binutils-gdb.git
4 years agoFix gdb.base/structs2.exp with Clang
Pedro Alves [Thu, 2 Jul 2020 18:32:40 +0000 (19:32 +0100)]
Fix gdb.base/structs2.exp with Clang

gdb.base/structs2.exp fails to run with Clang, because of:

 gdb compile failed, /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/structs2.c:16:14: warning:
       implicit conversion from 'int' to 'signed char' changes value from 130 to
       -126 [-Wconstant-conversion]
   param_reg (130, 120, 33000, 32000);
   ~~~~~~~~~  ^~~
 /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/structs2.c:16:24: warning:
       implicit conversion from 'int' to 'short' changes value from 33000 to
       -32536 [-Wconstant-conversion]
   param_reg (130, 120, 33000, 32000);
   ~~~~~~~~~            ^~~~~
 2 warnings generated.

 === gdb Summary ===

 # of untested testcases         1

Fix it by passing actual negative numbers.

gdb/testsuite/ChangeLog:

* gdb.base/structs2.c (main): Adjust second parem_reg call to
explicitly write negative numbers.
* gdb.base/structs2.exp: Adjust expected output.

4 years agoFix gdb.base/charset.exp with Clang
Pedro Alves [Thu, 2 Jul 2020 14:54:36 +0000 (15:54 +0100)]
Fix gdb.base/charset.exp with Clang

gdb.base/charset.exp fails to run with Clang, because of:

 gdb compile failed, /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/charset.c:144:20: warning:
       implicit conversion from 'int' to 'char' changes value from 162 to -94
       [-Wconstant-conversion]
11, 162, 17);
    ^~~
 /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/charset.c:151:16: warning:
       implicit conversion from 'int' to 'char' changes value from 167 to -89
       [-Wconstant-conversion]
167,
^~~
 /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/charset.c:168:16: warning:
       implicit conversion from 'int' to 'char' changes value from 167 to -89
       [-Wconstant-conversion]
167,
^~~
 3 warnings generated.

 === gdb Summary ===

 # of untested testcases         1

Fix it by changing init_string to take unsigned char parameters.

gdb/testsuite/ChangeLog:

* gdb.base/charset.c (init_string): Change all char parameters to
unsigned char parameters.

4 years agoRemove stale -DNO_PROTOTYPES bits from gdb testsuite
Pedro Alves [Fri, 3 Jul 2020 12:34:22 +0000 (13:34 +0100)]
Remove stale -DNO_PROTOTYPES bits from gdb testsuite

The gdb.base/call-sc.exp, gdb.base/structs.exp and
gdb.base/structs2.exp testcases still try compiling the sources with
-DNO_PROTOTYPES, but the corresponding sources don't have any #ifdef
NO_PROTOTYPES any longer.  Those were removed throughout years ago.

OTOH, gdb.base/ovlymgr.h does check for NO_PROTOTYPES, but no .exp
file compiles it with -DNO_PROTOTYPES.

gdb.base/reread.exp and gdb.base/varargs.exp set a 'prototypes'
global, which is a stale bit left behind when the "try-compiling
without and then with -DNO_PROTOTYPES" logic was around.

gdb/testsuite/ChangeLog:

* gdb.base/call-sc.exp (start_scalars_test): Use
prepare_for_testing and don't try compiling with -DNO_PROTOTYPES.
* gdb.base/overlays.c: Remove references to PARAMS.
* gdb.base/ovlymgr.h (PARAMS): Delete, and remove all references.
* gdb.base/reread.exp: Don't set 'prototypes' global.
* gdb.base/structs.exp (start_structs_test): Use
prepare_for_testing and don't try compiling with -DNO_PROTOTYPES.
* gdb.base/structs2.exp: Don't set 'prototypes' global.  Use
prepare_for_testing and don't try compiling with -DNO_PROTOTYPES.
Don't issue "set width 0".  Remove gdb_stop_suppressing_tests
call.
* gdb.base/varargs.exp: Don't set 'prototypes' global.

4 years agoRemove stale overlay testcase bits
Pedro Alves [Fri, 3 Jul 2020 12:34:21 +0000 (13:34 +0100)]
Remove stale overlay testcase bits

D10V support was removed years ago, but the gdb.base/d10vovly.c file
stayed behind.  Looking a bit closer, I can't find anywhere that
references gdb.base/m32rovly.c either.

Both gdb.base/m32rovly.c and gdb.base/d10vovly.c seem to be older
copies of gdb.base/ovlymgr.c, that are exactly the same, except for
some cosmetic differences, and for missing _ovly_debug_event.  Note
that gdb.base/ovlymgr.c has the #ifdef __M32R__ bits too.  Note also
that gdb.base/overlays.exp is currently only supported on m32r, and
that uses ovlymgr.c not gdb.base/m32rovly.c.

gdb/testsuite/ChangeLog:

* gdb.base/d10vovly.c: Delete.
* gdb.base/m32rovly.c: Delete.
* gdb.base/ovlymgr.c: Remove all code guarded by __D10V__.

4 years agoRe: Change readelf's display of symbol names
Alan Modra [Fri, 3 Jul 2020 05:59:13 +0000 (15:29 +0930)]
Re: Change readelf's display of symbol names

Fixes some fallout from git commit 0942c7ab94e5.

PR 26028
gas/
* testsuite/gas/ia64/unwind-ilp32.d: Add -T to readelf options.
gold/
* testsuite/Makefile.am (file_in_many_sections.stdout): Add -W
to readelf options.
* testsuite/Makefile.in: Regenerate.
ld/
* testsuite/ld-arm/arm-elf.exp (vxworks1): Pass --wide to readelf
when dumping relocs.
* testsuite/ld-i386/i386.exp (vxworks1): Likewise.
* testsuite/ld-sh/sh-vxworks.exp (vxworks1): Likewise.
* testsuite/ld-sparc/sparc.exp (vxworks1): Likewise.
* testsuite/ld-arm/vxworks1.rd: Adjust to suit.
* testsuite/ld-i386/vxworks1.rd: Adjust.
* testsuite/ld-sh/vxworks1.rd: Adjust.
* testsuite/ld-sparc/vxworks1.rd: Adjust.

4 years agoAutomatic date update in version.in
GDB Administrator [Fri, 3 Jul 2020 00:00:07 +0000 (00:00 +0000)]
Automatic date update in version.in

4 years agox86: Add SwapSources
H.J. Lu [Thu, 2 Jul 2020 13:30:00 +0000 (06:30 -0700)]
x86: Add SwapSources

We check register-only source operand to decide if two source operands of
VEX encoded instructions should be swapped.  But source operands in AMX
instructions with two source operands swapped are all register-only
operand.  Add SwapSources to indicate two source operands should be
swapped.

gas/

* config/tc-i386.c (build_modrm_byte): Check vexswapsources to
swap two source operands.

opcodes/

* i386-gen.c (opcode_modifiers): Add VexSwapSources.
* i386-opc.h (VexSwapSources): New.
(i386_opcode_modifier): Add vexswapsources.
* i386-opc.tbl: Add VexSwapSources to BMI2 and BMI instructions
with two source operands swapped.
* i386-tbl.h: Regenerated.

4 years agoSkip fill-1 gas test for MeP targets.
Nick Clifton [Thu, 2 Jul 2020 13:08:16 +0000 (14:08 +0100)]
Skip fill-1 gas test for MeP targets.

* testsuite/gas/all/fill-1.d: Skip for MeP targets.

4 years agoaarch64: Fix segfault on unicode symbols
Alex Coplan [Thu, 2 Jul 2020 12:53:07 +0000 (13:53 +0100)]
aarch64: Fix segfault on unicode symbols

This patch fixes a segfault which occurs when the AArch64 backend parses
a symbol operand that begins with a register name and ends with a
unicode byte (byte value > 127).

For example, the following input causes the crash:

x0é: udf x0é

gas/ChangeLog:

2020-07-02  Alex Coplan  <alex.coplan@arm.com>

* config/tc-aarch64.c (reg_name_p): Fix cast so that we don't
segfault on negative chars.
* testsuite/gas/aarch64/reglike-label-unicode-segv.d: New test.
* testsuite/gas/aarch64/reglike-label-unicode-segv.s: Input.

4 years agogdb: remove unused fetch_inferior_event and inferior_event_handler parameters
Simon Marchi [Thu, 2 Jul 2020 12:37:26 +0000 (08:37 -0400)]
gdb: remove unused fetch_inferior_event and inferior_event_handler parameters

I noticed that fetch_inferior_event receives the client_data parameter
from its caller, inferior_event_handler, but doesn't actually need it.
This patch removes it.  In turn, inferior_event_handler doesn't use its
parameter, so remove it too.

The `data` argument used when registering
remote_async_inferior_event_handler is changed to NULL, to avoid
confusion.  It could make people think that the value passed is used
somewhere, when in fact it's not.

gdb/ChangeLog:

* inf-loop.c (inferior_event_handler): Remove client_data param.
* inf-loop.h (inferior_event_handler): Likewise.
* infcmd.c (step_1): Adjust.
* infrun.c (proceed): Adjust.
(fetch_inferior_event): Remove client_data param.
(infrun_async_inferior_event_handler): Adjust.
* infrun.h (fetch_inferior_event): Remove `void *` param.
* linux-nat.c (handle_target_event): Adjust.
* record-btrace.c (record_btrace_handle_async_inferior_event):
Adjust.
* record-full.c (record_full_async_inferior_event_handler):
Adjust.
* remote.c (remote_async_inferior_event_handler): Adjust.

Change-Id: I3c2aa1eb0ea3e0985df096660d2dcd794674f2ea

4 years agoChange readelf's display of symbol names (when not in --wide mode) so that if they...
Nick Clifton [Thu, 2 Jul 2020 10:30:52 +0000 (11:30 +0100)]
Change readelf's display of symbol names (when not in --wide mode) so that if they are going to be truncated then "[...]" is displayed at the end.  Add a comment line option to disable this enhancement and restore the old behaviour.

PR 26028
binutils* readelf.c (print_symbol): Handle truncation of symbol names.
(options): Add -T/--silent-truncation option.
(parse_args): Handle the option.
(print_dynamic_symbol): Correct calculation of width available to
display symbol name.
* doc/binutils.texi: Document the -T option to readelf.
* NEWS: Mention the new feature.

gas * testsuite/gas/ia64/group-2.d: Add -T option to readelf
command line.
* testsuite/gas/ia64/unwind.d: Likewise.
* testsuite/gas/mmix/bspec-1.d: Likewise.
* testsuite/gas/mmix/bspec-2.d: Likewise.
* testsuite/gas/mmix/comment-1.d: Likewise.
* testsuite/gas/tic6x/scomm-directive-4.d: Likewise.

ld * testsuite/ld-powerpc/powerpc.exp: Add -T option to readelf
command line when running some tests.
* testsuite/ld-arm/arm-elf.exp: Likewise.
* testsuite/ld-mips-elf/mips-elf.exp: Likewise.
* testsuite/ld-mmix/local1.d: Likewise.
* testsuite/ld-mmix/local3.d: Likewise.
* testsuite/ld-mmix/local5.d: Likewise.
* testsuite/ld-mmix/local7.d: Likewise.
* testsuite/ld-powerpc/powerpc.exp: Likewise.

4 years agoMake tui_win_info::name pure virtual
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Make tui_win_info::name pure virtual

It seemed cleaner to me for tui_win_info::name to be pure virtual.
This meant adding a name method to the locator window; but this too
seems like an improvement.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-data.h (struct tui_win_info) <name>: Now pure virtual.
* tui/tui-stack.h (struct tui_locator_window) <name>: New method.

4 years agoRemove tui_gen_win_info
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Remove tui_gen_win_info

This merges the tui_gen_win_info base class with tui_win_info;
renaming the resulting class to tui_win_info.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-wingeneral.c (tui_win_info::refresh_window): Move from
tui_gen_win_info.
(tui_win_info::make_window): Merge with
tui_gen_win_info::make_window.
(tui_win_info::make_visible): Move from tui_gen_win_info.
* tui/tui-win.c (tui_win_info::max_width): Move from
tui_gen_win_info.
* tui/tui-layout.h (class tui_layout_window) <m_window>: Change
type.
<window_factory>: Likewise.
* tui/tui-layout.c (tui_win_info::resize): Move from
tui_gen_win_info.
(make_standard_window): Change return type.
(get_locator_window, tui_get_window_by_name): Likewise.
(tui_layout_window::apply): Remove a cast.
* tui/tui-data.h (MIN_WIN_HEIGHT): Move earlier.
(struct tui_win_info): Merge with tui_gen_win_info.
(struct tui_gen_win_info): Remove.

4 years agoDerive tui_locator_window from tui_win_info
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Derive tui_locator_window from tui_win_info

tui_locator_window is the last remaining concrete child class of
tui_gen_win_info.  It seems a bit cleaner to me to flatten the
hierarchy a bit; this patch prepares for that by changing
tui_locator_window to derive from tui_win_info.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-stack.h (struct tui_locator_window): Derive from
tui_win_info.
<do_scroll_horizontal, do_scroll_vertical>: New methods.
<can_box>: New method.

4 years agoRemove body of tui_locator_window constructor
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Remove body of tui_locator_window constructor

The tui_locator_window constructor initializes the first character of
two of its members.  However, this is actually an error, since these
were changed to be std::string.  This removes the erroneous code.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-stack.h (struct tui_locator_window): Remove body.

4 years agoDon't derive tui_data_item_window from tui_gen_win_info
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Don't derive tui_data_item_window from tui_gen_win_info

There's no deep reason that tui_data_item_window should derive from
tui_gen_win_info -- it currently uses a curses window to render, but
that isn't truly needed, and it adds some hacks to other parts of the
TUI.

This patch changes tui_data_item_window so that it doesn't have a base
class, and updates the register window.  This simplifies the code and
enables a subsequent cleanup.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-regs.c (tui_data_window::display_registers_from)
(tui_data_window::display_registers_from)
(tui_data_window::first_data_item_displayed)
(tui_data_window::delete_data_content_windows): Update.
(tui_data_window::refresh_window, tui_data_window::no_refresh):
Remove.
(tui_data_window::check_register_values): Update.
(tui_data_item_window::rerender): Add parameters.  Update.
(tui_data_item_window::refresh_window): Remove.
* tui/tui-data.h (struct tui_gen_win_info) <no_refresh>: No longer
virtual.
* tui/tui-regs.h (struct tui_data_item_window): Don't derive from
tui_gen_win_info.
<refresh_window, max_height, min_height>: Remove.
<rerender>: Add parameters.
<x, y, visible>: New members.
(struct tui_data_window) <refresh_window, no_refresh>: Remove.
<m_item_width>: New member.

4 years agoRename tui_data_item_window::item_no
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Rename tui_data_item_window::item_no

tui_data_item_window::item_no is misnamed -- it only can be used for a
register, but it references a "display" number as well.  (Based on
other comments I've seen in the past -- most since deleted -- I think
there were plans at one point to display variables in this window as
well.  However, this was never implemented.)

This patch renames this member to be more correct.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-regs.c (tui_data_window::show_register_group)
(tui_data_window::check_register_values): Update.
* tui/tui-regs.h (struct tui_data_item_window) <regno>: Rename
from item_no.

4 years agoRemove useless "if' from tui-regs.c
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Remove useless "if' from tui-regs.c

tui_data_window::show_register_group had a useless "if" -- the
condition could never be false.  This patch removes it.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-regs.c (tui_data_window::show_register_group): Remove
useless "if".

4 years agoRemove tui_data_window::name
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Remove tui_data_window::name

The "name" member of tui_data_window was set, but never used.  This
removes it.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-regs.c (tui_data_window::show_register_group): Update.
* tui/tui-regs.h (struct tui_data_item_window) <name>: Remove.

4 years agoMove some code out of tui-data.h
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Move some code out of tui-data.h

This moves some code out of tui-data.h, to more closely related
places.  Some unused forward declarations are also removed.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-stack.c (SINGLE_KEY): Move from tui-data.h
* tui/tui-winsource.h (enum tui_line_or_address_kind)
(struct tui_line_or_address): Move from tui-data.h.
* tui/tui-win.c (DEFAULT_TAB_LEN): Move from tui-data.h.
* tui/tui-data.h (DEFAULT_TAB_LEN): Move to tui-win.c.
(tui_cmd_window, tui_source_window_base, tui_source_window)
(tui_disasm_window): Don't declare.
(enum tui_line_or_address_kind, struct tui_line_or_address): Move
to tui-winsource.h.
(SINGLE_KEY): Move to tui-stack.c.

4 years agoRemove tui_expand_tabs
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Remove tui_expand_tabs

tui_expand_tabs only has a single caller.  This patch removes this
function, in favor of a tab-expanding variant of string_file.  This
simplifies the code somewhat.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-regs.h (struct tui_data_item_window) <content>: Now a
std::string.
* tui/tui-regs.c (class tab_expansion_file): New.
(tab_expansion_file::write): New method.
(tui_register_format): Change return type.  Use
tab_expansion_file.
(tui_get_register, tui_data_window::display_registers_from)
(tui_data_item_window::rerender): Update.
* tui/tui-io.h (tui_expand_tabs): Don't declare.
* tui/tui-io.c (tui_expand_tabs): Remove.

4 years agoUse complete_on_enum in tui_reggroup_completer
Tom Tromey [Thu, 2 Jul 2020 03:21:12 +0000 (21:21 -0600)]
Use complete_on_enum in tui_reggroup_completer

tui_reggroup_completer has an "XXXX" comment suggesting the use of
complete_on_enum.  This patch implements this suggestion.

gdb/ChangeLog
2020-07-01  Tom Tromey  <tom@tromey.com>

* tui/tui-regs.c (tui_reggroup_completer): Use complete_on_enum.

4 years agoAutomatic date update in version.in
GDB Administrator [Thu, 2 Jul 2020 00:00:07 +0000 (00:00 +0000)]
Automatic date update in version.in

4 years agoRecognize -1 as a tombstone value in .debug_line
Fangrui Song [Wed, 1 Jul 2020 19:31:44 +0000 (12:31 -0700)]
Recognize -1 as a tombstone value in .debug_line

LLD from 11 onwards (https://reviews.llvm.org/D81784) uses -1 to
represent a relocation in .debug_line referencing a discarded symbol.
Recognize -1 to fix gdb.base/break-on-linker-gcd-function.exp when the
linker is a newer LLD.

gdb/ChangeLog:

* dwarf2/read.c (lnp_state_machine::check_line_address): Test -1.

4 years agoPR26188, buff overflow in coff_find_nearest_line_with_names
Alan Modra [Wed, 1 Jul 2020 07:24:50 +0000 (16:54 +0930)]
PR26188, buff overflow in coff_find_nearest_line_with_names

PR 26188
* coffgen.c (coff_find_nearest_line_with_names): Sanity check
raw syment index before dereferencing.

4 years agoAllow reference form for DW_AT_associated and DW_AT_allocated attributes
Alok Kumar Sharma [Wed, 1 Jul 2020 06:23:09 +0000 (11:53 +0530)]
Allow reference form for DW_AT_associated and DW_AT_allocated attributes

Currently, GDB rejects the (die) reference form while it accepts exprloc
form. It is allowed in DWARF standard. "Table 7.5: Attribute encodings"
in DWARF5 standard. Flang compiler assigns (die) reference to
DW_AT_associated and DW_AT_allocated for some cases.

gdb/ChangeLog

* dwarf2/read.c (set_die_type): Removed conditions to restrict
forms for DW_AT_associated and DW_AT_allocated attributes,
which is already checked in function attr_to_dynamic_prop.

4 years agoRe: Remove x86 NaCl target support
Alan Modra [Wed, 1 Jul 2020 07:22:02 +0000 (16:52 +0930)]
Re: Remove x86 NaCl target support

Fixes accidental breakage of vxworks by commit bbd19b19e4.

* elf32-i386.c (elf_backend_object_p): Undef for vxworks.

4 years agoobsolete xc16x
Alan Modra [Tue, 30 Jun 2020 23:45:34 +0000 (09:15 +0930)]
obsolete xc16x

The xc16x md_apply_fix code is just so broken that in my opinion the
target should never have been accepted, and from a quick look at
commit logs for the target it appears that no one has ever contributed
fixes for anything.  This target has just been a 14 year burden on
global binutils and cgen maintainers.  That's not how free software is
supposed to work.

bfd/
* config.bfd: Obsolete xc16x.
gas/
* config/tc-xc16x.c (md_apply_fix): Add FIXME.

4 years agogas eqv-dot test fails
Alan Modra [Tue, 30 Jun 2020 23:44:45 +0000 (09:14 +0930)]
gas eqv-dot test fails

* testsuite/gas/all/eqv-dot.d: xfail targets that set linkrelax
in data sections, and mep.

4 years agoAutomatic date update in version.in
GDB Administrator [Wed, 1 Jul 2020 00:00:07 +0000 (00:00 +0000)]
Automatic date update in version.in

4 years agoUpdate readline/README to mention patchlevel
Tom Tromey [Tue, 30 Jun 2020 21:17:07 +0000 (15:17 -0600)]
Update readline/README to mention patchlevel

This updates readline/README to mention the current patchlevel, and
the "git am" approach to importing new upstream patches.

readline/ChangeLog
2020-06-30  Tom Tromey  <tom@tromey.com>

* README: Update instructions.

4 years agoproblems restoring the history file are not signaled correctly to the calling application
Chet Ramey [Fri, 7 Feb 2020 19:58:55 +0000 (14:58 -0500)]
problems restoring the history file are not signaled correctly to the calling application

4 years agoreading history entries with timestamps can result in joined entries
Chet Ramey [Fri, 7 Feb 2020 19:57:47 +0000 (14:57 -0500)]
reading history entries with timestamps can result in joined entries

4 years agofix problems moving back beyond start of history
Chet Ramey [Fri, 7 Feb 2020 19:56:45 +0000 (14:56 -0500)]
fix problems moving back beyond start of history

4 years agoReadline-8.0 patch 1: fix file descriptor leak with zero-length history file
Chet Ramey [Wed, 14 Aug 2019 15:28:48 +0000 (11:28 -0400)]
Readline-8.0 patch 1: fix file descriptor leak with zero-length history file

4 years agoFix test breakages caused by removal of gdb_py_test_multiple.
Philippe Waroquiers [Tue, 30 Jun 2020 16:40:21 +0000 (18:40 +0200)]
Fix test breakages caused by removal of gdb_py_test_multiple.

Tom de Vries detected that some python tests were broken as
they were still using gdb_py_test_multiple that was replaced
by gdb_test_multiline.  Repair these tests by using the new function.

gdb/testsuite/ChangeLog
2020-06-30  Philippe Waroquiers  <philippe.waroquiers@skynet.be>

* gdb.python/py-breakpoint.exp: use gdb_test_multiline instead
of gdb_py_test_multiple.
* gdb.python/py-cmd.exp: Likewise.
* gdb.python/py-events.exp: Likewise.
* gdb.python/py-function.exp: Likewise.
* gdb.python/py-inferior.exp: Likewise.
* gdb.python/py-infthread.exp: Likewise.
* gdb.python/py-linetable.exp: Likewise.
* gdb.python/py-parameter.exp: Likewise.
* gdb.python/py-value.exp: Likewise.

4 years agoRemove x86 NaCl target support
H.J. Lu [Tue, 30 Jun 2020 15:47:02 +0000 (08:47 -0700)]
Remove x86 NaCl target support

NaCl has been deprecated:

https://developer.chrome.com/native-client/migration

and NaCl will completely disappear in 2021:

https://lists.llvm.org/pipermail/llvm-dev/2020-April/141107.html

Remove x86 NaCl target support from bfd, binutils, gas and ld.

bfd/

* archures.c (bfd_mach_i386_nacl): Removed.
(bfd_mach_i386_i386_nacl): Likewise.
(bfd_mach_x86_64_nacl): Likewise.
(bfd_mach_x64_32_nacl): Likewise.
* config.bfd: Remove *-*-nacl* targets.
* configure.ac: Remove x86 NaCl target vectors.
* cpu-i386.c (bfd_arch_i386_onebyte_nop_fill): Removed.
(bfd_x64_32_nacl_arch): Likewise.
(bfd_x86_64_nacl_arch): Likewise.
(bfd_i386_nacl_arch): Likewise.
(bfd_x64_32_arch_intel_syntax): Updated.
* elf32-i386.c: Don't include "elf-nacl.h".
(elf_i386_nacl_plt): Removed.
(elf_i386_nacl_plt0_entry): Likewise.
(elf_i386_nacl_plt_entry): Likewise.
(elf_i386_nacl_pic_plt0_entry): Likewise.
(elf_i386_nacl_pic_plt_entry): Likewise.
(elf_i386_nacl_eh_frame_plt): Likewise.
(elf_i386_nacl_plt): Likewise.
(elf32_i386_nacl_elf_object_p): Likewise.
(elf_i386_get_synthetic_symtab): Updated.
(elf_i386_link_setup_gnu_properties): Likewise.
* elf64-x86-64.c: Don't include "elf-nacl.h".
(elf_x86_64_nacl_plt): Removed.
(elf64_x86_64_nacl_elf_object_p): Likewise.
(elf_x86_64_nacl_plt0_entry): Likewise.
(elf_x86_64_nacl_plt_entry): Likewise.
(elf_x86_64_nacl_eh_frame_plt): Likewise.
(elf_x86_64_nacl_plt): Likewise.
(elf32_x86_64_nacl_elf_object_p): Likewise.
(elf_x86_64_get_synthetic_symtab): Updated.
(elf_x86_64_link_setup_gnu_properties): Likewise.
* elfxx-x86.c (_bfd_x86_elf_link_setup_gnu_properties): Likewise.
* targets.c: Remove x86 NaCl target vectors.
* bfd-in2.h: Regenerated.
* configure: Likewise.

binutils/

* NEWS: Mention x86 NaCl target support removal.
* dwarf.c (init_dwarf_regnames_by_bfd_arch_and_mach): Remove
x86 NaCl target support.
* testsuite/binutils-all/elfedit-1.d: Likewise.
* testsuite/binutils-all/i386/i386.exp: Likewise.
* testsuite/binutils-all/x86-64/objects.exp: Likewise.
* testsuite/binutils-all/x86-64/pr23494a-x32.d: Likewise.
* testsuite/binutils-all/x86-64/pr23494a.d: Likewise.
* testsuite/binutils-all/x86-64/pr23494b-x32.d: Likewise.
* testsuite/binutils-all/x86-64/pr23494b.d: Likewise.
* testsuite/binutils-all/x86-64/pr23494c-x32.d: Likewise.
* testsuite/binutils-all/x86-64/pr23494c.d: Likewise.
* testsuite/binutils-all/x86-64/pr23494d-x32.d: Likewise.
* testsuite/binutils-all/x86-64/pr23494d.d: Likewise.
* testsuite/binutils-all/x86-64/pr23494e-x32.d: Likewise.
* testsuite/binutils-all/x86-64/pr23494e.d: Likewise.
* testsuite/binutils-all/x86-64/x86-64.exp: Likewise.

gas/

* NEWS: Mention x86 NaCl target support removal.
* config/tc-i386.c: Remove x86 NaCl target support.
* config/tc-i386.h: Likewise.
* configure.tgt: Likewise.
* testsuite/gas/i386/i386.exp: Likewise.
* testsuite/gas/i386/iamcu-1.d: Likewise.
* testsuite/gas/i386/iamcu-2.d: Likewise.
* testsuite/gas/i386/iamcu-3.d: Likewise.
* testsuite/gas/i386/iamcu-4.d: Likewise.
* testsuite/gas/i386/iamcu-5.d: Likewise.
* testsuite/gas/i386/k1om.d: Likewise.
* testsuite/gas/i386/l1om.d: Likewise.

ld/

* Makefile.am (ALL_EMULATION_SOURCES): Remove eelf_i386_nacl.c,
eelf32_x86_64_nacl.c, eelf_x86_64_nacl.c.
Remove x86 NaCl dep files.
* NEWS: Mention x86 NaCl target support removal.
* configure.tgt: Remove x86 NaCl target support.
* testsuite/ld-elf/binutils.exp: Likewise.
* testsuite/ld-elf/elf.exp: Likewise.
* testsuite/ld-elfvers/vers.exp: Likewise.
* testsuite/ld-i386/align-branch-1.d: Likewise.
* testsuite/ld-i386/export-class.exp: Likewise.
* testsuite/ld-i386/i386.exp: Likewise.
* testsuite/ld-i386/load1.d: Likewise.
* testsuite/ld-i386/pie1.d: Likewise.
* testsuite/ld-i386/pr12570a.d: Likewise.
* testsuite/ld-i386/pr12570b.d: Likewise.
* testsuite/ld-i386/pr19636-1d.d: Likewise.
* testsuite/ld-i386/pr19636-1l.d: Likewise.
* testsuite/ld-i386/pr19636-2c.d: Likewise.
* testsuite/ld-i386/pr19636-2d.d: Likewise.
* testsuite/ld-i386/pr19636-2e.d: Likewise.
* testsuite/ld-i386/pr20244-1a.d: Likewise.
* testsuite/ld-i386/pr20244-1b.d: Likewise.
* testsuite/ld-i386/pr20244-2a.d: Likewise.
* testsuite/ld-i386/pr20244-2b.d: Likewise.
* testsuite/ld-i386/pr20244-2c.d: Likewise.
* testsuite/ld-i386/pr20244-4a.d: Likewise.
* testsuite/ld-i386/pr20244-4b.d: Likewise.
* testsuite/ld-i386/pr21884.d: Likewise.
* testsuite/ld-ifunc/binutils.exp: Likewise.
* testsuite/ld-ifunc/ifunc-10-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-10-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-11-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-11-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-12-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-12-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-13-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-13-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-14a-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-14a-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-14b-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-14b-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-14c-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-14c-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-14d-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-14d-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-14e-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-14e-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-14f-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-14f-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-15-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-15-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-16-i386-now.d: Likewise.
* testsuite/ld-ifunc/ifunc-16-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-16-x86-64-now.d: Likewise.
* testsuite/ld-ifunc/ifunc-16-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-17a-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-17a-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-17b-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-17b-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-18a-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-18a-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-18b-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-18b-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-19a-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-19a-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-19b-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-19b-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-i386-now.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-local-i386-now.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-local-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-local-x86-64-now.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-local-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-x86-64-now.d: Likewise.
* testsuite/ld-ifunc/ifunc-2-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-20-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-20-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-21-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-21-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-22-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-22-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-5a-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-5a-local-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-5a-local-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-5a-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-5b-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-5b-local-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-5b-local-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-5b-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-5r-local-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-5r-local-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-6a-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-6a-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-6b-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-6b-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-7a-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-7a-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-7b-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-7b-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-8-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-8-x86-64.d: Likewise.
* testsuite/ld-ifunc/ifunc-9-i386.d: Likewise.
* testsuite/ld-ifunc/ifunc-9-x86-64.d: Likewise.
* testsuite/ld-ifunc/pr17154-i386-now.d: Likewise.
* testsuite/ld-ifunc/pr17154-i386.d: Likewise.
* testsuite/ld-ifunc/pr17154-x86-64-now.d: Likewise.
* testsuite/ld-ifunc/pr17154-x86-64.d: Likewise.
* testsuite/ld-plugin/lto.exp: Likewise.
* testsuite/ld-x86-64/align-branch-1.d: Likewise.
* testsuite/ld-x86-64/dwarfreloc.exp: Likewise.
* testsuite/ld-x86-64/line.exp: Likewise.
* testsuite/ld-x86-64/load1a.d: Likewise.
* testsuite/ld-x86-64/load1b.d: Likewise.
* testsuite/ld-x86-64/load1c.d: Likewise.
* testsuite/ld-x86-64/load1d.d: Likewise.
* testsuite/ld-x86-64/pie3.d: Likewise.
* testsuite/ld-x86-64/pr18160.d: Likewise.
* testsuite/ld-x86-64/pr19013-x32.d: Likewise.
* testsuite/ld-x86-64/pr19013.d: Likewise.
* testsuite/ld-x86-64/pr19636-2d.d: Likewise.
* testsuite/ld-x86-64/pr19636-2l.d: Likewise.
* testsuite/ld-x86-64/pr20253-1b.d: Likewise.
* testsuite/ld-x86-64/pr20253-1d.d: Likewise.
* testsuite/ld-x86-64/pr20253-1f.d: Likewise.
* testsuite/ld-x86-64/pr20253-1h.d: Likewise.
* testsuite/ld-x86-64/pr20253-1j.d: Likewise.
* testsuite/ld-x86-64/pr20253-1l.d: Likewise.
* testsuite/ld-x86-64/pr21884.d: Likewise.
* testsuite/ld-x86-64/pr22393-3a.rd: Likewise.
* testsuite/ld-x86-64/pr22393-3b.rd: Likewise.
* testsuite/ld-x86-64/tlsgd10.dd: Likewise.
* testsuite/ld-x86-64/tlsgd5.dd: Likewise.
* testsuite/ld-x86-64/tlsgd8.dd: Likewise.
* testsuite/ld-x86-64/x86-64.exp: Likewise.
* emulparams/elf32_x86_64_nacl.sh: Removed.
* emulparams/elf_i386_nacl.sh: Likewise.
* emulparams/elf_x86_64_nacl.sh: Likewise.
* testsuite/ld-i386/emit-relocs-nacl.rd: Likewise.
* testsuite/ld-i386/load1-nacl.d: Likewise.
* testsuite/ld-i386/pie1-nacl.d: Likewise.
* testsuite/ld-i386/plt-nacl.pd: Likewise.
* testsuite/ld-i386/plt-pic-nacl.pd: Likewise.
* testsuite/ld-i386/pr17709-nacl.rd: Likewise.
* testsuite/ld-i386/pr19636-1d-nacl.d: Likewise.
* testsuite/ld-i386/pr19636-2c-nacl.d: Likewise.
* testsuite/ld-i386/pr19636-2d-nacl.d: Likewise.
* testsuite/ld-i386/pr19636-2e-nacl.d: Likewise.
* testsuite/ld-i386/pr19827-nacl.rd: Likewise.
* testsuite/ld-i386/pr21884-nacl.d: Likewise.
* testsuite/ld-i386/pr21884-nacl.t: Likewise.
* testsuite/ld-i386/tlsbin-nacl.rd: Likewise.
* testsuite/ld-i386/tlsbin2-nacl.rd: Likewise.
* testsuite/ld-i386/tlsbindesc-nacl.rd: Likewise.
* testsuite/ld-i386/tlsdesc-nacl.rd: Likewise.
* testsuite/ld-i386/tlsgdesc-nacl.rd: Likewise.
* testsuite/ld-i386/tlsnopic-nacl.rd: Likewise.
* testsuite/ld-i386/tlspic-nacl.rd: Likewise.
* testsuite/ld-i386/tlspic2-nacl.rd: Likewise.
* testsuite/ld-x86-64/ilp32-4-nacl.d: Likewise.
* testsuite/ld-x86-64/load1a-nacl.d: Likewise.
* testsuite/ld-x86-64/load1b-nacl.d: Likewise.
* testsuite/ld-x86-64/load1c-nacl.d: Likewise.
* testsuite/ld-x86-64/load1d-nacl.d: Likewise.
* testsuite/ld-x86-64/pie3-nacl.d: Likewise.
* testsuite/ld-x86-64/plt-nacl.pd: Likewise.
* testsuite/ld-x86-64/pr17709-nacl.rd: Likewise.
* testsuite/ld-x86-64/pr19013-nacl.d: Likewise.
* testsuite/ld-x86-64/pr19636-2d-nacl.d: Likewise.
* testsuite/ld-x86-64/pr19827-nacl.rd: Likewise.
* testsuite/ld-x86-64/pr21884-nacl.d: Likewise.
* testsuite/ld-x86-64/pr21884-nacl.t: Likewise.
* testsuite/ld-x86-64/split-by-file-nacl.rd: Likewise.
* testsuite/ld-x86-64/tlsbin-nacl.rd: Likewise.
* testsuite/ld-x86-64/tlsbin2-nacl.rd: Likewise.
* testsuite/ld-x86-64/tlsbindesc-nacl.rd: Likewise.
* testsuite/ld-x86-64/tlsdesc-nacl.pd: Likewise.
* testsuite/ld-x86-64/tlsdesc-nacl.rd: Likewise.
* testsuite/ld-x86-64/tlsgdesc-nacl.rd: Likewise.
* testsuite/ld-x86-64/tlspic-nacl.rd: Likewise.
* testsuite/ld-x86-64/tlspic2-nacl.rd: Likewise.
* Makefile.in: Regenerated.
* po/BLD-POTFILES.in: Likewise.

4 years agoFix bug in quirk_rust_enum
Tom Tromey [Tue, 30 Jun 2020 13:57:35 +0000 (07:57 -0600)]
Fix bug in quirk_rust_enum

Tom de Vries pointed out that some Rust tests were failing after the
variant part rewrite.  He sent an executable, which helped track down
this bug.

quirk_rust_enum was passing 1 to alloc_rust_variant in one case.
However, a comment earlier says:

      /* We don't need a range entry for the discriminant, but we do
 need one for every other field, as there is no default
 variant.  */

In this case, we must pass -1 for this parameter.  That is what this
patch implements.

gdb/ChangeLog
2020-06-30  Tom Tromey  <tromey@adacore.com>

* dwarf2/read.c (quirk_rust_enum): Correctly call
alloc_rust_variant for default-less enum.

4 years agoDo not define basic_string_view::to_string
Tom Tromey [Tue, 30 Jun 2020 13:53:03 +0000 (07:53 -0600)]
Do not define basic_string_view::to_string

gdb's copy of basic_string_view includes a to_string method.  However,
according to cppreference, this is not a method on the real
std::basic_string_view:

https://en.cppreference.com/w/cpp/string/basic_string_view

This difference matters because gdb_string_view.h will use the
standard implementation when built with a C++17 or later.  This caused
PR build/26183.

This patch fixes the problem by changing the method to be a standalone
helper function, and then rewriting the uses.  Tested by rebuilding
with a version of GCC that defaults to C++17.

(Note that the build still is not clean; and also I noticed that the
libstdc++ string_view forbids the use of nullptr ... I wonder if gdb
violates that.)

gdb/ChangeLog
2020-06-30  Tom Tromey  <tromey@adacore.com>

PR build/26183:
* ada-lang.c (ada_lookup_name_info::ada_lookup_name_info): Use
gdb::to_string.

gdbsupport/ChangeLog
2020-06-30  Tom Tromey  <tromey@adacore.com>

PR build/26183:
* gdb_string_view.h (basic_string_view::to_string): Remove.
(gdb::to_string): New function.

4 years agoAdd support for attributes using DW_FORM_GNU_strp_alt
Nick Clifton [Tue, 30 Jun 2020 12:05:18 +0000 (13:05 +0100)]
Add support for attributes using DW_FORM_GNU_strp_alt

* dwarf.c (read_and_display_attr_value): Support
DW_FORM_GNU_strp_alt when used with DW_AT_dwo_name and
DW_AT_comp_dir.

4 years ago[gdb/testsuite] Handle early_flags in gdb_default_target_compile
Tom de Vries [Tue, 30 Jun 2020 07:23:09 +0000 (09:23 +0200)]
[gdb/testsuite] Handle early_flags in gdb_default_target_compile

In gdb_default_target_compile, we use dejagnu's default_target_compile, unless
we need support for languages that are not yet support in the used dejagnu
version, in which case we use a local default_target_compile:
gdb_default_target_compile_1.

However, there's another reason to use the local default_target_compile: when
early_flags is used, because there's no dejagnu release available yet
supporting this.

Fix this by detecting and handling early_flags in gdb_default_target_compile.

Tested on x86_64-linux.

gdb/testsuite/ChangeLog:

2020-06-30  Tom de Vries  <tdevries@suse.de>

PR testsuite/26175
* lib/future.exp (gdb_default_target_compile): Detect and handle
early_flags.

4 years agosolaris XPASS ld-elf/group1
Alan Modra [Tue, 30 Jun 2020 04:42:49 +0000 (14:12 +0930)]
solaris XPASS ld-elf/group1

* testsuite/ld-elf/group1.d: Don't xfail all solaris targets, just
ix86 and x86_64.

4 years agoDon't xfail nds32 for ld-elf/reloc-discard test
Alan Modra [Tue, 30 Jun 2020 04:17:32 +0000 (13:47 +0930)]
Don't xfail nds32 for ld-elf/reloc-discard test

* testsuite/ld-elf/reloc-discard.d: Don't xfail nds32.

4 years agoFix nds32*-elf XPASS readelf -wiaoRlL dw5
Alan Modra [Tue, 30 Jun 2020 03:45:29 +0000 (13:15 +0930)]
Fix nds32*-elf XPASS readelf -wiaoRlL dw5

The underlying fail was probably fixed with git commit fbaf61ad5.

* testsuite/binutils-all/readelf.exp (readelf_dump_test): Remove
xfail for nds32.

4 years agomicroblaze-linux XPASS ld-elf/var1
Alan Modra [Tue, 30 Jun 2020 03:52:52 +0000 (13:22 +0930)]
microblaze-linux XPASS ld-elf/var1

microblaze-linux uses the standard ELF script, microblaze-elf its own
script lacking an input section pattern needed to make this test pass.
Add the missing pattern for .data, in line with most other sections
that do have .* patterns.

* scripttempl/elfmicroblaze.sc (.data): Add .data.* entry.
* testsuite/ld-elf/var1.d: Don't xfail microblaze.

4 years agoRISC-V: Support debug and float CSR as the unprivileged ones.
Nelson Chu [Mon, 8 Jun 2020 02:54:53 +0000 (10:54 +0800)]
RISC-V: Support debug and float CSR as the unprivileged ones.

The unprivileged CSR should be controlled by other specific specs rather
than the privileged spec.  For example, the debug CSR should be controlled
by the debug spec, and the float CSR should be controlled by the float
spec.  User may use assembler options to choose what the debug and other
specs they want, or may encode the versions of specs into the architecture
string directly.  Since we haven't decided which one is better, we set the
defined and aborted versions of unprivileged CSR to PRIV_SPEC_CLASS_NONE
in the include/opcode/riscv-opc.h, to tell assembler don't check priv spec
versions for them.  However, these PRIV_SPEC_CLASS_NONE will be changed
to FLOAT_SPEC_CLASS_* and DEBUG_SPEC_CLASS_* in the future.

gas/
* config/tc-riscv.c (riscv_csr_class_check): Removed.  Move the
checking into riscv_csr_address.
(riscv_csr_version_check): Likewise.
(riscv_csr_address): New function.  Return the suitable CSR address
after checking the ISA dependency and versions.  Issue warnings if
we find any conflict and -mcsr-check is set.  CSR_CLASS_F and
CSR_CLASS_DEBUG are unprivileged CSR for now, so don't check the
priv spec versions for them.
(reg_csr_lookup_internal): Call riscv_csr_address to find the
suitable CSR address.

* testsuite/gas/riscv/priv-reg-fail-fext.d: Remove -mpriv-spec=1.11.
* testsuite/gas/riscv/priv-reg-fail-read-only-01.d: Likewise.
* testsuite/gas/riscv/priv-reg-fail-rv32-only.d: Likewise.
* testsuite/gas/riscv/priv-reg-fail-fext.l:  We don't care the
priv spec warnings here.  These warnings are added by accident.
Remove them and only focus on the ISA dependency warnings.
* testsuite/gas/riscv/priv-reg-fail-rv32-only.l: Likewise.
* testsuite/gas/riscv/priv-reg-fail-read-only-01.l: Likewise.
* testsuite/gas/riscv/priv-reg-fail-version-1p9.l: Updated since
dscratch0 and dscratch1 are regarded as the unprivileged CSR rather
than the privileged ones.
* testsuite/gas/riscv/priv-reg-fail-version-1p9p1.l: Likewise.
* testsuite/gas/riscv/priv-reg-fail-version-1p10.l: Likewise.
* testsuite/gas/riscv/priv-reg-fail-version-1p11.l: Likewise.
* testsuite/gas/riscv/priv-reg.s: Likewise.  Add missing debug CSR.
* testsuite/gas/riscv/priv-reg-version-1p9.d: Likewise.
* testsuite/gas/riscv/priv-reg-version-1p9p1.d: Likewise.
* testsuite/gas/riscv/priv-reg-version-1p10.d: Likewise.
* testsuite/gas/riscv/priv-reg-version-1p11.d: Likewise.
* testsuite/gas/riscv/csr-dw-regnums.d: Likewise.
* testsuite/gas/riscv/csr-dw-regnums.s: Likewise.

include/
* opcode/riscv-opc.h: Support the unprivileged CSR.  The versions
of the unprivileged CSR should be PRIV_SPEC_CLASS_NONE for now.
* opcode/riscv.h (enum riscv_csr_class): Add CSR_CLASS_DEBUG.

opcodes/
* riscv-dis.c (print_insn_args, case 'E'): Updated.  Let the
unprivileged CSR can also be initialized.

4 years agoRISC-V: Cleanup the include/opcode/riscv-opc.h.
Nelson Chu [Tue, 23 Jun 2020 06:23:51 +0000 (14:23 +0800)]
RISC-V: Cleanup the include/opcode/riscv-opc.h.

The include/opcode/riscv-opc.h file is no longer automatically generated,
so we remove the misleading comments and add new ones.  Besides, the CAUSE_*
macros and DECLARE_CAUSE are unused for binutils and gdb.  Therefore, remove
them, too.

include/
* opcode/riscv-opc.h: Cleanup and remove the unused macros.

4 years agoAutomatic date update in version.in
GDB Administrator [Tue, 30 Jun 2020 00:00:08 +0000 (00:00 +0000)]
Automatic date update in version.in

4 years agogdb: fix documentation of gdbarch_displaced_step_copy_insn
Simon Marchi [Mon, 29 Jun 2020 15:27:40 +0000 (11:27 -0400)]
gdb: fix documentation of gdbarch_displaced_step_copy_insn

I spotted something that looks wrong in the doc of
gdbarch_displaced_step_copy_insn.

It says that if the function returns NULL, it means that it has emulated
the behavior of the instruction and written the result to REGS.
However, it says below that the function may return NULL to indicate
that the instruction can't be single-stepped out-of-line, in which case
the core steps the instruction in-line.  The two are contradictory.

The right one is the latter, if the function returns NULL, the core
falls back to in-line stepping. I checked all the implementations of
this function and they all agree with this.

gdb/ChangeLog:

* gdbarch.sh (displaced_step_copy_insn): Update doc.
* gdbarch.h: Re-generate.

Change-Id: I98163cdd38970cde4c77680e249b10f5d2d5bf9b

4 years agogdb/testsuite: better handle failures in simavr board, reap simavr process
Simon Marchi [Mon, 29 Jun 2020 14:19:43 +0000 (10:19 -0400)]
gdb/testsuite: better handle failures in simavr board, reap simavr process

This patch makes a few adjustments to the simavr.exp to handle tests
that error out more gracefully.  I put all the changes in the same patch
because right now it's in a very bad shape, so it's very painful to do
small incremental adjustements.  I found that with these changes, it
becomes reasonably stable.

For example, the gdb.base/step-over-syscall.exp test is a bit buggy
(stuff for another patch...), in that it calls gdb_load (through
clean_restart) with a file that doesn't exist.  The gdb_load
implementation in simavr.exp gets called with a file that doesn't exist,
and simavr (expectedly) fails to start.

When this happens, we currently leave the $simavr_spawn_id variable set.
However, because the simavr process has terminated, its spawn id is
closed.  When the next test tries to close the previous connection to
simavr, it fails to, and this error is thrown:

    ERROR: close: spawn id exp86 not open
        while executing
    "close -i $simavr_spawn_id"
        (procedure "gdb_load" line 18)
        invoked from within

In other words, any test ran after step-over-syscall.exp will have
simavr.exp's gdb_load fail on it.

This patch tries to make sure that simavr.exp's gdb_load only leaves
simavr_spawn_id set if everything went fine.  If we couldn't start
simavr, don't see the expected output, or fail to connect to it, we
close the spawn id (if needed) and unset simavr_spawn_id.

As an additional precaution, I added a catch around the "close the
previous connection" code.  Ideally, this shouldn't be necessary, but I
bet there are other similar scenarios where we might try to close an
already close spawn id.  So I think displaying a warning is better than
messing up all following tests.

Also, the board never wait'ed for the simavr process, resulting in tons
of zombie simavr processes hanging around.  This patch adds some calls
to "wait" whenever we close the connection (or realize it is already
closed), which seems to fix the problem.

Finally I switched a `gdb_expect` to bare `expect`, where we wait for
the "listening" message from simavr.  I found it necessary because
gdb_expect (through remote_expect) adds a `-i <gdb spawn id> -timeout
10`.  So if for some reason the GDB process has crashed in the mean
time, this expect will unexpectedly error out with a `spawn id <gdb
spawn id> not open`.  Therefore, change `gdb_expect` to `expect` so that
we only deal with simavr's spawn id here.

Here are the results on TESTS="gdb.base/*.exp" before:

    # of expected passes 4816
    # of unexpected failures 4433
    # of known failures 2
    # of unresolved testcases 300
    # of untested testcases 143
    # of unsupported tests 7
    # of paths in test names 2
    # of duplicate test names 10

and after:

    # of expected passes 21957
    # of unexpected failures 1564
    # of expected failures 8
    # of unknown successes 1
    # of known failures 31
    # of unresolved testcases 532
    # of untested testcases 153
    # of unsupported tests 28
    # of paths in test names 2
    # of duplicate test names 32

I tested using simavr's current master (7c254e2081b5).

gdb/testsuite/ChangeLog:

* boards/simavr.exp (gdb_load): Catch errors when closing
previous connection.  Close connection, wait for process and
unset simavr_spawn_id on failure.

Change-Id: I695fc765e1c22e1d1dc0b08e0f5141244986b868

4 years agox86: Support VEX base opcode length > 1
H.J. Lu [Mon, 29 Jun 2020 13:28:31 +0000 (06:28 -0700)]
x86: Support VEX base opcode length > 1

Intel AMX instructions with 8-bit immediate opcode extension without
operands:

tilerelease, 0, 0x49c0, None, 2, CpuAMX_TILE|Cpu64, Vex|VexOpcode=1|No_bSuf|No_wSuf|No_lSuf|No_sSuf|No_qSuf|No_ldSuf, { 0 }

Update build_vex_prefix to support VEX base opcode length > 1.

* tc-i386.c (build_vex_prefix): Support VEX base opcode length > 1.
(md_assemble): Don't process ImmExt without operands.

4 years agotic6x: Call _bfd_elf_add_dynamic_tags
H.J. Lu [Mon, 29 Jun 2020 12:52:22 +0000 (05:52 -0700)]
tic6x: Call _bfd_elf_add_dynamic_tags

Update tic6x linker tests to expect C6000_DSBT_BASE, C6000_DSBT_SIZE and
C6000_DSBT_INDEX dynamic tags immediately before NULL dynamic tag.

bfd/

* elf32-tic6x.c (elf32_tic6x_size_dynamic_sections): Call
_bfd_elf_add_dynamic_tags.

ld/

* testsuite/ld-tic6x/shlib-1.rd: Move C6000_DSBT_BASE,
C6000_DSBT_SIZE and C6000_DSBT_INDEX dynamic tags to the last.
* testsuite/ld-tic6x/shlib-1b.rd: Likewise.
* testsuite/ld-tic6x/shlib-1r.rd: Likewise.
* testsuite/ld-tic6x/shlib-1rb.rd: Likewise.
* testsuite/ld-tic6x/shlib-app-1.rd: Likewise.
* testsuite/ld-tic6x/shlib-app-1b.rd: Likewise.
* testsuite/ld-tic6x/shlib-app-1r.rd: Likewise.
* testsuite/ld-tic6x/shlib-app-1rb.rd: Likewise.
* testsuite/ld-tic6x/shlib-noindex.rd: Likewise.
* testsuite/ld-tic6x/static-app-1.rd: Likewise.
* testsuite/ld-tic6x/static-app-1b.rd: Likewise.
* testsuite/ld-tic6x/static-app-1r.rd: Likewise.
* testsuite/ld-tic6x/static-app-1rb.rd: Likewise.

4 years ago[gdb/testsuite] Emit unresolved for unknown proc
Tom de Vries [Mon, 29 Jun 2020 11:59:19 +0000 (13:59 +0200)]
[gdb/testsuite] Emit unresolved for unknown proc

Since commit 26783bce15 "[gdb/testsuite] Don't abort testrun for invalid
command in test-case" we don't abort the testrun when encountering an invalid
command.  However, since we don't report errors in the summary, there's a
chance that the error goes unnoticed.

Make the invalid command error more visible by marking the test-case
unresolved, such that we have f.i.:
...
PASS: gdb.python/py-breakpoint.exp: test_bkpt_internal: Test watchpoint write
UNRESOLVED: gdb.python/py-breakpoint.exp: test_bkpt_eval_funcs: \
  testcase aborted due to invalid command name: gdb_py_test_multiple
ERROR: tcl error sourcing py-breakpoint.exp.
ERROR: invalid command name "gdb_py_test_multiple"
    while executing
  ...
                === gdb Summary ===

nr of expected passes            56
nr of unresolved testcases       1
...

Tested on x86_64-linux.

gdb/testsuite/ChangeLog:

2020-06-29  Tom de Vries  <tdevries@suse.de>

* lib/gdb.exp (unknown): Make test-case unresolved.

4 years ago[gdbserver] Add missing include of gdbsupport/agent.h
Tom de Vries [Mon, 29 Jun 2020 10:14:10 +0000 (12:14 +0200)]
[gdbserver] Add missing include of gdbsupport/agent.h

The file gdbserver/ax.h contains:
...
 #ifdef IN_PROCESS_AGENT
 #define debug_threads debug_agent
 #endif
...
but does not declare debug_agent.

Fix this by adding an include of gdbsupport/agent.h.

[ If this fix would have been in place before commit 8118159c69 "[gdbserver] Fix
Wlto-type-mismatch for debug_agent", we would have simply run into this build
breaker with a regular, non-lto build:
...
src/gdbserver/ax.cc:28:5: error: conflicting declaration 'int debug_agent'
 int debug_agent = 0;
     ^~~~~~~~~~~
In file included from src/gdbserver/ax.h:25:0,
                 from src/gdbserver/ax.cc:20:
src/gdbsupport/agent.h:47:13: note: previous declaration as 'bool debug_agent'
 extern bool debug_agent;
             ^~~~~~~~~~~
... ]

Tested on x86_64-linux.

gdbserver/ChangeLog:

2020-06-29  Tom de Vries  <tdevries@suse.de>

* ax.h: Include gdbsupport/debug_agent.h.

4 years ago[gdb/testsuite] Expect conformation question in gdb.server/solib-list.exp
Tom de Vries [Mon, 29 Jun 2020 09:02:06 +0000 (11:02 +0200)]
[gdb/testsuite] Expect conformation question in gdb.server/solib-list.exp

Before commit a8654e7d78 'Fixes PR 25475: ensure exec-file-mismatch "ask"
always asks in case of mismatch', there was a difference in behaviour in
test-case gdb.server/solib-list.exp.

If the executable did not contain debug info (as is usually the case), gdb
would detect a mismatch but not ask for confirmation:
...
(gdb) target remote localhost:2346^M
Remote debugging using localhost:2346^M
warning: Mismatch between current exec-file solib-list^M
and automatically determined exec-file /lib64/ld-2.26.so^M
exec-file-mismatch handling is currently "ask"^M
Reading symbols from /lib64/ld-2.26.so...^M
Reading symbols from /usr/lib/debug/lib64/ld-2.26.so.debug...^M
0x00007ffff7dd7ea0 in _start () at rtld.c:745^M
745     }^M
(gdb) PASS: gdb.server/solib-list.exp: non-stop 0: target remote
...

If the executable did contain debug info (as happens to be the case for
openSUSE), gdb would detect a mismatch and ask for confirmation:
...
(gdb) PASS: gdb.server/solib-list.exp: non-stop 0: file binfile
target remote localhost:2346^M
Remote debugging using localhost:2346^M
warning: Mismatch between current exec-file solib-list^M
and automatically determined exec-file /lib64/ld-2.26.so^M
exec-file-mismatch handling is currently "ask"^M
Load new symbol table from "/lib64/ld-2.26.so"? (y or n) y^M
Reading symbols from /lib64/ld-2.26.so...^M
Reading symbols from /usr/lib/debug/lib64/ld-2.26.so.debug...^M
0x00007ffff7dd7ea0 in _start () at rtld.c:745^M
745     }^M
(gdb) PASS: gdb.server/solib-list.exp: non-stop 0: target remote
...

After commit a8654e7d78, the confirmation is now also asked in case there's
no debug info.

Tighten the test-case by verifying that the confirmation question is asked, as
suggested in the log message of commit a8654e7d78:
...
we can remove the bypass introduced by Tom in 6b9374f1, in order to always
answer to the 'load' question.
...

gdb/testsuite/ChangeLog:

2020-06-29  Tom de Vries  <tdevries@suse.de>

PR gdb/25475
* gdb.server/solib-list.exp: Verify that the symbol reload
confirmation question is asked.

4 years agogas/testsuite: test-case for PR25331 (mmix reloc fixup bug)
Hans-Peter Nilsson [Mon, 29 Jun 2020 03:38:26 +0000 (05:38 +0200)]
gas/testsuite: test-case for PR25331 (mmix reloc fixup bug)

The bug manifested "only" for a 64-bit host:

pr25331.c: Assembler messages:
pr25331.c:430: Error: internal error: fixup not contained within frag
failed with: <pr25331.c: Assembler messages:
pr25331.c:430: Error: internal error: fixup not contained within frag>, no expected output
FAIL: gas/mmix/pr25331

gas:
PR gas/25331
* testsuite/gas/mmix/pr25331.d, testsuite/gas/mmix/pr25331.s: New test.

4 years agogas: Fix mmix fixups and TC_FX_SIZE_SLACK, PR25331
Hans-Peter Nilsson [Mon, 29 Jun 2020 03:32:02 +0000 (05:32 +0200)]
gas: Fix mmix fixups and TC_FX_SIZE_SLACK, PR25331

Finally; sorry for the delay.  There were a few false starts, where I
misinterpreted the error-messages and the comment that Alan added:
it's not the fix size that's too large (and the frag too small), it's
stating the wrong size of what will be "fixed up" - that of the actual
target value, not the size of the field that needs to be adjusted.
Comments added for clarity.

Test-suite committed separately.

gas:
PR gas/25331
* config/tc-mmix.c (md_assemble) <fixup for
BFD_RELOC_MMIX_BASE_PLUS_OFFSET>: This fixup affects 1 byte, not 8.
Also, set its fx_no_overflow.
(md_convert_frag) <case ENCODE_RELAX (STATE_PUSHJSTUB, STATE_ZERO)>:
Similarly this fixup affects 4 bytes, not 8 and needs its
fx_no_overflow set.
* config/tc-mmix.h (TC_FX_SIZE_SLACK): Don't define.

4 years agobinutils/dwarf.c: Correct an `index' global shadowing error for pre-4.8 GCC
Hans-Peter Nilsson [Mon, 29 Jun 2020 01:50:45 +0000 (03:50 +0200)]
binutils/dwarf.c: Correct an `index' global shadowing error for pre-4.8 GCC

In older gcc, shadowing a function name with a local variable name is
flagged as an error, certainly a bug but which is usually worked
around in binutils:

gcc -DHAVE_CONFIG_H -I. -I$SRC/binutils  -I. -I$SRC/binutils -I../bfd -I$SRC/binutils/../bfd -I$SRC/binutils/../include -DLOCALEDIR="\"/usr/local/share/locale\"" -Dbin_dummy_emulation=bin_vanilla_emulation  -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Wshadow -Werror -I$SRC/binutils/../zlib -g -O2 -MT dwarf.o -MD -MP -MF $depbase.Tpo -c -o dwarf.o $SRC/binutils/dwarf.c &&\
mv -f $depbase.Tpo $depbase.Po
cc1: warnings being treated as errors
$SRC/binutils/dwarf.c: In function 'display_debug_str_offsets':
$SRC/binutils/dwarf.c:6913: error: declaration of 'index' shadows a global declaration
/usr/include/string.h:309: error: shadowed declaration is here
make[4]: *** [dwarf.o] Error 1

See also GCC PR c/53066.  This is just another one that crept in since
I and others last had to use an old version.  The name "idx" was used
in the preceding function, display_debug_addr.  Also, it was declared
c99 style (after a statement in the block).  Committed as obvious.

binutils:
* dwarf.c (display_debug_str_offsets): Rename local variable
index to idx.  Move to top of function.

4 years agoasan: _bfd_pei_slurp_codeview_record use of uninit value
Alan Modra [Mon, 29 Jun 2020 00:21:07 +0000 (09:51 +0930)]
asan: _bfd_pei_slurp_codeview_record use of uninit value

Fixes some seriously careless code.  bfd_bread return value is
(bfd_size_type)-1 on error.  "if (bfd_bread (...) < 4)" does not check
for an error since bfd_size_type is unsigned.  In any case, I think we
should be reading and checking the requested length.

* peXXigen.c (_bfd_XXi_slurp_codeview_record): Properly check
return value of bfd_bread.  Don't read more than requested length.
Sanity check length.  Properly terminate file name.

4 years agoC++ comments
Alan Modra [Mon, 29 Jun 2020 00:37:56 +0000 (10:07 +0930)]
C++ comments

binutils isn't c99 (yet).  This replaces or removes some C++ style
comments.

bfd/
* arc-got.h: Use C style comments.
* coff-z80.c: Likewise.
* elf32-csky.c: Likewise.
* peXXigen.c: Likewise.
* elf32-m32c.c (m32c_elf_relax_delete_bytes): Remove commented out
code.
binutils/
* dwarf.c: Use C style comments.
* resrc.c: Likewise.
gas/
* config/tc-s12z.c: Use C style comments.
* config/tc-z80.c: Likewise.
* config/tc-xtensa.c (emit_ld_r_n): Remove commented out code.
include/
* coff/internal.h: Use C style comments.
* coff/pe.h: Likewise.
* elf/ppc64.h: Likewise.
opcodes/
* arm-dis.c: Use C style comments.
* cr16-opc.c: Likewise.
* ft32-dis.c: Likewise.
* moxie-opc.c: Likewise.
* tic54x-dis.c: Likewise.
* s12z-opc.c: Remove useless comment.
* xgate-dis.c: Likewise.

4 years agox86_64 k1om tests
Alan Modra [Mon, 29 Jun 2020 00:34:11 +0000 (10:04 +0930)]
x86_64 k1om tests

On x86_64-nacl we currently see
FAIL: Absolute non-overflowing relocs
FAIL: ld-x86-64/protected2-k1om
FAIL: ld-x86-64/protected3-k1om

This limits the tests as per the l1om variants.

* testsuite/ld-x86-64/abs-k1om.d: Run only on x86_64-*-linux*.
* testsuite/ld-x86-64/protected2-k1om.d: Likewise.
* testsuite/ld-x86-64/protected3-k1om.d: Likewise.

4 years agoAutomatic date update in version.in
GDB Administrator [Mon, 29 Jun 2020 00:00:07 +0000 (00:00 +0000)]
Automatic date update in version.in

4 years agoRemove "cmd_type" function
Tom Tromey [Sun, 28 Jun 2020 16:33:06 +0000 (10:33 -0600)]
Remove "cmd_type" function

The cmd_type function only has a single caller, which is in the CLI
implementation code.  This patch removes the function, and moves the
cmd_types enum definition from command.h to cli-decode.h, fixing an 18
year old FIXME.

gdb/ChangeLog
2020-06-28  Tom Tromey  <tom@tromey.com>

* command.h (cmd_types): Remove.
(cmd_type): Don't declare.
* cli/cli-decode.h (enum cmd_types): Uncomment.  No longer a
typedef.
* cli/cli-cmds.c (setting_cmd): Use cmd->type directly.
* cli/cli-decode.c (cmd_type): Remove.

4 years agoAutomatic date update in version.in
GDB Administrator [Sun, 28 Jun 2020 00:00:07 +0000 (00:00 +0000)]
Automatic date update in version.in

4 years agogdbsupport: add format attribute to print_xml_feature::add_line
Simon Marchi [Sat, 27 Jun 2020 18:38:43 +0000 (14:38 -0400)]
gdbsupport: add format attribute to print_xml_feature::add_line

Fixes this clang error:

      CXX      tdesc.o
    /home/smarchi/src/binutils-gdb/gdbsupport/tdesc.cc:444:25: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
      string_vappendf (tmp, fmt, ap);
                            ^~~

There is already a but about GCC not emitting this warning:

    https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82206

gdbsupport/ChangeLog:

* tdesc.h (class print_xml_feature) <add_line>: Add
ATTRIBUTE_PRINTF.

Change-Id: I7014075e83717f6d7e19d044a3675ff9981ebe17

4 years agoMake {get,set}_inferior_io_terminal inferior methods
Pedro Alves [Sat, 27 Jun 2020 13:56:05 +0000 (14:56 +0100)]
Make {get,set}_inferior_io_terminal inferior methods

This converts the get_inferior_io_terminal and
set_inferior_io_terminal free functions to inferior methods.

Since the related commands are called "tty", "{set,show}
inferior-tty", and MI's "-inferior-tty-{set,show}", to make the
connection between the commands and the code more obvious, the methods
are named set_tty/tty instead of set_io_terminal/io_terminal.

gdb/ChangeLog:

* fork-child.c (prefork_hook): Adjust.
* infcmd.c (set_inferior_io_terminal, get_inferior_io_terminal):
Delete.
(set_inferior_tty_command, show_inferior_tty_command): Adjust.
* inferior.c (inferior::set_tty, inferior::tty): New methods.
* inferior.h (set_inferior_io_terminal, get_inferior_io_terminal):
Remove declarations.
(struct inferior) <set_tty, tty>: New methods.
(struct inferior) <terminal>: Rename to ...
(struct inferior) <m_terminal>: ... this and make private.
* main.c (captured_main_1): Adjust.
* mi/mi-cmd-env.c (mi_cmd_inferior_tty_set): Adjust.
(mi_cmd_inferior_tty_show): Adjust.
* nto-procfs.c (nto_procfs_target::create_inferior): Adjust.
* windows-nat.c (windows_nat_target::create_inferior): Adjust.

4 years agoasan: readelf: use after free in process_archive
Alan Modra [Sat, 27 Jun 2020 03:17:45 +0000 (12:47 +0930)]
asan: readelf: use after free in process_archive

This tidies up in cases where fuzzed thin archives hit the error return
path in setup_nested_archive.

* elfcomm.c (setup_nested_archive): Set nested_arch->file to NULL
after freeing.
(release_archive): Set fields of arch to NULL after freeing.

4 years agoAutomatic date update in version.in
GDB Administrator [Sat, 27 Jun 2020 00:00:06 +0000 (00:00 +0000)]
Automatic date update in version.in

4 years agoMake test names unique in python.exp and guile.exp
Philippe Waroquiers [Mon, 22 Jun 2020 19:56:59 +0000 (21:56 +0200)]
Make test names unique in python.exp and guile.exp

Version 2, handles the comments of Simon and Pedro.

Note that gdb_test_multiline and gdb_py_test_multiple are using
the "input line" as the test name, and so when there is a duplicated
input line (such as a line containing "end"), we have duplicated test
names => as gdb_test_multiline and gdb_py_test_multiple are identical,
as indicated in FIXME, move this to gdb.exp, and make the test name unique
by adding the inputnr to the pass message for each input.

2020-06-26  Philippe Waroquiers  <philippe.waroquiers@skynet.be>

* lib/gdb.exp (gdb_test_multiline): New, moved from gdb-guile.exp,
have a input seq nr in each pass message.
        * lib/gdb-guile.exp (gdb_test_multiline): Move to gdb.exp.
* lib/gdb-python.exp (gdb_py_test_multiple): Remove.
* gdb.python/python.exp: Make test names unique,
use gdb_test_multiline instead of gdb_py_test_multiple,
use $gdb_test_name.
* gdb.guile/guile.exp: Make test names unique, use $gdb_test_name

4 years agox86: Process ImmExt without operands
H.J. Lu [Fri, 26 Jun 2020 17:24:59 +0000 (10:24 -0700)]
x86: Process ImmExt without operands

To support Intel AMX instructions with 8-bit immediate opcode extension,
but without operands:

tilerelease, 0, 0x49, 0xc0, 1, CpuAMX_TILE|Cpu64, Vex|VexOpcode=1|No_bSuf|No_wSuf|No_lSuf|No_sSuf|No_qSuf|No_ldSuf|ImmExt, { 0 }

process ImmExt without operands.

* config/tc-i386.c (md_assemble): Process ImmExt without
operands.

4 years agoi386-opc.tbl: Add a blank line
H.J. Lu [Fri, 26 Jun 2020 16:24:19 +0000 (09:24 -0700)]
i386-opc.tbl: Add a blank line

* i386-opc.tbl: Add a blank line.

4 years agox86: Correct VexSIB128 to VecSIB128
H.J. Lu [Fri, 26 Jun 2020 16:20:28 +0000 (09:20 -0700)]
x86: Correct VexSIB128 to VecSIB128

4 years agox86: Rename VecSIB to SIB for Intel AMX
H.J. Lu [Fri, 26 Jun 2020 15:24:30 +0000 (08:24 -0700)]
x86: Rename VecSIB to SIB for Intel AMX

Rename VecSIB to SIB to support Intel Advanced Matrix Extensions which
introduces instructions with a mandatory SIB byte which isn't a vector
SIB (VSIB).

gas/

* config/tc-i386.c (check_VecOperands): Replace vecsib with sib.
Replace VecSIB128, VecSIB256 and VecSIB512 with VECSIB128,
VECSIB256 and VECSIB512, respectively.
(build_modrm_byte): Replace vecsib with sib.

opcodes/

* i386-gen.c (opcode_modifiers): Replace VecSIB with SIB.
(VecSIB128): Renamed to ...
(VECSIB128): This.
(VecSIB256): Renamed to ...
(VECSIB256): This.
(VecSIB512): Renamed to ...
(VECSIB512): This.
(VecSIB): Renamed to ...
(SIB): This.
(i386_opcode_modifier): Replace vecsib with sib.
* i386-opc.tbl (VexSIB128): New.
(VecSIB256): Likewise.
(VecSIB512): Likewise.
Replace VecSIB=1, VecSIB=2 and VecSIB=3 with VexSIB128, VecSIB256
and VecSIB512, respectively.

4 years agolibctf: support platforms with separate libintl
Nick Alcock [Mon, 16 Dec 2019 14:17:26 +0000 (14:17 +0000)]
libctf: support platforms with separate libintl

We were not using the right configure machinery to spot libintl on
platforms where it was required, leading to the spurious failure of
various configure tests (e.g. for things like ELF support in BFD).

libctf/
* aclocal.m4: Add config/gettext-sister.m4: Shuffle into
alphabetical order.
* configure.ac: Add ZW_GNU_GETTEXT_SISTER_DIR.
* config.h.in: Regenerated.
* Makefile.in: Likewise.
* configure: Likewise.

4 years agolibctf: add some missing #includes.
Nick Alcock [Mon, 16 Dec 2019 11:38:57 +0000 (11:38 +0000)]
libctf: add some missing #includes.

Causes warnings on (at least) recent FreeBSD.

libctf/
* ctf-create.c: Include <unistd.h>.
* ctf-open-bfd.c: Likewise.

4 years agolibctf, elfcpp, gold: do not assume that <byteswap.h> contains bswap_*
Nick Alcock [Fri, 13 Dec 2019 15:19:17 +0000 (15:19 +0000)]
libctf, elfcpp, gold: do not assume that <byteswap.h> contains bswap_*

At least one C library (uclibc-ng) defines some of these only when
the compiler is GCC.  We might as well test for all three cases and
handle any of them being missing.

Very similar code exists in libctf and split between elfcpp and gold:
fix both.

(Also sync up elfcpp with a change made to libctf swap.h a few months
ago: since there is no out-of-line definition of the bswap replacements,
they should be declared static inline, not just inline, to prevent the
linker generating out-of-line references to them.)

PR libctf/25120
libctf/
* configure.ac: Check for bswap_16, bswap_32, and bswap_64 decls.
* swap.h (bswap_16): Do not assume that presence of <byteswap.h>
means this is declared.
(bswap_32): Likewise.
(bswap_64): Likewise.
(bswap_identity_64): Remove, unused.
* configure: Regenerated.
* config.h.in: Likewise.
gold/
* configure.ac: Check for bswap_16, bswap_32, and bswap_64 decls.
* configure: Regenerated.
* config.h.in: Likewise.
elfcpp/
* elfcpp_swap.h (bswap_16): Do not assume that presence of
<byteswap.h> means this is declared.  Make static inline, matching
recent change to libctf, since there is no non-inline definition
of these functions.
(bswap_32): Likewise.
(bswap_64): Likewise.

4 years agolibctf: work with compilers not supporting GNU C attributes
Nick Alcock [Fri, 13 Dec 2019 14:54:09 +0000 (14:54 +0000)]
libctf: work with compilers not supporting GNU C attributes

The obvious fallback __attribute__ stanza was missing.

Thanks to Harald van Dijk.

PR 25120
libctf/
* ctf-impl.h (_libctf_printflike_): Add non-GNU-C fallback.
(_libctf_unlikely_): Likewise.
(_libctf_unused): Likewise.
(_libctf_malloc_): Likewise.

4 years agolibctf: avoid nonportable __thread in CTF archive handling
Nick Alcock [Fri, 13 Dec 2019 14:48:07 +0000 (14:48 +0000)]
libctf: avoid nonportable __thread in CTF archive handling

This keeps archive searching threadsafe using the new bsearch_r that was
just added to libiberty.

PR25120
libctf/
* ctf-archive.c (search_nametbl): No longer global: declare...
(ctf_arc_open_by_name_internal): ... here. Use bsearch_r.
(search_modent_by_name): Take and use ARG for the nametbl.

4 years agolibctf, binutils: support CTF archives like objdump
Nick Alcock [Fri, 13 Dec 2019 12:01:12 +0000 (12:01 +0000)]
libctf, binutils: support CTF archives like objdump

objdump and readelf have one major CTF-related behavioural difference:
objdump can read .ctf sections that contain CTF archives and extract and
dump their members, while readelf cannot.  Since the linker often emits
CTF archives, this means that readelf intermittently and (from the
user's perspective) randomly fails to read CTF in files that ld emits,
with a confusing error message wrongly claiming that the CTF content is
corrupt.  This is purely because the archive-opening code in libctf was
needlessly tangled up with the BFD code, so readelf couldn't use it.

Here, we disentangle it, moving ctf_new_archive_internal from
ctf-open-bfd.c into ctf-archive.c and merging it with the helper
function in ctf-archive.c it was already using.  We add a new public API
function ctf_arc_bufopen, that looks very like ctf_bufopen but returns
an archive given suitable section data rather than a ctf_file_t: the
archive is a ctf_archive_t, so it can be called on raw CTF dictionaries
(with no archive present) and will return a single-member synthetic
"archive".

There is a tiny lifetime tweak here: before now, the archive code could
assume that the symbol section in the ctf_archive_internal wrapper
structure was always owned by BFD if it was present and should always be
freed: now, the caller can pass one in via ctf_arc_bufopen, wihch has
the usual lifetime rules for such sections (caller frees): so we add an
extra field to track whether this is an internal call from ctf-open-bfd,
in which case we still free the symbol section.

include/
* ctf-api.h (ctf_arc_bufopen): New.
libctf/
* ctf-impl.h (ctf_new_archive_internal): Declare.
(ctf_arc_bufopen): Remove.
(ctf_archive_internal) <ctfi_free_symsect>: New.
* ctf-archive.c (ctf_arc_close): Use it.
(ctf_arc_bufopen): Fuse into...
(ctf_new_archive_internal): ... this, moved across from...
* ctf-open-bfd.c: ... here.
(ctf_bfdopen_ctfsect): Use ctf_arc_bufopen.
* libctf.ver: Add it.
binutils/
* readelf.c (dump_section_as_ctf): Support .ctf archives using
ctf_arc_bufopen.  Automatically load the .ctf member of such
archives as the parent of all other members, unless specifically
overridden via --ctf-parent.  Split out dumping code into...
(dump_ctf_archive_member): ... here, as in objdump, and call
it once per archive member.
(dump_ctf_indent_lines): Code style fix.

4 years agolibctf: create: forwards are always in the namespace of their referent
Nick Alcock [Tue, 5 Nov 2019 17:57:55 +0000 (17:57 +0000)]
libctf: create: forwards are always in the namespace of their referent

The C namespace a forward is located in is always the same as the
namespace of the corresponding complete type: 'struct foo' is in the
struct namespace and does not collide with, say, 'union foo'.

libctf allowed for this in many places, but inconsistently: in
particular, forward *addition* never allowed for this, and was interning
forwards in the default namespace, which is always wrong, since you can
only forward structs, unions and enums, all of which are in their own
namespaces in C.

Forward removal needs corresponding adjustment to remove the names form
the right namespace, as does ctf_rollback.

libctf/
* ctf-create.c (ctf_add_forward): Intern in the right namespace.
(ctf_dtd_delete): Remove correspondingly.
(ctf_rollback): Likewise.

4 years agolibctf: create: ctf_add_type should hand back already-added non-SoUs
Nick Alcock [Tue, 5 Nov 2019 13:09:57 +0000 (13:09 +0000)]
libctf: create: ctf_add_type should hand back already-added non-SoUs

When we add a type from a dictionary and then try to add it again, we
should hand it back unchanged unless it is a structure, union or enum
with a different number of members.  That's what the comment says we do.

Instead, we hand it back unchanged *only* if it is a structure, union or
enum with the same number of members: non-structs, unions and enums are
unconditionally added.  This causes extreme type bloating and (in
conjunction with the bug fixed by the next commit) can easily lead to
the same type being mistakenly added to a dictionary more than once
(which, for forwards, was not banned and led to dictionary corruption).

libctf/
* ctf-create.c (ctf_add_type_internal): Hand back existing types
unchanged.

4 years agolibctf: create: don't add forwards if the type added already exists
Nick Alcock [Mon, 21 Oct 2019 10:33:19 +0000 (11:33 +0100)]
libctf: create: don't add forwards if the type added already exists

This is what ctf_add_forward is documented to do, but it's not what it
actually does: the code is quite happy to add forwards that duplicate
existing structs, etc.

This is obviously wrong and breaks both the nondeduplicating linker
and the upcoming deduplicator, as well as allowing ordinary callers of
ctf_add_type to corrupt the dictionary by just adding the same root-
visible forward more than once.

libctf/
* ctf-create.c (ctf_add_forward): Don't add forwards to
types that already exist.

4 years agolibctf: create: non-root-visible types should not appear in name tables
Nick Alcock [Mon, 21 Oct 2019 10:27:43 +0000 (11:27 +0100)]
libctf: create: non-root-visible types should not appear in name tables

We were accidentally interning newly-added and newly-opened
non-root-visible types into name tables, and removing names from name
tables when such types were removed.  This is very wrong: the whole
point of non-root-visible types is they do not go in name tables and
cannot be looked up by name.  This bug made non-root-visible types
basically identical to root-visible types, right back to the earliest
days of libctf in the Solaris era.

libctf/
* ctf-open.c (init_types): Only intern root-visible types.
* ctf-create.c (ctf_dtd_insert): Likewise.
(ctf_dtd_delete): Only remove root-visible types.
(ctf_rollback): Likewise.
(ctf_add_generic): Adjust.
(ctf_add_struct_sized): Adjust comment.
(ctf_add_union_sized): Likewise.
(ctf_add_enum): Likewise.
* ctf-impl.h (ctf_dtd_insert): Adjust prototype.

4 years agobinutils, ld: work with --disable-libctf
Nick Alcock [Mon, 30 Sep 2019 14:04:05 +0000 (15:04 +0100)]
binutils, ld: work with --disable-libctf

This unfortunately means conditionalizing out all the libctf code, but
the result is not too unbearably ugly, if a bit repetitive.  I have
stubbed out code in the !ENABLE_LIBCTF path to avoid extra redundant
ifdefs where it seems that might be helpful.  (The stubs are not too
disruptive, but I've tried to keep them on one line where possible to
avoid filling up the screen with stubs that nobody would care about.
If this is too much of a coding style violation I can change it.)

Changes since v2: use GCC_ENABLE rather than repeating all the
                  AC_ARG_ENABLE stuff over and over again.

ld/
* configure.ac [--enable-libctf]: New, default yes.
Set ENABLE_LIBCTF accordingly.
* Makefile.am [!ENABLE_LIBCTF]: Empty LIBCTF.
* configure: Regenerate.
* config.in: Regenerate.
* Makefile.in: Regenerate.
* aclocal.m4: Regenerate.
* ldlang.c (ctf_output): Conditionalize on ENABLE_LIBCTF.
(ldlang_open_ctf): Likewise.
(lang_merge_ctf): Likewise.
(ldlang_ctf_apply_strsym): Likewise.
(lang_write_ctf): Likewise.
(ldlang_write_ctf_late): Likewise.
(ldlang_open_ctf) [!ENABLE_LIBCTF]: Warn about the presence of CTF
sections.
(lang_merge_ctf) [!ENABLE_LIBCTF]: New stub.
(ldlang_ctf_apply_strsym) [!ENABLE_LIBCTF]: Likewise.
(lang_write_ctf) [!ENABLE_LIBCTF]: Likewise.
(ldlang_write_ctf_late) [!ENABLE_LIBCTF]: Likewise.
* ldelfgen.c (ldelf_emit_ctf_early): Conditionalize on
ENABLE_LIBCTF.
(struct ctf_strsym_iter_cb_arg): Likewise.
(ldelf_ctf_strtab_iter_cb): Likewise.
(ldelf_ctf_symbols_iter_cb): Likewise.
(ldelf_examine_strtab_for_ctf): Likewise.
(ldelf_emit_ctf_early) [!ENABLE_LIBCTF]: New stub.
(ldelf_examine_strtab_for_ctf) [!ENABLE_LIBCTF]: New stub.

binutils/
* configure.ac [--enable-libctf]: New, default yes.
Set ENABLE_LIBCTF accordingly.
* Makefile.am [!ENABLE_LIBCTF]: Empty LIBCTF and LIBCTF_NOBFD.
* configure: Regenerate.
* config.in: Regenerate.
* Makefile.in: Regenerate.
* aclocal.m4: Regenerate.
* objdump.c (usage): Conditionalize portions on ENABLE_LIBCTF.
(option_values): Likewise.
(long_options): Likewise.
(main): Likewise.
(dump_ctf_indent_lines): Conditionalize out when !ENABLE_LIBCTF.
(make_ctfsect): Likewise.
(dump_ctf_archive_member): Likewise.
(dump_ctf) [ENABLE_LIBCTF]: Likewise.
(dump_ctf) [!ENABLE_LIBCTF]: New empty stub.
* readelf.c (options): Conditionalize portions on ENABLE_LIBCTF.
(usage): Likewise.
(process_section_contents): Likewise.
(shdr_to_ctf_sect): Conditionalize out when !ENABLE_LIBCTF.
(dump_ctf_indent_lines): Likewise.
(dump_section_as_ctf) [ENABLE_LIBCTF]: Likewise.

4 years agoFix --enable-libctf and --disable-static
Nick Alcock [Wed, 9 Oct 2019 09:42:00 +0000 (10:42 +0100)]
Fix --enable-libctf and --disable-static

This fixes test runs and compilation when --disable-libctf,
--disable-static, or --enable-shared are passed.

Changes since v2: Use GCC_ENABLE and fix indentation.  Fix prototype
using 'void'.  Use 'unsupported' and gdb_caching_proc.

Changes since v3: Adapt to upstream changes providing skip_ctf_tests.

Changes since v4: Adapt to upstream changes in the seven months (!)
since I last looked at this.

gdb/ChangeLog
* configure.ac: Add --enable-libctf: handle --disable-static
properly.
* acinclude.m4: sinclude ../config/enable.m4.
* Makefile.in (aclocal_m4_deps): Adjust accordingly.
(LIBCTF): Substitute in.
(CTF_DEPS): New, likewise.
(CLIBS): libctf needs symbols from libbfd: move earlier.
(CDEPS): Use CTF_DEPS, not LIBCTF, now LIBCTF can include rpath
flags.
* ctfread.c: Surround in ENABLE_LIBCTF.
(elfctf_build_psymtabs) [!ENABLE_LIBCTF]: New stub.
* configure: Regenerate.
* config.in: Likewise.

gdb/testsuite/ChangeLog
* configure.ac: Add --enable-libctf.
* aclocal.m4: sinclude ../config/enable.m4.
* Makefile.in (site.exp): Add enable_libctf to site.exp.
* lib/gdb.exp (skip_ctf_tests): Use it.
* gdb.base/ctf-constvars.exp: Error message tweak.
* gdb.base/ctf-ptype.exp: Likewise.
* configure: Regenerate.

4 years agox86: make I disassembler macro available for new use
Jan Beulich [Fri, 26 Jun 2020 14:44:31 +0000 (16:44 +0200)]
x86: make I disassembler macro available for new use

The {Q|Q} construct on lgdt/lidt and the slightly different {Q|IQ}
struck me as odd when encountering. Realizing they both have the same
effect, let's free up the I macro by using the former form of construct
where needed (there aren't that many uses overall).

With there now being several multi-character macros also re-do "alt"
handling slightly: Terminate it when finding the closing brace, rather
than after the next single character. Also set the flag only when
actually in Intel syntax mode.

4 years agold/x86: actually invoke k1om tests
Jan Beulich [Fri, 26 Jun 2020 14:43:38 +0000 (16:43 +0200)]
ld/x86: actually invoke k1om tests

Unlike claimed by the ChangeLog entries added by 7a9068fe1646 ("Add
initial Intel K1OM support") these tests did never get enabled.

4 years agox86: fix processing of -M disassembler option
Jan Beulich [Fri, 26 Jun 2020 14:42:55 +0000 (16:42 +0200)]
x86: fix processing of -M disassembler option

Multiple -M options can be specified in any order. Therefore stright
assignment to fields affected needs to be avoided, such that earlier
options' effects won't be discarded. This was in particular a problem
for -Msuffix followed by certain of the other sub-options.

While updating documentation, take the liberty and also drop the
redundant mentioning of being able to comma-separate multiple options.

4 years agoFix -Wstring-compare testcase build failure
Gary Benson [Fri, 26 Jun 2020 13:53:28 +0000 (14:53 +0100)]
Fix -Wstring-compare testcase build failure

Clang fails to compile the file gdb/testsuite/gdb.cp/try_catch.cc
with the following error:
  warning: result of comparison against a string literal is
  unspecified (use strncmp instead) [-Wstring-compare]

This commit fixes the error, replacing the pointer comparison with
a call to strcmp.  This commit also adds a final check: the test
program is run to the final return statement, and the value of
"test" is checked to ensure it is still "true" at that point.

gdb/testsuite/ChangeLog:

* gdb.cp/try_catch.cc: Include string.h.
(main): Replace comparison against string literal with
strcmp, avoiding build failure with -Wstring-compare.
Add "marker test-complete".
* gdb.cp/try_catch.exp: Run the test to the above marker,
then verify that the value of "test" is still true.

4 years agoImprove documentation of which shell is used by GDB's shell commands
Eli Zaretskii [Fri, 26 Jun 2020 06:51:45 +0000 (09:51 +0300)]
Improve documentation of which shell is used by GDB's shell commands

gdb/doc/ChangeLog:

2020-06-26  Eli Zaretskii  <eliz@gnu.org>

* gdb.texinfo (Shell Commands): More accurate description of use
of $SHELL.  Reported by Sandra Loosemore <sandra@codesourcery.com>.

4 years agom68k: tag floating-point ABI used
Pat Bernardi [Thu, 25 Jun 2020 20:05:38 +0000 (16:05 -0400)]
m68k: tag floating-point ABI used

This patch adds GNU attribute support to m68k and utilises it to tag the
floating-point calling convention used (hard-float or soft-float). It enables
the linker to ensure linked objects use a consistent floating-point ABI and
allows tools like GDB to infer the ABI used from the ELF file. It is based on
similar work done for PowerPC.

bfd/
* elf32-m68k.c (m68k_elf_merge_obj_attributes): New function.
(elf32_m68k_merge_private_bfd_data): Merge GNU attributes.
binutils/
* readelf.c (display_m68k_gnu_attribute): New function.
(process_arch_specific): Call display_m68k_gnu_attribute for EM_68K.
gas/
* config/tc-m68k.c (m68k_elf_gnu_attribute): New function.
(md_pseudo_table): Handle "gnu_attribute".
* doc/as.texi: Document GNU attribute for M68K.
include/
* elf/m68k.h: Add enum for GNU object attribute with floating point
tag name and values.
ld/
* testsuite/ld-m68k/attr-gnu-4-0.s: New file.
* testsuite/ld-m68k/attr-gnu-4-1.s: Likewise.
* testsuite/ld-m68k/attr-gnu-4-2.s: Likewise.
* testsuite/ld-m68k/attr-gnu-4-00.d: Likewise.
* testsuite/ld-m68k/attr-gnu-4-01.d: Likewise.
* testsuite/ld-m68k/attr-gnu-4-02.d: Likewise.
* testsuite/ld-m68k/attr-gnu-4-10.d: Likewise.
* testsuite/ld-m68k/attr-gnu-4-11.d: Likewise.
* testsuite/ld-m68k/attr-gnu-4-12.d: Likewise.
* testsuite/ld-m68k/attr-gnu-4-20.d: Likewise.
* testsuite/ld-m68k/attr-gnu-4-21.d: Likewise.
* testsuite/ld-m68k/attr-gnu-4-22.d: Likewise.
* testsuite/ld-m68k/m68k.exp: Run the new tests.

4 years agoRISCV changes broke 32-bit --enable-targets=all
Alan Modra [Fri, 26 Jun 2020 01:22:30 +0000 (10:52 +0930)]
RISCV changes broke 32-bit --enable-targets=all

By the look of it, git commit 39ff0b812324 broke 32-bit host
--enable-targets=all binutils builds.

/usr/local/bin/ld: ../opcodes/.libs/libopcodes.a(riscv-dis.o): in function `parse_riscv_dis_option':
/home/alan/src/binutils-gdb/opcodes/riscv-dis.c:102: undefined reference to `riscv_get_priv_spec_class'
collect2: error: ld returned 1 exit status
Makefile:925: recipe for target 'objdump' failed

The problem is that elfxx-riscv.c is not built for a 32-bit host
without --enable-64-bit-bfd or unless RISCV is given specifically as a
target.  No such trimming of 64-bit only targets is done in opcodes.

One solution is to move these support functions to cpu-riscv.c, which
runs into "error: implicit declaration of function ‘xmalloc’".  Now,
xmalloc is not supposed to be used in libbfd or libopcodes - it's rude
to crash out of an application that calls libbfd or libopcodes
functions without giving it a chance to deal with out-of-memory
itself.  So I removed the xmalloc and instead used a fixed size
buffer.  If you are worried about adding 36 bytes for the buffer to
the riscv_get_priv_spec_class_from_numbers stack frame size, then you
have no idea of the likely xmalloc + malloc stack frame size!  Trying
to reduce memory usage is commendable, but in this instance
riscv_estimate_digit and malloc for a temp buffer uses a lot more
memory than a fixed max-size buffer.

* elfxx-riscv.c (struct priv_spec_t, priv_specs),
(riscv_get_priv_spec_class, riscv_get_priv_spec_class_from_numbers),
(riscv_get_priv_spec_name): Move to..
* cpu-riscv.c: ..here.
(riscv_get_priv_spec_class_from_numbers): Don't xmalloc temp buffer.
Use %u to print unsigned numbers.

4 years agoAutomatic date update in version.in
GDB Administrator [Fri, 26 Jun 2020 00:00:07 +0000 (00:00 +0000)]
Automatic date update in version.in

4 years agogdb: use make_unique_xstrdup in set_inferior_io_terminal
Simon Marchi [Thu, 25 Jun 2020 21:06:18 +0000 (17:06 -0400)]
gdb: use make_unique_xstrdup in set_inferior_io_terminal

gdb/ChangeLog:

* infcmd.c (set_inferior_io_terminal): Use make_unique_xstrdup.

Change-Id: I38b6e753f58947531fe4a293d574bc27ec128f47

4 years agogdb: make inferior::terminal a unique ptr
Simon Marchi [Thu, 25 Jun 2020 18:44:13 +0000 (14:44 -0400)]
gdb: make inferior::terminal a unique ptr

This changes the inferior::terminal field to be a unique pointer, so its
deallocation is automatically managed.

gdb/ChangeLog:

* inferior.h (struct inferior) <terminal>: Change type to
gdb::unique_xmalloc_ptr<char>.
* inferior.c (inferior::~inferior): Don't free inf->terminal.
* infcmd.c (set_inferior_io_terminal): Don't free terminal
field, adjust to unique pointer.
(get_inferior_io_terminal): Adjust to unique pointer.

Change-Id: Iedb6459b4f9eeae812b0cb9d514b5707d5107cdb

4 years agocpu: fix offset16 type, update c-calls in bpf.cpu
David Faust [Thu, 25 Jun 2020 18:34:29 +0000 (20:34 +0200)]
cpu: fix offset16 type, update c-calls in bpf.cpu

Correct the type of the offset16 field to HI, and simplify memory
accesses which use it. Also update c-calls in semantics for a
few instructions.

cpu/ChangeLog:

2020-06-25 David Faust  <david.faust@oracle.com>

* bpf.cpu (f-offset16): Change type from INT to HI.
(dxli): Simplify memory access.
(dxsi): Likewise.
(define-endian-insn): Update c-call in semantics.
(dlabs) Likewise.
(dlind) Likewise.

4 years agogdb/riscv: Loop over all registers for 'info all-registers'
Andrew Burgess [Tue, 16 Jun 2020 16:00:33 +0000 (17:00 +0100)]
gdb/riscv: Loop over all registers for 'info all-registers'

Currently the 'info all-registers' command only loops over those
registers that are known to GDB.  Any registers that are unknown, that
is, are mentioned in the target description, but are not something GDB
otherwise knows, will not be displayed.

This feels wrong, so this commit fixes this mistake.  The output of
'info all-registers' now matches 'info registers all'.

gdb/ChangeLog:

* riscv-tdep.c (riscv_print_registers_info): Loop over all
registers, not just the known core set of registers.

gdb/testsuite/ChangeLog:

* gdb.arch/riscv-tdesc-regs.exp: New test cases.

4 years agogdb/riscv: Record information about unknown tdesc registers
Andrew Burgess [Tue, 16 Jun 2020 13:53:12 +0000 (14:53 +0100)]
gdb/riscv: Record information about unknown tdesc registers

Making use of the previous commit, record information about unknown
registers in the target description, and use this to resolve two
issues.

1. Some targets (QEMU) are reporting three register fflags, frm, and
   fcsr, twice, once in the FPU feature, and once in the CSR feature.
   GDB does create two registers with identical names, but this
   is (sort of) fine, we only ever use the first one, and as both
   registers access the same target state things basically work OK.
   The only real problem is that the register names show up twice in
   'info registers all' output.

   In this commit we spot the duplicates of these registers and then
   return NULL when asked for the name of these registers.  This
   causes GDB to hide these registers from the user, fixing this
   problem.

2. Some targets (QEMU) advertise CSRs that GDB then can't read.  The
   problem is these targets also say these CSRs are part of the
   save/restore register groups.

   This means that before an inferior call GDB tries to save all of
   these CSRs, and a failure to read one causes the inferior call to
   be abandoned.

   We already work around this issue to some degree, known CSRs are
   removed from the save/restore groups, despite what the target might
   say.  However, any unknown CSRs are (currently) not removed in this
   way.

   After this commit we keep a log of the register numbers for all
   unknown CSRs, then when asked about the register groups, we
   override the group information for unknown CSRs, removing them from
   the save and restore groups.

gdb/ChangeLog:

* riscv-tdep.c (riscv_register_name): Return NULL for duplicate
fflags, frm, and fcsr registers.
(riscv_register_reggroup_p): Remove unknown CSRs from save and
restore groups.
(riscv_tdesc_unknown_reg): New function.
(riscv_gdbarch_init): Pass riscv_tdesc_unknown_reg to
tdesc_use_registers.
* riscv-tdep.h (struct gdbarch_tdep): Add
unknown_csrs_first_regnum, unknown_csrs_count,
duplicate_fflags_regnum, duplicate_frm_regnum, and
duplicate_fcsr_regnum fields.

gdb/testsuite/ChangeLog:

* gdb.arch/riscv-tdesc-regs.exp: Extend test case.

4 years agogdb: Extend target description processing of unknown registers
Andrew Burgess [Tue, 16 Jun 2020 12:14:04 +0000 (13:14 +0100)]
gdb: Extend target description processing of unknown registers

This commit adds a new step to the processing of a target description
done in tdesc_use_registers, this new step is about how unknown
registers are processed.

Currently an architecture looks through the target description and
calls tdesc_numbered_register for each register is was expecting (or
hoping) to find.  This builds up a map from GDB's register numbers to
the tdesc_reg object.  Later the architecture calls
tdesc_use_registers.

In tdesc_use_registers we build a hash with keys being all the
tdesc_reg object pointers, from this hash we remove all of the
tdesc_reg objects that were assigned register numbers using
tdesc_numbered_register.

Finally we walk through all of the tdesc_reg objects, and if it was
not already assigned a number we assign that register the next
available number.

The problem with this is that the architecture has no visibility of
which unknown registers exist, and which tdesc_feature the register
came from, in some cases this might be important.

For example, on RISC-V GDB overrides the use of
tdesc_register_reggroup_p, with riscv_register_reggroup_p to modify
some of the register group choices.  In this function GDB wants to
treat all registers from a particular feature in a certain way.  This
is fine for registers that GDB knows might be in that feature, but for
unknown registers the RISC-V parts of GDB have no easy way to figure
out which unknown registers exist, and what numbers they were
assigned.

We could figure this information out by probing the register
structures after calling tdesc_use_registers, but this would be
horrible, much better to have tdesc_use_registers tell the
architecture about unknown registers.

This is what this commit does.  A new phase of tdesc_use_registers,
just before the unknown registers are assigned a number, we loop over
each tdesc_reg object, if it has not been assigned a number then we
figure out what number would be assigned and then call back into the
architecture passing the tdesc_feature, register name, and the
proposed register number.

The architecture is free to return the proposed register number, or it
can return a different number (which has a result identical to having
called tdesc_numbered_register).  Alternatively the architecture can
return -1 to indicate the register should be numbered later.

After calling the callback for every tdesc_reg object any registers
still don't have a number assigned (because the architecture returned
-1), then a new register number is assigned, which might be different
from the proposed number that was suggested earlier.

This commit adds the general target-description parts of this
mechanism.  No targets are currently using this code.  The RISC-V
target will make use of this in the next commit.

There should be no user visible changes after this commit.

gdb/ChangeLog:

* target-descriptions.c (tdesc_use_registers): Add new parameter a
callback, use the callback (when not null) to help number unknown
registers.
* target-descriptions.h (tdesc_unknown_register_ftype): New typedef.
(tdesc_use_registers): Add extra parameter to declaration.

4 years agogdb/riscv: Improve support for matching against target descriptions
Andrew Burgess [Tue, 16 Jun 2020 09:02:09 +0000 (10:02 +0100)]
gdb/riscv: Improve support for matching against target descriptions

For the RISC-V target it is desirable if the three floating pointer
status CSRs fflags, frm, and fcsr can be placed into either the FPU
feature or the CSR feature.  This allows different targets to build
the features in a way that better reflects their target.

The change to support this within GDB is fairly simple, so this is
done in this commit, and some tests added to check this new
functionality.

gdb/ChangeLog:

* riscv-tdep.c (value_of_riscv_user_reg): Moved to here from later
in the file.
(class riscv_pending_register_alias): Likewise.
(riscv_register_feature::register_info): Change 'required_p' field
to 'required', and change its type.  Add 'check' member function.
(riscv_register_feature::register_info::check): Define new member
function.
(riscv_xreg_feature): Change initialisation of 'required' field.
(riscv_freg_feature): Likewise.
(riscv_virtual_feature): Likewise.
(riscv_csr_feature): Likewise.
(riscv_check_tdesc_feature): Take extra parameter, the csr
tdesc_feature, rewrite the function to use the new
riscv_register_feature::register_info::check function.
(riscv_gdbarch_init): Pass the csr tdesc_feature where needed.

gdb/testsuite/ChangeLog:

* gdb.arch/riscv-tdesc-loading-01.xml: New file.
* gdb.arch/riscv-tdesc-loading-02.xml: New file.
* gdb.arch/riscv-tdesc-loading-03.xml: New file.
* gdb.arch/riscv-tdesc-loading-04.xml: New file.
* gdb.arch/riscv-tdesc-loading.exp: New file.