| Age | Commit message (Collapse) | Author |
|---|
|
|
|
On 31/1/2011 Paul Guyot reported a bug in the native code compilation
of inexact equality/inequality tests between floats and integers. The
relevant test was:
f(X) ->
Y = X / 2,
Y == 0.
and hipe erroneously evaluated the calls f(0) and f(0.0) to 'false'.
The culprit was in the simplification code of the Icode range analysis
which used an erroneous test (lists:any/1 instead of lists:all/1).
|
|
|
|
* ks/bs_start_match-fails:
Fix erroneous fail info of a hipe_bs_primop
OTP-9036
|
|
* bjorn/beam-loader/OTP-9030: (43 commits)
c: Reduce memory footprint
erl_posix_msg: Reduce memory footprint
Introduce a few more variations of the move instructions
Combine a move + jump sequence into the move_jump instruction
Optimize and clean-up the exact equality/non-equality instructions
Optimize addition of a small integer to a variable
Introduce a special instruction for select_val with two values
Introduce a few more specialized put_list instructions
Eliminate the "put_list c n Dst" instructions
Eliminate the specific move_sd instruction
Eliminate use of GetArg1() in the badmatch and case_end instructions
Eliminate use of GetArg2() in the i_element instruction
Eliminate use of GetArg1() in the fast_element instruction
Eliminate use of GetArg1() in the jump_on_val* instructions
Eliminate use of GetArg1() in the select_val instruction
beam_emu: Eliminate sloppy use of tmp_arg1 and tmp_arg2
beam_emu: Don't inline helper functions into process_main()
beam_emu: Clean up calling of the error_handler module
Simplify a select_val instruction that selects only one value
Optimize creation of tuples
...
|
|
erts_debug:instructions/0 is useful for finding which specific
instructions that are not used at all.
|
|
|
|
Conflicts:
lib/stdlib/vsn.mk
|
|
|
|
|
|
The introduction of filenames being unicode binaries revealed a problem
in the type analysis of the native code compiler which resulted in an
infinite loop when compiling the 'filename' module.
In addition, the hard-coded type information for the built-in functions
of the 're' module was out-of-date, which resulted in erroneous type
information for 'filelib' functions being stored in the PLT.
|
|
|
|
|
|
The options for the ets:new/2 call has changed recently. There are
read_concurrency hints as well as compressed tables. The hipe/dialyzer
did not take this into account. This patch corrects the problem by
going through the documentation and altering the type specification.
The error fixed is with the dialyzer. When constructing an ETS table
with, e.g., {read_concurrency, true}, the dialyzer will report that the
caller of ets:new/2 will not return.
; First, we update the documentation from the ets man page.
; Second, we reorder the type specification of keypos so they appear
in the same order as in the documentation.
; Finally, we add the missing read_concurrency and compressed Tweak
options.
|
|
The introduction of filenames being unicode binaries revealed a problem
in the type analysis of the native code compiler which resulted in an
infinite loop when compiling the 'filename' module.
In addition, the hard-coded type information for the built-in functions
of the 're' module was out-of-date, which resulted in erroneous type
information for 'filelib' functions being stored in the PLT.
|
|
|
|
* ks/erl_bif_types-cleanup/OTP-8961:
Fix type information of 'file' and 'code' modules
Conflicts:
lib/hipe/cerl/erl_bif_types.erl
|
|
* pan/unicode-filenames/OTP-8887: (27 commits)
Test and correct filelib and filename
Add documentation to erlang.xml and slight correction to unicode_usage.xml
Add section about Unicode file names to stdlib users guide
Correct bug in file_name_SUITE making it fail on Unix instead of Windows7
Add documentation about raw filenames and Unicode file name translation mode
Make filelib not crash on re codepoints beyond 255 in re when filename is raw
Mend on_load_embedded testcase which did not handle windows links
Correct testcase regarding windows versions supporting soft links.
Teach filelib to use re in unicode mode when filenames are not raw
Treat soft links on Windows correctly in file_name_SUITE
Adapt new soft and hard link routines on Windos to Unicode
Corrected testcases broken by unicode filenames
Update preloaded prim_file
Teach prim_file not to accept atoms and not to throw exceptions
Adapt inet_drv to Visual Studio 2008
Teach spawn_executable about Unicode
Convert filenames read on MacOSX to canonical form
Teach file to accept codepoints beyond 255.
Add testcases
Correct shell utilities to handle unicode and possibly binaries
...
|
|
* lukas/hipe/decode_packet_type:
Update httppacket type to return both strings and binaries
|
|
Also corrected type-info for bifs
|
|
Dialyzer for a long time now has had hard-coded type information about
key functions of the 'file' and 'code' modules in 'erl_bif_types'.
Some of this information was not up-to-date according to the published
Erlang/OTP documentation, while some other part contained small errors.
Now that specs are available, this information should be moved to the
corresponding files, not be hard-coded in erl_bif_types.
This change takes out all information for the 'file' module and fixes
some small errors in type information for the 'code' module.
|
|
* ks/file_open-mode:
Add the {encoding, _} options to file:open/2's modes
|
|
* ks/hipe-patches:
hipe: Update types and specs
|
|
|
|
- Fixed pretty rare crash when taking the infimum of two tuple_sets
(initial fix by Kostis Sagonas but then a better fix was provided
by Stavros Aronis and this is the one adopted).
- Fixed pretty rare crash when using parameterized types containing unbound
variables (thanks to Nicolas Trangez for reporting it).
- Fixed pretty rare infinite loop when refining the types of an SCC whose
functions all returned none() (thanks to Stavros Aronis).
|
|
* ks/rec-erl_types:
Deeper unfolding of recursive types
|
|
The change of taking out some hard-coded information about the file module
uncovered that the unfolding of recursive types was probably too shallow.
This has been fixed by unfolding it until a deeper limit (REC_TYPE_LIMIT).
While at erl_types.erl also did some cleanup changes and fixed an erroneous
io:format message.
|
|
|
|
|
|
Although the {encoding, encoding()} set of options is documented in
the manual page of the 'file' module, they do not appear in the Mode
description on that page nor in the mode() type declaration.
The patch adds this information in both the code of the module and
the documentation of the module.
To avoid duplication, the declaration of the encoding() type is added
to the 'unicode' module where it most probably belongs.
While at it, added a proper declaration for posix(), took out the
now superfluous information about the types of file:open/2 from the
erl_bif_types module, and corrected the return type of file:open/2
so that it corresponds to the published documentation.
|
|
* ks/ets-update-counter:
Fix type of ets:update_counter/3 to correspond with the documentation
|
|
|
|
* ks/dialyzer-fixes:
Fix two errors in dialyzer
|
|
Changes which fix some problems reported by dialyzer users:
1. The handling of nested opaque types
2. The handling of remote types in record expressions used as types
While at it, updated dialyzer's RELEASE_NOTES to reflect these fixes
and did some cleanups to dialyzer_utils.erl so that it gets in sync
with dialyzer's main development branch.
|
|
|
|
Resolve name clash with auto-imported BIFs min/2 and max/2.
|
|
|
|
* pg/fix-hipe-load:
Fix hipe:load/1
OTP-8802
|
|
This fixes a problem in the HiPE compiler's BEAM-to-Icode
pass where it can crash due to the BEAM compiler's merging
of identical basic blocks with different exception handling
contexts. This bug has been latent, but is now exposed by
what appears to be more aggressive optimizations in BEAM.
The added comment explains things in more detail.
|
|
hipe:load/1 (and unexported hipe:load/2) just did not work. Calling it
would fail with a badmatch because only the hipe chunk was passed to
do_load/3's third parameter called WholeModule. Since this parameter is
then passed to beam_lib:all_chunks/1 which accepts the whole module as a
binary as well as a path to the beam file, and since a path is exactly
what we have in load/2, the fix consists in letting do_load/3 accept a
path and passing it from load/2.
|
|
elib_malloc is an alternate memory allocator that
is no longer possible to build.
|
|
Merging the three off-heap lists (binaries, funs and externals) into
one list. This reduces memory consumption by two words (pointers) per
ETS object.
|
|
The code handling this BIF did not take into account that
the fun in the first argument that could return []
and therefore the return value of the BIF could also be [].
|
|
|
|
* egil/eprof:
Fix eprof to handle error cases
Dialyzer cleanup for eprof
Fix fprof_SUITE to use new eprof API
Switch pattern API for eprof profile with mfa
Update eprof documentation
Update eprof_SUITE with more basic tests
Rename sorting choices for eprof
Teach eprof to align text output
Add dump functionality to eprof
Add log functionality to eprof
Update eprof tests to reflect new eprof
Add start timestamp to eprof profiling
Teach eprof to use the new breakpoints
OTP-8706 egil/eprof
eprof has been reimplemented with support in the Erlang virtual machine and
is now both faster (i.e. slows down the code being measured less) and
scales much better. In measurements we saw speed-ups compared to the old
eprof ranging from 6 times (for sequential code that only uses one
scheduler/core) up to 84 times (for parallel code that uses 8 cores).
|
|
|
|
|
|
* ks/dialyzer-R14-fixes:
Various changes to dialyzer-related files for R14.
OTP-8699 ks/dialyzer-R14-fixes
Various changes to dialyzer-related files for R14.
- Dialyzer properly supports the new attribute -export_type and checks that
remote types only refer to exported types. A warning is produced if some
files/applications refer to types defined in modules which are neither in
the PLT nor in the analyzed applications.
- Support for detecting data races involving whereis/1 and unregister/1.
- More precise identification of the reason(s) why a record construction
violates the types declared for its fields.
- Fixed bug in the handling of the 'or' guard.
- Better handling of the erlang:element/2 BIF.
- Complete handling of Erlang BIFs.
|
|
Bumped version number and wrote RELEASE_NOTES.
Included all changes described in release notes.
Some spec-related changes to some files in lib/hipe/cerl.
|
|
* bg/bif-types:
Add type information for erlang:garbage_collect_message_area/0
Add type information for erts_debug:* BIFs
Add type information for erlang:port_call/2
Add type information for erlang:display* BIFs
Add type information for the BIFs in the binary module
Introduce and use the t_endian() helper
Add type information for erlang:binary_part/2,3
Add type info for erlang:append/2 and erlang:subtract/2
Add type information for the erlang:adler32* BIFs
Add ppc64 return for erlang:system_info(hipe_architecture)
Test that all BIFs have known types
erl_types: Export is_erl_type/1
|
