From: Palmer Dabbelt <palmer@sifive.com>
To: andrew.burgess@embecosm.com
Cc: gdb-patches@sourceware.org, Jim Wilson <jimw@sifive.com>,
andrew.burgess@embecosm.com
Subject: Re: [PATCH] gdb/riscv: Don't error when decoding a 6 or 8 byte instruction
Date: Tue, 04 Jun 2019 23:49:00 -0000 [thread overview]
Message-ID: <mhng-cd2f82bc-a502-4bbc-add5-058da36f13e6@palmer-si-x1c4> (raw)
In-Reply-To: <20190604115054.25306-1-andrew.burgess@embecosm.com>
On Tue, 04 Jun 2019 04:50:54 PDT (-0700), andrew.burgess@embecosm.com wrote:
> If the RISC-V prologue scanner finds a 6 or 8 byte instruction we
> currently throw an internal error, which is not great for the user.
>
> A mechanism already exists in the prologue scanner to leave
> instructions marked as unknown so that we can stop the prologue scan
> without raising an error, this is used for all 2 and 4 byte
> instructions that are not part of the small set the prologue scanner
> actually understands.
>
> This commit changes GDB so that all 6 and 8 byte instructions are
> marked as unknown, rather than causing an error.
>
> gdb/ChangeLog:
>
> * riscv-tdep.c (riscv_insn::decode): Gracefully ignore
> instructions of lengths 6 or 8 bytes.
>
> gdb/testsuite/ChangeLog:
>
> * gdb.arch/riscv-unwind-long-insn-6.s: New file.
> * gdb.arch/riscv-unwind-long-insn-8.s: New file.
> * gdb.arch/riscv-unwind-long-insn.c: New file.
> * gdb.arch/riscv-unwind-long-insn.exp: New file.
> ---
> gdb/ChangeLog | 5 ++
> gdb/riscv-tdep.c | 10 ++--
> gdb/testsuite/ChangeLog | 7 +++
> gdb/testsuite/gdb.arch/riscv-unwind-long-insn-6.s | 45 +++++++++++++++++
> gdb/testsuite/gdb.arch/riscv-unwind-long-insn-8.s | 45 +++++++++++++++++
> gdb/testsuite/gdb.arch/riscv-unwind-long-insn.c | 25 ++++++++++
> gdb/testsuite/gdb.arch/riscv-unwind-long-insn.exp | 60 +++++++++++++++++++++++
> 7 files changed, 193 insertions(+), 4 deletions(-)
> create mode 100644 gdb/testsuite/gdb.arch/riscv-unwind-long-insn-6.s
> create mode 100644 gdb/testsuite/gdb.arch/riscv-unwind-long-insn-8.s
> create mode 100644 gdb/testsuite/gdb.arch/riscv-unwind-long-insn.c
> create mode 100644 gdb/testsuite/gdb.arch/riscv-unwind-long-insn.exp
>
> diff --git a/gdb/riscv-tdep.c b/gdb/riscv-tdep.c
> index 3fc86ab825..bae987cf66 100644
> --- a/gdb/riscv-tdep.c
> +++ b/gdb/riscv-tdep.c
> @@ -1385,10 +1385,12 @@ riscv_insn::decode (struct gdbarch *gdbarch, CORE_ADDR pc)
> m_opcode = OTHER;
> }
> else
> - internal_error (__FILE__, __LINE__,
> - _("unable to decode %d byte instructions in "
> - "prologue at %s"), m_length,
> - core_addr_to_string (pc));
> + {
> + /* This must be a 6 or 8 byte instruction, we don't currently decode
> + any of these, so just ignore it. */
> + gdb_assert (m_length == 6 || m_length == 8);
> + m_opcode = OTHER;
> + }
> }
>
> /* The prologue scanner. This is currently only used for skipping the
> diff --git a/gdb/testsuite/gdb.arch/riscv-unwind-long-insn-6.s b/gdb/testsuite/gdb.arch/riscv-unwind-long-insn-6.s
> new file mode 100644
> index 0000000000..b21b1e10f3
> --- /dev/null
> +++ b/gdb/testsuite/gdb.arch/riscv-unwind-long-insn-6.s
> @@ -0,0 +1,45 @@
> +/* Copyright 2019 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/>. */
> +
> + .option nopic
> + .text
> +
> + .align 1
> + .globl bar
> + .type bar, @function
> +bar:
> + tail 1f
> + .size bar, .-func
> +
> + .align 1
> + .globl func
> + .type func, @function
> +func:
> + /* A fake 6 byte instruction. This is never executed, but the
> + prologue scanner will try to decode it. These long
> + instructions are ISA extensions, I use .byte rather than an
> + actual instruction mnemonic so that the test can be compiled
> + with a toolchain that doesn't include any long instruction
> + extensions. */
> + .byte 0x1f, 0x00, 0x00, 0x00, 0x00, 0x00
> +1:
> + addi sp,sp,-16
> + sw s0,12(sp)
> + addi s0,sp,16
> + nop
> + lw s0,12(sp)
> + addi sp,sp,16
> + jr ra
> + .size func, .-func
> diff --git a/gdb/testsuite/gdb.arch/riscv-unwind-long-insn-8.s b/gdb/testsuite/gdb.arch/riscv-unwind-long-insn-8.s
> new file mode 100644
> index 0000000000..3fad07b59d
> --- /dev/null
> +++ b/gdb/testsuite/gdb.arch/riscv-unwind-long-insn-8.s
> @@ -0,0 +1,45 @@
> +/* Copyright 2019 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/>. */
> +
> + .option nopic
> + .text
> +
> + .align 1
> + .globl bar
> + .type bar, @function
> +bar:
> + tail 1f
> + .size bar, .-func
> +
> + .align 1
> + .globl func
> + .type func, @function
> +func:
> + /* A fake 8 byte instruction. This is never executed, but the
> + prologue scanner will try to decode it. These long
> + instructions are ISA extensions, I use .byte rather than an
> + actual instruction mnemonic so that the test can be compiled
> + with a toolchain that doesn't include any long instruction
> + extensions. */
> + .byte 0x3f, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
> +1:
> + addi sp,sp,-16
> + sw s0,12(sp)
> + addi s0,sp,16
> + nop
> + lw s0,12(sp)
> + addi sp,sp,16
> + jr ra
> + .size func, .-func
> diff --git a/gdb/testsuite/gdb.arch/riscv-unwind-long-insn.c b/gdb/testsuite/gdb.arch/riscv-unwind-long-insn.c
> new file mode 100644
> index 0000000000..d601e2d3d8
> --- /dev/null
> +++ b/gdb/testsuite/gdb.arch/riscv-unwind-long-insn.c
> @@ -0,0 +1,25 @@
> +/* Copyright 2019 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/>. */
> +
> +extern void func (void);
> +extern void bar (void);
> +
> +int
> +main ()
> +{
> + bar ();
> + func ();
> + return 0;
> +}
> diff --git a/gdb/testsuite/gdb.arch/riscv-unwind-long-insn.exp b/gdb/testsuite/gdb.arch/riscv-unwind-long-insn.exp
> new file mode 100644
> index 0000000000..e4bc489720
> --- /dev/null
> +++ b/gdb/testsuite/gdb.arch/riscv-unwind-long-insn.exp
> @@ -0,0 +1,60 @@
> +# Copyright 2019 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 tests GDB's ability to handle 6 and 8 byte instructions in the
> +# RISC-V prologue scanner. These instruction should be ignored, but
> +# should not result in an error that interrupts the debug session.
> +#
> +# Each of the files riscv-unwind-long-insn-*.s include a function
> +# (func) that contains a fake long instruction (6 or 8 bytes) in the
> +# prologue. We trick GDB into parsing the fake instruction by tail
> +# calling from a different function, 'bar' to the middle of 'func'.
> +
> +if {![istarget "riscv*-*-*"]} {
> + verbose "Skipping ${gdb_test_file_name}."
> + return
> +}
> +
> +foreach_with_prefix {insn_size} {6 8} {
> + standard_testfile riscv-unwind-long-insn.c \
> + riscv-unwind-long-insn-${insn_size}.s
> +
> + if {[prepare_for_testing "failed to prepare" $testfile \
> + "$srcfile $srcfile2" debug]} {
> + return -1
> + }
> +
> + if ![runto_main] then {
> + fail "can't run to main"
> + return 0
> + }
> +
> + gdb_breakpoint "bar"
> + gdb_continue_to_breakpoint "bar"
> +
> + # This next single instruction step takes us through a tail-call
> + # from 'bar' into 'func'.
> + gdb_test "si" "func \(\).*"
> +
> + # Now check that we have a sane backtrace.
> + gdb_test "bt" \
> + [multi_line \
> + "#0\[ \t\]*func \\\(\\\) at .*$srcfile2:\[0-9\]+" \
> + "#1\[ \t\]*$hex in main \\\(\\\) at .*$srcfile:\[0-9\]+"] \
> + "Backtrace to the main frame"
> +
> + # Finally finish, and we should end up back in main.
> + gdb_test "finish" "main \\\(\\\) at .*$srcfile:.*"
> +}
Thanks!
next prev parent reply other threads:[~2019-06-04 23:49 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-06-04 11:51 Andrew Burgess
2019-06-04 16:36 ` John Baldwin
2019-06-04 23:49 ` Palmer Dabbelt [this message]
2019-06-05 9:08 ` Andrew Burgess
2019-06-05 9:12 ` Pedro Alves
2019-06-05 9:38 ` Andrew Burgess
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=mhng-cd2f82bc-a502-4bbc-add5-058da36f13e6@palmer-si-x1c4 \
--to=palmer@sifive.com \
--cc=andrew.burgess@embecosm.com \
--cc=gdb-patches@sourceware.org \
--cc=jimw@sifive.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