1
@c Copyright 1996-2013 Free Software Foundation, Inc.
2
@c This is part of the GAS manual.
3
@c For copying conditions, see the file as.texinfo.
8
@chapter ARM Dependent Features
12
@node Machine Dependencies
13
@chapter ARM Dependent Features
19
* ARM Options:: Options
21
* ARM Floating Point:: Floating Point
22
* ARM Directives:: ARM Machine Directives
23
* ARM Opcodes:: Opcodes
24
* ARM Mapping Symbols:: Mapping Symbols
25
* ARM Unwinding Tutorial:: Unwinding
30
@cindex ARM options (none)
31
@cindex options for ARM (none)
35
@cindex @code{-mcpu=} command line option, ARM
36
@item -mcpu=@var{processor}[+@var{extension}@dots{}]
37
This option specifies the target processor. The assembler will issue an
38
error message if an attempt is made to assemble an instruction which
39
will not execute on the target processor. The following processor names are
84
@code{fa526} (Faraday FA526 processor),
85
@code{fa626} (Faraday FA626 processor),
104
@code{fa606te} (Faraday FA606TE processor),
105
@code{fa616te} (Faraday FA616TE processor),
106
@code{fa626te} (Faraday FA626TE processor),
107
@code{fmp626} (Faraday FMP626 processor),
108
@code{fa726te} (Faraday FA726TE processor),
130
@code{cortex-m0plus},
131
@code{ep9312} (ARM920 with Cirrus Maverick coprocessor),
132
@code{i80200} (Intel XScale processor)
133
@code{iwmmxt} (Intel(r) XScale processor with Wireless MMX(tm) technology coprocessor)
136
The special name @code{all} may be used to allow the
137
assembler to accept instructions valid for any ARM processor.
139
In addition to the basic instruction set, the assembler can be told to
140
accept various extension mnemonics that extend the processor using the
141
co-processor instruction space. For example, @code{-mcpu=arm920+maverick}
142
is equivalent to specifying @code{-mcpu=ep9312}.
144
Multiple extensions may be specified, separated by a @code{+}. The
145
extensions should be specified in ascending alphabetical order.
147
Some extensions may be restricted to particular architectures; this is
148
documented in the list of extensions below.
150
Extension mnemonics may also be removed from those the assembler accepts.
151
This is done be prepending @code{no} to the option that adds the extension.
152
Extensions that are removed should be listed after all extensions which have
153
been added, again in ascending alphabetical order. For example,
154
@code{-mcpu=ep9312+nomaverick} is equivalent to specifying @code{-mcpu=arm920}.
157
The following extensions are currently supported:
158
@code{crypto} (Cryptography Extensions for v8-A architecture, implies @code{fp+simd}),
159
@code{fp} (Floating Point Extensions for v8-A architecture),
160
@code{idiv} (Integer Divide Extensions for v7-A and v7-R architectures),
164
@code{mp} (Multiprocessing Extensions for v7-A and v7-R architectures),
165
@code{os} (Operating System for v6M architecture),
166
@code{sec} (Security Extensions for v6K and v7-A architectures),
167
@code{simd} (Advanced SIMD Extensions for v8-A architecture, implies @code{fp}),
168
@code{virt} (Virtualization Extensions for v7-A architecture, implies
173
@cindex @code{-march=} command line option, ARM
174
@item -march=@var{architecture}[+@var{extension}@dots{}]
175
This option specifies the target architecture. The assembler will issue
176
an error message if an attempt is made to assemble an instruction which
177
will not execute on the target architecture. The following architecture
178
names are recognized:
210
If both @code{-mcpu} and
211
@code{-march} are specified, the assembler will use
212
the setting for @code{-mcpu}.
214
The architecture option can be extended with the same instruction set
215
extension options as the @code{-mcpu} option.
217
@cindex @code{-mfpu=} command line option, ARM
218
@item -mfpu=@var{floating-point-format}
220
This option specifies the floating point format to assemble for. The
221
assembler will issue an error message if an attempt is made to assemble
222
an instruction which will not execute on the target floating point unit.
223
The following format options are recognized:
243
@code{vfpv3-d16-fp16},
256
@code{neon-fp-armv8},
258
@code{crypto-neon-fp-armv8}.
260
In addition to determining which instructions are assembled, this option
261
also affects the way in which the @code{.double} assembler directive behaves
262
when assembling little-endian code.
264
The default is dependent on the processor selected. For Architecture 5 or
265
later, the default is to assembler for VFP instructions; for earlier
266
architectures the default is to assemble for FPA instructions.
268
@cindex @code{-mthumb} command line option, ARM
270
This option specifies that the assembler should start assembling Thumb
271
instructions; that is, it should behave as though the file starts with a
272
@code{.code 16} directive.
274
@cindex @code{-mthumb-interwork} command line option, ARM
275
@item -mthumb-interwork
276
This option specifies that the output generated by the assembler should
277
be marked as supporting interworking.
279
@cindex @code{-mimplicit-it} command line option, ARM
280
@item -mimplicit-it=never
281
@itemx -mimplicit-it=always
282
@itemx -mimplicit-it=arm
283
@itemx -mimplicit-it=thumb
284
The @code{-mimplicit-it} option controls the behavior of the assembler when
285
conditional instructions are not enclosed in IT blocks.
286
There are four possible behaviors.
287
If @code{never} is specified, such constructs cause a warning in ARM
288
code and an error in Thumb-2 code.
289
If @code{always} is specified, such constructs are accepted in both
290
ARM and Thumb-2 code, where the IT instruction is added implicitly.
291
If @code{arm} is specified, such constructs are accepted in ARM code
292
and cause an error in Thumb-2 code.
293
If @code{thumb} is specified, such constructs cause a warning in ARM
294
code and are accepted in Thumb-2 code. If you omit this option, the
295
behavior is equivalent to @code{-mimplicit-it=arm}.
297
@cindex @code{-mapcs-26} command line option, ARM
298
@cindex @code{-mapcs-32} command line option, ARM
301
These options specify that the output generated by the assembler should
302
be marked as supporting the indicated version of the Arm Procedure.
305
@cindex @code{-matpcs} command line option, ARM
307
This option specifies that the output generated by the assembler should
308
be marked as supporting the Arm/Thumb Procedure Calling Standard. If
309
enabled this option will cause the assembler to create an empty
310
debugging section in the object file called .arm.atpcs. Debuggers can
311
use this to determine the ABI being used by.
313
@cindex @code{-mapcs-float} command line option, ARM
315
This indicates the floating point variant of the APCS should be
316
used. In this variant floating point arguments are passed in FP
317
registers rather than integer registers.
319
@cindex @code{-mapcs-reentrant} command line option, ARM
320
@item -mapcs-reentrant
321
This indicates that the reentrant variant of the APCS should be used.
322
This variant supports position independent code.
324
@cindex @code{-mfloat-abi=} command line option, ARM
325
@item -mfloat-abi=@var{abi}
326
This option specifies that the output generated by the assembler should be
327
marked as using specified floating point ABI.
328
The following values are recognized:
334
@cindex @code{-eabi=} command line option, ARM
335
@item -meabi=@var{ver}
336
This option specifies which EABI version the produced object files should
338
The following values are recognized:
344
@cindex @code{-EB} command line option, ARM
346
This option specifies that the output generated by the assembler should
347
be marked as being encoded for a big-endian processor.
349
@cindex @code{-EL} command line option, ARM
351
This option specifies that the output generated by the assembler should
352
be marked as being encoded for a little-endian processor.
354
@cindex @code{-k} command line option, ARM
355
@cindex PIC code generation for ARM
357
This option specifies that the output of the assembler should be marked
358
as position-independent code (PIC).
360
@cindex @code{--fix-v4bx} command line option, ARM
362
Allow @code{BX} instructions in ARMv4 code. This is intended for use with
363
the linker option of the same name.
365
@cindex @code{-mwarn-deprecated} command line option, ARM
366
@item -mwarn-deprecated
367
@itemx -mno-warn-deprecated
368
Enable or disable warnings about using deprecated options or
369
features. The default is to warn.
377
* ARM-Instruction-Set:: Instruction Set
378
* ARM-Chars:: Special Characters
379
* ARM-Regs:: Register Names
380
* ARM-Relocations:: Relocations
381
* ARM-Neon-Alignment:: NEON Alignment Specifiers
384
@node ARM-Instruction-Set
385
@subsection Instruction Set Syntax
386
Two slightly different syntaxes are support for ARM and THUMB
387
instructions. The default, @code{divided}, uses the old style where
388
ARM and THUMB instructions had their own, separate syntaxes. The new,
389
@code{unified} syntax, which can be selected via the @code{.syntax}
390
directive, and has the following main features:
394
Immediate operands do not require a @code{#} prefix.
397
The @code{IT} instruction may appear, and if it does it is validated
398
against subsequent conditional affixes. In ARM mode it does not
399
generate machine code, in THUMB mode it does.
402
For ARM instructions the conditional affixes always appear at the end
403
of the instruction. For THUMB instructions conditional affixes can be
404
used, but only inside the scope of an @code{IT} instruction.
407
All of the instructions new to the V6T2 architecture (and later) are
408
available. (Only a few such instructions can be written in the
409
@code{divided} syntax).
412
The @code{.N} and @code{.W} suffixes are recognized and honored.
415
All instructions set the flags if and only if they have an @code{s}
420
@subsection Special Characters
422
@cindex line comment character, ARM
423
@cindex ARM line comment character
424
The presence of a @samp{@@} anywhere on a line indicates the start of
425
a comment that extends to the end of that line.
427
If a @samp{#} appears as the first character of a line then the whole
428
line is treated as a comment, but in this case the line could also be
429
a logical line number directive (@pxref{Comments}) or a preprocessor
430
control command (@pxref{Preprocessing}).
432
@cindex line separator, ARM
433
@cindex statement separator, ARM
434
@cindex ARM line separator
435
The @samp{;} character can be used instead of a newline to separate
438
@cindex immediate character, ARM
439
@cindex ARM immediate character
440
Either @samp{#} or @samp{$} can be used to indicate immediate operands.
442
@cindex identifiers, ARM
443
@cindex ARM identifiers
444
*TODO* Explain about /data modifier on symbols.
447
@subsection Register Names
449
@cindex ARM register names
450
@cindex register names, ARM
451
*TODO* Explain about ARM register naming, and the predefined names.
453
@node ARM-Relocations
454
@subsection ARM relocation generation
456
@cindex data relocations, ARM
457
@cindex ARM data relocations
458
Specific data relocations can be generated by putting the relocation name
459
in parentheses after the symbol name. For example:
465
This will generate an @samp{R_ARM_TARGET1} relocation against the symbol
467
The following relocations are supported:
483
For compatibility with older toolchains the assembler also accepts
484
@code{(PLT)} after branch targets. On legacy targets this will
485
generate the deprecated @samp{R_ARM_PLT32} relocation. On EABI
486
targets it will encode either the @samp{R_ARM_CALL} or
487
@samp{R_ARM_JUMP24} relocation, as appropriate.
489
@cindex MOVW and MOVT relocations, ARM
490
Relocations for @samp{MOVW} and @samp{MOVT} instructions can be generated
491
by prefixing the value with @samp{#:lower16:} and @samp{#:upper16}
492
respectively. For example to load the 32-bit address of foo into r0:
495
MOVW r0, #:lower16:foo
496
MOVT r0, #:upper16:foo
499
@node ARM-Neon-Alignment
500
@subsection NEON Alignment Specifiers
502
@cindex alignment for NEON instructions
503
Some NEON load/store instructions allow an optional address
505
The ARM documentation specifies that this is indicated by
506
@samp{@@ @var{align}}. However GAS already interprets
507
the @samp{@@} character as a "line comment" start,
508
so @samp{: @var{align}} is used instead. For example:
511
vld1.8 @{q0@}, [r0, :128]
514
@node ARM Floating Point
515
@section Floating Point
517
@cindex floating point, ARM (@sc{ieee})
518
@cindex ARM floating point (@sc{ieee})
519
The ARM family uses @sc{ieee} floating-point numbers.
522
@section ARM Machine Directives
524
@cindex machine directives, ARM
525
@cindex ARM machine directives
528
@c AAAAAAAAAAAAAAAAAAAAAAAAA
530
@cindex @code{.2byte} directive, ARM
531
@cindex @code{.4byte} directive, ARM
532
@cindex @code{.8byte} directive, ARM
533
@item .2byte @var{expression} [, @var{expression}]*
534
@itemx .4byte @var{expression} [, @var{expression}]*
535
@itemx .8byte @var{expression} [, @var{expression}]*
536
These directives write 2, 4 or 8 byte values to the output section.
538
@cindex @code{.align} directive, ARM
539
@item .align @var{expression} [, @var{expression}]
540
This is the generic @var{.align} directive. For the ARM however if the
541
first argument is zero (ie no alignment is needed) the assembler will
542
behave as if the argument had been 2 (ie pad to the next four byte
543
boundary). This is for compatibility with ARM's own assembler.
545
@cindex @code{.arch} directive, ARM
546
@item .arch @var{name}
547
Select the target architecture. Valid values for @var{name} are the same as
548
for the @option{-march} commandline option.
550
Specifying @code{.arch} clears any previously selected architecture
553
@cindex @code{.arch_extension} directive, ARM
554
@item .arch_extension @var{name}
555
Add or remove an architecture extension to the target architecture. Valid
556
values for @var{name} are the same as those accepted as architectural
557
extensions by the @option{-mcpu} commandline option.
559
@code{.arch_extension} may be used multiple times to add or remove extensions
560
incrementally to the architecture being compiled for.
562
@cindex @code{.arm} directive, ARM
564
This performs the same action as @var{.code 32}.
567
@cindex @code{.pad} directive, ARM
568
@item .pad #@var{count}
569
Generate unwinder annotations for a stack adjustment of @var{count} bytes.
570
A positive value indicates the function prologue allocated stack space by
571
decrementing the stack pointer.
573
@c BBBBBBBBBBBBBBBBBBBBBBBBBB
575
@cindex @code{.bss} directive, ARM
577
This directive switches to the @code{.bss} section.
579
@c CCCCCCCCCCCCCCCCCCCCCCCCCC
581
@cindex @code{.cantunwind} directive, ARM
583
Prevents unwinding through the current function. No personality routine
584
or exception table data is required or permitted.
586
@cindex @code{.code} directive, ARM
587
@item .code @code{[16|32]}
588
This directive selects the instruction set being generated. The value 16
589
selects Thumb, with the value 32 selecting ARM.
591
@cindex @code{.cpu} directive, ARM
592
@item .cpu @var{name}
593
Select the target processor. Valid values for @var{name} are the same as
594
for the @option{-mcpu} commandline option.
596
Specifying @code{.cpu} clears any previously selected architecture
599
@c DDDDDDDDDDDDDDDDDDDDDDDDDD
601
@cindex @code{.dn} and @code{.qn} directives, ARM
602
@item @var{name} .dn @var{register name} [@var{.type}] [[@var{index}]]
603
@itemx @var{name} .qn @var{register name} [@var{.type}] [[@var{index}]]
605
The @code{dn} and @code{qn} directives are used to create typed
606
and/or indexed register aliases for use in Advanced SIMD Extension
607
(Neon) instructions. The former should be used to create aliases
608
of double-precision registers, and the latter to create aliases of
609
quad-precision registers.
611
If these directives are used to create typed aliases, those aliases can
612
be used in Neon instructions instead of writing types after the mnemonic
613
or after each operand. For example:
622
This is equivalent to writing the following:
628
Aliases created using @code{dn} or @code{qn} can be destroyed using
631
@c EEEEEEEEEEEEEEEEEEEEEEEEEE
633
@cindex @code{.eabi_attribute} directive, ARM
634
@item .eabi_attribute @var{tag}, @var{value}
635
Set the EABI object attribute @var{tag} to @var{value}.
637
The @var{tag} is either an attribute number, or one of the following:
638
@code{Tag_CPU_raw_name}, @code{Tag_CPU_name}, @code{Tag_CPU_arch},
639
@code{Tag_CPU_arch_profile}, @code{Tag_ARM_ISA_use},
640
@code{Tag_THUMB_ISA_use}, @code{Tag_FP_arch}, @code{Tag_WMMX_arch},
641
@code{Tag_Advanced_SIMD_arch}, @code{Tag_PCS_config},
642
@code{Tag_ABI_PCS_R9_use}, @code{Tag_ABI_PCS_RW_data},
643
@code{Tag_ABI_PCS_RO_data}, @code{Tag_ABI_PCS_GOT_use},
644
@code{Tag_ABI_PCS_wchar_t}, @code{Tag_ABI_FP_rounding},
645
@code{Tag_ABI_FP_denormal}, @code{Tag_ABI_FP_exceptions},
646
@code{Tag_ABI_FP_user_exceptions}, @code{Tag_ABI_FP_number_model},
647
@code{Tag_ABI_align_needed}, @code{Tag_ABI_align_preserved},
648
@code{Tag_ABI_enum_size}, @code{Tag_ABI_HardFP_use},
649
@code{Tag_ABI_VFP_args}, @code{Tag_ABI_WMMX_args},
650
@code{Tag_ABI_optimization_goals}, @code{Tag_ABI_FP_optimization_goals},
651
@code{Tag_compatibility}, @code{Tag_CPU_unaligned_access},
652
@code{Tag_FP_HP_extension}, @code{Tag_ABI_FP_16bit_format},
653
@code{Tag_MPextension_use}, @code{Tag_DIV_use},
654
@code{Tag_nodefaults}, @code{Tag_also_compatible_with},
655
@code{Tag_conformance}, @code{Tag_T2EE_use},
656
@code{Tag_Virtualization_use}
658
The @var{value} is either a @code{number}, @code{"string"}, or
659
@code{number, "string"} depending on the tag.
661
Note - the following legacy values are also accepted by @var{tag}:
662
@code{Tag_VFP_arch}, @code{Tag_ABI_align8_needed},
663
@code{Tag_ABI_align8_preserved}, @code{Tag_VFP_HP_extension},
665
@cindex @code{.even} directive, ARM
667
This directive aligns to an even-numbered address.
669
@cindex @code{.extend} directive, ARM
670
@cindex @code{.ldouble} directive, ARM
671
@item .extend @var{expression} [, @var{expression}]*
672
@itemx .ldouble @var{expression} [, @var{expression}]*
673
These directives write 12byte long double floating-point values to the
674
output section. These are not compatible with current ARM processors
677
@c FFFFFFFFFFFFFFFFFFFFFFFFFF
680
@cindex @code{.fnend} directive, ARM
682
Marks the end of a function with an unwind table entry. The unwind index
683
table entry is created when this directive is processed.
685
If no personality routine has been specified then standard personality
686
routine 0 or 1 will be used, depending on the number of unwind opcodes
690
@cindex @code{.fnstart} directive, ARM
692
Marks the start of a function with an unwind table entry.
694
@cindex @code{.force_thumb} directive, ARM
696
This directive forces the selection of Thumb instructions, even if the
697
target processor does not support those instructions
699
@cindex @code{.fpu} directive, ARM
700
@item .fpu @var{name}
701
Select the floating-point unit to assemble for. Valid values for @var{name}
702
are the same as for the @option{-mfpu} commandline option.
704
@c GGGGGGGGGGGGGGGGGGGGGGGGGG
705
@c HHHHHHHHHHHHHHHHHHHHHHHHHH
707
@cindex @code{.handlerdata} directive, ARM
709
Marks the end of the current function, and the start of the exception table
710
entry for that function. Anything between this directive and the
711
@code{.fnend} directive will be added to the exception table entry.
713
Must be preceded by a @code{.personality} or @code{.personalityindex}
716
@c IIIIIIIIIIIIIIIIIIIIIIIIII
718
@cindex @code{.inst} directive, ARM
719
@item .inst @var{opcode} [ , @dots{} ]
720
@itemx .inst.n @var{opcode} [ , @dots{} ]
721
@itemx .inst.w @var{opcode} [ , @dots{} ]
722
Generates the instruction corresponding to the numerical value @var{opcode}.
723
@code{.inst.n} and @code{.inst.w} allow the Thumb instruction size to be
724
specified explicitly, overriding the normal encoding rules.
726
@c JJJJJJJJJJJJJJJJJJJJJJJJJJ
727
@c KKKKKKKKKKKKKKKKKKKKKKKKKK
728
@c LLLLLLLLLLLLLLLLLLLLLLLLLL
730
@item .ldouble @var{expression} [, @var{expression}]*
733
@cindex @code{.ltorg} directive, ARM
735
This directive causes the current contents of the literal pool to be
736
dumped into the current section (which is assumed to be the .text
737
section) at the current location (aligned to a word boundary).
738
@code{GAS} maintains a separate literal pool for each section and each
739
sub-section. The @code{.ltorg} directive will only affect the literal
740
pool of the current section and sub-section. At the end of assembly
741
all remaining, un-empty literal pools will automatically be dumped.
743
Note - older versions of @code{GAS} would dump the current literal
744
pool any time a section change occurred. This is no longer done, since
745
it prevents accurate control of the placement of literal pools.
747
@c MMMMMMMMMMMMMMMMMMMMMMMMMM
749
@cindex @code{.movsp} directive, ARM
750
@item .movsp @var{reg} [, #@var{offset}]
751
Tell the unwinder that @var{reg} contains an offset from the current
752
stack pointer. If @var{offset} is not specified then it is assumed to be
755
@c NNNNNNNNNNNNNNNNNNNNNNNNNN
756
@c OOOOOOOOOOOOOOOOOOOOOOOOOO
758
@cindex @code{.object_arch} directive, ARM
759
@item .object_arch @var{name}
760
Override the architecture recorded in the EABI object attribute section.
761
Valid values for @var{name} are the same as for the @code{.arch} directive.
762
Typically this is useful when code uses runtime detection of CPU features.
764
@c PPPPPPPPPPPPPPPPPPPPPPPPPP
766
@cindex @code{.packed} directive, ARM
767
@item .packed @var{expression} [, @var{expression}]*
768
This directive writes 12-byte packed floating-point values to the
769
output section. These are not compatible with current ARM processors
772
@cindex @code{.pad} directive, ARM
773
@item .pad #@var{count}
774
Generate unwinder annotations for a stack adjustment of @var{count} bytes.
775
A positive value indicates the function prologue allocated stack space by
776
decrementing the stack pointer.
778
@cindex @code{.personality} directive, ARM
779
@item .personality @var{name}
780
Sets the personality routine for the current function to @var{name}.
782
@cindex @code{.personalityindex} directive, ARM
783
@item .personalityindex @var{index}
784
Sets the personality routine for the current function to the EABI standard
785
routine number @var{index}
787
@cindex @code{.pool} directive, ARM
789
This is a synonym for .ltorg.
791
@c QQQQQQQQQQQQQQQQQQQQQQQQQQ
792
@c RRRRRRRRRRRRRRRRRRRRRRRRRR
794
@cindex @code{.req} directive, ARM
795
@item @var{name} .req @var{register name}
796
This creates an alias for @var{register name} called @var{name}. For
803
@c SSSSSSSSSSSSSSSSSSSSSSSSSS
806
@cindex @code{.save} directive, ARM
807
@item .save @var{reglist}
808
Generate unwinder annotations to restore the registers in @var{reglist}.
809
The format of @var{reglist} is the same as the corresponding store-multiple
813
@exdent @emph{core registers}
814
.save @{r4, r5, r6, lr@}
815
stmfd sp!, @{r4, r5, r6, lr@}
816
@exdent @emph{FPA registers}
819
@exdent @emph{VFP registers}
820
.save @{d8, d9, d10@}
821
fstmdx sp!, @{d8, d9, d10@}
822
@exdent @emph{iWMMXt registers}
824
wstrd wr11, [sp, #-8]!
825
wstrd wr10, [sp, #-8]!
828
wstrd wr11, [sp, #-8]!
830
wstrd wr10, [sp, #-8]!
834
@cindex @code{.setfp} directive, ARM
835
@item .setfp @var{fpreg}, @var{spreg} [, #@var{offset}]
836
Make all unwinder annotations relative to a frame pointer. Without this
837
the unwinder will use offsets from the stack pointer.
839
The syntax of this directive is the same as the @code{add} or @code{mov}
840
instruction used to set the frame pointer. @var{spreg} must be either
841
@code{sp} or mentioned in a previous @code{.movsp} directive.
851
@cindex @code{.secrel32} directive, ARM
852
@item .secrel32 @var{expression} [, @var{expression}]*
853
This directive emits relocations that evaluate to the section-relative
854
offset of each expression's symbol. This directive is only supported
857
@cindex @code{.syntax} directive, ARM
858
@item .syntax [@code{unified} | @code{divided}]
859
This directive sets the Instruction Set Syntax as described in the
860
@ref{ARM-Instruction-Set} section.
862
@c TTTTTTTTTTTTTTTTTTTTTTTTTT
864
@cindex @code{.thumb} directive, ARM
866
This performs the same action as @var{.code 16}.
868
@cindex @code{.thumb_func} directive, ARM
870
This directive specifies that the following symbol is the name of a
871
Thumb encoded function. This information is necessary in order to allow
872
the assembler and linker to generate correct code for interworking
873
between Arm and Thumb instructions and should be used even if
874
interworking is not going to be performed. The presence of this
875
directive also implies @code{.thumb}
877
This directive is not neccessary when generating EABI objects. On these
878
targets the encoding is implicit when generating Thumb code.
880
@cindex @code{.thumb_set} directive, ARM
882
This performs the equivalent of a @code{.set} directive in that it
883
creates a symbol which is an alias for another symbol (possibly not yet
884
defined). This directive also has the added property in that it marks
885
the aliased symbol as being a thumb function entry point, in the same
886
way that the @code{.thumb_func} directive does.
888
@cindex @code{.tlsdescseq} directive, ARM
889
@item .tlsdescseq @var{tls-variable}
890
This directive is used to annotate parts of an inlined TLS descriptor
891
trampoline. Normally the trampoline is provided by the linker, and
892
this directive is not needed.
894
@c UUUUUUUUUUUUUUUUUUUUUUUUUU
896
@cindex @code{.unreq} directive, ARM
897
@item .unreq @var{alias-name}
898
This undefines a register alias which was previously defined using the
899
@code{req}, @code{dn} or @code{qn} directives. For example:
906
An error occurs if the name is undefined. Note - this pseudo op can
907
be used to delete builtin in register name aliases (eg 'r0'). This
908
should only be done if it is really necessary.
910
@cindex @code{.unwind_raw} directive, ARM
911
@item .unwind_raw @var{offset}, @var{byte1}, @dots{}
912
Insert one of more arbitary unwind opcode bytes, which are known to adjust
913
the stack pointer by @var{offset} bytes.
915
For example @code{.unwind_raw 4, 0xb1, 0x01} is equivalent to
918
@c VVVVVVVVVVVVVVVVVVVVVVVVVV
920
@cindex @code{.vsave} directive, ARM
921
@item .vsave @var{vfp-reglist}
922
Generate unwinder annotations to restore the VFP registers in @var{vfp-reglist}
923
using FLDMD. Also works for VFPv3 registers
924
that are to be restored using VLDM.
925
The format of @var{vfp-reglist} is the same as the corresponding store-multiple
929
@exdent @emph{VFP registers}
930
.vsave @{d8, d9, d10@}
931
fstmdd sp!, @{d8, d9, d10@}
932
@exdent @emph{VFPv3 registers}
933
.vsave @{d15, d16, d17@}
934
vstm sp!, @{d15, d16, d17@}
937
Since FLDMX and FSTMX are now deprecated, this directive should be
938
used in favour of @code{.save} for saving VFP registers for ARMv6 and above.
940
@c WWWWWWWWWWWWWWWWWWWWWWWWWW
941
@c XXXXXXXXXXXXXXXXXXXXXXXXXX
942
@c YYYYYYYYYYYYYYYYYYYYYYYYYY
943
@c ZZZZZZZZZZZZZZZZZZZZZZZZZZ
951
@cindex opcodes for ARM
952
@code{@value{AS}} implements all the standard ARM opcodes. It also
953
implements several pseudo opcodes, including several synthetic load
958
@cindex @code{NOP} pseudo op, ARM
964
This pseudo op will always evaluate to a legal ARM instruction that does
965
nothing. Currently it will evaluate to MOV r0, r0.
967
@cindex @code{LDR reg,=<label>} pseudo op, ARM
970
ldr <register> , = <expression>
973
If expression evaluates to a numeric constant then a MOV or MVN
974
instruction will be used in place of the LDR instruction, if the
975
constant can be generated by either of these instructions. Otherwise
976
the constant will be placed into the nearest literal pool (if it not
977
already there) and a PC relative LDR instruction will be generated.
979
@cindex @code{ADR reg,<label>} pseudo op, ARM
982
adr <register> <label>
985
This instruction will load the address of @var{label} into the indicated
986
register. The instruction will evaluate to a PC relative ADD or SUB
987
instruction depending upon where the label is located. If the label is
988
out of range, or if it is not defined in the same file (and section) as
989
the ADR instruction, then an error will be generated. This instruction
990
will not make use of the literal pool.
992
@cindex @code{ADRL reg,<label>} pseudo op, ARM
995
adrl <register> <label>
998
This instruction will load the address of @var{label} into the indicated
999
register. The instruction will evaluate to one or two PC relative ADD
1000
or SUB instructions depending upon where the label is located. If a
1001
second instruction is not needed a NOP instruction will be generated in
1002
its place, so that this instruction is always 8 bytes long.
1004
If the label is out of range, or if it is not defined in the same file
1005
(and section) as the ADRL instruction, then an error will be generated.
1006
This instruction will not make use of the literal pool.
1010
For information on the ARM or Thumb instruction sets, see @cite{ARM
1011
Software Development Toolkit Reference Manual}, Advanced RISC Machines
1014
@node ARM Mapping Symbols
1015
@section Mapping Symbols
1017
The ARM ELF specification requires that special symbols be inserted
1018
into object files to mark certain features:
1024
At the start of a region of code containing ARM instructions.
1028
At the start of a region of code containing THUMB instructions.
1032
At the start of a region of data.
1036
The assembler will automatically insert these symbols for you - there
1037
is no need to code them yourself. Support for tagging symbols ($b,
1038
$f, $p and $m) which is also mentioned in the current ARM ELF
1039
specification is not implemented. This is because they have been
1040
dropped from the new EABI and so tools cannot rely upon their
1043
@node ARM Unwinding Tutorial
1046
The ABI for the ARM Architecture specifies a standard format for
1047
exception unwind information. This information is used when an
1048
exception is thrown to determine where control should be transferred.
1049
In particular, the unwind information is used to determine which
1050
function called the function that threw the exception, and which
1051
function called that one, and so forth. This information is also used
1052
to restore the values of callee-saved registers in the function
1053
catching the exception.
1055
If you are writing functions in assembly code, and those functions
1056
call other functions that throw exceptions, you must use assembly
1057
pseudo ops to ensure that appropriate exception unwind information is
1058
generated. Otherwise, if one of the functions called by your assembly
1059
code throws an exception, the run-time library will be unable to
1060
unwind the stack through your assembly code and your program will not
1063
To illustrate the use of these pseudo ops, we will examine the code
1064
that G++ generates for the following C++ input:
1067
void callee (int *);
1078
This example does not show how to throw or catch an exception from
1079
assembly code. That is a much more complex operation and should
1080
always be done in a high-level language, such as C++, that directly
1081
supports exceptions.
1083
The code generated by one particular version of G++ when compiling the
1090
@ Function supports interworking.
1091
@ args = 0, pretend = 0, frame = 8
1092
@ frame_needed = 1, uses_anonymous_args = 0
1114
Of course, the sequence of instructions varies based on the options
1115
you pass to GCC and on the version of GCC in use. The exact
1116
instructions are not important since we are focusing on the pseudo ops
1117
that are used to generate unwind information.
1119
An important assumption made by the unwinder is that the stack frame
1120
does not change during the body of the function. In particular, since
1121
we assume that the assembly code does not itself throw an exception,
1122
the only point where an exception can be thrown is from a call, such
1123
as the @code{bl} instruction above. At each call site, the same saved
1124
registers (including @code{lr}, which indicates the return address)
1125
must be located in the same locations relative to the frame pointer.
1127
The @code{.fnstart} (@pxref{arm_fnstart,,.fnstart pseudo op}) pseudo
1128
op appears immediately before the first instruction of the function
1129
while the @code{.fnend} (@pxref{arm_fnend,,.fnend pseudo op}) pseudo
1130
op appears immediately after the last instruction of the function.
1131
These pseudo ops specify the range of the function.
1133
Only the order of the other pseudos ops (e.g., @code{.setfp} or
1134
@code{.pad}) matters; their exact locations are irrelevant. In the
1135
example above, the compiler emits the pseudo ops with particular
1136
instructions. That makes it easier to understand the code, but it is
1137
not required for correctness. It would work just as well to emit all
1138
of the pseudo ops other than @code{.fnend} in the same order, but
1139
immediately after @code{.fnstart}.
1141
The @code{.save} (@pxref{arm_save,,.save pseudo op}) pseudo op
1142
indicates registers that have been saved to the stack so that they can
1143
be restored before the function returns. The argument to the
1144
@code{.save} pseudo op is a list of registers to save. If a register
1145
is ``callee-saved'' (as specified by the ABI) and is modified by the
1146
function you are writing, then your code must save the value before it
1147
is modified and restore the original value before the function
1148
returns. If an exception is thrown, the run-time library restores the
1149
values of these registers from their locations on the stack before
1150
returning control to the exception handler. (Of course, if an
1151
exception is not thrown, the function that contains the @code{.save}
1152
pseudo op restores these registers in the function epilogue, as is
1153
done with the @code{ldmfd} instruction above.)
1155
You do not have to save callee-saved registers at the very beginning
1156
of the function and you do not need to use the @code{.save} pseudo op
1157
immediately following the point at which the registers are saved.
1158
However, if you modify a callee-saved register, you must save it on
1159
the stack before modifying it and before calling any functions which
1160
might throw an exception. And, you must use the @code{.save} pseudo
1161
op to indicate that you have done so.
1163
The @code{.pad} (@pxref{arm_pad,,.pad}) pseudo op indicates a
1164
modification of the stack pointer that does not save any registers.
1165
The argument is the number of bytes (in decimal) that are subtracted
1166
from the stack pointer. (On ARM CPUs, the stack grows downwards, so
1167
subtracting from the stack pointer increases the size of the stack.)
1169
The @code{.setfp} (@pxref{arm_setfp,,.setfp pseudo op}) pseudo op
1170
indicates the register that contains the frame pointer. The first
1171
argument is the register that is set, which is typically @code{fp}.
1172
The second argument indicates the register from which the frame
1173
pointer takes its value. The third argument, if present, is the value
1174
(in decimal) added to the register specified by the second argument to
1175
compute the value of the frame pointer. You should not modify the
1176
frame pointer in the body of the function.
1178
If you do not use a frame pointer, then you should not use the
1179
@code{.setfp} pseudo op. If you do not use a frame pointer, then you
1180
should avoid modifying the stack pointer outside of the function
1181
prologue. Otherwise, the run-time library will be unable to find
1182
saved registers when it is unwinding the stack.
1184
The pseudo ops described above are sufficient for writing assembly
1185
code that calls functions which may throw exceptions. If you need to
1186
know more about the object-file format used to represent unwind
1187
information, you may consult the @cite{Exception Handling ABI for the
1188
ARM Architecture} available from @uref{http://infocenter.arm.com}.