From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-wr1-f48.google.com (mail-wr1-f48.google.com [209.85.221.48]) by sourceware.org (Postfix) with ESMTPS id 57E24385737E; Thu, 12 May 2022 14:49:19 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 57E24385737E Authentication-Results: sourceware.org; dmarc=none (p=none dis=none) header.from=palves.net Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=gmail.com Received: by mail-wr1-f48.google.com with SMTP id t6so7629085wra.4; Thu, 12 May 2022 07:49:19 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:subject:date:message-id:mime-version :content-transfer-encoding; bh=/pBxxEmU/YJsR0EAlXVAceV12OuZDjfSD1T+Pap9je0=; b=HtGEPAXHlp0dZB6ubhL/tigpmhtUTmP/AX5H3HDZKOyBDShZnanJyfXd1xMY/d3PdA Di9P+JVmhAG5seZ77s36Tyz/gqQYPUZVUTK6JEgeckYnWvcMGcJg+KTAq60JuEuk7rai LILSHwYaXuxhYMIV4+RrstEN8s0+ok5tx4SIsf0DvwEjAPEYdECzsgu9zZ1MnyN0MyG6 gA+sBuh8I+fJqTuH8lX09mq7TNUyx3kYxjB6ii/ytm4VD/YhrbQcZaHSrRLLDbFffCxt XQ0jwN8A1HNAQ97big+l/li4pKpQTRrhyk4GOdAXYG8//7/5Upb56hlLmIfgZf/suWiQ cfMg== X-Gm-Message-State: AOAM532ckqiYzWsWK0uCeClLNxhaA42044j/YrcUDhdYOawY7BRcVQz5 36dUfgCgXweKb6tZzCyWJlgpzBluHNA= X-Google-Smtp-Source: ABdhPJxN1ouTHbMdtzIL4OVPPWzgEYTvKEQKii0SaaN0nk8SeOPxbnY7TrTTdfaKIs+zooIqukO6jQ== X-Received: by 2002:a5d:68c6:0:b0:20a:d654:6cae with SMTP id p6-20020a5d68c6000000b0020ad6546caemr37503wrw.564.1652366957489; Thu, 12 May 2022 07:49:17 -0700 (PDT) Received: from localhost ([2001:8a0:f924:2600:209d:85e2:409e:8726]) by smtp.gmail.com with ESMTPSA id u13-20020adfdd4d000000b0020c5253d913sm4217805wrm.95.2022.05.12.07.49.16 (version=TLS1_3 cipher=TLS_AES_128_GCM_SHA256 bits=128/128); Thu, 12 May 2022 07:49:16 -0700 (PDT) From: Pedro Alves To: gdb-patches@sourceware.org, binutils@sourceware.org Subject: [PATCH] Fix "gdb --write" with core files Date: Thu, 12 May 2022 15:49:15 +0100 Message-Id: <20220512144915.3270761-1-pedro@palves.net> X-Mailer: git-send-email 2.36.0 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Spam-Status: No, score=-9.6 required=5.0 tests=BAYES_00, FREEMAIL_FORGED_FROMDOMAIN, FREEMAIL_FROM, GIT_PATCH_0, HEADER_FROM_DIFFERENT_DOMAINS, KAM_DMARC_STATUS, KAM_SHORT, RCVD_IN_DNSWL_NONE, RCVD_IN_MSPIKE_H2, SPF_HELO_NONE, SPF_PASS, TXREP, T_SCC_BODY_TEXT_LINE autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) 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: Thu, 12 May 2022 14:49:22 -0000 If you load a core file into GDB with the --write option, or "set write on" (equivalent), and then poke memory expecting it to patch the core binary, you'll notice something odd -- the write seems to succeed, but in reality, it doesn't. The value you wrote doesn't persist. Like so: $ gdb -q --write -c testsuite/outputs/gdb.base/patch/gcore.test [New LWP 615986] Core was generated by `/home/pedro/gdb/build/gdb/testsuite/outputs/gdb.base/patch/patch'. Program terminated with signal SIGTRAP, Trace/breakpoint trap. #0 0x0000555555555131 in ?? () (gdb) p *(unsigned char *)0x0000555555555131 = 1 $1 = 1 '\001' (gdb) p *(unsigned char *)0x0000555555555131 $2 = 185 '\271' (gdb) Diffing hexdumps of before/after patching, reveals that a "0x1" was actually written somewhere in the file. The problem is that the "0x1" was written at the wrong offset in the file... That happens because _bfd_elf_set_section_contents does this to seek to the section's offset: pos = hdr->sh_offset + offset; if (bfd_seek (abfd, pos, SEEK_SET) != 0 || bfd_bwrite (location, count, abfd) != count) return false; ... and 'hdr->sh_offset' is zero, so we seek to just OFFSET, which is incorrect. The reason 'hdr->sh_offset' is zero is that kernel-generated core files normally don't even have a section header table (gdb-generated ones do, but that's more an accident than a feature), and indeed elf_core_file_p doesn't even try to read sections at all: /* Core files are simply standard ELF formatted files that partition the file using the execution view of the file (program header table) rather than the linking view. In fact, there is no section header table in a core file. The process status information (including the contents of the general register set) and the floating point register set are stored in a segment of type PT_NOTE. We handcraft a couple of extra bfd sections that allow standard bfd access to the general registers (.reg) and the floating point registers (.reg2). */ bfd_cleanup elf_core_file_p (bfd *abfd) My first approach was to fix it in _bfd_elf_set_section_contents directly -- if bfd->direction is both_direction, and updating a core, then jump straight to _bfd_generic_set_section_contents, which is documented to work exactly in this scenario: /* This generic function can only be used in implementations where creating NEW sections is disallowed. It is useful in patching existing sections in read-write files, though. See other set_section_contents functions to see why it doesn't work for new sections. */ bool _bfd_generic_set_section_contents (bfd *abfd, However, on second thought, it sounds like this is something that might need to be done too for other formats and/or ELF backends if they install custom versions of set_section_contents, or if when opening they also create handcrafted BFD sections, so I thought it might be better done in common code. And thus that is what this patch does. New GDB testcase included, which exercises both patching an executable and patching a core file. Patching an executable already works without this fix, because in that case BFD reads in the sections table. Still, we had no testcase for that yet. In fact, we have no "set write on" testcases at all, this is the first one. Tested on x86-64 GNU/Linux, gdb, ld, binutils, and gas. Change-Id: I0f49f58b48aabab2e269f2959b8fd8a7fe36fdce --- bfd/section.c | 26 ++++++++ gdb/testsuite/gdb.base/patch.c | 26 ++++++++ gdb/testsuite/gdb.base/patch.exp | 106 +++++++++++++++++++++++++++++++ 3 files changed, 158 insertions(+) create mode 100644 gdb/testsuite/gdb.base/patch.c create mode 100644 gdb/testsuite/gdb.base/patch.exp diff --git a/bfd/section.c b/bfd/section.c index 5a487ce6c6f..23d59385f5f 100644 --- a/bfd/section.c +++ b/bfd/section.c @@ -1507,6 +1507,24 @@ bfd_set_section_contents (bfd *abfd, && location != section->contents + offset) memcpy (section->contents + offset, location, (size_t) count); + /* When a BFD is opened for update, the backend's data structures + may not represent all the sections in the BFD. E.g., when + reading ELF core files, we don't even read their section header + table (elf_core_file_p), while the ELF backend's + set_section_contents implementation assumes it can find a BFD + section's file offset from the interned ELF section table info. + Core files do end up with BFD sections after + open/bfd_check_format, but those sections are made up by BFD and + don't exist in the actual core file, like the "load" and ".reg" + sections. To avoid potentially every backend having to handle + these scenarios, jump directly to the generic version here. */ + if (abfd->direction == both_direction) + { + BFD_ASSERT (abfd->output_has_begun); + return _bfd_generic_set_section_contents (abfd, section, + location, offset, count); + } + if (BFD_SEND (abfd, _bfd_set_section_contents, (abfd, section, location, offset, count))) { @@ -1592,6 +1610,14 @@ bfd_get_section_contents (bfd *abfd, return true; } + /* For consistency with bfd_set_section_contents. */ + if (abfd->direction == both_direction) + { + BFD_ASSERT (abfd->output_has_begun); + return _bfd_generic_get_section_contents (abfd, section, + location, offset, count); + } + return BFD_SEND (abfd, _bfd_get_section_contents, (abfd, section, location, offset, count)); } diff --git a/gdb/testsuite/gdb.base/patch.c b/gdb/testsuite/gdb.base/patch.c new file mode 100644 index 00000000000..c75f7800c94 --- /dev/null +++ b/gdb/testsuite/gdb.base/patch.c @@ -0,0 +1,26 @@ +/* Copyright 2022 Free Software Foundation, Inc. + + This file is part of GDB. + + 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 . */ + +/* Value != 0 so it doesn't ends up in .bss, which we wouldn't be able + to patch. */ +int extern_global = 1; + +int +main() +{ + return 0; +} diff --git a/gdb/testsuite/gdb.base/patch.exp b/gdb/testsuite/gdb.base/patch.exp new file mode 100644 index 00000000000..6d9c17ff740 --- /dev/null +++ b/gdb/testsuite/gdb.base/patch.exp @@ -0,0 +1,106 @@ +# Copyright 2022 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 . + +standard_testfile + +if {[build_executable "failed to prepare" $testfile $srcfile debug]} { + return -1 +} + +# Check that we can patch an executable. + +with_test_prefix "exec" { + clean_restart + + gdb_test_no_output "set write on" + + gdb_load $binfile + + gdb_test "p extern_global" " = 1" "read original value" + gdb_test "p extern_global = 2" " = 2" "modify value" + gdb_test "p extern_global" " = 2" "value modified" + + clean_restart $binfile + + gdb_test "p extern_global" " = 2" "value modified persisted" +} + +# Check that we can patch a core file. + +# Generate a core file. +with_test_prefix "gcore" { + clean_restart $binfile + + if {![runto_main]} { + return + } + + # Extract the value at PC, and add 1, letting it wrap if + # necessary. This is the value we will poke into memory. + set poke_value "" + gdb_test_multiple "p/x (*(unsigned char *) \$pc) + 1" "compute poke value" { + -re -wrap " = ($hex)" { + set poke_value $expect_out(1,string) + pass $gdb_test_name + } + } + if {$poke_value == ""} { + return + } + + set corefile [standard_output_file gcore.test] + set core_supported [gdb_gcore_cmd "$corefile" "save a corefile"] + if {!$core_supported} { + return + } +} + +# Load it into GDB with "set write on", and change the instruction at +# PC. +with_test_prefix "load core, write on" { + # Don't load a binary, we want to make sure we're patching the + # core, not the executable. + clean_restart + + gdb_test_no_output "set write on" + + set core_loaded [gdb_core_cmd "$corefile" "re-load generated corefile"] + if { $core_loaded == -1 } { + # No use proceeding from here. + return + } + + gdb_test "p/x *(char *) \$pc = $poke_value" \ + " = $poke_value" \ + "poke value" + gdb_test "p/x *(char *) \$pc" \ + " = $poke_value" \ + "value modified" +} + +# Re-load it into a new GDB session, now with "set write off", and +# confirm the value change persisted. +with_test_prefix "re-load modified core" { + # Don't load a binary, we want to make sure we've patched the + # core, not the executable. + clean_restart + + set core_loaded [gdb_core_cmd "$corefile" "re-load generated corefile"] + gdb_assert { $core_loaded != -1 } "core re-loaded" + + gdb_test "p/x *(char *) \$pc" \ + " = $poke_value" \ + "value modified persisted" +} base-commit: c8a9e88bf6ff32d90d082d07d3c5d12b938f8335 -- 2.36.0