summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* optimizer: remove root slot from basic-block.Kaz Kylheku2022-06-091-6/+3
| | | | | | | | | | | | | | The root slot is just supposed to be (car bl.list): the first basic block. Unfortunately, it's not maintained when bl.list is edited, which could cause bugs. This change makes no difference in the stdlib compiled files, other than of course the changes in optimize.tlo from this code change itself; so that is evidence suggesting this change is least not making anything worse. * stdlib/optimize.tl (basic-blocks): Remove root slot. (basic-blocks (link-graph, calc-liveness, elim-dead-code): Refer to (car bl.list) instead of bl.root.
* listener: bugfix: handle warnings around linenoise.Kaz Kylheku2022-06-081-0/+4
| | | | | | | | | | | * parser.c (repl): Also push and pop the warning handler around the linenoise call, like we do around evaluation. The reason is that when the library is used in source code form, it can generate warnings while loading. Loading can be triggered by completion from inside linenoise. Neglecting to handle warnings and throw a continue causes a problem because the warnings then fall victim to the listener's master catch. That then interferes with the load.
* string-out-stream: gc issue.Kaz Kylheku2022-06-061-3/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | The problem being addressed here showed up for me when compiling with either gcc-7 or gcc-11 on Ubuntu 18, using -fsanitize-undefined. A few test test cases run under --gc-debug were crashing. I narrowed it down to this small test case, whose correct output would be "1": ./txr --gc-debug -p '(sys:fmt-simple 1 : : : :)' "\x6700ACB0缻\x6700ACB0缻 " The issue doesn't have anything to do with -fsanitize-undefined; that just how it got reproduced by chance. I'm reasonably certain that in builds for which "make tests" passes, and the above test case doesn't repro, this issue is absent: the code got generated in such a way that it the string_own call doesn't cause the stream object to be reclaimed. * stream.c (get_string_from_stream): change the order of operations in the ownership transfer of the string from the stream to the returned string object. We capture the string in a local variable, and null out so->buf before calling string_own. The problem is that get_string_from_stream is called in a way where the caller no longer has any use for the stream, and so the object is no longer live. It's possible for the string_own call to cause the stream object to be garbage collected. Therefore, the object must not still be hanging on to the wchar_t * string, which was already transferred to the string object.
* txr: fix --free-all crash due to atexit order.Kaz Kylheku2022-06-053-14/+15
| | | | | | | | | | | | | | | | | | | | | | | | The --free-all feature is broken. It is needed for confirming lack of memory leaks during development. What breaks it is the atexit call to run_load_hooks in eval.c, which occurs after the atexit call to free_all in txr.c which blows everything away. * Makefile (tst/tests/019/load-hook.ok): Let's put --free-all on this test case, because it has to with the load hooks. Make no mistake though: everything crashes with --free-all, not this test case specifically. * txr.c (opt_free_all): New global variable. (free_all): Free resources only if opt_free_all is true. Lose the paranoid called flag. (main): Register free_all with atexit before callilng init. This ensures that it will be called after any atexit handlers registered as part of init, like the one in sysif.c and eval.c. (txr_main): The --free-all option now just sets opt_free_all. * txr.h (opt_free_all): Declared.
* doc: missing word in tree-insert section.Kaz Kylheku2022-06-011-0/+1
| | | | * txr.1: Add missing "If" at start of sentence.
* build: fix broken build when we don't HAVE_ZLIB.Kaz Kylheku2022-05-312-0/+21
| | | | | | | | | | | * parser.c (open_txr_file): Use liberal heaps of #if HAVE_ZLIB. If there is no Zlib, and the caller explicitly requests a .tlo.gz file to be loaded, then throw. Do not implicitly look for a .tlo.gz file. * stream.c (open_file): #if HAVE_ZLIB around a goto label that is only used out of HAVE_ZLIB code, to eliminate unused label warning.
* streams: use ~a for self string.Kaz Kylheku2022-05-311-3/+3
| | | | | | * stream.c (parse_mode, open_file, open_fileno): Since self is a string, use ~a to print it, or else quotes will appear.
* Version 277.txr-277Kaz Kylheku2022-05-316-1123/+1158
| | | | | | | | | | | | * RELNOTES: Updated. * configure (txr_ver): Bumped version. * stdlib/ver.tl (lib-version): Bumped. * txr.1: Bumped version and date. * txr.vim, tl.vim: Regenerated.
* crc32: name clash with Zlib on Android.Kaz Kylheku2022-05-312-3/+2
| | | | | | | * chksum.c (crc32): Stop-gap solution: make it static. Also, fix wrong self-string. * chksum.h (crc32): Declaration removed.
* windows: bundle Zlib in installer.Kaz Kylheku2022-05-311-0/+1
| | | | * inst.nsi: pull in cygz.dll.
* cygwin: bug: sh always uses cmd.exe.Kaz Kylheku2022-05-312-18/+7
| | | | | | | | | | * stream.c (sh): Use a single definition for this function, which uses the shell and shell_arg variables to use either /bin/sh -c or cmd.exe /c. We only want to use cmd.exe when running as a Windows native program on Cygnal. * tests/018/process.tl: Remove workaround from test case. This is what was causing the weirdness.
* gzio: unused variable warning.Kaz Kylheku2022-05-311-1/+0
| | | | * gzio.c (gzio_set_prop): Remove unused ops variable.
* gzio: fix multiple definition of gzio_stream_s.Kaz Kylheku2022-05-312-1/+3
| | | | | | | | Failed on Cygwin. * gzio.c (gzio_stream_s): Variable defined here. * gzio.h (gzio_stream_s): Turn to declaration with extern.
* buf: compression tests.Kaz Kylheku2022-05-303-3/+22
| | | | | | | | | | * buf.c (buf_compress): Let's use the level value of -1 if not specified, so Zlib defaults it to 6, or whatever. * tests/012/buf.tl: New tests. * txr.1: Note that -1 is a valid level value and that is the default.
* gzio: some tests.Kaz Kylheku2022-05-301-0/+53
| | | | * tests/018/gzip.tl: New file.
* New: spln and tokn functions.Kaz Kylheku2022-05-306-2/+78
| | | | | | | | | | | | | | | Instead of trying to work the new count parameter into the spl and tok functions, it's better to make new ones. * eval.c (eval_init): spln and tokn intrinsics registered. * lib.[ch] (spln, tokn): New functions. * tests/015/split.tl: New test cases. * txr.1: Documented. * stdlib/doc-syms.tl: Updated.
* tok-str: takes count argument.Kaz Kylheku2022-05-285-10/+52
| | | | | | | | | | | | | | * eval.c (eval_init): Update registration of tok-str. * lib.c (tok_str): New argument, count_opt. Implemented in the compat 155 case; what the heck. (tok): Pass nil to new parameter of tok_str. * lib.h (tok_str): Declaration updated. * tests/015/split.tl: New tests. * txr.1: Documented.
* gzio: remove unused variables.Kaz Kylheku2022-05-281-2/+0
| | | | * gzio.c (gzio_stream_print, gzio_get_prop): Remove unused locals.
* buf: compression and decompression functions.Kaz Kylheku2022-05-283-0/+119
| | | | | | | | | * buf.c (buf_compress, buf_decompress): New static functions. (buf_init): buf-compress and buf-decompress intrinsics registered. * txr.1: Documented. * stdlib/doc-syms.tl: Updated.
* buf: remove stray utf8 declaration.Kaz Kylheku2022-05-281-2/+0
| | | | * buf.c (utf8_dup_to_buf): stray declaration removed.
* tags: rename to avoid name clash.Kaz Kylheku2022-05-272-1/+1
| | | | | | | | * tags.tl: File renamed to txrtags.tl. * libtags.txr: load "txrtags" rather than "tags". And here is where we can explain the problem being fixed: (load "tags") loads the actual tags file, if it already exists, and not tags.tl.
* streams: remove workaround for older Cygwin bug.Kaz Kylheku2022-05-271-7/+0
| | | | | | * stream.c (se_putc): Remove a workaround for a Cygwin bug that was fixed in 2016 in 2.5.0. Here is the mailing list thread: https://sourceware.org/pipermail/cygwin/2016-March/226554.html
* doc: .tlo.gz suffix.Kaz Kylheku2022-05-271-9/+36
| | | | | * txr.1: In all places that mention .tlo in relation to loading, mention support for .tlo.gz suffix.
* load: support .tlo.gz files.Kaz Kylheku2022-05-271-9/+38
| | | | | | * parser.c (open_txr_file): If a file ends in .tlo.gz, open a gzio stream on it. Also search for .tlo.gz if .tlo is not found.
* gc: bug: fix compilation if HAVE_VALGRIND.Kaz Kylheku2022-05-271-1/+2
| | | | | * gc.c (mark_obj_maybe): Avoid declaration after statement, which is enforced when configured in --maintainer mode.
* gzio: include type in printed representation of stream.Kaz Kylheku2022-05-271-2/+16
| | | | | | * gzio.c (gzio_stream_print): New static function. (gzio_ops_rd, gzio_ops_wr): Wire in function instead of stream_print_op.
* gzio: bugfix: just return descr for :name property.Kaz Kylheku2022-05-271-1/+1
| | | | | | * gzio.c (gzio_get_prop): Don't play games with a computed :name; this property is relied upon for getting the path associated with an open file.
* doc: document gzip file mode.Kaz Kylheku2022-05-271-1/+44
| | | | | * txr.1: Document the "z" option letter, its optional digit arg, and the restrictions.
* gzio: support :byte-oriented prop.Kaz Kylheku2022-05-271-2/+10
| | | | | | | * gzio.c (gzio_get_prop, gzio_set_prop): Handle :byte-oriented symbol, connecting it to the is_byte_oriented flag. This is already implemented, by copy and paste from the stdio routines.
* gzio: remove real-time-mode flag.Kaz Kylheku2022-05-271-2/+0
| | | | | * gzio.c (struct gzio_handle): Remove is_real_time flag. (make_gzio_stream): Remove initialization of flag.
* I/O convenience functions get mode-opt argument.Kaz Kylheku2022-05-272-56/+136
| | | | | | | | | | | | | | | | | | | | | This is motivated by the desired to be able to specify gzip compression: e.g. (file-put-string "file.gz" "abc" "z") However, it has other uses. For instance with "x", we can exclusively create a file with content in one call. * stdlib/getput.tl (file-get, file-put, file-append, file-get-string, file-put-string, file-append-string, file-put-lines, file-append-lines, file-get-buf, file-put-buf, file-place-buf, file-append-buf, file-get-json, file-put-json, file-append-json, file-get-jsons, file-put-jsons, file-append-jsons): New option argument which combines with the open-file mode. * txr.1: Documented.
* open-fileno: support "z" flag for gzip.Kaz Kylheku2022-05-261-12/+27
| | | | | * stream.c (open_fileno): Use w_gzdopen_mode and make_gzio_stream to make a gzio stream if the gzip flag is present.
* gzio: support more modes in open-file.Kaz Kylheku2022-05-264-3/+49
| | | | | | | | | | | | * gzio.c (w_gzopen_mode): Use w_open_mode if available, in order to support most of the flags (including "x" which Zlib has, but which we are not passing through). * stream.c (w_open_mode): New function formed from w_fopen_mode content. (w_fopen_mode): Call w_open_mode. * stream.c (w_open_mode): Declared.
* First cut at new feature: gzip streams.Kaz Kylheku2022-05-266-9/+631
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * Makefile (OBJS): Conditionally include new gzio.o object file. * configure: Detect external zlib, setting up new have_zlib variable in config.make, HAVE_ZLIB in config.h, and also -lz in conf_ldlibs. * gzio.[ch]: New files, implementing the stream abstraction over the gzip file I/O routines in zlib. * stream.h (struct stdio_mode): New gzip flag and gzlevel bitfield to hold a value 0 to 9. (stdio_mode_init_blank, stdio_mode_init_r, stdio_mode_init_rpb): Update intializers to cover new bitfield members. * stream.c: Include <zlib.h> and "gzio.h" if HAVE_ZLIB. (do_parse_mode): Recognize new mode modifier letter "z", setting the gzip flag in the mode structure. If it's followed by a digit, set the gziplevel to that value. (format_mode): Don't output "b" letter for binary mode if gzip is set, because gzopen interprets "b" differently. Don't put out "t" if gzip is set. If gzip mode is specified, do put out the level. If gzip is set, and gziplevel is nonzero then encode the level: gzopen will understand it. (open_file): If gzip mode is requested, then open the file using gzopen mode, a new function in gzio.c. The return a gzio stream based on the returned gzip file handle. However, if we are reading, and the gzip stream indicates that it's not decompressing anything, then we close it and open the file using an ordinary stream. (stream_init): Call gzio_init if HAVE_ZLIB is true. This is done here because the module is integrated with stream.c, and also so that lib.c doesn't have to know about HAVE_ZLIB and <zlib.h>.
* open-file: show function name in bad modes diagnostic.Kaz Kylheku2022-05-261-1/+2
| | | | | | * stream.c (normalize_mode): Include the self symbol in the error message being thrown. This was not forgotten in the sister function normalize_mode_no_bin.
* stream-set-prop: return t on successful :name set.Kaz Kylheku2022-05-261-0/+1
| | | | | | * stream.c (stdio_set_prop): When setting the :name property, return t, to indicate it was recognized and stored, as is documented.
* streams: avoid double access to errno.Kaz Kylheku2022-05-261-2/+3
| | | | | * stream.c (stdio_maybe_read_error): Access errno once, and refer to the value.
* configure: don't exit on no mmap.Kaz Kylheku2022-05-261-2/+0
| | | | | | * configure: do not treat the failure to detect mmap as a fatal condition. This was just debugging commands accidentally left behind.
* tests: fix failing load-search test.Kaz Kylheku2022-05-261-1/+2
| | | | | | * tests/019/load-search.tl: skip a certain test if it is run as superuser; it fails because superuser is not affected by denied directory search and execute permissions.
* Version 276.txr-276Kaz Kylheku2022-05-247-699/+787
| | | | | | | | | | | | | | * RELNOTES: Updated. * configure (txr_ver): Bumped version. * stdlib/ver.tl (lib-version): Bumped. * txr.1: Bumped version and date. * txr.vim, tl.vim: Regenerated. * protsym.c: Likewise.
* ffi: reproduce odd GNU C behavior for aligned bitfields.Kaz Kylheku2022-05-243-7/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | | We've already taken care of imitating the situation that GNU C allows __attribute__((aligned(n))) to weaken the alignment of a bitfield, contrary to it being documented that align only strengthens alignment. Even a value of n == 1 is meaningful in that it can cause the bitfield to start allocating from a new byte. This patch corrects a newly discovered nuance: when a bitfield is attributed with a weaker alignment than its underlying type (e.g. uint32_t field marked with 2 byte alignment), the original type's alignment is still in effect for calculating the alignment of the structure, and the padding. * ffi.c (struct txr_ffi_type): New member oalign, for keeping track of the type's original alignment, prior to adjustment. (make_ffi_type_struct): For a named bitfield, take the oalign value into account when determining the most strict member alignment. (ffi_type_compile): When marking a type as aligned, the we remember the original alignment in atft->oalign. * tests/017/bitfields.tl: New test case, struct s16. * txr.1: Documented.
* fixup! ffi: couple of tests; assertion.Kaz Kylheku2022-05-241-5/+5
|
* doc: add a note about pack.Kaz Kylheku2022-05-241-0/+34
| | | | | * txr.1: pack with alignment values greater than 1 doesn't directly correspond to a single GNU C feature.
* ffi: couple of tests; assertion.Kaz Kylheku2022-05-242-0/+15
| | | | | | | | | | | | | * ffi.c (make_ffi_type_struct): Add check for impossible condition. The bits_alloc variable could only exceed bits_type (and thus cause the room variable to have a nonsensical, large value) if the bitfield allocation tried to continue allocating bits into an aligned unit, whose alignment exceeds the size of the underlying type. But in that case, tft->aligned would have to be true, and so the offset would have been aligned prior to this code, rendering bits_alloc zero. * tests/017/bitfields.tl: New tests.
* oop: corruption bugfix.Kaz Kylheku2022-05-231-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | * struct.c (alloc_seen): Fix under-sized alloca and memset, where the size being used is the number of elements, but should be the number of bytes. This showed up as a failure in the tests/012/oop-mi.tl on PPC64, uncovered by a completely unrelated change in eval.c. The unrelated code never executes during the failed test case; it just affects some memory contents of the stack somehow. The under-allocated array is used during structure type initialization to hold a bitmask which helps to suppress the duplicate initiation of base structures which are inherited multiple times (duplicate bases). I think why this shows up on PPC64 is the big endian order. The sizes involved are small. The test case has a couple of duplicate bases only, but a full 64 bit mask needs to be allocated for that since the mask is an array of ucnum. On little endian, it doesn't matter than only a single byte is allocated: that is enough for 8 duplicate bases, and because of the byte order, those bases land into the byte that is allocated. On big endian, the bases will be on the high part of the 64 bit word, and thus over in the area we didn't allocate. To reproduce an issue on little endian, you need an inheritance hierarchy with more duplicate bases.
* ffi: bitfield tests pass on PPC64.Kaz Kylheku2022-05-231-102/+53
| | | | | | * ffi.c (swap_get32, swap_put32, swap_get64, swap_put64): correct implementations for operating on little-endian words on big endian.
* ffi: bitfield tests and fixes.Kaz Kylheku2022-05-233-17/+615
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The bitfield allocation rules are wrong. Some of it is due to the recent changes which are based on incorrect analysis, but reverting things doesn't fix it. The idea that we compare the current member's alignment with the previous is wrong; it is not borne out by empirical tests with gcc. So we do a straight revert of that. In GNU C, an __attribute__((aligned (N))) attribute applied to a bitfield member will perform the requested alignment if, evidently, the bit field is already being placed into a new byte. (If the bit field is about to be packed into an existing byte, then there is a warning about the align attribute being ignored). Because we don't have alignment as a member attribute, but only as a type attribute, we must implement a flag which indicates that a type has had align applied to it (even if the alignment didn't change) so we can then honor this in the right place in the bitfield allocation code. * ffi.c (struct txr_ffi_type): New attribute flag, aligned. (make_ffi_type_struct): Remove the prev_align variable and all related logic. Consolidate all alignment into one place, which is done before we allocate the bitfield or regular member. We align if the new member isn't a bitfield, or even if it is a bitfield if it has the aligned attribute, or if the bitfield is changing endian compared to the previous member (our local rule, not from GNU C). (ffi_type_compile): The align and pack operators now set the aligned attribute, except in the (pack 1 ...) case which semantically denotes lack of alignment. * tests/017/bitfields.tl: New file. * txr.1: Documented.
* ffi: allow bitfields based on endian types.Kaz Kylheku2022-05-222-210/+598
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is very nice: you can now declaratively match a structure that uses bitfields, and comes from a different endian system. * ffi.c (ffi_kind_t): Replace FFI_KIND_NUM type with FFI_KIND_INT, FFI_KIND_UINT and FFI_KIND_FLO. Now using the tft->kind value, we can distinguish integer and floating types, and determine signedness. (struct txr_ffi_type): New flag, bigendian, which is set to 1 for all big types that are big endian. This is not just the endian types like be-int32, but natural types like int, if the underlying platform is big endian. (swap_get32, swap_put32, swap_get64, swap_put64): New static functions. (ffi_generic_swap_fat_ubit_put, ffi_generic_swap_fat_sbit_put, ffi_generic_swap_fat_ubit_get, ffi_generic_swap_fat_sbit_get, ffi_generic_swap_ubit_put, ffi_generic_swap_sbit_put, ffi_generic_swap_ubit_get, ffi_generic_swap_sbit_get): New static functions. (ffi_make_type_builtin): On big endian, set the bigendian flag on every type. For the endian types, this will get adjusted as an additional construction step. (make_ffi_type_endian): New static function. Calls ffi_make_type_builtin, and then initializes the bigendian flag with the given value. (make_ffi_type_struct, make_ffi_type_union): Because bitfields can have endiannness now, we can no longer refer to the machine's own endianness in laying them out; we have to look at the mtft->bigendian value. Furthermore, we need an additional rule: when a bitfield member follows a member that has different endian, a new allocation unit has to begin. (ffi_type_compile): the sbit and ubit types must set the type to FFI_KIND_INT and FFI_KIND_UINT now. For the big operator, we can simplify the error checking: instead of exhaustively comparing the type to all the supported integer types, we just now check whether it is FFI_KIND_INT or FFI_KIND_UINT. Here, if we detect that an endian bitfield has opposite byte order from the machine, then we instantiate the bitfield with the ffi_generic_swap_* virtual functions. These perform the required byte swapping accesses to the bitfield storage cells, so then the bit field manipulation code just works using the local integer representation (shifting and masking). Of course, the shift amount depends on the endian; and that is calculated at type creation time in make_ffi_type_struct. (ffi_init_types): Replace FFI_KIND_NUM with the appropriate constant for each affected type. In some cases, we have to calculate whether to use the INT or UINT one, for the types whose signedness is not specified. We switch all the endian types to new constructor make_ffi_type_endian, passing the required value of the bigendian flag. * txr.1: Documented.
* ffi: fat bitfield bugfix for big endian.Kaz Kylheku2022-05-211-2/+12
| | | | | | | * ffi.c (make_ffi_type_struct, make_ffi_type_union): We must take fat bitfields into account when calculating mtft->shift on big endian. We must subtract from bits_llint not bits_int for big endian.
* ffi: floating-point endian types: wrong Lisp type.Kaz Kylheku2022-05-211-4/+4
| | | | | | | | | * ffi.c (ffi_init_types): The be-float, be-double, le-float and le-double types wrongly have integer indicated as their Lisp type symbol; it should be float. This doesn't fix any actual bug; the Lisp type is only important for structures/unions. The conversion of be-double is driven by the behavior of the type's virtual functions, which don't look at the Lisp type symbol.
generated by cgit v1.2.3 (git 2.25.1) at 2025-03-19 15:22:25 +0000