The warning about discarded sections in elf_link_input_bfd doesn't
belong there since the code is dealing with symbols. Multiple symbols
in a discarded section will result in multiple identical warnings
about the section. Move the warning to a new function in ldlang.c.
The patch also tidies the warning quoting of section and file names,
consistently using `%pA' and `%pB'. I'm no stickler for one style of
section and file name quoting, but they ought to be consistent within
a warning, eg. see the first one fixed in ldlang.c, and when a warning
is emitted for multiple targets they all ought to use exactly the same
format string to reduce translation work. elf64-ppc.c loses the
build_one_stub errors since we won't get there before hitting the
fatal errors in size_one_stub.
bfd/
* elflink.c (elf_link_input_bfd): Don't warn here about
discarded sections.
* elf32-arm.c (arm_build_one_stub): Use consistent style in
--enable-non-contiguous-regions error.
* elf32-csky.c (csky_build_one_stub): Likewise.
* elf32-hppa.c (hppa_build_one_stub): Likewise.
* elf32-m68hc11.c (m68hc11_elf_build_one_stub): Likewise.
* elf32-m68hc12.c (m68hc12_elf_build_one_stub): Likewise.
* elf32-metag.c (metag_build_one_stub): Likewise.
* elf32-nios2.c (nios2_build_one_stub): Likewise.
* elfnn-aarch64.c (aarch64_build_one_stub): Likewise.
* xcofflink.c (xcoff_build_one_stub): Likewise.
* elf64-ppc.c (ppc_size_one_stub): Likewise.
(ppc_build_one_stub): Delete dead code.
ld/
* ldlang.c (lang_add_section): Use consistent style in
--enable-non-contiguous-regions warnings.
(size_input_section): Likewise.
(warn_non_contiguous_discards): New function.
(lang_process): Call it.
* testsuite/ld-arm/non-contiguous-arm.d: Update.
* testsuite/ld-arm/non-contiguous-arm4.d: Update.
* testsuite/ld-arm/non-contiguous-arm7.d: Add
--enable-non-contiguous-regions-warnings.
* testsuite/ld-arm/non-contiguous-arm7.err: New.
* testsuite/ld-powerpc/non-contiguous-powerpc.d: Update.
* testsuite/ld-powerpc/non-contiguous-powerpc64.d: Update.
this testcase wasn't correctly testing everything, it passed, even
though sections from an excluded file were included. Fixing this
reveals a problem in the new section selector. This fixes that as
well.
walk_wild_file, hence walk_wild_section and walk_wild_section_handler
aren't called with the prefix tree. Hence initialization of the latter
and all potential special cases for it aren't used anymore. That also
removes the need to handler_data[] and some associated helper functions.
So, remove all of that.
Now that we have a list of potentially matching sections per wild
statement we can actually pre-fill that one by going once over all input
sections and match their names against a prefix-tree that points to the
potentially matching wild statements.
So instead of looking at all sections names for each glob for each wild
statement we now look at the sections only once and then only check
against those globs that have a possibility of a match at all (usually
only one or two).
This pushes the whole section selection off the profiles.
The check_relocs callback (and others) might have created new
section behind our back and some of them (e.g. on powerpc the
"linker stubs" .got) need to come in front of all others, despite
being created late (a symptom would be "TOC opt*" failing on powerpc).
This resets all section matches before updating for newly created
sections (i.e. completely rebuilds the matches).
and remember the results. Before this the order of section matching
is basically:
foreach script-wild-stmt S
foreach pattern P of S
foreach inputfile I
foreach section S of I
match S against P
if match: do action for S
And this process is done three or four times: for each top-level call to
walk_wild() or wild(), that is: check_input_sections, lang_gc_sections,
lang_find_relro_sections and of course map_input_to_output_sections.
So we iterate over all sections of all files many many times (for each
glob). Reality is a bit more complicated (some special glob types don't
need the full iteration over all sections, only over all files), but
that's the gist of it.
For future work this shuffles the whole ordering a bit by lazily doing
the matching process and memoizing results, trading a little memory for
a 75% speedup of the overall section selection process.
This lazy resolution introduces a problem with sections added late
that's corrected in the next patch.
there's no reason why the tree-based variant can't always be used
when sorting is required, it merely needs to also support filename
sorting and have a fast path for insertion at end (aka rightmost tree
leaf).
The filename sorting isn't tested anywhere and the only scripttempl
that uses it is avr (for 'SORT(*)(.ctors)'), and I believe even there it
was a mistake. Either way, this adds a testcase for filename sorting as
well.
Then the non-BST based sorting can be simplified to only support
the fast case of no sorting required at all (at the same time renaming
the two variants to _sort and _nosort).
fnmatch is slow, so avoiding it in more cases is good. This implements
a more generic version of match_simple_wild which needs some
pre-processing of patterns. In particular it supports patterns of the
form PREFIX*SUFFIX (where all parts are optional), i.e. a super set of
what's handled now. Most section matchers of this form and hence don't
need any calls to fnmatch anymore.
We retain the implementation of match_simple_wild for the filename
matchers (they aren't called often enough to matter).
BFD_VMA_FMT can't be used in format strings that need to be
translated, because the translation won't work when the type of
bfd_vma differs from the machine used to compile .pot files. We've
known about this for a long time, but patches slip through review.
So just get rid of BFD_VMA_FMT, instead using the appropriate PRId64,
PRIu64, PRIx64 or PRIo64 and SCN variants for scanf. The patch is
mostly mechanical, the only thing requiring any thought is casts
needed to preserve PRId64 output from bfd_vma values, or to preserve
one of the unsigned output formats from bfd_signed_vma values.
Commit f493c2174e messed the formatting in linker map files,
particularly for 32-bit builds where a number of tests using map files
regressed. I should have noticed the BFD64 conditional printing of
spaces to line up output due to the original %V printing hex vmas with
16 digits when BFD64 and 8 digits when not. Besides that, it is nicer
to print 32-bit vmas for 32-bit targets. So change %V back to be
target dependent, now using bfd_sprintf_vma. Since minfo doesn't
return the number of chars printed, that means some places that
currently use %V must instead sprintf to a buffer in order to find the
length printed.
* ldmisc.h (print_spaces): Declare.
(print_space): Change to a macro.
* ldmisc.c (vfinfo): Use bfd_sprintf_vma for %V. Tidy %W case.
(print_space): Delete.
(print_spaces): New function.
* emultempl/aix.em (print_symbol): Use print_spaces.
* ldctor.c (ldctor_build_sets): Likewise.
* ldmain.c (add_archive_element): Likewise.
* ldlang.c (print_one_symbol, lang_print_asneeded): Likewise.
(print_output_section_statement, print_data_statement): Likewise.
(print_reloc_statement, print_padding_statement): Likewise.
(print_assignment): Likewise. Also replace %V printing of vmas
with printing to a buffer in order to properly format output.
(print_input_section, lang_one_common): Likewise.
PR 29389
bfd * bfd.c (BFD_CLOSED_BY_CACHE): New bfd flag.
* cache.c (bfd_cache_delete): Set BFD_CLOSED_BY_DELETE on the
closed bfd.
(bfd_cache_lookup_worker): Clear BFD_CLOSED_BY_DELETE on the newly
reopened bfd.
* opncls.c (bfd_set_filename): Refuse to change the name of a bfd
that has been closed by bfd_cache_delete. Mark changed bfds as
uncacheable.
* bfd-in2.h: Regenerate.
ld * ldlang.h (lang_input_statement_struct): Add sort_key field.
* emultempl/pe.em (after_open): If multiple import libraries refer
to the same bfd, store their names in the sort_key field.
* emultempl/pep.em (after_open): Likewise.
* ldlang.c (sort_filename): New function. Returns the filename to
be used when sorting input files.
(wild_sort): Use the sort_filename function.
These two macros print either a 16 digit hex number or an 8 digit
hex number. Unfortunately they depend on both target and host, which
means that the output for 32-bit targets may be either 8 or 16 hex
digits.
Replace them in most cases with code that prints a bfd_vma using
PRIx64. In some cases, deliberately lose the leading zeros.
This change some output, notably in base/offset fields of m68k
disassembly which I think looks better that way, and in error
messages. I've kept leading zeros in symbol dumps (objdump -t)
and in PE header dumps.
bfd/
* bfd-in.h (fprintf_vma, sprintf_vma, printf_vma): Delete.
* bfd-in2.h: Regenerate.
* bfd.c (bfd_sprintf_vma): Don't use sprintf_vma.
(bfd_fprintf_vma): Don't use fprintf_vma.
* coff-rs6000.c (xcoff_reloc_type_tls): Don't use sprintf_vma.
Instead use PRIx64 to print bfd_vma values.
(xcoff_ppc_relocate_section): Likewise.
* cofflink.c (_bfd_coff_write_global_sym): Likewise.
* mmo.c (mmo_write_symbols_and_terminator): Likewise.
* srec.c (srec_write_symbols): Likewise.
* elf32-xtensa.c (print_r_reloc): Similarly for fprintf_vma.
* pei-x86_64.c (pex64_dump_xdata): Likewise.
(pex64_bfd_print_pdata_section): Likewise.
* som.c (som_print_symbol): Likewise.
* ecoff.c (_bfd_ecoff_print_symbol): Use bfd_fprintf_vma.
opcodes/
* dis-buf.c (perror_memory, generic_print_address): Don't use
sprintf_vma. Instead use PRIx64 to print bfd_vma values.
* i386-dis.c (print_operand_value, print_displacement): Likewise.
* m68k-dis.c (print_base, print_indexed): Likewise.
* ns32k-dis.c (print_insn_arg): Likewise.
* ia64-gen.c (_opcode_int64_low, _opcode_int64_high): Delete.
(opcode_fprintf_vma): Delete.
(print_main_table): Use PRIx64 to print opcode.
binutils/
* od-macho.c: Replace all uses of printf_vma with bfd_printf_vma.
* objcopy.c (copy_object): Don't use sprintf_vma. Instead use
PRIx64 to print bfd_vma values.
(copy_main): Likewise.
* readelf.c (CHECK_ENTSIZE_VALUES): Likewise.
(dynamic_section_mips_val): Likewise.
(print_vma): Don't use printf_vma. Instead use PRIx64 to print
bfd_vma values.
(dump_ia64_vms_dynamic_fixups): Likewise.
(process_version_sections): Likewise.
* rddbg.c (stab_context): Likewise.
gas/
* config/tc-i386.c (offset_in_range): Don't use sprintf_vma.
Instead use PRIx64 to print bfd_vma values.
(md_assemble): Likewise.
* config/tc-mips.c (load_register, macro): Likewise.
* messages.c (as_internal_value_out_of_range): Likewise.
* read.c (emit_expr_with_reloc): Likewise.
* config/tc-ia64.c (note_register_values): Don't use fprintf_vma.
Instead use PRIx64 to print bfd_vma values.
(print_dependency): Likewise.
* listing.c (list_symbol_table): Use bfd_sprintf_vma.
* symbols.c (print_symbol_value_1): Use %p to print pointers.
(print_binary): Likewise.
(print_expr_1): Use PRIx64 to print bfd_vma values.
* write.c (print_fixup): Use %p to print pointers. Don't use
fprintf_vma.
* testsuite/gas/all/overflow.l: Update expected output.
* testsuite/gas/m68k/mcf-mov3q.d: Likewise.
* testsuite/gas/m68k/operands.d: Likewise.
* testsuite/gas/s12z/truncated.d: Likewise.
ld/
* deffilep.y (def_file_print): Don't use fprintf_vma. Instead
use PRIx64 to print bfd_vma values.
* emultempl/armelf.em (gld${EMULATION_NAME}_finish): Don't use
sprintf_vma. Instead use PRIx64 to print bfd_vma values.
* emultempl/pe.em (gld${EMULATION_NAME}_finish): Likewise.
* ldlang.c (lang_map): Use %V to print region origin.
(lang_one_common): Don't use sprintf_vma.
* ldmisc.c (vfinfo): Don't use fprintf_vma or sprintf_vma.
* pe-dll.c (pe_dll_generate_def_file): Likewise.
gdb/
* remote.c (remote_target::trace_set_readonly_regions): Replace
uses of sprintf_vma with bfd_sprintf_vma.
On seeing PR29369 my suspicion was naturally on a recent powerpc64
change, commit 0ab80031430e. Without a reproducer, I spent time
wondering what could have gone wrong, and while I doubt this patch
would have fixed the PR, there are some improvements that can be made
to cater for user silliness.
I also noticed that when -z relro -z now sections are created out of
order, with .got before .plt in the section headers but .got is laid
out at a higher address. That's due to the address expression for
.branch_lt referencing SIZEOF(.got) and so calling init_os (which
creates a bfd section) for .got before the .plt section is created.
Fix that by ignoring SIZEOF in exp_init_os. Unlike ADDR and LOADADDR
which need to reference section vma and lma respectively, SIZEOF can
and does cope with a missing bfd section by returning zero for its
size, which of course is correct.
PR 29369
* ldlang.c (exp_init_os): Don't create a bfd section for SIZEOF.
* emulparams/elf64ppc.sh (OTHER_RELRO_SECTIONS_2): Revise
.branch_lt address to take into account possible user sections
with alignment larger than 8 bytes.
A number of targets use assignments like:
. = DATA_SEGMENT_RELRO_END (SIZEOF (.got.plt) >= 12 ? 12 : 0, .);
(from i386) in linker scripts to put the end of the relro segment past
the header in .got.plt. Examination of testcases like those edited by
this patch instead sees the end of the relro segment being placed at
the start of .got.plt. For the i386 pie1 test:
[ 9] .got.plt PROGBITS 00002000 001000 00000c 04 WA 0 0 4
GNU_RELRO 0x000f90 0x00001f90 0x00001f90 0x00070 0x00070 R 0x1
A map file shows:
.dynamic 0x0000000000001f90 0x70
*(.dynamic)
.dynamic 0x0000000000001f90 0x70 tmpdir/pie1.o
0x0000000000001f90 _DYNAMIC
.got 0x0000000000002000 0x0
*(.got)
.got 0x0000000000002000 0x0 tmpdir/pie1.o
*(.igot)
0x0000000000002ff4 . = DATA_SEGMENT_RELRO_END (., (SIZEOF (.got.plt) >= 0xc)?0xc:0x0)
.got.plt 0x0000000000002000 0xc
*(.got.plt)
.got.plt 0x0000000000002000 0xc tmpdir/pie1.o
0x0000000000002000 _GLOBAL_OFFSET_TABLE_
The DATA_SEGMENT_RELRO_END value in the map file is weird too. All of
this is triggered by SIZEOF (.got.plt) being evaluated wrongly as
zero. Fix it by taking into account the action of
lang_reset_memory_regions during relaxation.
* ldexp.c (fold_name <SIZEOF>): Use rawsize if size has been reset.
* ldlang.c (lang_size_sections_1): Don't reset processed_vma here.
* testsuite/ld-i386/pie1.d: Adjust to suit.
* testsuite/ld-x86-64/pr20830a.d: Likewise.
* testsuite/ld-x86-64/pr20830b.d: Likewise.
* testsuite/ld-x86-64/pr21038a.d: Likewise.
* testsuite/ld-x86-64/pr21038b.d: Likewise.
* testsuite/ld-x86-64/pr21038c.d: 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.
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.
This is just a tidy, making the __ehdr_start symbol flag tweaks all in
one place.
* ldelf.c (ldelf_before_allocation): Don't set rel_from_abs
for __ehdr_start.
* ldlang.c (lang_symbol_tweaks): Set it here instead.
1. Compute the desired PT_GNU_RELRO segment base and find the maximum
section alignment of sections starting from the PT_GNU_RELRO segment.
2. Find the first preceding load section.
3. Don't add the 1-page gap between the first preceding load section and
the relro segment if the maximum page size >= the maximum section
alignment. Align the PT_GNU_RELRO segment first. Subtract the maximum
page size if therer is still a 1-page gap.
PR ld/28743
PR ld/28819
* ldlang.c (lang_size_relro_segment_1): Rewrite.
* testsuite/ld-x86-64/pr28743-1.d: New file.
* testsuite/ld-x86-64/pr28743-1.s: Likewise.
* testsuite/ld-x86-64/x86-64.exp: Run pr28743-1.
bfd/
* elf-bfd.h (UNDEFWEAK_NO_DYNAMIC_RELOC): Test linker_def.
ld/
* ldelf.c (ldelf_before_allocation): Don't force __ehdr_start
local and hidden here..
* ldlang.c (lang_symbol_tweaks): ..do so here instead and set
def_regular and linker_def for check_relocs. New function
extracted from lang_process.
Add a -z pack-relative-relocs option to enable DT_RELR and create a
relr.dyn section for DT_RELR. DT_RELR is implemented with the linker
relaxation infrastructure, but it doesn't require the --relax option
enabled. -z pack-relative-relocs implies -z combreloc. -z nocombreloc
implies -z nopack-relative-relocs.
-z pack-relative-relocs is chosen over the similar option in lld,
--pack-dyn-relocs=relr, to implement a glibc binary lockout mechanism
with a special glibc version symbol, to avoid random crashes of DT_RELR
binaries with the existing glibc binaries.
bfd/
* elf-bfd.h (elf_link_hash_table): Add srelrdyn.
* elflink.c (_bfd_elf_link_create_dynamic_sections): Create a
.relr.dyn section for DT_RELR.
include/
* bfdlink.h (bfd_link_info): Add enable_dt_relr.
ld/
* News: Mention -z pack-relative-relocs and
-z nopack-relative-relocs.
* ld.texi: Document -z pack-relative-relocs and
-z nopack-relative-relocs.
* ldelf.c (ldelf_after_parse): Disable DT_RELR if not building
PIE nor shared library. Add 3 spare dynamic tags for DT_RELR,
DT_RELRSZ and DT_RELRENT.
* ldlang.c (lang_relax_sections): Also enable relaxation if
DT_RELR is enabled.
* emulparams/elf32_x86_64.sh: Source dt-relr.sh.
* emulparams/elf_i386.sh: Likewise.
* emulparams/elf_x86_64.sh: Likewise.
* emulparams/dt-relr.sh: New file.
* scripttempl/elf.sc: Support .relr.dyn.
For the sake of DT_RELR.
bfd/
* elflink.c (elf_link_input_bfd): Don't set SEC_ELF_REVERSE_COPY
here. Move sanity checks to reverse copying code.
ld/
* ldlang.c (lang_add_section): Set SEC_ELF_REVERSE_COPY for
.ctors/.dtors in .init_array/.fini_array.
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.
It's not foolproof, for example we don't catch output to a linker
script, to a library specified with -l, or to an element of a thin
archive.
* ldlang.c (open_output): Exit with error on output file matching
an input file.
* testsuite/ld-misc/just-symbols.exp: Adjust ld -r test to suit.
* ld.texi: Document new output section type.
* ldgram.y: Add new token.
* ldlang.c: Handle the new flag.
* ldlang.h: Add readonly_section to list of section types.
* ldlex.l: Add a new identifier.
* testsuite/ld-scripts/output-section-types.t: New example linker script.
* testsuite/ld-scripts/output-section-types.d: Test driver.
* testsyute/ld-scripts/script.exp: Run the new test.
It's not enough to test that the output is ELF before casting
bfd_link_hash_entry to elf_link_hash_entry. Some ELF targets (d30v,
dlx, pj, s12z, xgate) use the generic linker support in bfd/linker.c
and thus their symbols are of type generic_link_hash_entry.
Not all of the places this patch touches can result in wrong accesses,
but I thought it worth ensuring that all occurrences of
elf_link_hash_entry in ld/ were obviously correct.
PR 27719
* ldlang.c (lang_mark_undefineds, undef_start_stop): Test that
the symbol hash table is the correct type before accessing
elf_link_hash_entry symbols.
* plugin.c (is_visible_from_outside): Likewise.
* emultempl/armelf.em (ld${EMULATION_NAME}_finish): Likewise.
* emultempl/solaris2.em (elf_solaris2_before_allocation): Likewise.
The exception for debug sections in clearing SEC_EXCLUDE when
relocatable was really for one specific debug section, so let's make
it do just that.
bfd/
PR 27590
* elf.c (_bfd_elf_make_section_from_shdr): Remove SHF_EXCLUDE
test for .gnu.debuglto*.
ld/
PR 27590
* ldlang.c (lang_gc_sections): Clear SEC_EXCLUDE when relocatable
for all sections except .stabstr.
The testcases added here show situations where synthesized start/stop
symbols don't cause their associated input sections to be marked.
Fixed with the elflink.c and ldlang.c changes.
bfd/
PR 27500
* elflink.c (_bfd_elf_gc_mark_rsec): Do special start/stop
processing not when start/stop symbol section is unmarked but
on first time a start/stop symbol is processed.
ld/
* ldlang.c (insert_undefined): Don't mark symbols here.
(lang_mark_undefineds): Do so here instead, new function.
(lang_process): Call lang_mark_undefineds.
* testsuite/ld-gc/start3.d,
* testsuite/ld-gc/start3.s: New test.
* testsuite/ld-gc/start4.d,
* testsuite/ld-gc/start4.s: New test.
* testsuite/ld-gc/gc.exp: Run them.
If a weak reference to a __start_foo or __stop_foo symbol ends up
having no definition due to all the foo sections being removed for
some reason, undef_start_stop currently makes the symbol strong
undefined. That risks a linker undefined symbol error. Fix that by
making the symbol undefweak and also undo some dynamic symbol state.
Note that saving the state of the symbol type at the time
lang_init_start_stop runs is not sufficient. The linker may have
merged in a shared library reference by that point and made what was
an undefweak in regular objects, a strong undefined. So it is
necessary to look at the ELF symbol flags to decide whether an
undefweak is the proper resolution.
Something probably should be done for COFF/PE too, but I'm unsure how
to do go about that.
* ldlang.c (undef_start_stop): For ELF make undefined start/stop
symbols undefweak if that was how they were referenced. Undo
dynamic state too.
The variable section in a CTF dict is meant to contain the types of
variables that do not appear in the symbol table (mostly file-scope
static declarations). We implement this by having the compiler emit
all potential data symbols into both sections, then delete those
symbols from the variable section that correspond to data symbols the
linker has reported.
Unfortunately, the check for this in ctf_serialize is wrong: rather than
checking the set of linker-reported symbols, we check the set of names
in the data object symtypetab section: if the linker has reported no
symbols at all (usually if ld -r has been run, or if a non-linker
program that does not use symbol tables is calling ctf_link) this will
include every single symbol, emptying the variable section completely.
Worse, when ld -r is in use, we want to force writeout of every
symtypetab entry on the inputs, in an indexed section, whether or not
the linker has reported them, since this isn't a final link yet and the
symbol table is not finalized (and may grow more symbols than the linker
has yet reported). But the check for this is flawed too: we were
relying on ctf_link_shuffle_syms not having been called if no symbols
exist, but that function is *always* called by ld even when ld -r is in
use: ctf_link_add_linker_symbol is the one that's not called when there
are no symbols.
We clearly need to rethink this. Using the emptiness of the set of
reported symbols as a test for ld -r is just ugly: the linker already
knows if ld -r is underway and can just tell us. So add a new linker
flag CTF_LINK_NO_FILTER_REPORTED_SYMS that is set to stop the linker
filtering the symbols in the symtypetab sections using the set that the
linker has reported: use the presence or absence of this flag to
determine whether to emit unindexed symtabs: we only remove entries from
the variable section when filtering symbols, and we only remove them if
they are in the reported symbol set, fixing the case where no symbols
are reported by the linker at all.
(The negative sense of the new CTF_LINK flag is intentional: the common
case, both for ld and for simple tools that want to do a ctf_link with
no ELF symbol table in sight, is probably to filter out symbols that no
linker has reported: i.e., for the simple tools, all of them.)
There's another wrinkle, though. It is quite possible for a non-linker
to add symbols to a dict via ctf_add_*_sym and then write it out via the
ctf_write APIs: perhaps it's preparing a dict for a later linker
invocation. Right now this would not lead to anything terribly
meaningful happening: ctf_serialize just assumes it was called via
ctf_link if symbols are present. So add an (internal-to-libctf) flag
that indicates that a writeout is happening via ctf_link_write, and set
it there (propagating it to child dicts as needed). ctf_serialize can
then spot when it is not being called by a linker, and arrange to always
write out an indexed, sorted symtypetab for fastest possible future
symbol lookup by name in that case. (The writeouts done by ld -r are
unsorted, because the only thing likely to use those symtabs is the
linker, which doesn't benefit from symtypetab sorting.)
Tests added for all three linking cases (ld -r, ld -shared, ld), with a
bit of testsuite framework enhancement to stop it unconditionally
linking the CTF to be checked by the lookup program with -shared, so
tests can now examine CTF linked with -r or indeed with no flags at all,
though the output filename is still foo.so even in this case.
Another test added for the non-linker case that endeavours to determine
whether the symtypetab is sorted by examining the order of entries
returned from ctf_symbol_next: nobody outside libctf should rely on
this ordering, but this test is not outside libctf :)
include/ChangeLog
2021-01-26 Nick Alcock <nick.alcock@oracle.com>
* ctf-api.h (CTF_LINK_NO_FILTER_REPORTED_SYMS): New.
ld/ChangeLog
2021-01-26 Nick Alcock <nick.alcock@oracle.com>
* ldlang.c (lang_merge_ctf): Set CTF_LINK_NO_FILTER_REPORTED_SYMS
when appropriate.
libctf/ChangeLog
2021-01-27 Nick Alcock <nick.alcock@oracle.com>
* ctf-impl.c (_libctf_nonnull_): Add parameters.
(LCTF_LINKING): New flag.
(ctf_dict_t) <ctf_link_flags>: Mention it.
* ctf-link.c (ctf_link): Keep LCTF_LINKING set across call.
(ctf_write): Likewise, including in child dictionaries.
(ctf_link_shuffle_syms): Make sure ctf_dynsyms is NULL if there
are no reported symbols.
* ctf-create.c (symtypetab_delete_nonstatic_vars): Make sure
the variable has been reported as a symbol by the linker.
(symtypetab_skippable): Mention relationship between SYMFP and the
flags.
(symtypetab_density): Adjust nonnullity. Exit early if no symbols
were reported and force-indexing is off (i.e., we are doing a
final link).
(ctf_serialize): Handle the !LCTF_LINKING case by writing out an
indexed, sorted symtypetab (and allow SYMFP to be NULL in this
case). Turn sorting off if this is a non-final link. Only delete
nonstatic vars if we are filtering symbols and the linker has
reported some.
* testsuite/libctf-regression/nonstatic-var-section-ld-r*:
New test of variable and symtypetab section population when
ld -r is used.
* testsuite/libctf-regression/nonstatic-var-section-ld-executable.lk:
Likewise, when ld of an executable is used.
* testsuite/libctf-regression/nonstatic-var-section-ld.lk:
Likewise, when ld -shared alone is used.
* testsuite/libctf-regression/nonstatic-var-section-ld*.c:
Lookup programs for the above.
* testsuite/libctf-writable/symtypetab-nonlinker-writeout.*: New
test, testing survival of symbols across ctf_write paths.
* testsuite/lib/ctf-lib.exp (run_lookup_test): New option,
nonshared, suppressing linking of the SOURCE with -shared.
This changes the initialisation of output sections so that it is
possible to create read-only sections fed only from linker script
BYTE, SHORT, LONG or QUAD. That currently isn't possible even for one
of the well-known ELF sections like .rodata, because once a section is
marked read/write that sticks. On the other hand if we start
read-only, well-known ELF sections end up read/write as appropriate.
For example .tdata will still be SHF_ALLOC + SHF_WRITE + SHF_TLS.
PR 26378
* ldlang.c (map_input_to_output_sections): Start with a read-only
section for data statements.
* testsuite/ld-elf/size-2.d: Adjust to suit.
This moves the SHF_LINK_ORDER sorting from bfd_elf_final_link to
the linker which means generic ELF targets now support SHF_LINK_ORDER
and we cope with odd cases that require resizing of output sections.
The patch also fixes two bugs in the current implementation,
introduced by commit cd6d537c48. The pattern test used by that
commit meant that sections matching something like
"*(.IA_64.unwind* .gnu.linkonce.ia64unw.*)" would not properly sort a
mix of sections matching the two wildcards. That commit also assumed
a stable qsort.
bfd/
PR 27160
* section.c (struct bfd_section): Remove pattern field.
(BFD_FAKE_SECTION): Adjust to suit.
* bfd-in2.h: Regenerate.
* elflink.c (compare_link_order, elf_fixup_link_order): Delete.
(bfd_elf_final_link): Don't call elf_fixup_link_order.
ld/
PR 27160
* ldlang.h (lang_output_section_statement_type): Add data field.
(lang_input_section_type, lang_section_bst_type): Add pattern field.
(statement_list): Declare.
(lang_add_section): Adjust prototype.
* emultempl/aarch64elf.em: Adjust lang_add_section calls.
* emultempl/armelf.em: Likewise.
* emultempl/beos.em: Likewise.
* emultempl/cskyelf.em: Likewise.
* emultempl/hppaelf.em: Likewise.
* emultempl/m68hc1xelf.em: Likewise.
* emultempl/metagelf.em: Likewise.
* emultempl/mipself.em: Likewise.
* emultempl/mmo.em: Likewise.
* emultempl/msp430.em: Likewise.
* emultempl/nios2elf.em: Likewise.
* emultempl/pe.em: Likewise.
* emultempl/pep.em: Likewise.
* emultempl/ppc64elf.em: Likewise.
* emultempl/spuelf.em: Likewise.
* emultempl/vms.em: Likewise.
* ldelf.c: Likewise.
* ldelfgen.c: Include ldctor.h.
(struct os_sections): New.
(add_link_order_input_section, link_order_scan): New functions.
(compare_link_order, fixup_link_order): New functions.
(ldelf_map_segments): Call link_order_scan and fixup_link_order.
* ldlang.c (statement_list): Make global.
(output_section_callback_fast): Save pattern in tree node.
(lang_add_section): Add pattern parameter, save in lang_input_section.
(output_section_callback_tree_to_list): Adjust lang_add_section calls.
(lang_insert_orphan, output_section_callback): Likewise.
(ldlang_place_orphan): Likewise.
(gc_section_callback): Don't set section->pattern
* testsuite/ld-elf/pr26256-2a.d: Don't xfail generic.
* testsuite/ld-elf/pr26256-3b.d: Likewise.
* testsuite/ld-elf/pr26256-2b.d: Likewise. notarget xgate.
For non-relocatable link with SHF_LINK_ORDER inputs, allow mixed indirect
and data inputs with ordered and unordered inputs:
1. Add pattern to bfd_section for the matching section name pattern in
linker script and update BFD_FAKE_SECTION.
2. Sort the consecutive bfd_indirect_link_order sections with the same
pattern to allow linker script to overdide input section order.
3. Place unordered sections before ordered sections.
4. Change the offsets of the indirect input sections only.
bfd/
PR ld/26256
* elflink.c (compare_link_order): Place unordered sections before
ordered sections.
(elf_fixup_link_order): Add a link info argument. Allow mixed
ordered and unordered input sections for non-relocatable link.
Sort the consecutive bfd_indirect_link_order sections with the
same pattern. Change the offsets of the bfd_indirect_link_order
sections only.
(bfd_elf_final_link): Pass info to elf_fixup_link_order.
* section.c (bfd_section): Add pattern.
(BFD_FAKE_SECTION): Initialize pattern to NULL.
* bfd-in2.h: Regenerated.
gas/
PR ld/26256
* config/obj-elf.c (obj_elf_change_section): Also filter out
SHF_LINK_ORDER.
ld/
PR ld/26256
* ldlang.c (gc_section_callback): Set pattern.
* testsuite/ld-elf/pr26256-1.s: New file.
* testsuite/ld-elf/pr26256-1.t: Likewise.
* testsuite/ld-elf/pr26256-1a.d: Likewise.
* testsuite/ld-elf/pr26256-1b.d: Likewise.
* testsuite/ld-elf/pr26256-2.s: Likewise.
* testsuite/ld-elf/pr26256-2.t: Likewise.
* testsuite/ld-elf/pr26256-2a.d: Likewise.
* testsuite/ld-elf/pr26256-2b-alt.d: Likewise.
* testsuite/ld-elf/pr26256-2b.d: Likewise.
* testsuite/ld-elf/pr26256-3.s: Likewise.
* testsuite/ld-elf/pr26256-3a.d: Likewise.
* testsuite/ld-elf/pr26256-3a.t: Likewise.
* testsuite/ld-elf/pr26256-3b.d: Likewise.
* testsuite/ld-elf/pr26256-3b.t: Likewise.
The failure on this PR is due to using the same bfd section for
multiple output sections. Commit 21401fc7bf managed to create
duplicate linker script output section statements, but not the actual
bfd sections.
PR 27100
* ldlang.h (lang_output_section_statement_type): Add dup_output.
* ldlang.c (lang_output_section_statement_lookup): Set dup_output.
(init_os): Test dup_output rather than constraint.
* testsuite/ld-scripts/pr27100.d,
* testsuite/ld-scripts/pr27100.s,
* testsuite/ld-scripts/pr27100.t: New test.
* testsuite/ld-scripts/data.exp: Run it. Don't exclude aout here.
* testsuite/ld-scripts/data.d: Do so here instead.
* testsuite/ld-scripts/fill.d: Likewise.
* testsuite/ld-scripts/fill16.d: Likewise.
Previously, ld merged duplicate output sections if such existed in
scripts, except for those with a constraint of SPECIAL. This makes
scripts with duplicate output section statements create duplicate
output sections in the linker output file.
* ldlang.c (lang_output_section_statement_lookup): Change "create"
parameter to a tristate, if 2 then always create a new output
section statement. Update all callers, with
lang_enter_output_section_statement using "2".
(map_input_to_output_sections): Don't ignore SPECIAL constraint
here.
* ldlang.h (lang_output_section_statement_type): Update prototype.
(lang_output_section_find): Update.
