Mingw and Cygwin builds install the DLLs in the application directory,
not the library directory, so ensure that one is created for them when
installing the DLLs.
Fixes#7653
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7654)
We only had the main 'install' target depend on 'all'. This changes
the dependencies so targets like install_dev, install_runtime_libs,
install_engines and install_programs depend on build targets that are
correspond to them more specifically. This increases the parallel
possibilities.
Fixes#7466
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7583)
When trying 'make -j{n} install', you may occasionally run into
trouble because to sub-targets (install_dev and install_runtime) try
to install the same shared libraries. That makes parallel install
difficult.
This is solved by dividing install_runtime into two parts, one for
libraries and one for programs, and have install_dev depend on
install_runtime_libs instead of installing the shared runtime
libraries itself.
Fixes#7466
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7583)
This adds a keyword SUBDIRS for build.info, to be used like this:
SUBDIRS=foo bar
This tells Configure that it should look for 'build.info' in the
relative subdirectories 'foo' and 'bar' as well.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7558)
This simple fix allows the following construct:
PROGRAMS=foo
SOURCE[foo]=foo.c bar.c
DEFINE[foo]=FOO=1 BAR=0
These will trickle down to the build of object files, so building
foo.o and bar.o will be done with these options: -DFOO=1 -DBAR=0
(exact syntax depending on platform, of course)
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7553)
Sometimes, some specific program or object file might need an extra
macro definition of its own. This allows that to be easily done.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7553)
This makes sure that any resulting directory target in the build files
also depend on object files meant for shared libraries.
As a side effect, we move the production of the dirinfo structure from
common.tmpl to Configure, to make it easier to check the result.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7452)
This excluses user from additional PATH adjustments in case NDK has
llvm-ar.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7443)
03ad7c009e failed if one didn't pass
explicit -D__ANDROID_API__=N :-(
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7443)
We made the build of foo.obj depend on foo.d, meaning the latter gets
built first. Unfortunately, the way the compiler works, we are forced
to redirect all output to foo.d, meaning that if the source contains
an error, the build fails without showing those errors.
We therefore remove the dependency and force the build of foo.d to
always happen after build of foo.obj.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7469)
This also gave enough reason to collect the stuff that's common for
all iOS config targets into the template "ios-common".
Fixes#7318
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7403)
When building shared libraries on Windows, we had a clash between
'libcrypto.lib' the static routine library and 'libcrypto.lib' the
import library.
We now change it so the static versions of our libraries get '_static'
appended to their names. These will never get installed, but can
still be used for our internal purposes, such as internal tests.
When building non-shared, the renaming mechanism doesn't come into
play. In that case, the static libraries 'libcrypto.lib' and
'libssl.lib' are installed, just as always.
Fixes#7492
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7496)
We only had targets for the "simple" shared library names (libfoo.so
and not libfoo.so.x.y on Unix, import library libfoo.lib but not
libfoo.dll on Windows). This has created some confusion why it wasn't
possible to rebuild the less "simple" name directly (just as an
example, someone who mistook the import library on Windows for a
static library, removed the DLL and then found it was difficult to
rebuild directly), so we change the target to include all possible
names.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7451)
When looking at configured macro definitions, we must look at both
what comes from the config target AND what comes from user
configuration.
Fixes#7396
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/7402)
Some modules are built with case insensitive (uppercase) symbols on
VMS. This needs to be reflected in the export symbol vector.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7347)
Rewrite util/mknum.pl to become cleaner, and to use the separate
generic C header parsing module, as well as the separate ordinals
manipulation module.
Adapt the build files.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7191)
Move the .num updating functionality to util/mknum.pl.
Rewrite util/mkdef.pl to create .def / .map / .opt files exclusively,
using the separate ordinals reading module.
Adapt the build files.
Adapt the symbol presence test.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7191)
The generation of linker scripts was badly balanced, as all sorts of
platform dependent stuff went into the top build.info, when that part
should really be made as simply and generic as possible.
Therefore, we move a lot of the "magic" to the build files templates,
since they are the place for platform dependent things. What remains
is to parametrize just enough in the build.info file to generate the
linker scripts correctly for each associated library.
"linker script" is a term usually reserved for certain Unix linkers.
However, we only use them to say what symbols should be exported, so
we use the term loosely for all platforms. The internal extension is
'.ld', and is changed by the build file templates as appropriate for
each target platform.
Note that this adds extra meaning to the value of the shared_target
attribute.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7333)
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/7277)
With the change to have separate object files by intent, VMS name
mangling gets done differently. While we previously had that for
libraries only, we must now turn that on generally for our programs,
because some of them depend in internal libraries where mangled names
are all that there is.
Dynamic modules are still built with non-mangled names, which is good
enough to show that it's possible to build with our public libraries
using our public headers.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7208)
The possibility to do this was killed when we started producing object
file names with encoded intention (and possibly different builds), and
leads to build errors.
With that, 'libobj2shlib' is renamed to 'obj2shlib' to reflect this
design change. The old name is still used if the new one isn't
available, for the sake of backward compatibility.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7198)
This is in preparation for having separate CFLAGS variables for static
and for shared library builds.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7159)
This will allow to have different object files for different products,
even if they share the same source code, and possibly different builds
for those different object files.
For example, one can have something like this:
SOURCES[libfoo]=cookie.c
INCLUDES[libfoo]=include/foo
SOURCES[libbar]=cookie.c
INCLUDES[libbar]=include/bar
This would mean that the object files and libraries would be build
somewhat like this:
$(CC) -Iinclude/foo -o libfoo-lib-cookie.o cookie.c
$(AR) $(ARFLAGS) libfoo.a libfoo-lib-cookie.o
$(CC) -Iinclude/bar -o libbar-lib-cookie.o cookie.c
$(AR) $(ARFLAGS) libbar.a libbar-lib-cookie.o
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7159)
Instead, use the include settings from the products later in the process,
making it possible to have different includes for two different libraries
that share the same source code.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7159)
-path is non-portable extension, fortunately it's possible to express
.git subdirectory exclusion with -prune.
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7004)
This allows the original path to be displayed when it's shown
to be invalid, so the user can relate without question.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6925)
Extra slashes in paths are permissible in Unix-like platforms...
however, when compared with the result from 'which', which returns
canonical paths, the comparison might fail even though the compared
paths may be equivalent. We make the NDK path canonical internally to
ensure the equivalence compares as equal, at least for the most
trivial cases.
Fixes#6917
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6924)
On the same note, change the 'NASM not found' message to give specific
advice on how to handle the failure.
Fixes#6765
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6771)
The result is that we don't have to produce different names on
different platforms, and we won't have confusion on Windows depending
on if the script was built with mingw or with MSVC.
Partial fix for #3254
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6764)
To avoid the possibility that someone creates rem.exe, rem.bat or
rem.cmd, simply don't use it. In the cases it was used, it was to
avoid empty lines, but it turns out that nmake handles those fine, so
no harm done.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/6686)
It seems that nmake first tries to run executables on its own, and
only pass commands to cmd if that fails. That means it's possible to
have nmake run something like 'echo.exe' when the builtin 'echo'
command was expected, which might give us unexpected results.
To get around this, we create our own echoing script and call it
explicitly from the nmake makefile.
Fixes#6670
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/6686)
OpenSSL 1.1.0 supports the use of this environment variable for
passing to the build files. For the sake of backward compatibility,
we keep it.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/6668)
This makes AIX build procedure behave more like e.g. Solaris. Most
notably this makes it possible to pass -Wl,-R,'$(LIBRPATH)' at config
time to embed installation destination as library search path into
openssl binary. This doesn't imply that other applications have to be
linked with -bsvr4, they are free to choose whatever appropriate for
given circumstances.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6487)
AIX treats its shared libraries in unique manner, by placing multiple
shared objects of different versions and bitnesses, into .a file.
So far we have been naively linking with version-less libcrypto|ssl.so,
which poses long-term maintenance problems. One could choose to link
straight with libcrypto.so.X.Y [or libcrypto.X.Y.so], but it would be
inconsistent with the way AIX [or Unix] does things.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6487)
Add irix-common template that covers even irix-shared from shared-info.pl.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6536)
- Print positive feedback in the case when 'make doc-nits' finds no errors.
- Other than before, keep the 'doc-nits' output file only in case of errors
and remove it if it is empty.
- Declare 'doc-nits' as a phony make target to facilitate rerunning
'make doc-nits' without having to remove the output file first.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6517)
Add aix-common template that covers even aix-shared from shared-info.pl,
add -bsymbolic to shared_ldflags.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6453)
The flag was apparently omitted in switch from Makefile.shared to
shared-info.pl. Do put it back! And in the process move all
solaris-shared flags from shared-info.pl to solaris-common.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6446)
Move ios targets to 15-ios.conf and modernize by deploying xcrun.
This excuses user from looking for paths and setting environment
variables. [Thanks to @0neday for hint.]
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6410)
This adds the possibility to exclude files by regexp in util/copy.pl
Partial fix for #3254
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6303)
LINK can outsmart itself and choose to not update export .lib upon
corresponding .dll re-link. Since dependency is between .lib and all
.obj-s, re-compilation of any .obj makes NMAKE relink .dll and all
.exe-s over and over...
Reviewed-by: Rich Salz <rsalz@openssl.org>
deletion of *.exp files in krb5 sub-module.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6186)
config probe doesn't work in cross-compile scenarios or with clang.
In addition consolidate -Qunused-arguments handling.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6174)
Rationale for enforcing no-engine is because of disconnect between
compile-time config and run-time, which is a per-application sandbox
directory which one can't predict in advance. Besides, none of the
bundled engines actually give an edge on iOS...
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6106)
This didn't get built anyway for gcc because it was detected as a cross
compile. But it did get built for clang - even though this is still a cross
compile build. This disables it in all cases for Android.
Fixes#5748
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/6057)
Configurations/90-team.conf isn't for public consumption, so we rename
it to 90-team.norelease.conf and make sure 'make dist' and 'make tar'
don't include it in the tarball.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5836)
Purpose of build_all_generated is to execute all the rules that require
perl, so that one can copy the tree to system with compiler but without
perl. This commit removes last dependencies on perl.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5929)
Computing the value of the GENERATED variable in the build file
templates is somewhat overcomplicated, and because of possible
duplication errors, changes are potentially error prone.
Looking more closely at how this list is determined, it can be
observed that the exact list of files to check is consistently
available in all the values found in the %unified_info tables
'depends', 'sources' and 'shared_sources', and all that's needed is to
filter those values so only those present as keys in the 'generate'
table are left.
This computation is also common for all build files, so due to its
apparent complexity, we move it to common0.tmpl, with the result left
in a global variable (@generated), to be consumed by all build file
templates.
common0.tmpl is included among the files to process when creating
build files, but unlike common.tmpl, it comes first of all.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5930)
HP-UX provides sockets symbols with incompatible prototypes under same
name. This caused problems in 64-bit builds. Additional macros force
unambiguous symbols with unambiguous prototypes.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5742)
We started using $(CPP) instead of $(CC) -E, with the assumption that
CPP would be predefined. This is, however, not always true, and
rather depends on the 'make' implementation. Furthermore, on
platforms where CPP=cpp or something else other than '$(CC) -E',
there's a risk that it won't understand machine specific flags that we
pass to it. So it turns out that trying to use $(CPP) was a mistake,
and we therefore revert that use back to using $(CC) -E directly.
Fixes#5867
Note: this affects config targets that use Alpha, ARM, IA64, MIPS,
s390x or SPARC assembler modules.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5872)
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5625)
The LIBZ macro definition was already quoted in BASE_windows, then got
quotified once more in windows-makefile.tmpl. That's a bit too much
quotations, ending up with the compiler being asked to define the
macro |"LIBZ=\"ZLIB1\""| (no, not the macro LIBZ with the value
"ZLIB1"). This is solved by removing the extra quoting in
BASE_windows.
Along with this, change the quotation of macro definitions and include
file specification, so we end up with things like -I"QuotedPath" and
-D"Macro=\"some weird value\"" rather than "-IQuotedPath" and
"-DMacro=\"some weird value\"".
Fixes#5827
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5833)
picker() is type agnostic, but its output consumer is not. Or rather
it doesn't work if picker() picks nothing when consumer expects
array. So ensure array is returned when array is expected.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5770)
This is quick-n-dirty ad-hoc solution, the problem asks for more
elegant one...
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5755)
.res files weren't treated consistently. They weren't included at all
in the build of a shared library, and were named inconsistently; .res
sometimes, and .res.o otherwise. Now we standardise it to .res.o,
which is the recommended way with GNU tools.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5730)
Although it deviates from the actual prototype of DSO_dsobyaddr(), this
is now ISO C compliant and gcc -Wpedantic accepts the code.
Added DATA segment checking to catch ptrgl virtual addresses. Avoid
memleaks with every AIX/dladdr() call. Removed debug-fprintf()s.
Added test case for DSO_dsobyaddr(), which will eventually call dladdr().
Removed unecessary AIX ifdefs again.
The implementation can only lookup function symbols, no data symbols.
Added PIC-flag to aix*-cc build targets.
As AIX is missing a dladdr() implementation it is currently uncertain our
exit()-handlers can still be called when the application exits. After
dlclose() the whole library might have been unloaded already.
Signed-off-by: Matthias Kraft <makr@gmx.eu>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5668)
Since they intend to omit gcc, it's more appropriate to simply detect
if there is NDK's clang on PATH, as opposite to requiring to specify it
with CC=clang (and looking for it on PATH).
Also detect NDK version and default to armv7-a for NDK>16.
Address failure to recognize -D__ADNDROID_API__=N in CPPFLAGS.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5613)
In the end, it's more efficient to only have one perl instance (that
loads configdata.pm) dealing with dependency files than running one
(that still loads configdata.pm) for each such file.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5631)
It seems that only gcc -MMD produces dependency files that are "sane"
for our needs. For all other methods, some post processing is needed:
- 'makedepend' (Unix) insists that object files are located in the
same spot as the source file.
- 'cl /Zs /showIncludes' (Visual C) has "Note: including file: " where
we'd like to see the object.
- 'CC/DECC' (VMS) insists that the object file is located in the
current directory, i.e. it strips away all directory information.
So far, we've managed this (except for the VMS case) with individual
uncommented perl command lines directly in the build file template.
We're now collecting these diverse hacks into one perl script that
takes an argument to tell what kind of input to expect and that
massages whatever it gets on STDIN and outputs the result on STDOUT.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5591)
