PC-relative relocs typically use the addend in adjusting what they are
relative to. For example:
bcl 20,31,1f
1: mflr 12
addi 12,12,xxx-1b
generates "R_PPC64_REL16 xxx+0x4" for the addi (when little-endian).
The addend reflects the fact that you want the offset relative to the
previous insn not the current one in this case.
So the question is, will we ever want to do something like that for an
instruction using R_PPC64_GOT_PCREL34? I thought so at the time I
first implemented support in ld but at the time I think the hardware
was possibly going to support pcrel+offset+reg addressing. In which
case you might want something like:
load_big_offset_into_r2
pld 3,sym-big_offset@got@pcrel(2)
which would be a way of supporting more than 8G offsets from code to
the GOT. We could do the same with
load_big_offset_into_r2
pla 9,sym-big_offset@got@pcrel
ldx 3,9,2
However, this is really a poor version of TOC-pointer relative code.
So let's go with an addend on R_PPC64_GOT_PCREL34 meaning that
sym+addend should be put in a GOT entry, and the relocation calculate
the pc-relative offset to that GOT entry.
Note that this is an extension to the ABI, which says (by the
expression given for GOT relocs) that non-zero addends on GOT and PLT
relocs are ignored. This is true for all GOT/PLT relocs, not just the
pcrel ones.
* elf64-ppc.c (ppc64_elf_check_relocs): Interpret an addend in
GOT_PCREL and PLT_PCREL relocs as affecting the value stored
in the GOT/PLT entry rather than affecting the offset to that
GOI/PLT entry.
(ppc64_elf_edit_toc, ppc64_elf_relocate_section): Likewise.
The loads and stores handled in the second instruction of a sequence
marked by R_PPC64_PCREL_OPT may be a prefix instruction. For example:
pld ra,symbol@got@pcrel
0:
pld rt,off(ra)
.reloc 0b-8,R_PPC64_PCREL_OPT,(.-8)-(0b-8)
can be optimised to
pld rt,symbol+off@pcrel
pnop
* elf64-ppc.c (xlate_pcrel_opt): Handle prefix loads and stores
in second instruction.
(ppc64_elf_relocate_section): Likewise.
This patch fixes the worst of the cases where libbfd might terminate
a program due to calling xstrdup or xmalloc. I've also fixed some
error paths that didn't clean up properly.
PR 24955
* libbfd-in.h (bfd_strdup): New inline function.
* archive.c (_bfd_get_elt_at_filepos): Use bfd_strdup. Close
bfd on error.
* elfcode.h (_bfd_elf_bfd_from_remote_memory): Use bfd_strdup.
* opncls.c (bfd_fopen): Use bfd_strdup. Close fd and stream
on error.
(bfd_openstreamr): Use bfd_strdup.
(bfd_openr_iovec, bfd_openw, bfd_create): Likewise.
* plugin.c (try_load_plugin): Use bfd_malloc.
* libbfd.h: Regenerate.
Once the executable is started, info variables can show thousands
of variables belonging to glibc (see below).
This long list of variables then causes the test to fail, due
to expect's buffer overflow:
Running /bd/home/philippe/gdb/git/build_binutils-gdb/gdb/testsuite/../../../binutils-gdb/gdb/testsuite/gdb.base/info-var.exp ...
ERROR: internal buffer is full.
ERROR: internal buffer is full.
ERROR: internal buffer is full.
Fix this by testing 'info variables' without running the executable.
gdb ./info-var
...
Reading symbols from ./info-var...
(gdb) |info variables|wc
27 54 971
(gdb) start
Temporary breakpoint 1 at 0x1129: file /bd/home/philippe/gdb/git/build_binutils-gdb/gdb/testsuite/../../../binutils-gdb/gdb/testsuite/gdb.base/info-var-f1.c, line 23.
Starting program: /bd/home/philippe/gdb/git/build_binutils-gdb/gdb/testsuite/outputs/gdb.base/info-var/info-var
Temporary breakpoint 1, main ()
at /bd/home/philippe/gdb/git/build_binutils-gdb/gdb/testsuite/../../../binutils-gdb/gdb/testsuite/gdb.base/info-var-f1.c:23
23 return global_var + get_offset() + f1_var;
(gdb) |info variables|wc
4334 14581 130738
(gdb)
gdb/testsuite/ChangeLog
2019-09-04 Philippe Waroquiers <philippe.waroquiers@skynet.be>
* gdb.base/info-var.exp: Test info variables without running
to main, to avoid expect's buffer overflow.
Some domain_enum values were not handled in print_partial_symbols
which means that 'maintenance print psymbols' would print '<invalid
domain>' when it shouldn't have.
gdb/ChangeLog:
* psymtab.c (print_partial_symbols): Handle missing domain_enum
values MODULE_DOMAIN and COMMON_BLOCK_DOMAIN.
In Ada, the programmer can request that a range type with a non-zero
base be stored in the minimal number of bits required for the range.
This is done by biasing the values; so, for example, a range of -7..-4
may be stored as two bits with a bias of -7.
This patch implements this for gdb. It is done by adding a bias to
struct range_bounds and then adjusting a few spots to handle this.
The test case is written to use -fgnat-encodings=minimal, but a future
compiler patch will change the compiler to emit DW_AT_GNU_bias with
-fgnat-encodings=gdb. It seemed good to get the gdb patch in first.
Tested on x86-64 Fedora 29; plus a variety of targets using AdaCore's
internal test suite.
gdb/ChangeLog
2019-09-03 Tom Tromey <tromey@adacore.com>
* ada-valprint.c (ada_val_print_num): Don't recurse for range
types.
(has_negatives): Unbias a range type bound.
* dwarf2read.c (read_subrange_type): Handle DW_AT_GNU_bias.
* gdbtypes.c (operator==): Handle new field.
(create_range_type): Add "bias" parameter.
(create_static_range_type, resolve_dynamic_range): Update.
* gdbtypes.h (struct range_bounds) <bias>: New member.
(create_range_type): Add bias parameter.
* printcmd.c (print_scalar_formatted): Unbias range types.
* value.c (unpack_long): Unbias range types.
(pack_long): Bias range types.
gdb/testsuite/ChangeLog
2019-09-03 Tom Tromey <tromey@adacore.com>
* gdb.ada/bias.exp: New file.
* gdb.ada/bias/bias.adb: New file.
* gdb.ada/print_chars.exp: Add regression test.
* gdb.ada/print_chars/foo.adb (My_Character): New type.
(MC): New variable.
PR 24958
* mmix-dis.c (MAX_REG_NAME_LEN): Define.
(MAX_SPEC_REG_NAME_LEN): Define.
(struct mmix_dis_info): Use defined constants for array lengths.
(get_reg_name): New function.
(get_sprec_reg_name): New function.
(print_insn_mmix): Use new functions.
There is a long standing bug in the Arm toolchain where invalid
stap probes get created due to the probes referring to symbols which
have been resolved away.
More details are here:
https://bugzilla.redhat.com/show_bug.cgi?id=1196181
When these invalid probes are present, GDB will create the breakpoints
and then fail to stop. The errors are only spotted the first time
GDB stops, which is too late.
The solution is to ensure the arguments for all the probes are
resolved before using them.
This fixes >100 timeouts when running break-interp.exp when using
bad probes.
gdb/ChangeLog:
* solib-svr4.c (svr4_find_and_create_probe_breakpoints): Check all
probe arguments.
The probe function get_argument_count does not need a frame, only
the current gdbarch. Switch the code to pass gdbarch instead.
No functional changes.
gdb/ChangeLog:
* break-catch-throw.c (fetch_probe_arguments): Use gdbarch.
* dtrace-probe.c (dtrace_probe::get_argument_count): Likewise.
* probe.c (probe_safe_evaluate_at_pc) (compute_probe_arg)
(compile_probe_arg): Likewise.
* probe.h (get_argument_count): Likewise.
* solib-svr4.c (solib_event_probe_action): Likewise.
* stap-probe.c (stap_probe::get_argument_count): Likewise.
Move the bulk of svr4_create_solib_event_breakpoints into a new
function to simplify the logic. No functional changes.
gdb/ChangeLog:
* solib-svr4.c (svr4_find_and_create_probe_breakpoints): Move
code to here...
(svr4_create_solib_event_breakpoints): ...from here.
This fixes a small leak of debug_filename. bfd_openr copies the file
name since git commit 1be5090bca.
PR 11983
* dwarf2.c (_bfd_dwarf2_slurp_debug_info): Free debug_filename
on success. Tidy.
The linker doesn't allocate memory space for sections that are only SEC_ALLOC
and SEC_THREAD_LOCAL. See the IS_TBSS test in ld/ldlang.c. So we need to
pretend that .tdata.dyn sections have contents to get the right result. It
will be marked this way anyways if there is a .tdata section to merge with.
bfd/
PR 23825
* elfnn-riscv.c (riscv_elf_create_dynamic_sections): Add SEC_LOAD,
SEC_DATA, and SEC_HAS_CONTENTS to .tdata.dyn section.
This was noticed while trying to test the compiler -msave-restore support.
Putting non-pic code in a shared library gives a linker error, but doesn't
stop the build.
rohan:2030$ cat libtmp.c
extern int sub2 (int);
int sub (int i) { return sub2 (i + 10); }
rohan:2031$ cat libtmp2.c
extern int sub (int);
int sub2 (int i) { return sub (i + 10); }
rohan:2032$ riscv64-unknown-linux-gnu-gcc --shared -o libtmp.so libtmp.c
rohan:2033$ riscv64-unknown-linux-gnu-gcc --shared -o libtmp2.so libtmp2.c libtmp.so
/home/jimw/FOSS/install-riscv64/lib/gcc/riscv64-unknown-linux-gnu/8.3.0/../../../../riscv64-unknown-linux-gnu/bin/ld: /tmp/cctrsIBe.o(.text+0x18): unresolvable R_RISCV_CALL relocation against symbol `sub'
rohan:2034$ echo $?
0
rohan:2035$ ls -lt libtmp2.so
-rwxr-xr-x 1 jimw jimw 6912 Aug 30 14:32 libtmp2.so
rohan:2036$
The patch fixes this by forcing a linker error. I now get this.
ohan:2059$ sh tmp.script
/home/jimw/FOSS/BINUTILS/X-riscv64-linux/ld/ld-new: libtmp2.o(.text+0x18): unresolvable R_RISCV_CALL relocation against symbol `sub'
/home/jimw/FOSS/BINUTILS/X-riscv64-linux/ld/ld-new: final link failed: bad value
rohan:2060$ echo $?
1
rohan:2061$ ls -lt libtmp2.so
ls: cannot access 'libtmp2.so': No such file or directory
bfd/
* elfnn-riscv.c (riscv_elf_relocate_section): For unresolvable reloc
error, call bfd_set_error, set ret to FALSE, and goto out label.
Rationale: https://sourceware.org/ml/gdb-patches/2019-08/msg00651.html
This very simple patch removes the "\nError: " suffix from the warning
message printed by nat/fork-inferior.c:trace_start_error. This proved
to just pollute the screen, causing things like:
Starting program: /usr/bin/true
warning: Could not trace the inferior process.
Error:
warning: ptrace: Permission denied
This "Error: " string is not useful at all, and can confuse things,
therefore let's just remove it and simplify the resulting messages:
Starting program: /usr/bin/true
warning: Could not trace the inferior process.
warning: ptrace: Permission denied
gdb/ChangeLog:
2019-08-30 Sergio Durigan Junior <sergiodj@redhat.com>
* nat/fork-inferior.c (trace_start_error): Remove "\nError: "
suffix from warning message.
The TUI has two duplicate "re-render this window" methods, "rerender"
and "refresh_all". They differ only slightly in semantics, so I
wanted to see if they could be unified.
After looking into this, I decided that refresh_all was not needed.
There are 4 calls to tui_refresh_all_win (the only caller of this
method):
1. tui_enable. This sets the layout, which renders the windows.
2. tui_cont_sig. Here, I think it's sufficient to simply redraw the
current window contents from the curses backing store, because gdb
state didn't change while it was suspended
3. tui_dispatch_ctrl_char. This is the C-l handler, and here it's
explicitly enough to just refresh the screen (as above).
4. tui_refresh_all_command. This is the command equivalent of C-l.
So, this patch removes this method entirely and simplifies
tui_refresh_all_win.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.h (struct tui_source_window_base)
<refresh_all>: Don't declare.
* tui/tui-winsource.c (tui_source_window_base::refresh_all):
Remove.
* tui/tui-win.c (tui_refresh_all_win): Don't call refresh_all or
tui_show_locator_content.
* tui/tui-regs.h (struct tui_data_window) <refresh_all>: Don't
declare.
* tui/tui-regs.c (tui_data_window::refresh_all): Remove.
* tui/tui-data.h (struct tui_win_info) <refresh_all>: Don't
declare.
tui_cont_sig does not need to call wrefresh, because this is already
done by tui_refresh_all_win.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-io.c (tui_cont_sig): Don't call wrefresh.
This move _initialize_tui_stack to the end of tui-stack.c, per the gdb
style; and then removes two unnecessary forward declarations.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-stack.c (_initialize_tui_stack): Move later.
Remove unnecessary forward declarations.
This changes tui_locator_window::set_locator_fullname to re-render the
locator window, so that the callers don't need to do this.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-stack.c (tui_locator_window::set_locator_fullname): Call
rerender.
(tui_update_locator_fullname, tui_show_frame_info): Don't call
tui_show_locator_content.
This swaps the bodies ot tui_show_locator_content and
tui_locator_window::rerender, so that the latter does the work, and
the former is now just an exported convenience wrapper.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-stack.c (tui_show_locator_content): Move lower. Rewrite.
(tui_locator_window::rerender): Rewrite using body of previous
tui_show_locator_content.
This changes tui_set_locator_fullname and tui_set_locator_info to be
methods on tui_locator_window. This enables some subsequent
cleannups.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-stack.h (struct tui_locator_window) <set_locator_info,
set_locator_fullname>: New methods.
* tui/tui-stack.c (tui_locator_window::set_locator_fullname):
Rename from tui_set_locator_fullname.
(tui_locator_window::set_locator_info): Rename from
tui_set_locator_info. Return bool.
(tui_update_locator_fullname, tui_show_frame_info): Update.
show_layout calls tui_refresh_all in one case. However, it doesn't
need to any more, because the resize method on each window will also
update the contents.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (show_layout): Don't call tui_refresh_all.
The call to touchwin in tui_gen_win_info::refresh_window was an
artifact of some earlier refactorings. Testing shows it isn't needed
any more -- I believe it was only ever needed for the data item window
display problem; but that's been solved more locally.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (tui_gen_win_info::refresh_window): Don't
call touchwin.
box_win can't be called with a NULL window, or with an invisible
window. So, the NULL checks in that function can be removed.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (box_win): Assume win_info and
win_info->handle cannot be NULL.
This patch starts with the observation that the code in
tui_data_window::display_registers_from can all be replaced with a
call to resize. To make this work propertly, it also changes
tui_display_register to be the "rerender" method on
tui_data_item_window.
The refresh_window method is needed due to the use of nested windows
here. The ncurses man page makes it sound like this is not very well
supported; and experience bears this out: negelecting the touchwin
call in this path will cause the register window to blank when
switching focus.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-regs.h (struct tui_data_item_window) <rerender,
refresh_window>: Declare.
* tui/tui-regs.c (tui_data_window::display_registers_from): Call
resize.
(tui_data_item_window::rerender): Rename from
tui_display_register.
(tui_data_item_window::refresh_window): New method.
* tui/tui-layout.c (tui_gen_win_info::resize): Do nothing on
no-op.
This changes tui_data_window so that the data members are private.
This required the addition of a simple accessor method in one case.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-regs.h (struct tui_data_window) <regs_content,
regs_column_count, current_group>: Move later. Now private.
<get_current_group>: New method.
* tui/tui-regs.c (tui_reg_command): Update.
* tui/tui-layout.c (tui_set_layout): Update.
This patch removes a call to erase_data_content in refresh_all and
then removes some other calls that are more clearly unnecessary once
one follows calls from that point.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_data_window::display_registers_from_line)
(tui_data_window::rerender): Don't call
check_and_display_highlight_if_needed.
(tui_data_window::refresh_all): Remove call to
erase_data_content.
A few methods in tui_data_window check whether the contents are empty;
but all the callers already check this, so these calls can be removed.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_data_window::last_regs_line_no)
(tui_data_window::display_registers_from)
(tui_data_window::display_reg_element_at_line)
(tui_data_window::display_registers_from_line): Remove checks of
"empty".
tui_data_window::rerender clears the data item windows, and then calls
display_all_data. However, that method only does anything if the
contents are not empty. So, display_all_data can be renamed and the
wrapper removed.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-regs.h (struct tui_data_window) <display_all_data>:
Don't declare.
* tui/tui-regs.c (tui_data_window::show_registers): Call
rerender.
(tui_data_window::rerender): Rename from display_all_data.
(tui_data_window::rerender): Remove old implementation.
NO_DATA_STRING shouldn't be used. It's referenced in a single spot,
in tui_data_window::display_all_data. This patch removes the use and
replaces it with the more correct text. A later patch (though not in
this series) will remove this call entirely, when it's more obviously
correct to do so.
gdb/ChangeLog
2019-08-30 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_data_window::display_all_data): Change
text.
* tui/tui-data.h (NO_DATA_STRING): Remove define.
We need to copy BFD_COMPRESS, BFD_DECOMPRESS and BFD_COMPRESS_GABI flags
for thin archive.
PR ld/24951
* archive.c (_bfd_get_elt_at_filepos): Copy BFD_COMPRESS,
BFD_DECOMPRESS and BFD_COMPRESS_GABI flags for thin archive.
This patch make changes to the assembler to encode MVE VMOV instruction "a" same as "b".
a: VMOV<c><q> <Dd>, <Dm>
b: VMOV<c><q>.F64 <Dd>, <Dm>
gas/ChangeLog:
2019-08-30 Srinath Parvathaneni <srinath.parvathaneni@arm.com>
* config/tc-arm.c (do_neon_mov): Modify "if" statement.
* testsuite/gas/arm/mve-vmov-bad-3.d: New test.
* testsuite/gas/arm/mve-vmov-bad-3.l: Likewise.
* testsuite/gas/arm/mve-vmov-bad-3.s: Likewise.
Move FASTMATH to the right enum.
2019-08-30 Claudiu Zissulescu <claziss@gmail.com>
* opcode/arc.h (FASTMATH): Move it from insn_class_t to
insn_subclass_t enum.
This testcase was originally for PR gdb/15415, a problem with the
"run" command expanding symlinks in the name of the program being run.
It does not correctly distinguish between files on build, host, and
target, and it is not clear if it would be testing anything useful in
configurations where "run" is not being used.
2019-08-29 Sandra Loosemore <sandra@codesourcery.com>
* gdb.base/argv0-symlink.exp: Run only on native target
and local host.
This was reported by Bernhard Wodok, along with a patch to fix the
issue. I adjusted the patch a bit, and I'm submitting the patch on
his behalf.
According to Bernhard, the issue can be reproduced by doing:
1. start gdb
2. enter 'target remote :2345'
3. observe that it throws a "connection refused" error immediately
instead of waiting and throwing a timeout error
I.e., I believe it can be reproduced by our current tests, which is
why I'm not proposing any extra tests here (well, I don't use nor have
any Windows system to test this, so...).
The problem happens because, on ser-tcp:wait_for_connect, we call
'gdb_select' passing 0 as its first argument, which, when using MinGW,
ends up using the 'gdb_select' version from mingw-hdep.c, and when the
first argument is 0 this means that WaitForMultipleObjects will be
called with 0 as its first argument as well. According to the MS API
docs, this is forbidden:
https://docs.microsoft.com/en-us/windows/win32/api/synchapi/nf-synchapi-waitformultipleobjects
The proposed fix is simple: we just call Sleep when N == 0 (and when
TIMEOUT is non-NULL), and return 0. It makes sense to me.
Both Bernhard and Paul Carroll confirmed that the fix works. I'm
Cc'ing Bernhard in case you have any questions about the patch.
OK?
gdb/ChangeLog:
2019-08-29 Bernhard Wodok <barto@gmx.net>
Sergio Durigan Junior <sergiodj@redhat.com>
PR win32/24284
* mingw-hdep.c (gdb_select): Handle case when 'n' is zero.
The gdb.fortran/info-types.exp test-case passes with gcc 7 (though not on
openSUSE, due to the extra debug info) and fails with gcc 4.8 and gcc 8.
Fix the gdb_test regexp to fix all those cases.
gdb/testsuite/ChangeLog:
2019-08-29 Tom de Vries <tdevries@suse.de>
* gdb.fortran/info-types.exp: Fix gdb_test regexp to allow more
diverse debug info.
* lib/fortran.exp (fortran_int8): New proc, based on fortran_int4.
This removes a restriction on various R_PPC_EMB relocations that has
been present for ppc32 since 1996-04-26 git commit e25a798839. As far
as I know, only those relocs that would require addressing via r2 for
.sdata2/.sbss2 access are disallowed in shared libraries.
PR 24697
* elf32-ppc.c (ppc_elf_check_relocs): Call bad_shared_reloc
when !bfd_link_executable for R_PPC_EMB_SDA2I16 and
R_PPC_EMB_SDA2REL. Don't call bad_shared_reloc for any other
reloc.
We can easily support an offset on the second instruction of a
sequence marked with R_PPC64_PCREL_OPT. For example,
pla ra,symbol@pcrel
ld rt,off(ra)
can be optimised to
pld rt,symbol+off@pcrel
nop
* elf64-ppc.c (xlate_pcrel_opt): Add poff parameter. Allow offset
on second insn, return it in poff.
(ppc64_elf_relocate_section): Add offset to paddi addend for
PCREL_OPT.
This was broken when I changed how we compute the value for the gp register.
It used to be computed inside the sdata section. Now it is computed at the
end which makes it an abs section symbol. There is code that tries to use
the alignment of the section that the gp value is in, but this does not work
if it is in the abs section, as the abs section has alignment of 1 byte.
There are people using alternative linker scripts that still define it in the
sdata section, so the code is still useful. Thus adding a check to disable
this when gp is in the abs section.
bfd/
* elfnn-riscv.c (_bfd_riscv_relax_lui): Add check to exclude abs
section when setting max_alignment. Update comment.
(_bfd_riscv_relax_pc): Likewise.
I don't see a need to calculate "ptr = start + uvalue" then compare
"ptr" with "start" and "end". Given "start <= end" on entry, the
"uvalue" comparison with "max_uvalue" ought to be sufficient to ensure
"start + uvalue" is bounded by "start" and "end" regardless of the
size of pointers and the unsigned dwarf_vma integer type.
* dwarf.c (check_uvalue): Remove unnecessary pointer checks.
BFD was leaking memory in bfd_check_format_matches. As part of
deciding the proper format of an archive, BFD looks at the format of
the first file stored. That file's bfd was left open for reasons
given in a comment removed in git commit 0e71e4955c that said:
/* We ought to close `first' here, but we can't, because
we have no way to remove it from the archive cache.
It's close to impossible to figure out when we can
release bfd_ardata. FIXME. */
Well, things have changed since that comment was true and we now can
remove files from the archive cache. Closing the first file is good
and cures some of the leaks. Other leaks are caused by
bfd_check_format_matches throwing away bfd tdata before trying a new
match. That lost the element cache set up when format checking the
first element in the archive. The easiest and cleanest fix is to
simply disable the caching when checking the first element.
PR 24891
* bfd.c (struct bfd): Add no_element_cache.
* archive.c (_bfd_get_elt_at_filepos): Don't add element to
archive cache when no_element_cache.
(bfd_generic_archive_p): Set no_element_cache when opening first
element to check format. Close first element too.
(do_slurp_bsd_armap): Don't zero ardata->cache here.
* bfd-in2.h: Regenerate.