pound-emu/archived-ballistic
1
0
Fork 0
mirror of https://github.com/pound-emu/ballistic.git synced 2026-01-31 01:15:21 +01:00
Code Issues Packages Projects Releases Wiki Activity
Files
main
archived-ballistic/spec/arm64_xml/iform-p.dtd
Ronald Caesar 26a677f8b4 decoder: Add ARM specification docs 
Signed-off-by: Ronald Caesar <github43132@proton.me>
2025-12-12 18:11:36 -04:00

890 lines
30 KiB
DTD
Raw Permalink Blame History

<!--
Instruction Form (iform) Description XML language
Copyright (c) 2010-2022 Arm Limited or its affiliates. All rights reserved.
This document is Non-Confidential.
-->
<!--
Docvars: many elements have a child element 'docvars' enclosing a
list of 'docvar', each defining a key and value pairs (perhaps
implying instr-class=advsimd and isa=a64).
They provide information defined by the source material
authors (perhaps architects defining instruction encodings and
aliases in spreadsheet pages). This information is intended to be
helpful for those filtering or grouping instructions: they
frequently correlate with encoding or assembler template, but use
those more 'precise' things when precision is required.
In many cases larger containing objects have docvars which
contains whatever key/value pairs are common to all descendant
encodings.
Values are not free text: they are symbolic. However, somewhere
there will be a 'docvars.txt' (or similar name) file, and that also
defines usable-in-documentation phrases associated with each
symbolic value.
The same sort of job used to be done with tags:
Each tag was a word or short phrase expressing something about the
object concerned.
-->
<!ELEMENT docvars (docvar+)>
<!ELEMENT docvar EMPTY>
<!ATTLIST docvar key CDATA #REQUIRED value CDATA #REQUIRED>
<!ENTITY % simpleinline "#PCDATA">
<!ENTITY % inline "#PCDATA | a | anchor | txt">
<!--
Formatted text is a block of AML formatted text.
-->
<!ENTITY % formatted_words "asm-code|instruction|literal|xref|b|arm-defined-word|parameter|sup|sub|binarynumber|hexnumber|syntax|field|value|function|enum|enumvalue|url">
<!--
Formatted text is a block of AML formatted text.
-->
<!ENTITY % formatted_text
"(#PCDATA|para|note|list|%formatted_words;)*">
<!--
<para>: A paragraph of text.
A paragraph is a line of text that may or may not be
formatted. The <para> tag itself is *not* generated as a result of
AML. The generator should use this tag when emitting multiple
paragraphs.
-->
<!ELEMENT para (#PCDATA|%formatted_words;|image)*>
<!--
<para>: Non-architectural information.
A note contains non-architectural explanatory content.
-->
<!ELEMENT note (para|list|%formatted_words;)*>
<!--
<image>: an image
-->
<!ELEMENT image EMPTY>
<!ATTLIST image file CDATA #REQUIRED label CDATA #REQUIRED>
<!--
This is the list of AML constructs that have been introduced in the A32/T32
doc files. Expect them to change until we stabilize which ones we actually want.
-->
<!ELEMENT instruction (#PCDATA)*>
<!ATTLIST instruction param0 CDATA #IMPLIED>
<!ELEMENT list (listitem+)>
<!ATTLIST list
type (ordered|unordered|var|param) #REQUIRED
role (standard|compressed|break|wide) #IMPLIED
language CDATA #IMPLIED>
<!ELEMENT listitem ((term*|param*), content)>
<!ELEMENT term (#PCDATA|%formatted_words;)*>
<!ELEMENT param (#PCDATA|%formatted_words;)*>
<!ELEMENT content (#PCDATA|list|para|%formatted_words;)*>
<!ELEMENT arm-defined-word (#PCDATA)*>
<!ELEMENT asm-code (#PCDATA)*>
<!ELEMENT code (#PCDATA)*>
<!ELEMENT sup (#PCDATA)*>
<!ELEMENT sub (#PCDATA)*>
<!ELEMENT binarynumber (#PCDATA)*>
<!ELEMENT hexnumber (#PCDATA)*>
<!ELEMENT syntax (#PCDATA)*>
<!ELEMENT field (#PCDATA)*>
<!ELEMENT value (#PCDATA)*>
<!ELEMENT parameter (#PCDATA)*> <!-- TO BE REMOVED -->
<!ELEMENT literal (#PCDATA)*> <!-- TO BE REMOVED -->
<!ELEMENT function (#PCDATA)*>
<!ELEMENT enum (#PCDATA)*>
<!ELEMENT enumvalue (#PCDATA)*>
<!ELEMENT b (#PCDATA)*> <!-- TO BE REMOVED -->
<!ELEMENT xref (#PCDATA)*>
<!ATTLIST xref linkend CDATA #REQUIRED>
<!ELEMENT url (#PCDATA)*>
<!--
<textsection>: a page of formatted text.
This is simply a formatted page of text.
-->
<!ELEMENT textsection (para|list)*>
<!ATTLIST textsection title CDATA #REQUIRED>
<!--
<instructionsection>: the whole instruction page.
Pseudocode is optional because it isn't typically present on
alias pages.
Attributes:
id: an ID for the collection in this file, unique across the
files.
title: an overall title, used as the top-title for a web page.
-->
<!-- There can acually be 0, 1 or 2 ps_section elements inside the
<instructionsection>, but that can't be represented here.
ps_section elements may also have constrained_unpredictables
after them. -->
<!ELEMENT instructionsection (docvars, heading, desc, affected_by_sme*, operationalnotes?,
(alias_list | aliasto)?,
classes, aliasmnem*, explanations?,
aliastablehook?,
(ps_section | constrained_unpredictables)*, exceptions?)>
<!ATTLIST instructionsection id CDATA #REQUIRED
title CDATA #REQUIRED
tags CDATA #IMPLIED
type (alias|instruction) #IMPLIED>
<!--
<a>: generally a hotlink (<anchor> used for definitions).
Attributes:
href: HTML-style in-file reference for links between items on the
page.
link: in-file reference for other files, used in conjunction with
file: file part of a reference. You can form an HTML reference
from @file#@link
hover: recommended hover text, where implemented (like HTML
'title'). A very short description of the thing linked to.
-->
<!ELEMENT a (%simpleinline;)*>
<!ATTLIST a href CDATA #IMPLIED
link CDATA #IMPLIED file CDATA #IMPLIED
hover CDATA #IMPLIED>
<!--
<anchor>: like HTML <a name="xx">, a target for a hotlink.
Attributes:
link: the hotlink ID.
file: obsolete, always empty. Deprecated
hover: hover text. Filled in for pseudocode definitions,
where it might reasonably be suggesting the text to be used
by linkers, though this is not in fact used.
-->
<!ELEMENT anchor (#PCDATA)>
<!ATTLIST anchor link CDATA #REQUIRED
file CDATA #IMPLIED
hover CDATA #IMPLIED>
<!--
<heading>: title for top of file and as used in alphabetic indexes
-->
<!ELEMENT heading (%inline;)*>
<!--
<desc>: descriptive text for the top of the instruction page
Comes in three parts, and only 'brief' is required.
-->
<!ELEMENT desc (brief, authored?, description?, encodingnotes?, syntaxnotes?, alg*, (longer, alg*)?, status?, predicated?, uses_dit?, takes_movprfx?, takes_pred_movprfx?, is_wide_zm?, affected_by_sme?)>
<!--
<brief>: wraps first sentence summary of the instructions on this
page. Suitable for use in an index.
Attributes are related to the generation of this data:
synth = 'single' this text might have been the result of a merge,
but in this case there was only one choice of input text, so
it's reliable
checked = 'yes': autogenerated as 'checked="no"', should be
cleaned by a human editor.
enclist: rarely conserved, but legal. A comma-separated list of
the names of encodings whose descriptions were originally merged
to create this description.
-->
<!ELEMENT brief %formatted_text;>
<!ATTLIST brief synth (single) #IMPLIED
checked (yes) #IMPLIED
enclist CDATA #IMPLIED>
<!ELEMENT description (#PCDATA|list|para|%formatted_words;)*>
<!ELEMENT status (#PCDATA)>
<!ELEMENT predicated (#PCDATA)>
<!--
<uses_dit>: instruction timing is data-insensitive if PSTATE.DIT is 1.
Attributes:
condition: when present, instruction timing is only data-independent
if PSTATE.DIT is 1 AND this condition is satisfied.
-->
<!ELEMENT uses_dit (#PCDATA)>
<!ATTLIST uses_dit condition CDATA #IMPLIED>
<!ELEMENT takes_movprfx (#PCDATA)>
<!ELEMENT takes_pred_movprfx (#PCDATA)>
<!ELEMENT is_wide_zm (#PCDATA)>
<!--
<alg>: algebraic description of instruction
Attributes:
howmany: counts the number of <alg> elements, if
there are more than one.
-->
<!ELEMENT alg (#PCDATA)>
<!ATTLIST alg howmany CDATA #IMPLIED>
<!--
<longer>: more description text than <brief>, where required
-->
<!ELEMENT longer %formatted_text;>
<!--
<authored>:
-->
<!ELEMENT authored %formatted_text;>
<!--
<encodingnotes>:
-->
<!ELEMENT encodingnotes %formatted_text;>
<!--
<affected_by_sme>: are subsequent instructions which depend on this
instruction's output delayed by FEAT_SME and Streaming
mode? Absent if no.
Attributes:
output: the output of this instruction.
-->
<!ELEMENT affected_by_sme EMPTY>
<!ATTLIST affected_by_sme output CDATA #REQUIRED>
<!--
<operationalnotes>:
-->
<!ELEMENT operationalnotes %formatted_text;>
<!--
<syntaxnotes>:
-->
<!ELEMENT syntaxnotes %formatted_text;>
<!-- Alias pages and their cross references -->
<!--
<alias_list>: list of aliases made from this instruction.
Present even if empty.
Attributes:
howmany: counts the number of <aliasref> elements, zero if none.
-->
<!ELEMENT alias_list (alias_list_intro?, aliasref*, alias_list_outro?)>
<!ATTLIST alias_list howmany CDATA #REQUIRED>
<!--
<alias_list_intro>: introductory text to a paragraph listing dependent
aliases.
-->
<!ELEMENT alias_list_intro (#PCDATA)>
<!--
<aliasref>: a dependent alias, whose contents are the title of the
alias page.
Attributes:
aliaspageid: the unique ID of the alias' page.
aliasfile: the filename, relative to this one, of the alias' file.
hover: the description of the alias' page (as it might appear in
an index), a useful source for browser hover text.
punct: (purely for constructing list text), what to append to the
text when forming a list, usually a comma or full-stop.
-->
<!ELEMENT aliasref (text, aliaspref+)>
<!ATTLIST aliasref aliaspageid CDATA #REQUIRED
aliasfile CDATA #REQUIRED
hover CDATA #IMPLIED
punct CDATA #REQUIRED>
<!--
<aliaspref>: this element contains the conditions under which this alias
is the preferred disassembly.
Attributes:
labels: textual label mapping to the encoding this alias applies to.
-->
<!ELEMENT aliaspref (%inline;)*>
<!ATTLIST aliaspref labels CDATA #IMPLIED>
<!--
<alias_list_outro>: additional text that appears following the list
of aliases that appears after the <alias_list_intro>.
-->
<!ELEMENT alias_list_outro (#PCDATA | text | aliastablelink)*>
<!--
<aliastablelink>: this element specifies where to insert a link to the
alias table.
-->
<!ELEMENT aliastablelink EMPTY>
<!--
<aliasto>: this element appears on a page describing one or more
aliases, and references the instruction page which
contains the instructions to which these are aliases.
Attributes:
refiform: the file name (relative to this one) of the instruction
iform page
iformid: the unique ID of the instruction iform page
-->
<!ELEMENT aliasto (#PCDATA)>
<!ATTLIST aliasto refiform CDATA #REQUIRED
iformid CDATA #REQUIRED>
<!--
<asmtemplate>: assembler template line
Attributes:
role: distinguishes asmtemplate lines provided for aliases (both
as a prototype, and as a description of what the alias maps to).
-->
<!ELEMENT asmtemplate (%inline; | text)*>
<!ATTLIST asmtemplate role (alias_prototype | alias_equivalent_to) #IMPLIED
comment CDATA #IMPLIED>
<!--
<aliasmnem>: alias section relating to one mnemonic which aliases to
instructions on this page.
Attributes:
mnemonic: the instruction mnemonic concerned.
id: a unique ID for this section across all the files in the 'site'
heading: text distinguishing this from other alias sections on
this page (so it is usable as a heading to the aliast description).
iformqual: any parenthetical part of the instruction page header,
which might be a good way to distinguish the aliases here from others.
-->
<!ELEMENT aliasmnem (desc?, (docvars?, (alias+ | aliases+)))>
<!ATTLIST aliasmnem mnemonic CDATA #REQUIRED
id CDATA #REQUIRED
iformqual CDATA #IMPLIED
heading CDATA #IMPLIED>
<!--
<aliases>: wraps around a possible list of different aliases,
alternative ways to write an instruction which corresponds to an
encoding on this page.
Attributes:
conditions: where the interpretation of an alias depends on
parameter values, this explains when this particular
interpretation applies. "conditions" also appears in each
alias, but use of the value from here is preferred.
-->
<!ELEMENT aliases (alias+)>
<!ATTLIST aliases conditions CDATA #IMPLIED>
<!--
<alias>: an alias is an instruction which expands to another (probably
more complex) one. They're expressed as a couple of assembler
templates: how you write it, and how it comes out.
Attributes:
encname: the unique encoding name of the related encoding.
assembler_prototype, equivalent_to: plain text of the alias
prototype and the assembler version of the equivalent
instruction. If you want plain text, these versions are
preferred: the text in the content of the following
<asmtemplate>s should be identical (but is littered with
cross-reference and decoration tags).
enctag: tag list from relevant encoding (no obsolete: see "tags"
comment above)
conditions: repeats information from 'aliases' for backward
compatibility reasons. Please use the value from the parent
'aliases' element.
description: a brief description of this particular form of
alias. These descriptions are merged into the <desc>
description components at the top of the <aliasmnem>.
-->
<!ELEMENT alias (docvars?, asmtemplate, asmtemplate, conditionexpr?)>
<!ATTLIST alias encname CDATA #REQUIRED
equivalent_to CDATA #REQUIRED
assembler_prototype CDATA #REQUIRED
enctag CDATA #IMPLIED
conditions CDATA #IMPLIED
description CDATA #IMPLIED>
<!ELEMENT conditionexpr (pstext)>
<!--
<c>: one element of a <box>
Attributes:
colspan: how many bit-positions this element takes up. It's
omitted when it just occupies one bit position.
-->
<!ELEMENT c (#PCDATA)>
<!ATTLIST c colspan CDATA #IMPLIED>
<!--
<box>: defines an instruction bitfield
If the bitfield defines a binary value, it will be recorded
bit-by-bit in the content of <c> child elements.
Attributes:
name: optional field name
hibit: highest bit number
width: field width in bits, can be omitted if it's just one bit.
usename: is a hint that a field name ought to be
displayed, even if a bit-value is also specified.
settings: integer attribute counts the number of bit-values specified,
so also tells you whether the bit-values need to be displayed.
psbits: contains any bit-setting relevant to the
pseudocode encoding (named by <regdiagram psname="xxx"...>, but
which has been overwritten to help present this page's
instruction encodings. If present, it will most often just be
'x's, indicating that the field wasn't set at all by the
'psname'd encoding. But it could be a multibit setting ("100x")
or even a not-equal-to condition "!= 1x1".
constraint: a complex condition (typically !=).
-->
<!ELEMENT box (c*)>
<!ATTLIST box hibit CDATA #REQUIRED
name CDATA #IMPLIED
width CDATA #IMPLIED
usename (1) #IMPLIED
settings CDATA #IMPLIED
psbits CDATA #IMPLIED
constraint CDATA #IMPLIED
comp CDATA #IMPLIED>
<!--
<regdiagram>: describes the field breakdown of an encoding,
as a series of boxes in high-first, left-to-right order. The
regdiagram found on a page corresponds to the encoding node
associated with the operation pseudocode. A particular
encoding's bit-encoding is defined by both the <regdiagram> and
the encoding's own <box>es
Attributes:
form: selects 16-bit, 2 x 16-bit or 32-bit organisation
tworows: alerts you if any box has both a name and bit-values,
which are likely to have to be rendered in two rows.
encname: obsolete field for compatibility only.
psname: the unique name of the associated pseudocode. Always
present (marked as '#IMPLIED' for compatibility reasons).
constraint: any additional constraints that have not been
described in the diagram.
-->
<!ELEMENT regdiagram (box*)>
<!ATTLIST regdiagram form (16 | 16x2 | 32) #REQUIRED
tworows (1) #IMPLIED
encname CDATA #IMPLIED
psname CDATA #IMPLIED
constraint CDATA #IMPLIED>
<!--
<encoding>: defines an instruction encoding, the basic unit of the
instruction set. It contains information to refine the encoding
(starting from the regdiagram in the same class, above) and an
assembler template
Attributes:
name: encoding ID, unique across the whole directory of XML files.
oneofinclass: how many encodings are sharing this <iclass>
oneof: how many encodings are to be found in this whole file
label: text which differentiates this encoding from others in the
class, generated from the tags.
bitdiffs: an encoding may be a refinement of the generic encoding
shown in the <iclass>'s <regdiagram>. This encoding is formally
presented in <box> elements, but this attribute provides a
human-friendly description of any extra encoding.
tags: a comma-separated list of tags (with tags common to all
elements in the class omitted).
-->
<!ELEMENT encoding (docvars?, arch_variants?, box*, asmtemplate+, equivalent_to?)>
<!ATTLIST encoding name CDATA #REQUIRED oneofinclass CDATA #REQUIRED
oneof CDATA #REQUIRED label CDATA #REQUIRED
bitdiffs CDATA #IMPLIED
tags CDATA #IMPLIED>
<!--
<equivalent_to>: assembler templates on alias pages are mapped to
encodings on the associate instruction page, with a condition.
-->
<!ELEMENT equivalent_to (asmtemplate, aliascond)>
<!ELEMENT aliascond (%inline;)*>
<!--
<classes>: wraps around a set of iclasses, but first of all you get a
<classintro> paragraph (to be in the same <iclass> two
encodings need to share decode pseudocode and an encoding diagram).
Many pages have just one iclass, but when they don't
there is some text in <classintro> which explains what the
iclasses are.
Attributes:
count: at least 2
-->
<!ELEMENT classesintro (%inline;)*>
<!ATTLIST classesintro count CDATA #IMPLIED>
<!ELEMENT classes (classesintro?, iclass+)>
<!--
<iclassintro>: currently unused. May vanish. But it is a place one
could put any introductory text applying to all instruction
encodings in the <iclass>
Attributes:
count: number of encodings but the same information (the number
of encodings in the iclass) is available in
<iclass ... no_encodings="xx"..>
-->
<!ELEMENT iclassintro (txt | a)*>
<!ATTLIST iclassintro count CDATA #IMPLIED>
<!--
<iclass>: a class of instructions sharing a register diagram.
Unless they're aliases, they'll also have some decode pseudocode.
The building block of the page.
Attributes:
oneof: how many <iclass>es are in the <class>.
name: when oneof is > 1, this will be a distinctive heading for
this iclass among all the iclasses.
no_encodings: how many <encoding> sections are to be found herein.
id: a name which is unique among the <iclass>es in this file.
isa: instruction set name (same as the unique ID of the
instruction set root in the encoding hierarchy).
-->
<!ELEMENT iclass (docvars?, iclassintro?, arch_variants?, regdiagram, encoding+, ps_section?, constrained_unpredictables?)>
<!ATTLIST iclass name CDATA #REQUIRED
id CDATA #REQUIRED
oneof CDATA #REQUIRED
no_encodings CDATA #REQUIRED
isa CDATA #REQUIRED>
<!--
<arch_variants>: wrapping multiple <arch_variant>s
-->
<!ELEMENT arch_variants (arch_variant*)>
<!--
<arch_variant>: a variant of the Arm architecture
that this instruction encoding is defined for. e.g. "Armv8-M".
Attributes:
name: text describing the variant.
-->
<!ELEMENT arch_variant EMPTY>
<!ATTLIST arch_variant name CDATA #IMPLIED
feature CDATA #IMPLIED>
<!--
<explanations>: wrapping multiple symbol <explanation>s
Attributes:
scope: historical, earlier layouts had explanations in different places
-->
<!ELEMENT explanations (explanation*)>
<!ATTLIST explanations scope (all) #REQUIRED>
<!--
<explanation>: an explanation for a single symbol. This can be
provided as simple unstructured text (<account>) or as a table of
values (<definition>).
Attributes:
enclist: is a comma-separated list of encoding IDs to which this
explanation applies;
symboldefcount: when a symbol is defined more than once
(different explanations apply to different encodings or aliases)
this counts up through the definitions starting at number 1.
Useful to renderers which choose not to repeat the symbol with
the different definitions.
tags: a comma-separated taglist (see top of file), which can be
used to produce meaningful differentiating text between
multiple appearances of the symbol. Optional, but always
provided if symboldefcount > 1).
-->
<!ELEMENT explanation (symbol, (account | definition), arch_variants?)>
<!ATTLIST explanation enclist CDATA #REQUIRED
tags CDATA #IMPLIED
symboldefcount CDATA #REQUIRED>
<!--
<symbol>: defines the assembler-language variable 'symbol' being
explained.
Attributes:
link: defines the anchor text used by references to
this symbol on the same page (why isn't it called 'anchor'?
would have been better).
-->
<!ELEMENT symbol (%inline;)*>
<!ATTLIST symbol link CDATA #IMPLIED>
<!--
<account>: used for an unstructured-text explanation of the
symbol. For historical reasons, the text is wrapped inside a
redundant <intro> sub-element.
Attributes:
encodedin: instruction bit-field specifier which says exactly
which bits are involved in this account (same as field in
<definition> below)
-->
<!ELEMENT intro %formatted_text;>
<!ELEMENT account (docvars?, intro)>
<!ATTLIST account encodedin CDATA #IMPLIED>
<!--
<definition>: a tabulated description relating to a symbol.
Introduced by some unstructured text, optionally followed by
some too.
Attributes:
encodedin: instruction bit-field specifier which says exactly
which bits are involved in this definition (same as in
<account>, above).
tabulatedwith: seems to be historical, but will be kept for now.
-->
<!ELEMENT definition (intro, table, after?)>
<!ATTLIST definition encodedin CDATA #REQUIRED
tabulatedwith CDATA #IMPLIED>
<!ELEMENT after (%inline;)*>
<!--
<table>: table used to provide a symbol description by
listing off possible bitfield values against the text to
substitute for the symbol. The table may have a 'class', which
in this case is always 'valuetable'
However, this table is deliberately (which may mean imperfectly)
modelled on a "standard" XML table recommendation for DocBook
or some such.
-->
<!ELEMENT table (tgroup+)>
<!ATTLIST table class (valuetable) #REQUIRED>
<!ELEMENT tgroup (thead, tbody)>
<!ATTLIST tgroup cols CDATA #REQUIRED>
<!ELEMENT thead (row+)>
<!ELEMENT tbody (row+)>
<!ELEMENT row (entry+)>
<!--
<entry>: table cell, required to have a 'class'
Attributes:
iclasslink, iclassfile: used to generate a hotlink
around the entry's text, typically back to the encoding index.
bitwidth: (only for the header row of bitfield classes), the
bitwidth of the field.
-->
<!ELEMENT entry (%inline; | arch_variants)*>
<!ATTLIST entry class (symbol | bitfield | feature) #REQUIRED
bitwidth CDATA #IMPLIED
iclasslink CDATA #IMPLIED
iclasslinkfile CDATA #IMPLIED>
<!--
<aliastablehook>: anchor used to position the table of aliases on browsable
XML pages.
Attributes:
anchor: the name of the anchor for the table
-->
<!ELEMENT aliastablehook (#PCDATA)>
<!ATTLIST aliastablehook anchor CDATA #REQUIRED>
<!--
<ps_section>: a number of pieces of related pseudocode.
Attributes:
howmany: how many <ps> sections are underneath:
may be useful to XSL scripts. Will always be "1"
except in the 'shared_pseudocode' page.
-->
<!ELEMENT ps_section (ps+)>
<!ATTLIST ps_section howmany CDATA #REQUIRED>
<!--
<ps>: piece of pseudocode consisting of (in theory) multiple
<pstext>s, each being the pseudocode for that particular
repository section (In practice, there is only going to be one
<pstext> per <ps>.
Attributes:
name: a name which uniquely identifies this piece of
pseudocode within the pseudocode source repository.
When multiple encodings share identical "postdecode"
and/or "encoding" pseudocode, the shared name will
be chosen randomly from one of the encodings, but
by construction they must all contain identical text.
mylink: decode pseudocode gets a link anchor
which is in fact just the name with '.' substituted for '/',
so it can be used as legal link text (probably useless)
enclabels: obsolete, I think. Never set to anything
sections: the number of <pstext> sections wrapped here. Should
always be "1" (separate sections, eg 'postdecode' and 'execute')
get their own ps_section
secttype: what sort of pseudocode is this, usable for
a browsable/printable section heading.
-->
<!ELEMENT ps (pstext)>
<!ATTLIST ps name CDATA #REQUIRED
mylink CDATA #IMPLIED
enclabels CDATA #IMPLIED
sections CDATA #REQUIRED
secttype CDATA #IMPLIED>
<!ELEMENT psdoc (#PCDATA)>
<!--
<pstext>: a chunk of pseudocode. Pseudocode is text in which
line-endings and indentation matter.
Attributes:
section: suitable for use as a title, if ever there was a need
for one (XXX though see <ps secttype>)
rep_section: the name for this pseudocode repository subsection
"decode" - per-encoding decode, at least one per file
"postdecode" - common/shared decode, at most one per file
"execute" - common execute/operation, exactly one per file
So to identify a piece of pseudocode look at
<ps name="xx"> and <pstext rep_section="xxx">).
mayhavelinks: in general pseudocode text is full of links to
definitions and stuff. This warns the renderer.
-->
<!ELEMENT pstext (%inline;)*>
<!ATTLIST pstext section CDATA #REQUIRED
rep_section CDATA #IMPLIED
mayhavelinks (1) #IMPLIED>
<!--
<constrained_unpredictables>: describes what happens in CONSTRAINED
UNPREDICTABLE situations.
Attributes:
encoding: The encoding with which the UNPREDICTABLE situations are
associated, e.g. "T1".
ps_block: The pseudocode block with which the UNPREDICTABLE situations are
associated, e.g. "Operation".
-->
<!ELEMENT constrained_unpredictables (cu_case+)>
<!ATTLIST constrained_unpredictables encoding CDATA #IMPLIED
ps_block CDATA #IMPLIED>
<!--
<cu_case>: describes what happens in a CONSTRAINED
UNPREDICTABLE situation with a particular condition.
-->
<!ELEMENT cu_case (cu_cause, cu_type+, arch_variants?)>
<!--
<cu_cause>: describes the condition under which a CONSTRAINED
UNPREDICTABLE situation occurs. This is expressed using pseudocode.
-->
<!ELEMENT cu_cause (pstext)>
<!--
<cu_type>: describes a possible outcome of a CONSTRAINED
UNPREDICTABLE situation.
Attributes:
constraint: the ID of an outcome, e.g. "Constraint_UNKNOWN".
constraint_text: text description for an outcome. Not recommended,
but made available for those cases where no ID has been assigned.
-->
<!ELEMENT cu_type (cu_type_variable*,cu_type_text*)>
<!ATTLIST cu_type constraint CDATA #IMPLIED>
<!--
<cu_type_variable>: describes a variable used in a constraint.
Attributes:
name: the name of the variable, e.g. "bitstring".
value: the value of the variable, e.g. "hw1[3:0]".
-->
<!ATTLIST cu_type_variable name CDATA #REQUIRED
value CDATA #REQUIRED>
<!--
<cu_type_text>: Free text describing a constraint.
-->
<!ELEMENT cu_type_text (#PCDATA)*>
<!--
<exceptions>: a list of exceptions that this instruction can cause.
Exceptions are grouped together in <exception_group> elements.
<exception_group> elements contain <exception> elements.
Attributes:
group_name: this group of exceptions only applies in certain
circumstances.
-->
<!ELEMENT exceptions (exception_group*)>
<!ELEMENT exception_group (exception+)>
<!ATTLIST exception_group group_name CDATA #IMPLIED>
<!--
<exception>: data about a particular exception this instruction
can cause.
Attributes:
register: the system register in which a bit or bits are set
when this particular exception is caused.
field: the field in @register which is set to a value when
this particular exception is caused. (If not present,
assume that any of the register fields in the register
being set indicates this exception has been caused.)
value: the value that @field in @register is set to when
this particular exception is caused. (If not present,
assume that the field is set to 1.)
name: if @register is not present, the exception will instead
be described using the name it has in the source pseudocode.
This is a fallback and so will only be present in few cases.
-->
<!ELEMENT exception EMPTY>
<!ATTLIST exception register CDATA #IMPLIED
field CDATA #IMPLIED
value CDATA #IMPLIED
name CDATA #IMPLIED>
<!--
<txt>: used to wrap text in sections and avoid mixed
mode. Probably a mistake and no longer required.
Attributes:
class: can be used to mark substrings for markup purposes.
-->
<!ELEMENT txt (%inline;)*>
<!ATTLIST txt class CDATA #IMPLIED>
<!--
<text>: wrap PCDATA to avoid mixed mode. Historical... deprecated
-->
<!ELEMENT text (#PCDATA)>
Reference in New Issue View Git Blame Copy Permalink