* tests/gpgscm/scheme.c (op_code_info): Fix type, add forward
declaration.
(check_arguments): New function.
(Eval_cycle): Use the new function.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/opdefines.h: Make tags mandatory.
* tests/gpgscm/scheme.c: Likewise.
* tests/gpgscm/scheme.h: Likewise.
--
Tags provide a constant-time lookup mechanism for almost every object.
This is useful for the interpreter itself, and the code for tags is
tiny.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (opexe_6): Merge into 'opexe_0'.
* tests/gpgscm/opdefines.h: Adapt.
--
Having separate functions to execute opcodes reduces our ability to
thread the code and prevents the dispatch_table from being moved to
rodata.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (opexe_5): Merge into 'opexe_0'.
* tests/gpgscm/opdefines.h: Adapt.
--
Having separate functions to execute opcodes reduces our ability to
thread the code and prevents the dispatch_table from being moved to
rodata.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (opexe_4): Merge into 'opexe_0'.
* tests/gpgscm/opdefines.h: Adapt.
--
Having separate functions to execute opcodes reduces our ability to
thread the code and prevents the dispatch_table from being moved to
rodata.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (opexe_3): Merge into 'opexe_0'.
* tests/gpgscm/opdefines.h: Adapt.
--
Having separate functions to execute opcodes reduces our ability to
thread the code and prevents the dispatch_table from being moved to
rodata.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (opexe_2): Merge into 'opexe_0'.
* tests/gpgscm/opdefines.h: Adapt.
--
Having separate functions to execute opcodes reduces our ability to
thread the code and prevents the dispatch_table from being moved to
rodata.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (opexe_1): Merge into 'opexe_0'.
* tests/gpgscm/opdefines.h: Adapt.
--
Having separate functions to execute opcodes reduces our ability to
thread the code and prevents the dispatch_table from being moved to
rodata.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/Makefile.am (gpgscm_SOURCES): Add new file.
* tests/gpgscm/scheme-private.h (struct cell): Move number to the top
of the union so that we can initialize it.
(struct scheme): Remove 'integer_segment'.
* tests/gpgscm/scheme.c (initialize_small_integers): Remove function.
(small_integers): New variable.
(MAX_SMALL_INTEGER): Compute.
(mk_small_integer): Adapt.
(mark): Avoid marking objects already marked. This allows us to run
the algorithm over objects in the rodata section if they are already
marked.
(scheme_init_custom_alloc): Remove initialization.
(scheme_deinit): Remove deallocation.
* tests/gpgscm/small-integers.h: New file.
--
Allocate small integers from a fixed pool in the rodata section. This
spares us the initialization, and deduplicates integers across
different processes. It also makes the integers immutable, increasing
memory safety.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme-private.h (struct scheme): Make 'vptr' const.
* tests/gpgscm/scheme.c (num_zero): Statically initialize and turn
into constant.
(num_one): Likewise.
(charnames): Change type so that it can be stored in rodata.
(is_ascii_name): Adapt slightly.
(assign_proc): Make argument const char *.
(op_code_info): Make some fields const char *.
(tests): Make const.
(dispatch_table): Make const. At least it can be made read-only after
relocation.
(Eval_Cycle): Adapt slightly.
(vtbl): Make const.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme-private.h (struct scheme): Remove fixed-size
arrays for cell segments, replace them with a pointer to the new
'struct cell_segment' instead.
* tests/gpgscm/scheme.c (struct cell_segment): New definition.
(_alloc_cellseg): Allocate the header within the segment, return a
pointer to the header.
(_dealloc_cellseg): New function.
(alloc_cellseg): Insert the segments into a list.
(_get_cell): Allocate a new segment if less than a quarter of
CELL_SIGSIZE is recovered during garbage collection.
(initialize_small_integers): Adapt callsite.
(gc): Walk the list of segments.
(scheme_init_custom_alloc): Remove initialization of removed field.
(scheme_deinit): Adapt deallocation.
--
Previously the number of cells that could be allocated was a
compile-time limit. Remove this limit.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme-private.h (struct cell): Use uintptr_t for
'_flags'. This way, '_flags' has the size of a machine word.
--
The compact vector representation introduced in 49e2ae65 requires that
we can tell apart pointers and type flags. This did not work on
64-bit big-endian architectures.
Fixes a crash on 64-bit big-endian architectures.
Hat-tip-to: gniibe
Fixes-commit: 49e2ae65e892f93be7f87cfaae3392b50a99e4b1
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (opexe_5): Do not modify the integer in-place
while printing an vector. Integer objects may be shared, so they must
not be mutated.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (get_vector_object): Initialize unused slots
at the end of vectors.
--
They should not be used for anything, but let's just initialize them
to something benign to be sure.
GnuPG-bug-id: 3014
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (copy_value): New function.
(mk_tagged_value): Use new function.
(opexe_4): Likewise for OP_SAVE_FORCED.
--
Occasionally, we need to copy a value from one location in the storage
to another. Scheme objects are fine. Some primitive objects,
however, require finalization, usually to free resources.
For these values, we either make a copy or acquire a reference.
Fixes e.g. a double free if a delayed expression evaluating to a
string is forced.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (procnum): Procedures always have an integer
number, so we can safely use the cheaper 'ivalue_unchecked'.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/tests.scm (mkdtemp-autoremove): New function that
cleans up at interpreter shutdown.
(run-tests-parallel): Use the new function.
(run-tests-sequential): Likewise.
(make-environment-cache): Execute setup with an temporary working
directory.
--
Make sure to remove all resources created in the filesystem even if
the test runner is interrupted. Make sure to remove anything that the
setup script creates.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/ffi.c (do_get_temp_path): New function.
(ffi_init): Make function available.
* tests/gpgscm/tests.scm (mkdtemp): Use the new function.
Fixes-commit: 06f1f163e96f1039304fd3cf565cf9de1ca45849
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/tests.scm (mkdtemp): Create temporary directories in
'/tmp' on UNIX, or in '%Temp' on Windows.
* tests/migrations/common.scm (run-test): Turn error into a warning.
* tests/openpgp/defs.scm (start-agent): Likewise.
--
This fixes the problem of GnuPG components being unable to communicate
because of too long GnuPG home directories in important build
environments like the Debian build servers despite the use of socket
directories.
This reverts d75d20909d9f60d33ffd210def92278c0f383aad.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (oblist_add_by_name): We now always get a
slot. Simplify accordingly.
(oblist_find_by_name): Always return the slot.
(vector_elem_slot): New function.
(new_slot_spec_in_env): We now always get a slot. Remove parameter
'env'. Simplify accordingly.
(find_slot_spec_in_env): Always return a slot.
(new_slot_in_env): Adapt callsite.
(opexe_0): Likewise.
(opexe_1): Likewise.
(scheme_define): Likewise.
--
Now that the ill-devised immediate values framework is gone, there is
no need to tag the pointers in vectors anymore. Therefore, we can
always return a pointer to the slot in the hash table lookup
functions.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (IMMEDIATE_TAG): Remove macro.
(is_immediate): Likewise.
(set_immediate): Likewise.
(clr_immediate): Likewise.
(enum scheme_types): Set the LSB in every value.
(fill_vector): Adapt.
(vector_elem): Likewise.
(set_vector_elem): Likewise.
(mark): Likewise.
(gc): Test for the LSB to tell typeflags apart from pointers stored in
the same memory location.
--
Supporting immediate values would require invasive changes to the
interpreter and is likely not worth the trouble. On the other hand,
tagging pointers in vectors complicated the hash table implementation
needlessly. Therefore, I remove this again.
This fixes a crash on big endian architectures.
GnuPG-bug-id: 2996
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/tests.scm (test::scm): Add a setup argument.
(test::binary): Likewise.
(run-tests-parallel): Remove setup parameter.
(run-tests-sequential): Likewise.
(make-environment-cache): New function that handles the cache
protocol.
* tests/gpgme/run-tests.scm: Adapt accordingly.
* tests/gpgsm/run-tests.scm: Likewise.
* tests/migrations/run-tests.scm: Likewise.
* tests/openpgp/run-tests.scm: Likewise.
--
This change allows us to have different environments for tests. This
is needed to run more GPGME tests, and to increase concurrency while
running all tests.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c: Define local strlwr only when HAVE_STRLWR is
not defined in config.h.
* tests/gpgscm/scheme-config.h: Remove hack.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/tests.scm (mkdtemp): Do not magically obey the
environment variable 'TMP', make sure to always return an absolute
path.
* tests/gpgme/Makefile.am (TMP): Drop variable.
(TESTS_ENVIRONMENT): Drop 'TMP'.
* tests/gpgme/gpgme-defs.scm (create-gpgmehome): Start the agent. Do
not create private key store, the agent does that for us.
* tests/gpgsm/Makefile.am (TMP): Drop variable.
(TESTS_ENVIRONMENT): Drop 'TMP'.
* tests/gpgme/gpgme-defs.scm (create-gpgsmhome): Start the agent. Do
not create private key store, the agent does that for us.
* tests/migrations/Makefile.am (TMP): Drop variable.
(TESTS_ENVIRONMENT): Drop 'TMP'.
* tests/migrations/common.scm (gpgconf): New variable.
(run-test): Create and remove socket directory.
* tests/migrations/extended-pkf.scm (src-tarball): Remove variable.
(setup): Remove function.
(trigger-migration): Likewise.
Use 'run-test' to execute the test.
* tests/migrations/from-classic.scm (src-tarball): Remove variable.
(setup): Remove function.
Use 'run-test' to execute the tests.
* tests/openpgp/Makefile.am (TMP): Drop variable.
(TESTS_ENVIRONMENT): Drop 'TMP'.
* tests/openpgp/README: Do not mention 'TMP'.
* tests/openpgp/defs.scm (with-home-directory): New macro.
(create-legacy-gpghome): Do not create private key store, the agent
does that for us.
(start-agent): Make sure to terminate the right agent with 'atexit'.
--
Previously, the test suite relied upon creating home directories in
'/tmp'. This has been problematic in some build environments,
although POSIX mandates that '/tmp' must be available.
We now rely on 'gpgconf --create-socketdir' to create a suitable
socket directory for us. This allows us to get rid of some cruft. It
also aligns the environment the tests are run in closer with the
environment that we intend that GnuPG runs in.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (port_increment_current_line): Avoid creating
the same integer if the delta is zero. This happens a lot during
parsing, and puts pressure on the memory allocator.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (opexe_5): Only increment the line number on
newlines.
Fixes-commit: 7cc57e2c63d0fa97569736419db5c76117e7685b
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme-private.h (struct port): Move location
information out of the union.
* tests/gpgscm/scheme.c (mark): All ports need marking now.
(gc): Likewise all ports on the load stack.
(port_clear_location): Adapt accordingly. Also, add an empty function
for !SHOW_ERROR_LINE.
(port_increment_current_line): Likewise.
(port_reset_current_line): Drop function in favor of...
(port_init_location): ... this new function.
(file_push): Simplify.
(file_pop): Likewise.
(port_rep_from_filename): Likewise.
(port_rep_from_file): Likewise.
(port_rep_from_string): Also initialize the location.
(port_rep_from_scratch): Likewise.
(port_close): Simplify and generalize.
(skipspace): Likewise.
(token): Likewise.
(_Error_1): Generalize.
(opexe_5): Likewise.
(scheme_deinit): Simplify and generalize.
(scheme_load_named_file): Likewise.
(scheme_load_string): Also initialize the location.
--
This change tracks the location of source code loaded from non-file
ports that is used in error messages. It also simplifies the code
quite a bit.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/init.scm (vm-history-print): Check that the tag added
to expressions when parsing source files matches the expected format.
* tests/gpgscm/lib.scm (assert): Likewise.
--
This makes the error handling more robust. We saw the assumption
about the format of the tags being violated on one build system, and
it obscured the view on the underlying problem.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/ffi.c (do_get_time): New function.
(ffi_init): Expose new function.
* tests/gpgscm/ffi.scm (get-time): Document new function.
* tests/gpgscm/time.scm: New file.
* tests/openpgp/quick-key-manipulation.scm: Use the new facilities to
check the expiration times of created keys.
* tests/openpgp/tofu.scm: Use the new module.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (oblist_initial_value): Increase the size of
the hash table based on the number of symbols used after initializing
the interpreter.
(new_frame_in_env): Increase the size of the hash table based on the
number of variables in the global environement.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (pointercmp): New function.
(new_slot_spec_in_env): Add and use slot for insertions.
(find_slot_spec_in_env): New variant of 'find_slot_in_env' that
returns the slot on failures.
(find_slot_in_env): Express using the new function.
(new_slot_in_env): Update callsite.
(opexe_0): Optimize lookup-or-insert.
(opexe_1): Likewise.
(scheme_define): Likewise.
--
Optimize environment lookups by keeping the lists in the hash table or
the list sorted. Optimize the insertions by passing the slot computed
by the lookup to the insertion.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (oblist_find_by_name): Keep the list of
symbols sorted, return the slot where a new symbol must be inserted on
lookup failures.
(oblist_add_by_name): Add the new symbol at the given slot.
(mk_symbol): Adjust callsite.
(gensym): Likewise.
(assign_syntax): Likewise.
--
Optimize symbol lookups by keeping the lists in the hash table (or the
list if compiled with USE_OBJECT_LIST) sorted by the symbol names.
Optimize the insertions by passing the slot computed by the lookup to
the insertion.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme-private.h (struct cell): Add a compact vector
representation.
* tests/gpgscm/scheme.c (vector_length): Use new representation.
(vector_size): New macro.
(get_vector_object): Use the new representation.
(fill_vector): Likewise.
(vector_elem): Likewise.
(set_vector_elem): Likewise.
(mark): Likewise.
(gc): Likewise. Be careful not to confuse immediate values for type
flags.
(finalize_cell): Vectors now require finalization.
--
Previously, vectors were represented using consecutive cons cells,
wasting one word per cell for the type information. Fix that by using
a flat array.
Previously, a vector of size N required 1 + (n + 1) / 2 cells. Now it
uses 1 + (n - 1 + 2) / 3 cells.
Signed-off-by: Justus Winter <justus@g10code.com>
* tests/gpgscm/scheme.c (IMMEDIATE_TAG): New macro.
({is,set,clr}_immediate): Likewise.
(enum scheme_types): Make type tags disjoint from immediate values.
(TYPE_BITS): We need one more bit now.
(ADJ,T_MASKTYPE): Compute values.
--
Immediate values are disjoint from all type tags and pointers,
allowing us to store immediate values in all memory locations.
Signed-off-by: Justus Winter <justus@g10code.com>