From: Joel Brobecker <brobecker@adacore.com>
To: Jan Vrany via Gdb-patches <gdb-patches@sourceware.org>
Cc: lsix@lancelotsix.com, Jan Vrany <jan.vrany@labware.com>,
Joel Brobecker <brobecker@adacore.com>
Subject: Re: [PATCH v2 2/2] ppc: recognize all program traps
Date: Sat, 27 Nov 2021 10:58:54 +0400 [thread overview]
Message-ID: <YaHXLrVm2lNh3hId@adacore.com> (raw)
In-Reply-To: <20211124130926.2412617-3-jan.vrany@labware.com>
On Wed, Nov 24, 2021 at 01:09:26PM +0000, Jan Vrany via Gdb-patches wrote:
> Permanent program breakpoints (ones inserted into the code) other than
> the one GDB uses for POWER (0x7fe00008) did not result in stop but
> caused GDB to loop infinitely.
>
> This was because GDB did not recognize trap instructions other than
> "trap". For example, "tw 12, 4, 4" was not be recognized, causing GDB
> to loop forever.
>
> This commit fixes this by providing POWER specific hook
> (gdbarch_program_breakpoint_here_p) recognizing all tw, twi, td and tdi
> instructions.
>
> Tested on Linux on PowerPC e500 and on QEMU PPC64le.
> ---
> gdb/rs6000-tdep.c | 65 ++++++++++++++++++++++++
> gdb/testsuite/gdb.arch/powerpc-trap.exp | 67 +++++++++++++++++++++++++
> gdb/testsuite/gdb.arch/powerpc-trap.s | 30 +++++++++++
> gdb/testsuite/gdb.arch/ppc64-trap.exp | 67 +++++++++++++++++++++++++
> gdb/testsuite/gdb.arch/ppc64-trap.s | 32 ++++++++++++
> 5 files changed, 261 insertions(+)
> create mode 100644 gdb/testsuite/gdb.arch/powerpc-trap.exp
> create mode 100644 gdb/testsuite/gdb.arch/powerpc-trap.s
> create mode 100644 gdb/testsuite/gdb.arch/ppc64-trap.exp
> create mode 100644 gdb/testsuite/gdb.arch/ppc64-trap.s
>
> diff --git a/gdb/rs6000-tdep.c b/gdb/rs6000-tdep.c
> index 43880fa4426..ed6838b9e16 100644
> --- a/gdb/rs6000-tdep.c
> +++ b/gdb/rs6000-tdep.c
> @@ -6247,6 +6247,69 @@ ppc_process_record (struct gdbarch *gdbarch, struct regcache *regcache,
> return 0;
> }
>
> +/* Used for matching tw, twi, td and tdi instructions for POWER. */
> +
> +static constexpr uint32_t TX_INSN_MASK = 0xFC0007FF;
> +static constexpr uint32_t TW_INSN = 0x7C000008;
> +static constexpr uint32_t TD_INSN = 0x7C000088;
> +
> +static constexpr uint32_t TXI_INSN_MASK = 0xFC000000;
> +static constexpr uint32_t TWI_INSN = 0x0C000000;
> +static constexpr uint32_t TDI_INSN = 0x08000000;
> +
> +static inline bool
> +is_tw_insn (uint32_t insn)
> +{
> + return (insn & TX_INSN_MASK) == TW_INSN;
> +}
> +
> +static inline bool
> +is_twi_insn (uint32_t insn)
> +{
> + return (insn & TXI_INSN_MASK) == TWI_INSN;
> +}
> +
> +static inline bool
> +is_td_insn (uint32_t insn)
> +{
> + return (insn & TX_INSN_MASK) == TD_INSN;
> +}
> +
> +static inline bool
> +is_tdi_insn (uint32_t insn)
> +{
> + return (insn & TXI_INSN_MASK) == TDI_INSN;
> +}
> +
> +/* Implementation of gdbarch_program_breakpoint_here_p for POWER. */
> +
> +static bool
> +rs6000_program_breakpoint_here_p (gdbarch *gdbarch, CORE_ADDR address)
> +{
> + gdb_byte target_mem[PPC_INSN_SIZE];
> +
> + /* Enable the automatic memory restoration from breakpoints while
> + we read the memory. Otherwise we may find temporary breakpoints, ones
> + inserted by GDB, and flag them as permanent breakpoints. */
> + scoped_restore restore_memory
> + = make_scoped_restore_show_memory_breakpoints (0);
Small coding standard violation: Indentation in this case should
remain 2 characters.
> +
> + if (target_read_memory (address, target_mem, PPC_INSN_SIZE) == 0)
> + {
> + uint32_t insn = (uint32_t)extract_unsigned_integer (
> + target_mem, PPC_INSN_SIZE, gdbarch_byte_order_for_code (gdbarch));
Small coding standard violations:
- You need a space after the "(uint32_t)" cast
- For the opening parenthesis in the call to extract_unsigned_integer,
it should go on the next line, indented by 2 characters:
uint32_t insn = (uint32_t) extract_unsigned_integer
(target_mem, PPC_INSN_SIZE, gdbarch_byte_order_for_code (gdbarch));
Other than that, the patch looks good to me.
> + /* Check if INSN is a TW, TWI, TD or TDI instruction. There
> + are multiple choices of such instructions with different registers
> + and / or immediate values but they all cause a break. */
> + if (is_tw_insn (insn) || is_twi_insn (insn) || is_td_insn (insn)
> + || is_tdi_insn (insn))
> + return true;
> + }
> +
> + return false;
> +}
> +
> /* Initialize the current architecture based on INFO. If possible, re-use an
> architecture from ARCHES, which is a list of architectures already created
> during this debugging session.
> @@ -7109,6 +7172,8 @@ rs6000_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
> rs6000_breakpoint::kind_from_pc);
> set_gdbarch_sw_breakpoint_from_kind (gdbarch,
> rs6000_breakpoint::bp_from_kind);
> + set_gdbarch_program_breakpoint_here_p (gdbarch,
> + rs6000_program_breakpoint_here_p);
>
> /* The value of symbols of type N_SO and N_FUN maybe null when
> it shouldn't be. */
> diff --git a/gdb/testsuite/gdb.arch/powerpc-trap.exp b/gdb/testsuite/gdb.arch/powerpc-trap.exp
> new file mode 100644
> index 00000000000..86f8e61b89b
> --- /dev/null
> +++ b/gdb/testsuite/gdb.arch/powerpc-trap.exp
> @@ -0,0 +1,67 @@
> +# Copyright 2020-2021 Free Software Foundation, Inc.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 3 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with this program. If not, see <http://www.gnu.org/licenses/>.
> +#
> +# This file is part of the gdb testsuite.
> +
> +# Test if GDB stops at various trap instructions inserted into
> +# the code.
> +
> +if { ![istarget powerpc-*] } {
> + verbose "Skipping ${gdb_test_file_name}."
> + return
> +}
> +
> +standard_testfile .s
> +if {[prepare_for_testing "failed to prepare" ${testfile} ${srcfile}]} {
> + return -1
> +}
> +
> +if {![runto_main]} {
> + untested "could not run to main"
> + return -1
> +}
> +
> +# Number of expected SIGTRAP's to get. This needs to be kept in sync
> +# with the source file.
> +set expected_traps 3
> +set keep_going 1
> +set count 0
> +
> +# Make sure we have a lower timeout in case GDB doesn't support a particular
> +# instruction. Such instruction will cause GDB to loop infinitely.
> +while {$keep_going} {
> + # Continue to next program breakpoint instruction.
> + gdb_test_multiple "continue" "trap instruction $count causes SIGTRAP" {
> + -re "Program received signal SIGTRAP, Trace/breakpoint trap.*$gdb_prompt $" {
> + pass $gdb_test_name
> +
> + # Advance PC to next instruction
> + gdb_test "set \$pc = \$pc + 4" "" "advance past trap instruction $count"
> +
> + incr count
> + }
> + # We've reached the end of the test.
> + -re "exited with code 01.*$gdb_prompt $" {
> + set keep_going 0
> + }
> + timeout {
> + fail $gdb_test_name
> + set keep_going 0
> + }
> + }
> +}
> +
> +# Verify we stopped at the expected number of SIGTRAP's.
> +gdb_assert {$count == $expected_traps} "all trap instructions triggered"
> diff --git a/gdb/testsuite/gdb.arch/powerpc-trap.s b/gdb/testsuite/gdb.arch/powerpc-trap.s
> new file mode 100644
> index 00000000000..c503cb1e33e
> --- /dev/null
> +++ b/gdb/testsuite/gdb.arch/powerpc-trap.s
> @@ -0,0 +1,30 @@
> +/* This testcase is part of GDB, the GNU debugger.
> +
> + Copyright 2021 Free Software Foundation, Inc.
> +
> + This program is free software; you can redistribute it and/or modify
> + it under the terms of the GNU General Public License as published by
> + the Free Software Foundation; either version 3 of the License, or
> + (at your option) any later version.
> +
> + This program is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + GNU General Public License for more details.
> +
> + You should have received a copy of the GNU General Public License
> + along with this program. If not, see <http://www.gnu.org/licenses/>. */
> +
> +/* To test if GDB stops at various trap instructions inserted into
> + the code. */
> +
> +.global main
> +.type main,function
> +main:
> + ori 0, 0, 0
> + trap
> + tw 12, 2, 2
> + twi 31, 3, 3
> + ori 0, 0, 0
> + blr
> +
> diff --git a/gdb/testsuite/gdb.arch/ppc64-trap.exp b/gdb/testsuite/gdb.arch/ppc64-trap.exp
> new file mode 100644
> index 00000000000..ffd640c681d
> --- /dev/null
> +++ b/gdb/testsuite/gdb.arch/ppc64-trap.exp
> @@ -0,0 +1,67 @@
> +# Copyright 2020-2021 Free Software Foundation, Inc.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 3 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with this program. If not, see <http://www.gnu.org/licenses/>.
> +#
> +# This file is part of the gdb testsuite.
> +
> +# Test if GDB stops at various trap instructions inserted into
> +# the code.
> +
> +if { ![istarget powerpc64-*] } {
> + verbose "Skipping ${gdb_test_file_name}."
> + return
> +}
> +
> +standard_testfile .s
> +if {[prepare_for_testing "failed to prepare" ${testfile} ${srcfile}]} {
> + return -1
> +}
> +
> +if {![runto_main]} {
> + untested "could not run to main"
> + return -1
> +}
> +
> +# Number of expected SIGTRAP's to get. This needs to be kept in sync
> +# with the source file.
> +set expected_traps 5
> +set keep_going 1
> +set count 0
> +
> +# Make sure we have a lower timeout in case GDB doesn't support a particular
> +# instruction. Such instruction will cause GDB to loop infinitely.
> +while {$keep_going} {
> + # Continue to next program breakpoint instruction.
> + gdb_test_multiple "continue" "trap instruction $count causes SIGTRAP" {
> + -re "Program received signal SIGTRAP, Trace/breakpoint trap.*$gdb_prompt $" {
> + pass $gdb_test_name
> +
> + # Advance PC to next instruction
> + gdb_test "set \$pc = \$pc + 4" "" "advance past trap instruction $count"
> +
> + incr count
> + }
> + # We've reached the end of the test.
> + -re "exited with code 01.*$gdb_prompt $" {
> + set keep_going 0
> + }
> + timeout {
> + fail $gdb_test_name
> + set keep_going 0
> + }
> + }
> +}
> +
> +# Verify we stopped at the expected number of SIGTRAP's.
> +gdb_assert {$count == $expected_traps} "all trap instructions triggered"
> diff --git a/gdb/testsuite/gdb.arch/ppc64-trap.s b/gdb/testsuite/gdb.arch/ppc64-trap.s
> new file mode 100644
> index 00000000000..b307018f3ec
> --- /dev/null
> +++ b/gdb/testsuite/gdb.arch/ppc64-trap.s
> @@ -0,0 +1,32 @@
> +/* This testcase is part of GDB, the GNU debugger.
> +
> + Copyright 2021 Free Software Foundation, Inc.
> +
> + This program is free software; you can redistribute it and/or modify
> + it under the terms of the GNU General Public License as published by
> + the Free Software Foundation; either version 3 of the License, or
> + (at your option) any later version.
> +
> + This program is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + GNU General Public License for more details.
> +
> + You should have received a copy of the GNU General Public License
> + along with this program. If not, see <http://www.gnu.org/licenses/>. */
> +
> +/* To test if GDB stops at various trap instructions inserted into
> + the code. */
> +
> +.global main
> +.type main,function
> +main:
> + ori 0, 0, 0
> + trap
> + tw 12, 2, 2
> + twi 31, 3, 3
> + td 12, 2, 2
> + tdi 31, 3, 3
> + ori 0, 0, 0
> + blr
> +
> --
> 2.30.2
>
--
Joel
next prev parent reply other threads:[~2021-11-27 6:58 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-11-23 15:42 [PATCH 1/2] ppc: use 'trap' ('tw, 31, 0, 0', 0x7fe00008) as breakpoint instruction Jan Vrany
2021-11-23 15:42 ` [PATCH 2/2] ppc: recognize all program traps Jan Vrany
2021-11-24 10:43 ` Lancelot SIX
2021-11-24 10:57 ` Lancelot SIX
2021-11-24 13:09 ` [PATCH v2 0/2] " Jan Vrany
2021-11-24 13:09 ` [PATCH v2 1/2] ppc: use 'trap' ('tw, 31, 0, 0', 0x7fe00008) as breakpoint instruction Jan Vrany
2021-11-27 6:45 ` Joel Brobecker
2021-11-29 11:50 ` Jan Vrany
2021-12-01 9:23 ` Joel Brobecker
2021-11-24 13:09 ` [PATCH v2 2/2] ppc: recognize all program traps Jan Vrany
2021-11-27 6:58 ` Joel Brobecker [this message]
2021-11-30 12:17 ` Pedro Alves
2021-12-01 13:52 ` Jan Vrany
2021-12-01 14:30 ` [PATCH v3 0/2] " Jan Vrany
2021-12-01 14:30 ` [PATCH v3 1/2] ppc: use "trap" ("tw, 31, 0, 0") as breakpoint instruction Jan Vrany
2021-12-04 10:16 ` Joel Brobecker
2021-12-01 14:30 ` [PATCH v3 2/2] ppc: recognize all program traps Jan Vrany
2021-12-04 10:18 ` Joel Brobecker
2021-12-07 23:30 ` Pedro Alves
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=YaHXLrVm2lNh3hId@adacore.com \
--to=brobecker@adacore.com \
--cc=gdb-patches@sourceware.org \
--cc=jan.vrany@labware.com \
--cc=lsix@lancelotsix.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).