From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from lndn.lancelotsix.com (vps-42846194.vps.ovh.net [IPv6:2001:41d0:801:2000::2400]) by sourceware.org (Postfix) with ESMTPS id 4CA01385840E for ; Wed, 24 Nov 2021 10:43:57 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 4CA01385840E Received: from ubuntu.lan (unknown [IPv6:2a02:390:9086::635]) by lndn.lancelotsix.com (Postfix) with ESMTPSA id 258E480D5D; Wed, 24 Nov 2021 10:43:56 +0000 (UTC) Date: Wed, 24 Nov 2021 10:43:51 +0000 From: Lancelot SIX To: Jan Vrany Cc: gdb-patches@sourceware.org Subject: Re: [PATCH 2/2] ppc: recognize all program traps Message-ID: <20211124104312.4tclxcwyrvnhgqwv@ubuntu.lan> References: <20211123154237.2335848-1-jan.vrany@labware.com> <20211123154237.2335848-2-jan.vrany@labware.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20211123154237.2335848-2-jan.vrany@labware.com> X-Greylist: Sender succeeded SMTP AUTH, not delayed by milter-greylist-4.5.11 (lndn.lancelotsix.com [0.0.0.0]); Wed, 24 Nov 2021 10:43:56 +0000 (UTC) X-Spam-Status: No, score=-11.6 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, KAM_SHORT, SPF_HELO_NONE, SPF_PASS, TXREP autolearn=ham autolearn_force=no version=3.4.4 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on server2.sourceware.org X-BeenThere: gdb-patches@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Gdb-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 24 Nov 2021 10:44:00 -0000 Hi, I do not know PPC so I will not comment on what actually done, but I have minor remarks noted below. On Tue, Nov 23, 2021 at 03:42:37PM +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 | 66 +++++++++++++++++++++++ > gdb/testsuite/gdb.arch/powerpc-trap.exp | 72 +++++++++++++++++++++++++ > gdb/testsuite/gdb.arch/powerpc-trap.s | 30 +++++++++++ > gdb/testsuite/gdb.arch/ppc64-trap.exp | 72 +++++++++++++++++++++++++ > gdb/testsuite/gdb.arch/ppc64-trap.s | 32 +++++++++++ > 5 files changed, 272 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..eeafca1ff99 100644 > --- a/gdb/rs6000-tdep.c > +++ b/gdb/rs6000-tdep.c > @@ -6247,6 +6247,70 @@ 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) > +{ > + const uint32_t insn_len = 4; > + gdb_byte target_mem[4]; You could probably use 'gdb_byte target_mem[insn_len]' here. > + > + /* 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); > + > + if (target_read_memory (address, target_mem, insn_len) == 0) > + { > + uint32_t insn = (uint32_t)extract_unsigned_integer ( > + target_mem, insn_len, gdbarch_byte_order_for_code (gdbarch)); > + > + /* 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 +7173,8 @@ 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..4e86bd4b9a1 > --- /dev/null > +++ b/gdb/testsuite/gdb.arch/powerpc-trap.exp > @@ -0,0 +1,72 @@ > +# 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 . > +# > +# 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 nex instruction > + gdb_test "set \$pc = \$pc + 4" "" "advance past trap instruction $count" > + > + # # Insert a breakpoint at the program breakpoint instruction so > + # # GDB can step over it. > + # gdb_test "break" \ > + # "Breakpoint $decimal at $hex: file .*$srcfile, line $decimal.*" \ > + # "insert breakpoint at trap instruction $count" The part above is commented. Is this something you forgot to remove before creating your patch? > + incr count > + } > + # We've reached the end of the test. > + -re "exited with code 01.*$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 . */ > + > +/* 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..1341808b9fd > --- /dev/null > +++ b/gdb/testsuite/gdb.arch/ppc64-trap.exp > @@ -0,0 +1,72 @@ > +# 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 . > +# > +# 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 nex instruction > + gdb_test "set \$pc = \$pc + 4" "" "advance past trap instruction $count" > + > + # # Insert a breakpoint at the program breakpoint instruction so > + # # GDB can step over it. > + # gdb_test "break" \ > + # "Breakpoint $decimal at $hex: file .*$srcfile, line $decimal.*" \ > + # "insert breakpoint at trap instruction $count" Similar remark here, did you forget to remove this? Best, Lancelot. > + incr count > + } > + # We've reached the end of the test. > + -re "exited with code 01.*$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 . */ > + > +/* 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 >