xref: /freebsd/contrib/llvm-project/lld/docs/WebAssembly.rst (revision cfd6422a5217410fbd66f7a7a8a64d9d85e61229)
1WebAssembly lld port
2====================
3
4The WebAssembly version of lld takes WebAssembly binaries as inputs and produces
5a WebAssembly binary as its output.  For the most part it tries to mimic the
6behaviour of traditional ELF linkers and specifically the ELF lld port.  Where
7possible the command line flags and the semantics should be the same.
8
9
10Object file format
11------------------
12
13The WebAssembly object file format used by LLVM and LLD is specified as part of
14the WebAssembly tool conventions on linking_.
15
16This is the object format that the llvm will produce when run with the
17``wasm32-unknown-unknown`` target.
18
19Usage
20-----
21
22The WebAssembly version of lld is installed as **wasm-ld**.  It shared many
23common linker flags with **ld.lld** but also includes several
24WebAssembly-specific options:
25
26.. option:: --no-entry
27
28  Don't search for the entry point symbol (by default ``_start``).
29
30.. option:: --export-table
31
32  Export the function table to the environment.
33
34.. option:: --import-table
35
36  Import the function table from the environment.
37
38.. option:: --export-all
39
40  Export all symbols (normally combined with --no-gc-sections)
41
42.. option:: --export-dynamic
43
44  When building an executable, export any non-hidden symbols.  By default only
45  the entry point and any symbols marked as exports (either via the command line
46  or via the `export-name` source attribute) are exported.
47
48.. option:: --global-base=<value>
49
50  Address at which to place global data.
51
52.. option:: --no-merge-data-segments
53
54  Disable merging of data segments.
55
56.. option:: --stack-first
57
58  Place stack at start of linear memory rather than after data.
59
60.. option:: --compress-relocations
61
62  Relocation targets in the code section are 5-bytes wide in order to
63  potentially accommodate the largest LEB128 value.  This option will cause the
64  linker to shrink the code section to remove any padding from the final
65  output.  However because it affects code offset, this option is not
66  compatible with outputting debug information.
67
68.. option:: --allow-undefined
69
70  Allow undefined symbols in linked binary.
71
72.. option:: --import-memory
73
74  Import memory from the environment.
75
76.. option:: --initial-memory=<value>
77
78  Initial size of the linear memory. Default: static data size.
79
80.. option:: --max-memory=<value>
81
82  Maximum size of the linear memory. Default: unlimited.
83
84By default the function table is neither imported nor exported, but defined
85for internal use only.
86
87Behaviour
88---------
89
90In general, where possible, the WebAssembly linker attempts to emulate the
91behaviour of a traditional ELF linker, and in particular the ELF port of lld.
92For more specific details on how this is achieved see the tool conventions on
93linking_.
94
95Function Signatures
96~~~~~~~~~~~~~~~~~~~
97
98One way in which the WebAssembly linker differs from traditional native linkers
99is that function signature checking is strict in WebAssembly.  It is a
100validation error for a module to contain a call site that doesn't agree with
101the target signature.  Even though this is undefined behaviour in C/C++, it is not
102uncommon to find this in real-world C/C++ programs.  For example, a call site in
103one compilation unit which calls a function defined in another compilation
104unit but with too many arguments.
105
106In order not to generate such invalid modules, lld has two modes of handling such
107mismatches: it can simply error-out or it can create stub functions that will
108trap at runtime (functions that contain only an ``unreachable`` instruction)
109and use these stub functions at the otherwise invalid call sites.
110
111The default behaviour is to generate these stub function and to produce
112a warning.  The ``--fatal-warnings`` flag can be used to disable this behaviour
113and error out if mismatched are found.
114
115Exports
116~~~~~~~
117
118When building a shared library any symbols marked as ``visibility=default`` will
119be exported.
120
121When building an executable, only the entry point (``_start``) and symbols with
122the ``WASM_SYMBOL_EXPORTED`` flag are exported by default.  In LLVM the
123``WASM_SYMBOL_EXPORTED`` flag is set by the ``wasm-export-name`` attribute which
124in turn can be set using ``__attribute__((export_name))`` clang attribute.
125
126In addition, symbols can be exported via the linker command line using
127``--export``.
128
129Finally, just like with native ELF linker the ``--export-dynamic`` flag can be
130used to export symbols in the executable which are marked as
131``visibility=default``.
132
133Imports
134~~~~~~~
135
136By default no undefined symbols are allowed in the final binary.  The flag
137``--allow-undefined`` results in a WebAssembly import being defined for each
138undefined symbol.  It is then up to the runtime to provide such symbols.
139
140Alternatively symbols can be marked in the source code as with the
141``import_name`` and/or ``import_module`` clang attributes which signals that
142they are expected to be undefined at static link time.
143
144Garbage Collection
145~~~~~~~~~~~~~~~~~~
146
147Since WebAssembly is designed with size in mind the linker defaults to
148``--gc-sections`` which means that all unused functions and data segments will
149be stripped from the binary.
150
151The symbols which are preserved by default are:
152
153- The entry point (by default ``_start``).
154- Any symbol which is to be exported.
155- Any symbol transitively referenced by the above.
156
157Weak Undefined Functions
158~~~~~~~~~~~~~~~~~~~~~~~~
159
160On native platforms, calls to weak undefined functions end up as calls to the
161null function pointer.  With WebAssembly, direct calls must reference a defined
162function (with the correct signature).  In order to handle this case the linker
163will generate function a stub containing only the ``unreachable`` instruction
164and use this for any direct references to an undefined weak function.
165
166For example a runtime call to a weak undefined function ``foo`` will up trapping
167on ``unreachable`` inside and linker-generated function called
168``undefined:foo``.
169
170Missing features
171----------------
172
173- Merging of data section similar to ``SHF_MERGE`` in the ELF world is not
174  supported.
175- No support for creating shared libraries.  The spec for shared libraries in
176  WebAssembly is still in flux:
177  https://github.com/WebAssembly/tool-conventions/blob/master/DynamicLinking.md
178
179.. _linking: https://github.com/WebAssembly/tool-conventions/blob/master/Linking.md
180