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.
There are quite a few ubsan warnings in gas. This one disappears with
a code tidy.
* read.c (s_app_line): Rename 'l' to 'linenum'. Avoid ubsan
warning.
BFD generally doesn't handle anything but a power of two section
alignment, and ELF sh_addralign is required to be an integral power of
two (or zero) by the ELF spec. Of course this is ignored by fuzzers,
and because bfd_log2 rounds up, we can end up with alignment_power
being 32 on a 32-bit object or 64 on a 64-bit object. That then
triggers ubsan warnings in places like bfd_update_compression_header
where we want to convert from alignment_power back to an alignment.
I suppose we could reject object files that have non-compliant
sh_addralign, but I think it's also reasonable to use the greatest
power of two divisor of sh_addralign, ie. the rightmost 1 bit.
* elf.c (_bfd_elf_make_section_from_shdr): Use greatest power
of two divisor of sh_addralign.
(_bfd_elf_assign_file_position_for_section): Likewise.
(assign_file_positions_for_non_load_sections): Likewise.
There was an omission on 3e6dc39ed7 "sim/testsuite: Set
global_cc_os also when no compiler is found"; global_cc_os
wasn't set for other than the primary target, which means
that the "unguarded" use of global_cc_os in
testsuite/cris/c/c.exp caused the dreaded "ERROR: can't read
"global_cc_os": no such variable" when e.g. configuring for
pru-elf and doing "make check-sim". Better initializing
both variables at the top to default values, rather than
adding another single 'set global_cc_os ""', to reduce the
risk of not setting them properly if or when that
if-statement-chain is made longer.
sim/testsuite:
* lib/sim-defs.exp (sim_init_toolchain): Default
global_cc_os and global_cc_works properly, before if-chain.
Add has_sib to struct instr_info and use SIB info only if ins->has_sib
is true.
PR binutils/28892
* i386-dis.c (instr_info): Add has_sib.
(get_sib): Set has_sib.
(OP_E_memory): Replace havesib with ins->has_sib.
(OP_VEX): Use ins->sib.index only if ins->has_sib is true.
It is possible for a compiler to optimize a function in a such ways that
the function does not follow the calling convention of the target. In
such situation, the compiler can use the DW_AT_calling_convention
attribute with the value DW_CC_nocall to tell the debugger that it is
unsafe to call the function. The DWARF5 standard states, in 3.3.1.1:
> If the value of the calling convention attribute is the constant
> DW_CC_nocall, the subroutine does not obey standard calling
> conventions, and it may not be safe for the debugger to call this
> subroutine.
Non standard calling convention can affect GDB's assumptions in multiple
ways, including how arguments are passed to the function, how values are
returned, and so on. For this reason, it is unsafe for GDB to try to do
the following operations on a function with marked with DW_CC_nocall:
- call / print an expression requiring the function to be evaluated,
- inspect the value a function returns using the 'finish' command,
- force the value returned by a function using the 'return' command.
This patch ensures that if a command which relies on GDB's knowledge of
the target's calling convention is used on a function marked nocall, GDB
prints an appropriate message to the user and does not proceed with the
operation which is unreliable.
Note that it is still possible for someone to use a vendor specific
value for the DW_AT_calling_convention attribute for example to indicate
the use of an alternative calling convention. This commit does not
prevent this, and target dependent code can be adjusted if one wanted to
support multiple calling conventions.
Tested on x86_64-Linux, with no regression observed.
Change-Id: I72970dae68234cb83edbc0cf71aa3d6002a4a540
Add an argument to the get_return_value function to indicate the symbol
of the function the debuggee is returning from. This will be used by
the following patch.
Since the function return type can be deduced from the symbol remove the
value_type argument which becomes redundant.
No user visible change after this patch.
Tested on x86_64-linux.
Change-Id: Idf1279f1f7199f5022738a6679e0fa63fbd22edc
Co-authored-by: Simon Marchi <simon.marchi@polymtl.ca>
...when we know we have a working compiler. This will reduce
the risk of faulty edits by exposing them rather than hiding
them as "unresolved". It also harmonizes behavior with that of
run_sim_test.
* c/c.exp: Mark C tests failing compilation test errors.
Calls to basename were added here as part of commit
e1e1ae6e9b "sim: testsuite: fix objdir handling", but that
commit missed adding "#include <libgen.h>" or the equivalent
GNU extension, see basename(3). Fixing that shows a logical
error in the change to openpf1.c; the non-/-prefixed
code-path was changed instead of the "/"-prefixed code-path,
which is the one executed after that commit.
For "newlib" these tests failed linking after that commit.
Recent newlib has the (asm-renamed) GNU-extension-variant of
basename, but we're better off not using it at all.
Unfortunately, compilation failures for C tests run by the
machinery in c.exp are currently just marked "unresolved",
in contrast to C and assembler tests run by calling
run_sim_test.
The interaction of calling with the full program-path vs.
use of --sysroot exposes a consistency problem: when
--sysroot is used, argv[0] isn't the path by which the
program can find itself. It's undecided whether argv[0] for
the program running in the simulator should be edited
(related to the naked argument to the simulator before
passing on to the simulated program) to remove a leading
--sysroot. Either way, such a change would be out of scope
for this commit.
* c/stat3.c (mybasename): New macro. Use it instead of basename.
* c/openpf1.c: Correct basename-related change and update related
comment.
With commit 7a259895bb "sim: testsuite: expand arch specific
toolchain settings", trying to use out-of-tree ld and as at test-time
broke for the "primary target", like when testing a release-tarball.
Subsequent to that commit, all assembler tests without in-tree-built
tools FAIL, getting errors when trying to call
$(abs_builddir)/../gas/as-new. But, that isn't the actual culprint;
it's actually it's its immediate predecessor, commit 8996c21067
"sim: testsuite: setup per-port toolchain settings for multitarget
build", which hardcodes in-tree-paths to those tools instead of
considering e.g. $(<X>_FOR_TARGET), the preferred overridable variable
for single-target builds, as set up by the toplevel Makefile.
This commit calls GCC_TARGET_TOOL (a deceptive name; gcc-specific
features aren't used) from toplev/config/acx.m4, somewhat like calls
in toplev/configure.ac but without the NCN_STRICT_CHECK_TARGET_TOOLS
step, for each X to find a value for $(<X>_FOR_TARGET). N.B.: in-tree
tools still override any ${target}-${tool} found in $PATH, i.e. only
previously broken builds are affected.
The variables $(<X>_FOR_TARGET) are usually overridden by the toplevel
Makefile to the same value or better, but has to be set here too, as
automake "wants" Makefiles to be self-contained (you get an error
pointing out that the variable may be empty). If it hadn't been for
that, SIM_AC_CHECK_TOOLCHAIN_FOR_PRIMARY_TARGET would not be needed.
This detail should only (positively) affect users invoking "make
check" in sim/ instead of "make check-sim" (or "make check") at the
toplevel. Now the output from "configure" matches the target tools
actually used by sim at test-time, for the "primary target".
Using $(CC) for "example-" targets CC_FOR_TARGET is not changed, as
that appears to be a deliberate special-case.
Note that all tools still have to be installed and present in
$PATH at configure-time to be properly used at test-time.
sim:
* m4/sim_ac_toolchain.m4 (SIM_AC_CHECK_TOOLCHAIN_FOR_PRIMARY_TARGET):
New defun.
(SIM_TOOLCHAIN_VARS): Call it using AC_REQUIRE, and use variables
AS_FOR_TARGET, LD_FOR_TARGET and CC_FOR_TARGET instead of hard-coded
values.
* Makefile.in, configure: Regenerate.
With --disable-sim-hardware (--enable-sim-hardware=no),
whose default was changed to --enable-sim-hardware(=yes) in
commit 34cf511206, building for cris-elf fails as
sim_hw_parse then doesn't exist.
A cris-elf simulator configured for --enable-sim-hardware
(or the default after to the mentioned commit) runs about
2.5x slower than one configured --disable-sim-hardware.
A further 2-5% performance regression was not investigated.
When sim_hw_parse doesn't exist, --cris-900000xx can't be
supported. The best action here is to remove it completely,
so its absence can be identified through --help, but
avoiding littering the code with "#if WITH_HW".
sim/cris:
* sim-if.c (cris_options) [WITH_HW]: Conditionalize
support of option --cris-900000xx.
(sim_open) [WITH_HW]: Conditionalize sim_hw_parse
call.
Apply the new run_sim_test option "require" as in "#require
simoption --cris-900000xx" for all tests using that option.
This allows a clean test-suite-run for a build with
--disable-sim-hardware, where that option is not supported,
by skipping those tests as "untested".
sim/testsuite/cris:
* asm/io1.ms, asm/io2.ms, asm/io3.ms, asm/io6.ms,
asm/io7.ms: Call "#require: simoption --cris-900000xx".
Simulator features can be present or not, typically
depending on different-valued configure options, like
--enable-sim-hardware[=off|=on]. To avoid failures in
test-suite-runs when testing such configurations, a new
predicate is needed, as neither "target", "progos" nor
"mach" fits cleanly.
The immediate need was to check for presence of a simulator
option, but rather than a specialized "requires-simoption:"
predicate I thought I'd handle the general (parametrized)
need, so here's a generic predicate machinery and a (first)
predicate to use together with it; checking whether a
particular option is supported, by looking at "run --help"
output. This was inspired by the check_effective_target_
machinery in the gcc test-suite.
Multiple "requires: <requirement> <parameter>" form a list of
predicates (with parameters), to be used as a conjunction.
sim/testsuite:
* lib/sim-defs.exp (sim_check_requires_simoption): New function.
(run_sim_test): Support "requires: <requirement> <parameter>".
For reasons that remain largely to be investigated (besides
the apparent lack of synchronization between two processes),
this test fails randomly, with two different sets of common
outputs. Curiously, that doesn't happen for the other
similar tests. There's a comment that mentions this, though
that doesn't make it a sustainable part of a test-suite.
(Known-blinking tests should be disabled until fixed.)
sim/testsuite/cris:
* hw/rv-n-cris/irq1.ms: Disable by use of a never-matched
"progos" value.
Commit a39487c668 "sim: cris: use -sim with C tests for cris-elf
targets" caused " -sim" to be appended to CFLAGS_FOR_TARGET for
cris*-*-elf, where testing had until then relied on
"RUNTESTFLAGS=--target_board=cris-sim" being passed when running "make
check-sim", adding the right options. While "-sim" happens to work,
the baseboard-file cris-sim.exp uses "-sim3" so for consistency use
that instead.
Then commit b42f20d2ac "sim: testsuite: drop most specific istarget
checks" caused " -sim" to be appended for *all* targets, which just
doesn't work. For example, for crisv32-linux-gnu, that's not a
recognized option and will cause a dejagnu error and further testing
in c.exp will be aborted.
While cris-sim.exp appends "-static" for *-linux-gnu, further changes
in the test-suite have caused "linux"-specific tests to break, so that
part will be tended to separately.
But, save and restore CFLAGS_FOR_TARGET around the modification and
use where needed, to not have the CRIS-specific modification affect a
continuing test-run (possibly for other targets).
sim/testsuite/cris:
* c/c.exp (CFLAGS_FOR_TARGET): Replace appended option " -sim"
with " -sim3", but do it conditionally for newlib targets. Save
and restore CFLAGS_FOR_TARGET in saved_CFLAGS_FOR_TARGET such
that it doesn't affect the value of CFLAGS_FOR_TARGET outside
c.exp.
If we don't set this variable, it doesn't exist, and using "#progos:"
in an assembler-file will cause an error rather than just skipping the
test, viz:
Running /src/sim/testsuite/cris/hw/rv-n-cris/rvc.exp ...
ERROR: tcl error sourcing /src/sim/testsuite/cris/hw/rv-n-cris/rvc.exp.
ERROR: can't read "global_cc_os": no such variable
while executing
"if { $opts(progos) != "" && $opts(progos) != $global_cc_os } {
untested $subdir/$name
return
}"
(procedure "run_sim_test" line 102)
Neither the commit introducing progos, nor the top comment
in run_sim_test, mentions progos as intended only for C
tests, or that its use must be gated on $global_cc_works !=
0, so (not) setting it in the no-working-compiler path seems
just overlooked.
Allowing it to be used for assembler tests makes it usable
for e.g. an always-false predicate and in expressions in
.exp files without gating on $global_cc_works != 0.
With this patch, global_cc_os is set to "", just as for "unknown OS".
sim/testsuite:
* lib/sim-defs.exp (sim_init_toolchain): Set global_cc_os also when
no working target C compiler is found.
In 5ee0bc23a6 "sim: clean up bfd_vma printing" there was
an additional introduction of PRIx32 and PRIu32 but just in
sim/cris/sim-if.c. One type of bug was fixed in commit
d16ce6e4d5 "sim: cris: fix memory setup typos" but one
remained; the PRIu32 usage is wrong, as hex output is
desired; note the 0x prefix.
Without this fix, you'll see output like:
memory map 0:0x4000..0x5fff (8192 bytes) overlaps 0:0x0..0x16383 (91012 bytes)
program stopped with signal 6 (Aborted).
for some C programs, like some of the ones in the sim/cris/c
testsuite from where the example is taken (freopen2.c).
The bug behavior was with memory allocation. With an
attempt to allocate memory using the brk syscall such that
the room up to the next 8192-byte "page boundary" wasn't
sufficient, the simulator memory allocation machinery horked
on a consistency error when trying to allocate a memory
block to raise the "end of the data segment": there was
already memory allocated at that address.
Unfortunately, none of the programs in sim/cris/asm exposed
this bug at the time, but an assembler test-case is
committed after this fix.
sim/cris:
* sim-if.c (sim_open): Correct PRIu32 to PRIx32.
This commit moves the two Python functions that are used for styling
into a new module, gdb.styling, there's then a small update in
python.c so GDB can find the functions in their new location.
The motivation for this change is purely to try and reduce the clutter
in the top-level gdb module, and encapsulate related functions into
modules. I did ponder documenting these functions as part of the
Python API, however, doing so would effectively "fix" the API, and I'm
still wondering if there's improvements that could be made, also, the
colorize function is only called in some cases now that GDB prefers
libsource-highlight, so it's not entirely sure how this would work as
part of a user facing API.
Still, despite these functions never having been part of a documented
API, it is possible that a user out there has overridden these to, in
some way, customize how GDB performs styling. Moving the function as
I propose in this patch could break things for that user, however,
fixing this breakage is trivial, and, as these functions were never
documented, I don't think we should be obliged to not break user code
that relies on them.
This commit adds styling support to the disassembler output, as such
two new commands are added to GDB:
set style disassembler enabled on|off
show style disassembler enabled
In this commit I make use of the Python Pygments package to provide
the styling. I did investigate making use of libsource-highlight,
however, I found the highlighting results to be inferior to those of
Pygments; only some mnemonics were highlighted, and highlighting of
register names such as r9d and r8d (on x86-64) was incorrect.
To enable disassembler highlighting via Pygments, I've added a new
extension language hook, which is then implemented for Python. This
hook is very similar to the existing hook for source code
colorization.
One possibly odd choice I made with the new hook is to pass a
gdb.Architecture through, even though this is currently unused. The
reason this argument is not used is that, currently, styling is
performed identically for all architectures.
However, even though the Python function used to perform styling of
disassembly output is not part of any documented API, I don't want
to close the door on a user overriding this function to provide
architecture specific styling. To do this, the user would inevitably
require access to the gdb.Architecture, and so I decided to add this
field now.
The styling is applied within gdb_disassembler::print_insn, to achieve
this, gdb_disassembler now writes its output into a temporary buffer,
styling is then applied to the contents of this buffer. Finally the
gdb_disassembler buffer is copied out to its final destination stream.
There's a new test to check that the disassembler output includes some
escape sequences, though I don't check for specific colours; the
precise colors will depend on which instructions are in the
disassembler output, and, I guess, how pygments is configured.
The only negative change with this commit is how we currently style
addresses in GDB.
Currently, when the disassembler wants to print an address, we call
back into GDB, and GDB prints the address value using the `address`
styling, and the symbol name using `function` styling. After this
commit, if pygments is used, then all disassembler styling is done
through pygments, and this include the address and symbol name parts
of the disassembler output.
I don't know how much of an issue this will be for people. There's
already some precedent for this in GDB when we look at source styling.
For example, function names in styled source listings are not styled
using the `function` style, but instead, either GNU Source Highlight,
or pygments gets to decide how the function name should be styled.
If the Python pygments library is not present then GDB will continue
to behave as it always has, the disassembler output is mostly
unstyled, but the address and symbols are styled using the `address`
and `function` styles, as they are today.
However, if the user does `set style disassembler enabled off`, then
all disassembler styling is switched off. This obviously covers the
use of pygments, but also includes the minimal styling done by GDB
when pygments is not available.
Don't change indirect symbol defined in IR to undefined if it is
referenced from shared object.
bfd/
PR ld/28879
* elflink.c (_bfd_elf_merge_symbol): Don't change indirect
symbol defined in IR to undefined if it is referenced from
shared object.
ld/
PR ld/28879
* testsuite/ld-plugin/lto.exp: Run PR ld/28879 tests.
* testsuite/ld-plugin/pr28879a.cc: New file.
* testsuite/ld-plugin/pr28879b.cc: Likewise.
The better to see any code that accesses expld.dataseg.
* ldexp.c (fold_segment_end): Remove seg parameter. Adjust calls.
(fold_segment_align, fold_segment_relro_end): Likewise.
* ldlang.c (lang_size_segment): Likewise.
(lang_size_relro_segment_1, lang_find_relro_sections_1): Likewise.
Now that ld properly aligns the end of the relro segment, the hack to
make relro work on powerpc can disappear.
bfd/
* bfd.c (bfd_emul_get_commonpagesize): Remove relro param.
Don't return bed->relropagesize.
* elf-bfd.h (struct elf_backend_data): Remove relropagesize.
* elfxx-target.h (ELF_RELROPAGESIZE): Remove.
* elf32-ppc.c (ELF_RELROPAGESIZE): Don't define.
* elf64-ppc.c: Likewise.
* bfd-in2.h: Regenerate.
ld/
* ldemul.c (after_parse_default): Adjust
bfd_emul_get_commonpagesize call.
x86 treats MAXPAGESIZE as a memory optimisation parameter, actual
hardware paging is always COMMPAGESIZE of 4k. Use COMMONPAGESIZE for
the end of the relro segment alignment.
The previous patch regresses pr18176, increasing the testcase file
size from 322208 to 2099872 bytes. Fixing this on x86 will require
introducing a gap after the end of the relro segment (of up to
relropagesize-1 bytes).
PR 28824
PR 18176
* ld.h (ld_config_type): Add relro_use_commonpagesize field.
* ldexp.c (fold_segment_align): Set relropagesize depending on
relro_use_commonpagesize.
* emultempl/elf-x86.em (elf_x86_create_output_section_statements):
Set relro_use_commonpagesize.
* testsuite/ld-x86-64/pr18176.d: xfail.
Background
==========
There are constraints on layout of binaries to meet demand paging and
memory protection requirements. Demand paged binaries must have file
offset mod pagesize equal to vma mod pagesize. Memory protection
(executable, read, write status) can only change at page boundaries.
The linker's MAXPAGESIZE variable gives the page size for these layout
constraints.
In a typical basic executable with two memory segments, text (RE) and
data (RW), the data segment must start on a different page to the
last text segment page. For example, with 64k pages and a small
executable of 48k text and 1k data, the text segment might start at
address 0x10000 and data at 0x20000 for a total of two 64k memory
pages. Demand paging would require the image on disk to be 64k+1k
in size. We can do better than that. If the data segment instead
starts at 0x2c000 (the end of the text segment plus one 64k page) then
there are still only two memory pages, but the disk image is now
smaller, 48k+1k in size. This is why the linker normally starts the
data segment at the end of the text segment plus one page. That
simple heuristic isn't ideal in all cases. Changing our simple
example to one with 64k-1 text size, following that heuristic would
result in data starting at 0x2ffff. Now we have two 64k memory data
pages for a data segment of 1k! If the data segment instead started
at 0x30000 we'd get a single data segment page at the cost of 1 byte
extra in the disk image, which is likely a good trade-off. So the
linker does adjust the simple heuristic. Just how much disk image
size increase is allowed is controlled by the linker's COMMONPAGESIZE
variable.
A PT_GNU_RELRO segment overlays the initial part of the data segment,
saying that those pages should be made read-only after relocation by
the dynamic loader. Page granularity for memory protection means that
the end of the relro segment must be at a page boundary.
The problem
===========
Unfortunately most targets currently only align the end of the relro
segment to COMMONPAGESIZE. That results in only partial relro
protection if an executable is running with MAXPAGESIZE pages, since
any part of the relro segment past the last MAXPAGESIZE boundary can't
be made read-only without also affecting sections past the end of the
relro segment. I believe this problem arose because x86 always runs
with 4k (COMMPAGESIZE) memory pages, and therefore using a larger
MAXPAGESIZE on x86 is for reasons other than the demand paging and
memory page protection boundary requirements.
The solution
============
Always end the relro segment on a MAXPAGESIZE boundary, except for
x86. Note that the relro segment, comprising of sections at the start
of the data segment, is sized according to how those sections are laid
out. That means the start of the relro segment is fixed relative to
its end. Which also means the start of the data segment must be at a
fixed address mod MAXPAGESIZE. So for relro the linker can't play
games with the start of the data segment to save disk space. At
least, not without introducing gaps between the relro sections. In
fact, because the linker was starting layout using its simple
heuristic of starting the data segment at the end of the text segment
plus one page, it was sometimes introducing page gaps for no reason.
See pr28743.
PR 28824
PR 28734
* ldexp.c (fold_segment_align): When relro, don't adjust up by
offset within page. Set relropagesize.
(fold_segment_relro_end): Align to relropagesize.
* ldexp.h (seg_align_type): Rename pagesize to commonpagesize.
Add relropagesize. Comment.
* ldlang.c (lang_size_segment): Adjust to suit field renaming.
(lang_size_relro_segment_1): Align relro_end using relropagesize.
