* module/language/tree-il/primitives.scm (*interesting-primitive-names*):
(*primitive-accessors*): Add string->utf8, utf8->string, and
string-utf8-length.
(primitive-module): New public function, moved here from (language
tree-il compile-bytecode).
* module/language/tree-il/compile-bytecode.scm: Use primitive-module
from (language tree-il primitives).
* module/language/tree-il/peval.scm (peval): A bugfix: load primitives
from their proper module. Allows bytevector primitives to fold.
* module/language/cps/guile-vm/reify-primitives.scm:
* module/language/cps/effects-analysis.scm:
* module/language/cps/types.scm
* module/language/tree-il/primitives.scm:
* module/language/tree-il/cps-primitives.scm:
* module/language/tree-il/effects.scm (make-effects-analyzer):
Add string->utf8, utf8->string, and string-utf8-length.
* module/language/tree-il/compile-cps.scm (string->utf8)
(string-utf8-length, utf8->string): New custom lowerers, including type
checks and an unboxed result for string-utf8-length.
* module/system/vm/assembler.scm:
* libguile/intrinsics.h:
* libguile/intrinsics.c: Because string-utf8-length returns an unboxed
value, we need an intrinsic for it; go ahead and add an intrinsic for
string->utf8 and utf8->string too, as we will likely be able to use
these in the future.
Previously, these instructions were compiled directly. However, the JIT
also wants to fuse comparisons with branches, and an intervening "drop"
breaks that. Instead we take a different approach and rely on the
compiler only emitting push/pop/drop in a peephole sort of way, and we
can just simulate the temp stack space and dispatch directly.
* libguile/jit.c (compile_push, compile_push_slow, compile_pop):
(compile_pop_slow, compile_drop, compile_drop_slow): Crash if these are
called.
(COMPILE_WIDE_OP1, COMPILE_WIDE_OP2, COMPILE_WIDE_OP3, COMPILE_WIDE_OP4)
(COMPILE_WIDE_OP5):
(COMPILE_WIDE_DOP1, COMPILE_WIDE_DOP2, COMPILE_WIDE_DOP3)
(COMPILE_WIDE_DOP4, COMPILE_WIDE_DOP5): New helpers. I didn't manage to
share implementation with COMPILE_*.
(COMPILE_WIDE_*): New variants of compilers that take their operands
from an array instead of parsing the inline operands, relying on
convention for how the compiler emits these instructions.
(parse_wide_operands): New helper to parse out "wide" operands.
(compile1, compile_slow_path): If we see a "push", parse wide operands.
(compile): Move label capture for the slow path into compile_slow_path,
because wide operands may skip some instructions.
* libguile/jit.c (compile_*): Instead of using the minimum sized types
that can represent the instruction's operand, use uint32_t. This will
allow us to handle push/pop/drop without moving the SP.
* libguile/symbols.c (scm_symbol_to_string, scm_string_to_symbol):
Remove some confusing documentation that assumes that Guile is
case-insensitive, and which uses a word that may be perceived as a slur.
* module/ice-9/test.scm:
* test-suite/tests/r4rs.test: Rename a test to avoid using a slur.
Recognize `raise-exception` in the same way we recognize `throw`, though
it is a bit less optimized and the boot story is not as complicated.
* doc/ref/vm.texi (Non-Local Control Flow Instructions):
* libguile/jit.c (compile_unreachable):
(compile_unreachable_slow):
* libguile/vm-engine.c (VM_NAME):
* module/language/cps/compile-bytecode.scm (compile-function):
* module/system/vm/assembler.scm (emit-unreachable): Add new
"unreachable" instruction, inserted after a call to non-continuable
`raise-exception`.
* module/language/tree-il/compile-cps.scm (raise-exception):
* module/language/tree-il/primitives.scm
(*interesting-primitive-names*): Recognize raise-exception, and if it is
called with just one argument, prune that branch of the control-flow
graph.
In https://gcc.gnu.org/wiki/Visibility, they recommend _WIN32 and
__CYGWIN__ as the test defines for dllexport.
Also, logic is incorrect since HAVE_VISIBILITY can be defined but zero.
* libguile/scm.h (SCM_API): modify #ifdef logic
clearenv() may not be provided by non-glibc systems. As a fallback,
just set environ to NULL.
* libguile/posix.c (scm_environ)[!HAVE_CLEARENV]: add fallback logic
for clearenv()
Clang uses a different format for inline assembly. Also, as noted
in the comment, this register usage is likely moot.
* libguile/vm-engine.c (JT_REG)[__GNUC__ && !__clang __]: define to empty
'nice' is marked as 'warn_unused_result' on some versions
of Ubuntu which causes make distcheck to fail.
This implements the error checking logic exactly as POSIX suggests
to silence the warning.
* libguile/posix.c (scm_nice): new error checking logic.
When calling `environ', Guile set the global variable `environ' to a
list allocated with the GC. Strings in it are also allocated with the
GC.
However, if an user call the Scheme setenv() procedure, the resulting
call to putenv() in libc might reallocate `environ' to a new pointer
while copying sub-pointers owned by Guile in it.
This results in the GC marking these strings for reclamation when they
are actually still present in `environ'. Thus, the values in the
environment are now undefined.
To fix this, Guile should only manipulate the `environ' using the
standard libc functions. This ensures that concurrent modification of
it is safe in multi-threaded program. Therefore, the procedure
`environ' now call the libc clearenv() procedure to purge the
environment. Then, the desired values are put in `environ' using
scm_putenv(). At the end, no GC allocated memory is put in `environ'.
Also, since `environ' can be changed at anytime in a multi-thread
program, emit a warning stipulating that the result is undefined
behavior if multiple threads are created in the program. Consider for
example a thread iterating over `environ' while another one do a call to
putenv(). The latter would do a realloc() on `environ' and thus the old
array read by the former now contains garbage.
On system where clearenv() is not present, an atomic store of NULL with
sequential consistency to `environ' should be sufficient but see the
NOTES of clearenv(3).
* libguile/posix.c (scm_environ): Do not store GC allocated memory in
environ.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* libguile/exceptions.c (exception_epoch_fluid): Rename from
active_exception_handlers_fluid.
(scm_dynwind_throw_handler): Increment exception epoch instead of
resetting active exception handlers.
(scm_init_exceptions): Update.
* module/ice-9/boot-9.scm (with-exception-handler): Rework to associate
an "epoch" fluid with each exception handler.
(with-throw-handler): Establish a new epoch, during the execution of a
throw handler.
(raise-exception): Rework to avoid capturing a list of exception
handlers, and to use epochs as a way to know which handlers have already
been examined and which are on the dispatch stack.
* test-suite/tests/exceptions.test ("throwing within exception
handlers"): New test.
This also makes soft ports suspendable.
* am/bootstrap.am (SOURCES): Add (ice-9 soft-ports).
* libguile/init.c (scm_i_init_guile): No need to init vports.
* libguile/vports.c: Call out to (ice-9 soft-ports).
* libguile/vports.h: Remove internal scm_init_vports.
* module/ice-9/boot-9.scm (the-scm-module): Import (ice-9 soft-ports).
Really this enlarges the boot closure a bit, so we should probably
refactor.
* module/ice-9/soft-ports.scm: New file.
* libguile/r6rs-ports.c: Call out to Scheme instead of defining here.
* libguile/r6rs-ports.h: Put custom binary port decls together, to
deprecate later.
* module/ice-9/binary-ports.scm: Re-implement custom binary ports in
terms of custom ports.
Custom ports are a kind of port that exposes the C port type interface
directly to Scheme. In this way the full capability of C is available
to Scheme, and also the read and write functions can be tail-called from
Scheme (via port-read / port-write).
* libguile/custom-ports.c:
* libguile/custom-ports.h:
* module/ice-9/custom-ports.scm: New files.
* libguile/init.c:
* libguile/Makefile.am:
* am/bootstrap.am: Add to the build.
* doc/ref/api-io.texi: Update the manual.
In Guile 3.0.9, 'system*' would no longer open /dev/null for file
descriptors 0, 1, and 2 when its 'current-input-port',
'current-output-port', or 'current-output-port' is not bound to a file
port. This patch reinstates that behavior.
Fixes <https://bugs.gnu.org/63024>.
* libguile/posix.c (piped_process): Open /dev/null to use as in/out/err
if the corresponding port is not backed by a file descriptor.
* test-suite/tests/posix.test ("system*")["https://bugs.gnu.org/63024"]:
New test.
* NEWS: Update.
Co-authored-by: Ludovic Courtès <ludo@gnu.org>
Fixes https://debbugs.gnu.org/63279. The issue was that if the producer
thunk caused a backtrace, pretty-printing the call-with-values frame
would segfault because there was an unininitialized slot on the stack.
For functions produced by the compiler this wouldn't be a problem
because there are stack maps, but primitives require that all slots on a
pending stack frame be packed (no uninitialized values) and tagged (all
SCM values, no unboxed values).
* test-suite/tests/error-handling.test: New test.
* test-suite/Makefile.am: Add new file.
* libguile/vm.c (define_vm_builtins): Fix call-with-values to have a
more compact stack.
* configure.ac: Check for 'posix_spawn_file_actions_addclosefrom_np'.
* libguile/posix.c (HAVE_ADDCLOSEFROM): New macro.
(close_inherited_fds): Wrap in #ifdef HAVE_ADDCLOSEFROM.
(do_spawn) [HAVE_ADDCLOSEFROM]: Use 'posix_spawn_file_actions_addclosefrom_np'.
This reverts 9332b63240, thereby
reinstating the performance issue in <https://bugs.gnu.org/59321>.
This optimization was subject to race conditions in multi-threaded code:
new file descriptors could pop up at any time and thus leak in the
child.
* libguile/posix.c (close_inherited_fds): Remove.
(close_inherited_fds_slow): Rename to...
(close_inherited_fds): ... this.
Fixes <https://bugs.gnu.org/61095>.
Reported by Omar Polo <op@omarpolo.com>.
* libguile/posix.c (close_inherited_fds_slow): On systems other than
GNU/Linux, call 'addclose' only when 'fcntl' succeeds on MAX_FD.
* NEWS: Update.
Noticed while investigating a migration to utf-8 strings. After making
changes that routed non-ascii symbol hashing through this function,
encoding-iso88597.test began intermittently failing because it would
traverse trailing garbage when u8_strnlen reported 8 chars instead of 4.
Change the scm_i_str2symbol and scm_i_str2uninterned_symbol internal
hash type to unsigned long to explicitly match the scm_i_string_hash
result type.
* libguile/hash.c (scm_i_utf8_string_hash): Call u8_mbsnlen not u8_strnlen.
* libguile/symbols.c (scm_i_str2symbol, scm_i_str2uninterned_symbol):
Use unsigned long for scm_i_string_hash result.
* test-suite/standalone/.gitignore: Add test-hashing.
* test-suite/standalone/Makefile.am: Add test-hashing.
* test-suite/standalone/test-hashing.c: Add.
Fixes <https://bugs.gnu.org/61073>.
* libguile/posix.c (FDES_FROM_PORT_OR_INTEGER): When OBJ is not an
integer, use 'SCM_VALIDATE_OPFPORT' before using 'SCM_FPORT_FDES'.
* test-suite/tests/posix.test ("spawn")["non-file port argument"]: New
test.
Fixes <https://bugs.gnu.org/60971>.
Reported by lloda <lloda@sarc.name> and Greg Troxel <gdt@lexort.com>.
On macOS and NetBSD, 'WEXITSTATUS' expects an lvalue so the expression
passed to 'verify' would be invalid.
* libguile/posix.c: Move 'verify' assertion within #ifdef.
This is a followup to f859e0f58b, which
led to warnings on GNU/Linux:
threads.c:358:43: warning: 'scm_i_current_thread' initialized and declared 'extern'
* libguile/threads.c (scm_i_current_thread): Make 'SCM_INTERNAL'
conditional.
* module/rnrs/bytevectors/gnu.scm: New file.
* am/bootstrap.am (SOURCES): Add it.
* libguile/bytevectors.c (scm_bytevector_slice): New function.
* libguile/bytevectors.h (scm_bytevector_slice): New declaration.
* test-suite/tests/bytevectors.test ("bytevector-slice"): New tests.
* doc/ref/api-data.texi (Bytevector Slices): New node.
Fixes <https://bugs.gnu.org/59321>.
Reported by <hylophile@posteo.de>.
Some systems provide "/proc/self/fd" which is a directory containing an
entry for each open file descriptor in the current process. We use this
to limit the number of close() calls needed to ensure file descriptors
aren't leaked to the child process when forking.
* libguile/posix.c (close_inherited_fds_slow):
(close_inherited_fds): New static helper functions.
(scm_spawn_process): Attempt to close inherited file descriptors
efficiently using 'close_inherited_fds', falling back to the brute-force
approach in 'close_inherited_fds_slow'.
* NEWS: Update.
If we got interrupted while waiting on our condition variable, we unlock
the kernel mutex momentarily while executing asynchronous operations
before putting us back into the waiting queue.
However, we have to retry acquiring the mutex before getting back into
the queue, otherwise it's possible that we wait indefinitely since
nobody could be the owner for a while.
* libguile/threads.c (lock_mutex): Try acquring the mutex after signal
interruption.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Fixes <https://bugs.gnu.org/59021>.
Previously, the stack allocated in 'capture_stack' and stored in
'p->stack_bottom' could be retained, leading to heap growth.
* libguile/vm.c (capture_stack): Make a single 'scm_gc_malloc' call
instead of two.
* configure.ac: Detect if ‘openat’ is defined.
* libguile/filesys.c
(flags_to_mode): Extract from ...
(scm_mode): ... here.
(scm_open_fdes_at, scm_openat): Define the Scheme bindings.
* libguile/filesys.h (scm_open_fdes_at, scm_openat): Make them part
of the API.
* doc/ref/posix.texi (File System): Document them.
* test-suite/tests/filesys.test ("openat"): Test ‘openat’.
* libguile/syscalls.h (openat_or_openat64): Decide between ‘openat’
and ‘openat64’.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* configure.ac: Detect if ‘fstatat’ is defined.
* libguile/filesys.c (scm_statat): Define a Scheme binding to ‘fstatat’.
* libguile/filesys.h (scm_statat): Make it part of the C API.
* doc/ref/posix.texi (File System): Document it.
* libguile/syscalls.h (fstatat_or_fstatat64): Choose between ‘fstatat’
and ‘fstatat64’.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* configure.ac: Detect whether ‘fchownat’ is available.
* libguile/filesys.c (scm_chownat): Define a Scheme binding to
‘fchownat’ when available.
* libguile/filesys.h (scm_chownat): Make it part of the API.
* doc/ref/posix.texi (File System): Document it.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
‘unlinkat’ is used for both unlinking regular files
and removing empty directories.
* configure.ac: Detect if ‘unlinkat’ exists.
* doc/ref/posix.texi (File System): Document why there is no
‘rmdirat’ procedure, and document the ‘delete-file-at’ procedure.
* libguile/filesys.c
(scm_rmdir): Adjust the docstring here as well.
(scm_delete_file_at): Define a Scheme binding to ‘unlinkat’.
* libguile/filesys.h (scm_delete_file_at): Make ‘scm_delete_file_at’
part of the C API.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* configure.ac: Detect existence of fchmodat.
* libguile/filesys.c (scm_chmodat): New procedure.
* libguile/filesys.h (scm_chmodat): Make it part of the API.
* test-suite/tests/filesys.test ("chmodat"): Test it.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* configure.ac: Detect if ‘renameat’ is defined.
* libguile/filesys.c (scm_renameat): Define a Scheme binding
to the ‘renameat’ system call.
* doc/ref/posix.texi (File System): Document it.
* libguile/filesys.h (scm_renameat): Make it part of the C API.
* test-suite/tests/filesys.test ("rename-file-at"): New tests.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
