From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 23505 invoked by alias); 24 Apr 2014 18:39:02 -0000 Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org Received: (qmail 23492 invoked by uid 89); 24 Apr 2014 18:39:01 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-1.3 required=5.0 tests=AWL,BAYES_00,FREEMAIL_FROM,RCVD_IN_DNSWL_LOW,SPF_PASS,UNSUBSCRIBE_BODY autolearn=no version=3.3.2 X-HELO: mail-ve0-f174.google.com Received: from mail-ve0-f174.google.com (HELO mail-ve0-f174.google.com) (209.85.128.174) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with (AES128-SHA encrypted) ESMTPS; Thu, 24 Apr 2014 18:39:00 +0000 Received: by mail-ve0-f174.google.com with SMTP id oz11so3355524veb.5 for ; Thu, 24 Apr 2014 11:38:57 -0700 (PDT) MIME-Version: 1.0 X-Received: by 10.58.23.6 with SMTP id i6mr2323043vef.12.1398364737807; Thu, 24 Apr 2014 11:38:57 -0700 (PDT) Received: by 10.58.243.98 with HTTP; Thu, 24 Apr 2014 11:38:57 -0700 (PDT) In-Reply-To: <20140424183510.GI7588@redacted.bos.redhat.com> References: <20140424183510.GI7588@redacted.bos.redhat.com> Date: Thu, 24 Apr 2014 18:39:00 -0000 Message-ID: Subject: Re: [PATCHv4] aarch64: detect atomic sequences like other ll/sc architectures From: Andrew Pinski To: Kyle McMartin Cc: "gdb-patches@sourceware.org" , Joel Brobecker Content-Type: text/plain; charset=UTF-8 X-IsSubscribed: yes X-SW-Source: 2014-04/txt/msg00506.txt.bz2 On Thu, Apr 24, 2014 at 11:35 AM, Kyle McMartin wrote: > Once more with feeling. :-) Fixed ChangeLog entries and whitespace. > > 2014-04-24 Kyle McMartin > > gdb: > * aarch64-tdep.c (aarch64_software_single_step): New function. > (aarch64_gdbarch_init): Handle single stepping of atomic sequences > with aarch64_software_single_step. > > gdb/testsuite: > * gdb.arch/aarch64-atomic-inst.c: New file. > * gdb.arch/aarch64-atomic-inst.exp: New file. > > --- > gdb/aarch64-tdep.c | 77 ++++++++++++++++++++++++++ > gdb/testsuite/gdb.arch/aarch64-atomic-inst.c | 48 ++++++++++++++++ > gdb/testsuite/gdb.arch/aarch64-atomic-inst.exp | 48 ++++++++++++++++ > 3 files changed, 173 insertions(+) > create mode 100644 gdb/testsuite/gdb.arch/aarch64-atomic-inst.c > create mode 100644 gdb/testsuite/gdb.arch/aarch64-atomic-inst.exp > > diff --git a/gdb/aarch64-tdep.c b/gdb/aarch64-tdep.c > index bba10d8..b64bbde 100644 > --- a/gdb/aarch64-tdep.c > +++ b/gdb/aarch64-tdep.c > @@ -2509,6 +2509,82 @@ value_of_aarch64_user_reg (struct frame_info *frame, const void *baton) > } > > > +/* Implement the "software_single_step" gdbarch method, needed to > + single step through atomic sequences on AArch64. */ > + > +static int > +aarch64_software_single_step (struct frame_info *frame) > +{ > + struct gdbarch *gdbarch = get_frame_arch (frame); > + struct address_space *aspace = get_frame_address_space (frame); > + enum bfd_endian byte_order = gdbarch_byte_order (gdbarch); > + const int insn_size = 4; > + const int atomic_sequence_length = 16; /* Instruction sequence length. */ > + CORE_ADDR pc = get_frame_pc (frame); > + CORE_ADDR breaks[2] = { -1, -1 }; > + CORE_ADDR loc = pc; > + CORE_ADDR closing_insn = 0; > + uint32_t insn = read_memory_unsigned_integer (loc, insn_size, byte_order); I think this is broken for big-endian aarch64. Instructions are always in little-endian mode so we really should be reading them always as little-endian. Thanks, Andrew Pinski > + int index; > + int insn_count; > + int bc_insn_count = 0; /* Conditional branch instruction count. */ > + int last_breakpoint = 0; /* Defaults to 0 (no breakpoints placed). */ > + > + /* Look for a Load Exclusive instruction which begins the sequence. */ > + if (!decode_masked_match (insn, 0x3fc00000, 0x08400000)) > + return 0; > + > + for (insn_count = 0; insn_count < atomic_sequence_length; ++insn_count) > + { > + int32_t offset; > + unsigned cond; > + > + loc += insn_size; > + insn = read_memory_unsigned_integer (loc, insn_size, byte_order); > + > + /* Check if the instruction is a conditional branch. */ > + if (decode_bcond (loc, insn, &cond, &offset)) > + { > + if (bc_insn_count >= 1) > + return 0; > + > + /* It is, so we'll try to set a breakpoint at the destination. */ > + breaks[1] = loc + offset; > + > + bc_insn_count++; > + last_breakpoint++; > + } > + > + /* Look for the Store Exclusive which closes the atomic sequence. */ > + if (decode_masked_match (insn, 0x3fc00000, 0x08000000)) > + { > + closing_insn = loc; > + break; > + } > + } > + > + /* We didn't find a closing Store Exclusive instruction, fall back. */ > + if (!closing_insn) > + return 0; > + > + /* Insert breakpoint after the end of the atomic sequence. */ > + breaks[0] = loc + insn_size; > + > + /* Check for duplicated breakpoints, and also check that the second > + breakpoint is not within the atomic sequence. */ > + if (last_breakpoint > + && (breaks[1] == breaks[0] > + || (breaks[1] >= pc && breaks[1] <= closing_insn))) > + last_breakpoint = 0; > + > + /* Insert the breakpoint at the end of the sequence, and one at the > + destination of the conditional branch, if it exists. */ > + for (index = 0; index <= last_breakpoint; index++) > + insert_single_step_breakpoint (gdbarch, aspace, breaks[index]); > + > + return 1; > +} > + > /* 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. > @@ -2624,6 +2700,7 @@ aarch64_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches) > set_gdbarch_breakpoint_from_pc (gdbarch, aarch64_breakpoint_from_pc); > set_gdbarch_cannot_step_breakpoint (gdbarch, 1); > set_gdbarch_have_nonsteppable_watchpoint (gdbarch, 1); > + set_gdbarch_software_single_step (gdbarch, aarch64_software_single_step); > > /* Information about registers, etc. */ > set_gdbarch_sp_regnum (gdbarch, AARCH64_SP_REGNUM); > diff --git a/gdb/testsuite/gdb.arch/aarch64-atomic-inst.c b/gdb/testsuite/gdb.arch/aarch64-atomic-inst.c > new file mode 100644 > index 0000000..9a73c7a > --- /dev/null > +++ b/gdb/testsuite/gdb.arch/aarch64-atomic-inst.c > @@ -0,0 +1,48 @@ > +/* This file is part of GDB, the GNU debugger. > + > + Copyright 2008-2014 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 . */ > + > +int main(void) > +{ > + unsigned long tmp, cond; > + unsigned long dword = 0; > + > + /* Test that we can step over ldxr/stxr. This sequence should step from > + ldxr to the following __asm __volatile. */ > + __asm __volatile ("1: ldxr %0,%2\n" \ > + " cmp %0,#1\n" \ > + " b.eq out\n" \ > + " add %0,%0,1\n" \ > + " stxr %w1,%0,%2\n" \ > + " cbnz %w1,1b" \ > + : "=&r" (tmp), "=&r" (cond), "+Q" (dword) \ > + : : "memory"); > + > + /* This sequence should take the conditional branch and step from ldxr > + to the return dword line. */ > + __asm __volatile ("1: ldxr %0,%2\n" \ > + " cmp %0,#1\n" \ > + " b.eq out\n" \ > + " add %0,%0,1\n" \ > + " stxr %w1,%0,%2\n" \ > + " cbnz %w1,1b\n" \ > + : "=&r" (tmp), "=&r" (cond), "+Q" (dword) \ > + : : "memory"); > + > + dword = -1; > +__asm __volatile ("out:\n"); > + return dword; > +} > diff --git a/gdb/testsuite/gdb.arch/aarch64-atomic-inst.exp b/gdb/testsuite/gdb.arch/aarch64-atomic-inst.exp > new file mode 100644 > index 0000000..377aebc > --- /dev/null > +++ b/gdb/testsuite/gdb.arch/aarch64-atomic-inst.exp > @@ -0,0 +1,48 @@ > +# Copyright 2008-2014 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, write to the Free Software > +# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. > +# > +# This file is part of the gdb testsuite. > + > +# Test single stepping through atomic sequences beginning with > +# a ldxr instruction and ending with a stxr instruction. > + > +if {![istarget "aarch64*"]} { > + verbose "Skipping ${gdb_test_file_name}." > + return > +} > + > +standard_testfile > +if { [prepare_for_testing ${testfile}.exp ${testfile} ${srcfile}] } { > + return -1 > +} > + > +if ![runto_main] { > + untested "could not run to main" > + return -1 > +} > + > +gdb_breakpoint "[gdb_get_line_number "ldxr"]" \ > + "Breakpoint $decimal at $hex" \ > + "Set the breakpoint at the start of the sequence" > + > +gdb_test "continue" "Continuing.*Breakpoint $decimal.*" \ > + "Continue until breakpoint" > + > +gdb_test "next" ".*__asm __volatile.*" \ > + "Step through the ldxr/stxr sequence" > + > +gdb_test "next" ".*return dword.*" \ > + "Stepped through sequence through conditional branch" > -- > 1.9.0 >