There are many more places that copy an uninitialised expressionS to a
symbol via symbol_set_value_expression and make_expr_symbol. This
patch focuses on general gas code that does that, and a few backends.
Note that unlike the i386 case that oss-fuzz found, it is likely that
the tc-alpha.c, tc-ppc.c and tc-tic54x.c changes are not fixing bugs,
alpha and tic54x because they don't use X_md, ppc because it carefully
handles X_md. Also, as an example an O_constant expression should
only ever have its X_add_number field accessed, therefore the other
fields can stay uninitialised. However, I think that copying
uninitialised struct fields around is not good practice. If nothing
else it can be confusing when examining symbols under gdb.
I also replaced gen-sframe.c "#ifdef SFRAME_FRE_TYPE_SELECTION_OPT"
with "if (SFRAME_FRE_TYPE_SELECTION_OPT)" so code in the false
branches is compiled and thus less likely to bitrot. (As far as I can
see, SFRAME_FRE_TYPE_SELECTION_OPT is always 1.)
* cgen.c (expr_build_binary): Use structure initialiser to
ensure all fields of expression are initialised.
* config/obj-coff.c (obj_coff_val): Likewise.
* config/tc-alpha.c (add_to_link_pool): Likewise.
* config/tc-i386-intel.c (i386_intel_simplify): Likewise.
* config/tc-mips.c (fix_loongson2f_jump, load_register),
(load_address, add_got_offset, add_got_offset_hilo),
(macro_build_branch_likely, macro, mips16_macro),
(s_cpload, s_cpsetup, s_cprestore, s_cpreturn): Likewise.
* config/tc-ppc.c (ppc_function): Likewise.
* config/tc-tic54x.c (tic54x_field): Likewise.
* dw2gencfi.c (output_cfi_insn): Likewise.
* expr.c (expr_build_uconstant): Likewise.
* read.c (s_mri_common): Likewise.
* gen-sframe.c (create_fre_start_addr_exp),
(create_func_info_exp, output_sframe_row_entry): Likewise.
Don't conditionally compile via SFRAME_FRE_TYPE_SELECTION_OPT.
* cgen.c (gas_cgen_parse_operand): Use md_expr_init_rest.
* config/tc-microblaze.c (microblaze_s_weakext): Likewise.
* ecoff.c (ecoff_directive_weakext, ecoff_stab): Likewise.
* read.c (pseudo_set): Likewise.
This patch removes clean_up_expression which runs just before operand()
returns. clean_up_expression sets as yet uninitialised fields of
expressionS. Well, it sets fields based on the value of X_op,
trusting that others have been written, and has one notable exception:
X_md is not initialised.
Instead initialise expressionS fully inside operand(), which is called
at the start of expr(), and introduce md_expr_init for the odd
backends that want to mess with X_md.
This is in response to an oss-fuzz report that read.c:pseudo_set calls
expr() leaving exp.X_md uninitialised and can copy that to a symbol
via symbol_set_value_expression. tc-i386-intel.c:565 is one place
that later tests the uninitialised X_md.
* config/tc-z80.h (md_expr_init, md_expr_init_rest): Define.
* config/tc-microblaze.h: Likewise.
* expr.c (clean_up_expression): Delete.
(operand): Init expression early.
(expr): Use md_expr_init_rest to init X_md when necessary.
This removes many unnecessary NULL casts. I'm also adding a few arg
casts in concat calls, to make the code consistent. Advice from quite
a few years ago was that it's better to use the exact type for args
corresponding to function ellipses, in case NULL is defined as plain
0. (I think that happened with some early 64-bit systems. Plain NULL
ought to be OK nowadays.)
Interestingly emit_leb128_expr() already assumes X_unsigned is properly
set for O_big. Adjust its conversion-to-bignum to respect the incoming
flag, and have convert_to_bignum() correctly set it on output.
It further can't be quite right that convert_to_bignum() depends on
anything other than the incoming expression. Therefore adjust
emit_expr_with_reloc() to be in line with the other invocation.
This also requires an adjustment for SH, which really should have been
part of 762acf217c ("gas: maintain O_constant signedness in more
cases").
PR gas/32721
In this bogus piece of code distilled from fuzzing and slightly edited:
A=%eax|%!
Y=A
Z=A
or $6,Z
the first of the equates with A on the rhs changes A's section (due to
the use of S_GET_VALUE()), from expression to register, thus yielding Y
in the expression section (and X_op being O_symbol), but Z in the
register section (and X_op being O_register with X_add_value being -1).
There shouldn't be random O_register expressions, though, for targets
setting md_register_arithmetic to false. Plus both Y and Z would better
be exchangeable.
In pseudo_set() wire handling of O_symbol expressions referencing a
symbol in the expression section to that of other stuff ending up in
this section.
Also avoid bogus O_register expressions to be created, for targets
setting md_register_arithmetic to false: S_GET_VALUE() would resolve
any arithmetic, which must not happen for such targets. To be on the
safe side for such targets, also amend resolve_register(). Correct
another earlier oversight there too (affecting at least Z80), by using
the new expr_copy() helper there as well.
Undo 46b9f07dfe ("PR 32721, internal error in
tc-i386.c:parse_register"), albeit without losing the simplification it
did.
... by way of introducing LEX_EOL and LEX_EOS. As a prereq convert the
remaining open-coded accesses.
The Alpha change is actually a functional one: The array slot for '!'
having been set to 1 is very unlikely to have been correct. 1 means "end
of line", when surely "end of statement" was always meant.
Let's extend lex_type[] to also cover whitespace, then having a simple
macro to uniformly recognize both blanks and tabs (and \r when it's not
EOL) as such.
In macro.c use sb_skip_white() as appropriate, instead of open-coding
it.
... by purging dot_{frag,value}. Right now these two and dot_symbol are
updated independently, which can't be quite right. Centralize .-related
information in dot_symbol, updating it also where previously
dot_{frag,value} were updated. Since S_GET_VALUE() can't be used to
retrieve what used to be dot_value, introduce a new helper to fetch both
frag and offset.
Deferring expression evaluation is often necessary. However, the value
current_location() records typically is intended to represent the
location at the point of use of the expression, with the exception being
.eqv (or its == equivalent). Change how expr_defer behaves in this
regard, and introduce a special mode just for pseudo_set() to use.
Introduce a predicate to cover both "deferred" modes, and use it
everywhere except in current_location(), where only the new mode wants
checking for.
Dividing the largest possible negative value by -1 generally is UB, for
the result not being representable at least in commonly used binary
notation. This UB on x86, for example, is a Floating Point Exception on
Linux, i.e. resulting in an internal error (albeit only when
sizeof(valueT) == sizeof(void *); the library routine otherwise involved
apparently deals with the inputs quite okay).
Leave original values unaltered for division by 1; this may matter down
the road, in case we start including X_unsigned and X_extrabit in
arithmetic. For the same reason treat modulo by 1 the same as modulo by
-1.
The quad and octa tests have more relaxed expecations than intended, for
X_unsigned and X_extrabit not being taken into account [yet]. The upper
halves can wrongly end up as all ones (for .octa, when !BFD64, even the
upper three quarters). Yet it makes little sense to address this just
for div/mod by ±1. quad-div2 is yet more special, to cover for most
32-bit targets being unable to deal with forward-ref expressions in
.quad even when BFD64; even ones being able to (like x86) then still
don't get the values right.
Commit 4a826962e7 changed its behavior without saying why, and without
putting in place any testcase demonstrating the required behavior.
Firmly latch the current position unless deferred-evaluation mode is in
effect.
There's no point involving symbol_clone_if_forward_ref(), just for it to
replace dot_symbol by one obtained from symbol_temp_new_now(). For the
abs-section case also produce a slightly more "complete" (as in: all
potentially relevant fields filled) expression by going through
expr_build_uconstant().
Move the function next to current_location(), for it to be easier to see
the (dis)similarities. Correct the function's comment while there.
Once properly set, it's only ever holding the same value as "radix".
Even if there was some plan with it, that plan hasn't made it anywhere
in over 20 years.
I don't view it as sensible to be more lax when it comes to references
to (uses of) such labels compared to their definition: The latter has
been limited to decimal numerics, while the former permitted any radix.
Beyond that leading zeroes on such labels aren't helpful either. Imo
labels and their use sites would better match literally, to avoid
confusion.
As it turns out, one z80 testcase actually had such an odd use of labels
where definition and use don't match in spelling. That testcase is being
adjusted accordingly.
While there also adjust a comment on a local variable in
integer_constant().
Adds two new external authors to etc/update-copyright.py to cover
bfd/ax_tls.m4, and adds gprofng to dirs handled automatically, then
updates copyright messages as follows:
1) Update cgen/utils.scm emitted copyrights.
2) Run "etc/update-copyright.py --this-year" with an extra external
author I haven't committed, 'Kalray SA.', to cover gas testsuite
files (which should have their copyright message removed).
3) Build with --enable-maintainer-mode --enable-cgen-maint=yes.
4) Check out */po/*.pot which we don't update frequently.
First of all the respective original changes didn't deal with just 0
having such a suffix - this needs additional logic outside of
integer_constant(). Further bogus suffixes having more than two L-s
were accepted, while valid suffixes with U following the L(s) weren't.
Finally respective tests were introduced for Sparc only.
Reviewed-by: Neal Frager <neal.frager@amd.com>
Give backends a chance to see these, just as they can see binary ones.
Most of those which use this hook already cope with NULL being passed
for the left operand (typically because of checking the operator first).
Adjust the two which don't.
Take the opportunity and also document the hook.
Unary '~' doesn't really produce an unsigned result. Neither does
subtraction (unless taking operand values into consideration). And an
abstract operator applied to two operands which aren't both unsigned
can't be assumed to yield an unsigned result; exceptions are
- shifts, where only signedness of the left hand operand matters,
- comparisons, which - unlike unary '!' - produce signed results (they
deliver 0 or ~0, as opposed to '!', which yields 0 or 1),
- logical operators (yielding 0 or 1 and hence treated like unary '!').
While doing this (specifically while extending the all/quad testcase),
update .quad and .8byte documentation: With 64-bit architectures now
being common, it is highly inappropriate to state that these directives
unconditionally require bignums.
There are two problems: symbol_equated_p() doesn't recognize equates of
registers, and S_CAN_BE_REDEFINED() goes by section rather than by
expression type. Both together undermine .eqv and .equiv clearly meaning
to guard the involved symbols against re-definition (both ways).
To compensate pseudo_set() now using O_symbol and S_CAN_BE_REDEFINED()
now checking for O_register,
- for targets creating register symbols through symbol_{new,create}() ->
symbol_init() -> S_SET_VALUE() (alpha, arc, dlx, ia64, m68k, mips,
mmix, tic4x, tic54x, plus anything using cgen or itbl-ops), have
symbol_init() set their expressions to O_register,
- x86'es parse_register() also can't go by section anymore when
trying to "look through" equates; probably symbol_equated_p() should
have been used there from the beginning, if only that had worked for
equates of registers,
- various targets need to "look through" equates when parsing insn
operands (which also helps transitive forward equates); perhaps even
more ought to, but many don't look to consider the possibility of
register equates in the first place.
This was uncovered by code reported in PR gas/30274 (duplicating
PR gas/30272), except that there .eqv was used when really .equ was
meant. Therefore that bug report is addressed here only in so far as
gas wouldn't crash anymore; the code there still won't assemble
successfully, just that now the issues there are properly diagnosed.
... out of mainline code, grouping together the two case labels. This
then also make more obvious that the comment there applies to both forms
of shifts.
Callers which want to make further parsing attempts at the buffer passed
to the function need to be aware that due to the potential of string
concatenation the input buffer may be altered in ways beyond what can be
undone by putting back at *input_line_pointer the character that the
function returns.
Even a unary '+' has to be considered arithmetic; at least on x86 in
Intel Syntax mode otherwise bogus insn operands may be accepted.
Convert this specific case to binary + (i.e. 0 + <register>). (An
implication is that md_operator(,1,) would need to deal with arch-
specific equivalents of unary '+' is a similar way, if such an arch-
specific variant would be specified in the first place.)
To avoid duplicating what make_expr_symbol() does to construct a
constant-zero expression, simply make its previously local variable a
file-scope static one. This way there's also no need to invoke
clean_up_expression().
In 162c6aef1f ("gas: fold symbol table entries generated for
.startof.() / .sizeof.()") I screwed up quite badly, inverting the case
sensitive and case insensitive comparison functions.
"Local labels are never absolute" says the comment. Except when they
are. Testcase
.offset
0:
a=0b
I don't see any particular reason to disallow local labels inside
struct definitions, so delete the comment and assertions.
* expr.c (integer_constant): Delete local label assertions.
This is the assertion
know (*input_line_pointer != ' ');
after calling operand.
The usual exit from operand calls SKIP_ALL_WHITESPACE.
* expr.c (operand): Call SKIP_ALL_WHITESPACE after call to expr.
The newer update-copyright.py fixes file encoding too, removing cr/lf
on binutils/bfdtest2.c and ld/testsuite/ld-cygwin/exe-export.exp, and
embedded cr in binutils/testsuite/binutils-all/ar.exp string match.
bfd * coff-i386.c (in_reloc_p): Add R_SECTION.
(howto_table): Add R_SECTION.
(coff_pe_i386_relocation_section): Add support for R_SECTION.
(coff_i386_reloc_type_lookup): Add support for
BFD_RELOC_16_SECCIDX.
* coff-x86_64.c (in_reloc_p): Add R_SECTION.
(howto_table): Add R_SECTION.
(coff_pe_amd64_relocation_section): Add support for R_SECTION.
(coff_amd64_reloc_type_lookup): Add support for
BFD_RELOC_16_SECCIDX.
* reloc.c: Add BFD_RELOC_16_SECIDX.
* bfd-in2.h: Regenerate.
* libbfd.h: Regenerate.
gas * config/tc-i386.c (pe_directive_secidx): New function.
(md_pseudo_table): Add support for secidx.
(x86_cons_fix_new): Likewise.
(tc_gen_reloc): Likewise.
* expr.c (op_rank): Add O_secidx.
* expr.h (operatorT): Likewise.
* symbols.c (resolve_symbol_value): Add support for O_secidx.
* testsuite/gas/i386/secidx.s: New test source file.
* testsuite/gas/i386/secidx.d: New test driver file.
* testsuite/gas/i386/i386.exp: Run new test.
include * coff/i386.h: Define R_SECTION.
* coff/x86_64.h: Likewise.
ld * testsuite/ld-pe/secidx1.s: New test source file.
* testsuite/ld-pe/secidx2.s: New test source file.
* testsuite/ld-pe/secidx.d: New test driver file.
* testsuite/ld-pe/secidx_64.d: New test driver file.
* testsuite/ld-pe/pe.exp: Add new tests.
operand() is not a place that should be calling ignore_rest_of_line.
ignore_rest_of_line shouldn't increment input_line_pointer if already
at buffer limit.
* expr.c (operand): Don't call ignore_rest_of_line.
* read.c (s_mri_common): Likewise.
(ignore_rest_of_line): Don't increment input_line_pointer if
already at buffer_limit.
Much of the gas source and older BFD source use "long" for function
parameters and variables, when other types would be more appropriate.
This patch fixes one of those cases. Dollar labels and numeric local
labels do not need large numbers. Small positive itegers are usually
all that is required. Due to allowing longs, it was possible for
fb_label_name and dollar_label_name to overflow their buffers.
* symbols.c: Delete unnecessary forward declarations.
(dollar_labels, dollar_label_instances): Use unsigned int.
(dollar_label_defined, dollar_label_instance): Likewise.
(define_dollar_label): Likewise.
(fb_low_counter, fb_labels, fb_label_instances): Likewise.
(fb_label_instance_inc, fb_label_instance): Likewise.
(fb_label_count, fb_label_max): Make them size_t.
(dollar_label_name, fb_label_name): Rewrite using sprintf.
* symbols.h (dollar_label_defined): Update prototype.
(define_dollar_label, dollar_label_name): Likewise.
(fb_label_instance_inc, fb_label_name): Likewise.
* config/bfin-lex.l (yylex): Remove unnecessary casts.
* expr.c (integer_constant): Likewise.
* read.c (read_a_source_file): Limit numeric label range to int.
Strange effects can result from the present handling, e.g.:
.if 1
"backslash\\":
.endif
yields first (correctly) "missing closing `"'" but then also "invalid
character '\' in mnemonic" and further "end of file inside conditional".
Symbols names ending in \ are in principle not expressable with that
scheme.
Instead of recording whether a backslash was seen, inspect the
subsequent character right away. Only accept \\ (meaning a single
backslash in the resulting symbol name) and \" (meaning an embedded
double quote in the resulting symbol name) for now, warning about any
other combination.
While perhaps not necessary immediately, also permit concatenated
strings to form a symbol name. This may become useful if going forward
we would want to support \<octal> or \x<hex> sequences, where closing
and re-opening quotes can be useful to delimit such sequences.
The ELF "Multibyte symbol names" test gets switched away from using
.set, as that would now also mean excluding nios2 and pru. By using
.equiv instead, even the existing #notarget can be dropped. (For h8300
the .section directive additionally needs attributes specified, to avoid
a target specific warning.)
The result of running etc/update-copyright.py --this-year, fixing all
the files whose mode is changed by the script, plus a build with
--enable-maintainer-mode --enable-cgen-maint=yes, then checking
out */po/*.pot which we don't update frequently.
The copy of cgen was with commit d1dd5fcc38ead reverted as that commit
breaks building of bfp opcodes files.
When the same such construct is used multiple times in a source file,
there's still no need to emit a separate symbol each time. Under the
assumption that there won't be many of these, use a simple array
lookup method to record previously used symbols.
When built on a 32-bit host without --enable-64-bit-bfd, powerpc-linux
and other 32-bit powerpc targeted binutils fail to assemble some
power10 prefixed instructions with 34-bit fields. A typical error
seen when running the testsuite is
.../gas/testsuite/gas/ppc/prefix-pcrel.s:10: Error: bignum invalid
In practice this doesn't matter for addresses: 32-bit programs don't
need or use the top 2 bits of a d34 field when calculating addresses.
However it may matter when loading or adding 64-bit constants with
paddi. A power10 processor in 32-bit mode still has 64-bit wide GPRs.
So this patch enables limited support for O_big PowerPC operands, and
corrects sign extension of 32-bit constants using X_extrabit.
* config/tc-ppc.c (insn_validate): Use uint64_t for operand values.
(md_assemble): Likewise. Handle bignum operands.
(ppc_elf_suffix): Handle O_big. Remove unnecessary input_line_pointer
check.
* expr.c: Delete unnecessary forward declarations.
(generic_bignum_to_int32): Return uint32_t.
(generic_bignum_to_int64): Return uint64_t. Compile always.
(operand): Twiddle X_extrabit for unary '~'. Set X_unsigned and
clear X_extrabit for unary '!'.
* expr.h (generic_bignum_to_int32): Declare.
(generic_bignum_to_int64): Declare.
* testsuite/gas/ppc/prefix-pcrel.s,
* testsuite/gas/ppc/prefix-pcrel.d: Add more instructions.
In an upcoming commit, I need to be able to set the prefix used
to introduce hexadecimal literal constants using a command line
flag. This is not currently possible, because the switch which
determines this (LITERAL_PREFIXDOLLAR_HEX) is a macro set at
build time.
This change substitutes it for a variable to be set at start up.
gas/ChangeLog:
* expr.c (literal_prefix_dollar_hex): New variable.
(operand)[case '$']: Use the new variable instead of the old macro.
Also, move this instance of "case '$'" next to the other one, and
enable it only in the complementary proprocessor case.
* expr.h (literal_prefix_dollar_hex): Declare it.
* config/tc-epiphany.c (md_begin): Assign literal_prefix_dollar_hex.
* config/tc-ip2k.c: ditto
* config/tc-mt.c: ditto
* config/tc-epiphany.h (LITERAL_PREFIXDOLLAR_HEX): Remove macro definition.
* config/tc-ip2k.h: ditto
* config/tc-mt.h: ditto
