OCaml's signal handlers need to know the state of the processor when the signal was received. The approach standardized by the Single Unix specification is to use 3-argument signal handing functions and the SA_SIGINFO flag to sigaction. However, as the Linux man page for sigaction says,
Undocumented:
Before the introduction of SA_SIGINFO, it was also possible to get some
additional information, namely by using a sa_handler with a second
argument of type struct sigcontext. See the relevant Linux kernel
sources for details. This use is obsolete now.
For historical reasons, the i386, PowerPC and s390x Linux ports of OCaml still use the undocumented, obsolete approach, while the other Linux ports use the modern SA_SIGINFO approach.
For consistency and future-proofing, this PR updates the i386, PowerPC64, and s390x Linux ports to use the modern approach to signal handling. PowerPC32 was left as before because of technical subtleties and lack of hardware to test changes.
The new code for PowerPC 64 bits includes the trick proposed in PR#1795 to support Musl in addition to Glibc on ppc64le.
The test takes 14s to run every time we run the checksuite, and it
does not seem to serve an easily identifiable purpose. In theory
similar tests could help detect a performance regression in
exhaustiveness checking, but this test seems to only degrade in
constant factor (#9152 proposed to use ocamlc.opt instead of ocamlc to
run it, but it was declined as "hiding" potential problems, which
suggests constant-factor differences). We do not monitor the testsuite
for 10s timing difference, so it is unlikely that we would notice
a constant-factor difference during automated testing.
(This is a small, independent readability fix.)
ARCH_SPECIFIC was moved into Makefile.compilerlibs by mistake; it is
not used in Makefile.compilerlibs (in particular the
ARCH_SPECIFIC_ASMCOMP_* variables do not depend on it), and it is used
in the main Makefile.
Before we would include it if present, but support building without
it. Unless I misunderstand something, this is unnecessarily complex:
- If we want to support building without it, we may as well
do without it all the time (in particular it makes it easier
to spot bugs which occurs when it is missing).
- Certain makefile rules (see config.status) work by being added
as dependencies to Makefile.build_config, and requiring this file
only when actually needed makes it possible to simplify those.
(simplifications not included in this commit)
I reviewed all 'clean' rules to check that they do not depend on
variables defined in the configure. On the other hand, the main
makefiles do a lot of conditional tests on those variables,
so --warn-undefined-variables warns quite a bit. (This could be solved
by a Makefile.no_config rule that mirrors Makefile.config with dummy
definitions.)
As discussed among the core OCaml developers.
This commit adds a Github action that "pings" issues that have been inactive for one year and schedules them for closing 30 days later if no action has been taken.
On FreeBSD, libbfd is - similarly to OpenBSD - not installed by default, but
can be added via the devel/libbfd port. This will install libbfd into the
/usr/local prefix, thus configure needs to look there for include and library
It's the macOS default installed version still. The dependency generation
inadvertently relies on behaviour introduced in GNU make 3.82 a decade ago.
The fix in otherlibs/systhreads/Makefile also corrects missing
NATIVE_CPPFLAGS when generating st_stubs.n.d, so st_stubs.n.o now
correctly depends on caml/stacks.h instead of caml/stack.h
The $(wildcard *.h) should only be there with
--disable-dependency-generation, since otherwise gcc -MM will be
determining exactly which header files should be checked.
Only the "old school build" test uses --disable-dependency-generation.
This is also tested on both Travis and AppVeyor, so we have good release
coverage checking of this option.
caml_alloc returns initialised blocks for tag < No_scan_tag. Otherwise,
initialise the blocks as necessary.
For Abtract_tag, Double_tag and Double_array_tag, the initial contents
are irrelevant.
Uninitialised Custom_tag objects are difficult to use correctly. Hence,
reject custom block allocations through Obj.new_block.
For String_tag, the last byte encodes the string length. Hence, reject
zero-length string objects. Initialise the last byte which encodes the
length to ensure non-negative lengths for uninitialised strings.
Xavier Leroy
Gabriel Scherer
Christophe Troestler
Jacques-Henri Jourdan
Florian Angeletti
Xavier Leroy
David Allsopp
octachron
Hannes Mehnert
Josh Berdine
Josh Berdine
KC Sivaramakrishnan