32 files. This leads to multiple issues
discussed later. Some extra headers are now needed to be staged since the
64bit WORLDTMP is not referenced at all for headers. The 64bit WORLDTMP
is still used via PATH for build tools. Overriding the default
target/arch is retained in the CC/CXX overrides.
This allows reverting the LDSCRIPT rewriting in installworld from r296921 and
r235122, thus allowing read-only objdirs to work for installing again.
This removes the need for _LDSCRIPTROOT.
This allows progressing the change to always use --sysroot for the build
rather than only relying on the cross-compiler's default sysroot. The
work for that is in D3970 and needed to resolve WITHOUT_CROSS_COMPILER
not using a --sysroot [1].
PR: 196193 [1]
Sponsored by: EMC / Isilon Storage Division
Specifically this fixes /usr/lib/libc.so stripping the paths to the
libraries. The reason for this in r266227 was both likely because ld(1) did
not fully respect --sysroot until r291226 and because of the lib32
build. The lib32 build does not use --sysroot into the /usr/lib32 path,
rather it only uses -L and -B into the /usr/lib32 path and --sysroot
into the normal (64bit) /usr/lib. The _LDSCRIPTROOT was added with
the ldscript support in bsd.lib.mk so that it builds a 32-bit-sysrooted pathed
ldscript in the object directory and then installs a normal unprefixed
version in installworld. This commit also fixes the rebuild during
install which was broken in r266227. This commit would break DIRDEPS_BUILD
build of lib32 but it does not currently have a way to build it anyhow.
For example, before this change we had in /usr/lib/libc.so:
GROUP ( libc.so.7 libc_nonshared.a libssp_nonshared.a )
Now it is restored to pre-r266227:
GROUP ( /lib/libc.so.7 /usr/lib/libc_nonshared.a /usr/lib/libssp_nonshared.a )
The motivation for this is in testing of lld.
From emaste:
lld does not have built-in search paths (e.g. /lib, /usr/lib) and relies on
-L arguments passed by the caller. As the linker is nearly always invoked
from the clang driver this is fine other than the fact that /usr/lib/libc.so
is an ldscript that refers to libc.so.7 which is in /lib, not /usr/lib.
PR: 207980
Reported by: emaste
Submitted by: emaste (based on)
Differential Revision: https://reviews.freebsd.org/D5637
Currently the base.txz distribution does not get the BSD.debug.dist mtree
extracted into it. So if you start from that and then try to build a 3rd-party
application outside of buildworld it will by-default try installing the
debug files into a missing directory if they are being installed into /usr/lib.
Check for the existence before forcing the directory to be created rather than
the older way of running a shell command with test -d || mkdir -p always.
Reported by: HardenedBSD (https://github.com/HardenedBSD/secadm/issues/23)
Sponsored by: EMC / Isilon Storage Division
Differential Revision: https://reviews.freebsd.org/D5411
This was a regression in r295985.
bsd.dep.mk adds to SRCS for dtrace probes, yacc grammars and some
others.
The code that is moving is planned to be removed once FAST_DEPEND is
default (and the only option) though since FAST_DEPEND doesn't use this.
Pointyhat to: bdrewery
Sponsored by: EMC / Isilon Storage Division
This allows 'make analyze' or 'make OBJ.clang-analyzer' to run the
Clang static analyzer and present results on stdout.
Obtained from: NetBSD (CVS Rev. 1.3)
Sponsored by: EMC / Isilon Storage Division
Differential Revision: https://reviews.freebsd.org/D5449
r96164 added them to avoid recursing twice with _SUBDIR. That issue was
fixed in bsd.subdir.mk in r291635 for all targets.
Sponsored by: EMC / Isilon Storage Division
Rather than depend on .depend not existing, check the actual
.depend.OBJ file that will be used for that object. If it doesn't
exist then use the guessed dependencies.
FAST_DEPEND may never have a .depend file. Not having one means all of the
previous logic would over-depend all object files on all headers which is not
what we wanted. It also means that if a .depend is generated before a build
is done for _EXTRADEPEND (such as for PROG or LIB) then all of these
dependencies would not be used since the .depend wasn't generated from mkdep
and the real .depend.* files are not generated until the build.
Sponsored by: EMC / Isilon Storage Division
This was a regression in r290629, which was revealed partly in r294360.
Once 'make depend' has ran it will generate all headers already. Thus
even with FAST_DEPEND lacking proper dependencies before building, it
will not have any missing headers. Once objects are compiled the depend
files will be generated with proper dependencies.
Sponsored by: EMC / Isilon Storage Division
POSIX requires for the c99 compiler.
(In fact, our c99(1) already ignores -lxnet; but our make(1) doesn't set
${CC} correctly, and our cc(1) treats xnet like any other library.)
Reviewed by: kib
This will save time generating dependency files that we didn't expect
due to cases where SRCS!=OBJS or for building custom targetted objects
in Makefiles that do not end up in the DEPENDOBJS list.
This uses a bmake trick to modify CFLAGS based on ${.TARGET}. A
.PARSEDIR check is done for the sake of MFC safety.
MFC after: 2 weeks
Sponsored by: EMC / Isilon Storage Division
has not ran yet.
This fixes building objects directly, or skipping 'make depend', not generating
required headers first. This case did work without FAST_DEPEND so there's no
reason it should not work here as well.
An example of this can be seen building in gnu/usr.bin/binutils/libbfd
without running 'make depend' first to generate config.h.
Sponsored by: EMC / Isilon Storage Division
MFC after: 3 weeks
X-MFC-With: r290433
Similar to the original reason for these dependency hints to be added,
in r124637, the missing-dependency file case can lead to building of the
wrong source.
A clear example of this is in gnu/lib/libstdc++ where the .PATH contains
both contrib/gcc and contrib/libstdc++/src.
contrib/gcc has a debug.c.
contrib/libstdc++/src has a debug.cc.
When building for the objects of debug.o, debug.So, and debug.po, it is
ambiguous for which src file to use due to the suffix transformation
rules, even though the proper one is listed first in .PATH.
This was normally avoided due to these dependency hints for the initial
build, and then mkdep would add an explicit 'debug.o: debug.cc'
dependency into the .depend file. WITH_FAST_DEPEND does not generate
the .depend file with these, but puts them into .depend.debug.o instead.
Rather than extending the exists() check to each object's .depend.*
file, just enable the hint when when using WITH_FAST_DEPEND. It fixes
the problem and seems to be safe enough to use since it is mapping SRCS
back to OBJS, rather than letting make make assumptions from OBJS to
SRCS.
A similar check mapping objects to headers is present in some mk files
but was not extended here for FAST_DEPEND since it has not yet been
found to be a problem.
X-MFC-With: r290433
MFC after: 3 weeks
Sponsored by: EMC / Isilon Storage Division
ccache is mostly beneficial for frequent builds where -DNO_CLEAN is not
used to achieve a safe pseudo-incremental build. This is explained in
more detail upstream [1] [2]. It incurs about a 20%-28% hit to populate the
cache, but with a full cache saves 30-50% in build times. When combined with
the WITH_FAST_DEPEND feature it saves up to 65% since ccache does cache the
resulting dependency file, which it does not do when using mkdep(1)/'CC
-E'. Stats are provided at the end of this message.
This removes the need to modify /etc/make.conf with the CC:= and CXX:=
lines which conflicted with external compiler support [3] (causing the
bootstrap compiler to not be built which lead to obscure failures [4]),
incorrectly invoked ccache in various stages, required CCACHE_CPP2 to avoid
Clang errors with parenthesis, and did not work with META_MODE.
The option name was picked to match the existing option in ports. This
feature is available for both in-src and out-of-src builds that use
/usr/share/mk.
Linking, assembly compiles, and pre-processing avoid using ccache since it is
only overhead. ccache does nothing special in these modes, although there is
no harm in calling it for them.
CCACHE_COMPILERCHECK is set to 'content' when using the in-tree bootstrap
compiler to hash the content of the compiler binary to determine if it
should be a cache miss. For external compilers the 'mtime' option is used
as it is more efficient and likely to be correct. Future work may optimize the
'content' check using the same checks as whether a bootstrap compiler is needed
to be built.
The CCACHE_CPP2 pessimization is currently default in our devel/ccache
port due to Clang requiring it. Clang's -Wparentheses-equality,
-Wtautological-compare, and -Wself-assign warnings do not mix well with
compiling already-pre-processed code that may have expanded macros that
trigger the warnings. GCC has so far not had this issue so it is allowed to
disable the CCACHE_CPP2 default in our port.
Sharing a cache between multiple checkouts, or systems, is explained in
the ccache manual. Sharing a cache over NFS would likely not be worth
it, but syncing cache directories between systems may be useful for an
organization. There is also a memcached backend available [5]. Due to using
an object directory outside of the source directory though you will need to
ensure that both are in the same prefix and all users use the same layout. A
possible working layout is as follows:
Source: /some/prefix/src1
Source: /some/prefix/src2
Source: /some/prefix/src3
Objdir: /some/prefix/obj
Environment: CCACHE_BASEDIR='${SRCTOP:H}' MAKEOBJDIRPREFIX='${SRCTOP:H}/obj'
This will use src*/../obj as the MAKEOBJDIRPREFIX and tells ccache to replace
all absolute paths to be relative. Using something like this is required due
to -I and -o flags containing both SRC and OBJDIR absolute paths that ccache
adds into its hash for the object without CCACHE_BASEDIR.
distcc can be hooked into by setting CCACHE_PREFIX=/usr/local/bin/distcc.
I have not personally tested this and assume it will not mix well with
using the bootstrap compiler.
The cache from buildworld can be reused in a subdir by first running
'make buildenv' (from r290424).
Note that the cache is currently different depending on whether -j is
used or not due to ccache enabling -fdiagnostics-color automatically if
stderr is a TTY, which bmake only does if not using -j.
The system I used for testing was:
WITNESS
Build options: -j20 WITH_LLDB=yes WITH_DEBUG_FILES=yes WITH_CCACHE_BUILD=yes
DISK: ZFS 3-way mirror with very slow disks using SSD l2arc/log.
The arc was fully populated with src tree files and ccache objects.
RAM: 76GiB
CPU: Intel(R) Xeon(R) CPU L5520 @2.27GHz
2 package(s) x 4 core(s) x 2 SMT threads = hw.ncpu=16
The WITH_FAST_DEPEND feature was used for comparison here as well to show
the dramatic time savings with a full cache.
buildworld:
x buildworld-before
+ buildworld-ccache-empty
* buildworld-ccache-full
% buildworld-ccache-full-fastdep
# buildworld-fastdep
+-------------------------------------------------------------------------------+
|% * # +|
|% * # +|
|% * # xxx +|
| |A |
| A|
| A |
|A |
| A |
+-------------------------------------------------------------------------------+
N Min Max Median Avg Stddev
x 3 3744.13 3794.31 3752.25 3763.5633 26.935139
+ 3 4519 4525.04 4520.73 4521.59 3.1104823
Difference at 95.0% confidence
758.027 +/- 43.4565
20.1412% +/- 1.15466%
(Student's t, pooled s = 19.1726)
* 3 1823.08 1827.2 1825.62 1825.3 2.0785572
Difference at 95.0% confidence
-1938.26 +/- 43.298
-51.5007% +/- 1.15045%
(Student's t, pooled s = 19.1026)
% 3 1266.96 1279.37 1270.47 1272.2667 6.3971113
Difference at 95.0% confidence
-2491.3 +/- 44.3704
-66.1952% +/- 1.17895%
(Student's t, pooled s = 19.5758)
# 3 3153.34 3155.16 3154.2 3154.2333 0.91045776
Difference at 95.0% confidence
-609.33 +/- 43.1943
-16.1902% +/- 1.1477%
(Student's t, pooled s = 19.0569)
buildkernel:
x buildkernel-before
+ buildkernel-ccache-empty
* buildkernel-ccache-empty-fastdep
% buildkernel-ccache-full
# buildkernel-ccache-full-fastdep
@ buildkernel-fastdep
+-------------------------------------------------------------------------------+
|# @ % * |
|# @ % * x + |
|# @ % * xx ++|
| MA |
| MA|
| A |
| A |
|A |
| A |
+-------------------------------------------------------------------------------+
N Min Max Median Avg Stddev
x 3 571.57 573.94 571.79 572.43333 1.3094401
+ 3 727.97 731.91 728.06 729.31333 2.2492295
Difference at 95.0% confidence
156.88 +/- 4.17129
27.4058% +/- 0.728695%
(Student's t, pooled s = 1.84034)
* 3 527.1 528.29 528.08 527.82333 0.63516402
Difference at 95.0% confidence
-44.61 +/- 2.33254
-7.79305% +/- 0.407478%
(Student's t, pooled s = 1.02909)
% 3 400.4 401.05 400.62 400.69 0.3306055
Difference at 95.0% confidence
-171.743 +/- 2.16453
-30.0023% +/- 0.378128%
(Student's t, pooled s = 0.954969)
# 3 201.94 203.34 202.28 202.52 0.73020545
Difference at 95.0% confidence
-369.913 +/- 2.40293
-64.6212% +/- 0.419774%
(Student's t, pooled s = 1.06015)
@ 3 369.12 370.57 369.3 369.66333 0.79033748
Difference at 95.0% confidence
-202.77 +/- 2.45131
-35.4225% +/- 0.428227%
(Student's t, pooled s = 1.0815)
[1] https://ccache.samba.org/performance.html
[2] http://www.mail-archive.com/ccache@lists.samba.org/msg00576.html
[3] https://reviews.freebsd.org/D3484
[5] https://github.com/jrosdahl/ccache/pull/30
PR: 182944 [4]
MFC after: 3 weeks
Sponsored by: EMC / Isilon Storage Division
Relnotes: yes
This speeds up buildworld by 16% on my system and buildkernel by 35%.
Rather than calling mkdep(1), which is just a wrapper around 'cc -E',
use the modern -MD -MT -MF flags to gather and generate dependencies during
compilation. This flag was introduced in GCC "a long time ago", in GCC 3.0,
and is also supported by Clang. (It appears that ICC also supports this but I
do not have access to test it). This avoids running the preprocessor *twice*
for every build, in both 'make depend' and 'make all'. This is especially
noticeable when using ccache since it does not cache preprocessor results from
mkdep(1) / 'cc -E', but still speeds up compilation with the -MD flags.
For 'make depend' a tree-walk is still done to ensure that all DPSRCS
are generated when expected, and that beforedepend/afterdepend and
_EXTRADEPEND are all still respected. In time this may change but for now
I've been conservative. The time for a tree-walk with -j combined with
SUBDIR_PARALLEL is not significant. For example, it takes about 9 seconds
with -j15 to walk all of src/ for 'make depend' now on my system.
A .depend file is still generated with the various rules that apply to
the final target, or custom rules. Otherwise there are now
per-built-object-file .depend files, such as .depend.filename.o. These
are included directly by make rather than populating .depend with a loop
and .depend lines, which only added overhead to the now almost-NOP 'make
depend' phase.
Before this I experimented with having mkdep(1) called in parallel per-file.
While this improved the kernel and lib/libc 'make depend' phase, it resulted
in slower build times overall.
The -M flags are removed from CFLAGS when linking since they have no effect.
Enabling this by default, for src or out-of-src, can be done once more testing
has been done, such as a ports exp-run, and with more compilers.
The system I used for testing was:
WITNESS
Build options: -j20 WITH_LLDB=yes WITH_DEBUG_FILES=yes WITH_FAST_DEPEND=yes
DISK: ZFS 3-way mirror with very slow disks using SSD l2arc/log.
The arc was fully populated with src tree files.
RAM: 76GiB
CPU: Intel(R) Xeon(R) CPU L5520 @2.27GHz
2 package(s) x 4 core(s) x 2 SMT threads = hw.ncpu=16
buildworld:
x buildworld-before
+ buildworld-fastdep
+-------------------------------------------------------------------------------+
|+ |
|+ |
|+ xx x|
| |_MA___||
|A |
+-------------------------------------------------------------------------------+
N Min Max Median Avg Stddev
x 3 3744.13 3794.31 3752.25 3763.5633 26.935139
+ 3 3153.34 3155.16 3154.2 3154.2333 0.91045776
Difference at 95.0% confidence
-609.33 +/- 43.1943
-16.1902% +/- 1.1477%
(Student's t, pooled s = 19.0569)
buildkernel:
x buildkernel-before
+ buildkernel-fastdep
+-------------------------------------------------------------------------------+
|+ x |
|++ xx|
| A||
|A| |
+-------------------------------------------------------------------------------+
N Min Max Median Avg Stddev
x 3 571.57 573.94 571.79 572.43333 1.3094401
+ 3 369.12 370.57 369.3 369.66333 0.79033748
Difference at 95.0% confidence
-202.77 +/- 2.45131
-35.4225% +/- 0.428227%
(Student's t, pooled s = 1.0815)
Sponsored by: EMC / Isilon Storage Division
MFC after: 3 weeks
Relnotes: yes
Before this, if a file was installed to DESTDIR/some/dir and that directory
was missing due to not having ran 'make distrib-dirs' yet, the file would
be installed as 'some/dir'. For something like bsd.incs.mk with INCLUDEDIR
being a sub-directory of /usr/include, this could result in all of the headers
being installed to a file rather than getting a directory of them.
Now it will error that the file/directory does not exist rather than hide
the issue.
Another option being discussed is to implement GNU's install -D flag which
would auto create any missing directories.
This is a mitigation of the problem. The proper order to the build is to
run 'make distrib-dirs' first, but that can be forgotten if building from
a sub-directory after updating the source code to the latest revision.
MFC after: 2 weeks
Sponsored by: EMC / Isilon Storage Division
This limits CLEANFILES removal to just bsd.obj.mk now and removes the need
for NOPATH_FILES.
This reverts r96529 which was done due to the command line being too long
for libc. Since then all architectures now use 256k for ARG_MAX (r170102).
Regardless of that, the libc CLEANFILES is only 72k now. Others
may be larger but not likely to hit the limit. If needed, we can improve
the bsd.obj.mk clean: target to split up the list via bmake features.
This also removes some workarounds that are no longer needed.
- a.out removal
- OBJS.tmp, which has not been needed since r117080.
- *.so, which has not been needed since a .so->.So rename in r42450.
This also fixes STATICOBJS and SHLIB_LINK not being in the .NOPATH list.
Sponsored by: EMC / Isilon Storage Division
