From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx2.freebsd.org (mx2.freebsd.org [IPv6:2610:1c1:1:606c::19:2]) by sourceware.org (Postfix) with ESMTPS id 3229C38485B2 for ; Wed, 18 May 2022 13:58:06 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 3229C38485B2 Authentication-Results: sourceware.org; dmarc=none (p=none dis=none) header.from=FreeBSD.org Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=FreeBSD.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2610:1c1:1:606c::19:1]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) client-signature RSA-PSS (4096 bits)) (Client CN "mx1.freebsd.org", Issuer "R3" (verified OK)) by mx2.freebsd.org (Postfix) with ESMTPS id A8B238587B; Wed, 18 May 2022 13:58:03 +0000 (UTC) (envelope-from jhb@FreeBSD.org) Received: from smtp.freebsd.org (smtp.freebsd.org [IPv6:2610:1c1:1:606c::24b:4]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256 client-signature RSA-PSS (4096 bits) client-digest SHA256) (Client CN "smtp.freebsd.org", Issuer "R3" (verified OK)) by mx1.freebsd.org (Postfix) with ESMTPS id 4L3F1l42fPz4f4P; Wed, 18 May 2022 13:58:03 +0000 (UTC) (envelope-from jhb@FreeBSD.org) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=freebsd.org; s=dkim; t=1652882283; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=Qc/NaU8hHpu05qEDKS2Nr0oTM85o3ctrxsUCtm8LfOg=; b=Yys5wAEQ4h1pTs0vXWiBDnkkeo0Iv956ozALJ7XS53Kg0u5dskmSgadrmKKPpzOFrdnu+H yJvr/BUqjtMT8Jk342oU30T8yEHO+ovzN0u1h0aWnXl9iSCzlpLZ0uuw7y6vSLWaYgd6HS D8z7BDw1Lpl2DKwC5hyno6PSKOhaliqU5MFzGUXBnkl+UkL6ojsOJ1i602xnKUA8EOIP31 2+UhTHeZe6wONCtuPIN5BPElNOW7x5JonrZG7bW+mFXMZf08IuxwiZLta4HzbwtrvMAPjY owqrptgEZUI1KJEIYUr6315hQ/pYPuQThSGPhJ8Q0J/63WJn6Q3NjNsMy/Hxlg== Received: from [IPV6:2601:648:8680:ed60:8ce3:c94a:ed81:31cb] (unknown [IPv6:2601:648:8680:ed60:8ce3:c94a:ed81:31cb]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (Client did not present a certificate) (Authenticated sender: jhb) by smtp.freebsd.org (Postfix) with ESMTPSA id F0F7D22400; Wed, 18 May 2022 13:58:02 +0000 (UTC) (envelope-from jhb@FreeBSD.org) Message-ID: <10cac250-853a-c03e-9a1e-e545c98fa428@FreeBSD.org> Date: Wed, 18 May 2022 06:58:00 -0700 MIME-Version: 1.0 User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:91.0) Gecko/20100101 Thunderbird/91.9.0 Subject: Re: [PATCH, v4] [AArch64] MTE corefile support Content-Language: en-US To: Luis Machado , gdb-patches@sourceware.org References: <20220331140343.9047-1-luis.machado@arm.com> <20220503215632.914608-1-luis.machado@arm.com> <0fbdfe5d-ba10-d2ec-58d2-6fbb300f485b@arm.com> From: John Baldwin In-Reply-To: <0fbdfe5d-ba10-d2ec-58d2-6fbb300f485b@arm.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=freebsd.org; s=dkim; t=1652882283; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=Qc/NaU8hHpu05qEDKS2Nr0oTM85o3ctrxsUCtm8LfOg=; b=YwsyTMjA+1RlbJOtcD0nVTaU5zPwukd7wpMMHW/XIwyJvpCvaExjR9YACp/pN7W+EMc78T ZkAoJkPO1vX6GVd+2U8K4ywiJU6S1iw5r7CPue5cNuqAj68C+IjBu4UeOPj0OuVRFk3E2d 7LI/7GqrRimD/UY0MyoBT5p9r8JWNJrUZMQQAzPTCrptNMEb3Ac0P1P1IQ8esQGDufcxDb lCW6OQRuZKTy8ol9UFSH/+8ovnupWWiS0lryYcBB7cyq/ROd3WC3szaSWwek8UYuXBVO5R hcizYxI3YP3UcAFQePMzOzn6Yw0t2UxlCD4yt7vQA9Bs6dtbvfQsEg3PNW5dPQ== ARC-Seal: i=1; s=dkim; d=freebsd.org; t=1652882283; a=rsa-sha256; cv=none; b=nNIMyCgOWyTx7setwLadtMt31/5i0XwnSVw9fsn4OmJl05Z+ONmsA26TjPicL6fxMrtHNL mh7WF5RtdHGF7DvMHD+OnEMvnXI0+5kzsWl9ENyZeD6llRqotwpfrpxe0M4Tz5xyPQ0ic/ htnt1iQJQQuHEu4+U3qe6GgA9JSlxtvT4Hc+VDBovfh0T+yhfHMXS3h0rS/tT70LTRPIpH fYIYeCFob3RYdiSHNpAJfIA0nEyoV6RxNNKz9InmN1tMoRu+X/4Y1glVpi7QsUEZx9++AX jk2FT2iudlSM7Gtw13BxN8z0G6LunQ5NS6sKgwbQ9Z1nFdchd3eMKFB7SWT0Pw== ARC-Authentication-Results: i=1; mx1.freebsd.org; none X-Spam-Status: No, score=-12.8 required=5.0 tests=BAYES_00, DKIMWL_WL_HIGH, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, KAM_SHORT, NICE_REPLY_A, 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: Wed, 18 May 2022 13:58:12 -0000 From my perspective this looks ok. On 5/18/22 5:46 AM, Luis Machado via Gdb-patches wrote: > Ping? The binutils patch has been approved, but I'd like to push both at the same time. > > On 5/3/22 22:56, Luis Machado via Gdb-patches wrote: >> v4: >> >> - Updated documentation (added cross-references). >> - Updated the segment name from PT_ARM_MEMTAG_MTE to >> PT_AARCH64_MEMTAG_MTE. >> >> v3: >> >> - Updated NEWS and documentation to be more thorough. >> >> v2: >> >> - Rework memory tag section handling to use generic section fields. >> >> -- >> >> Teach GDB how to dump memory tags for AArch64 when using the gcore command >> and how to read memory tag data back from a core file generated by GDB >> (via gcore) or by the Linux kernel. >> >> The format is documented in the Linux Kernel documentation [1]. >> >> Each tagged memory range (listed in /proc//smaps) gets dumped to its >> own PT_AARCH64_MEMTAG_MTE segment. A section named ".memtag" is created for each >> of those segments when reading the core file back. >> >> To save a little bit of space, given MTE tags only take 4 bits, the memory tags >> are stored packed as 2 tags per byte. >> >> When reading the data back, the tags are unpacked. >> >> I've added a new testcase to exercise the feature. >> >> Build-tested with --enable-targets=all and regression tested on aarch64-linux >> Ubuntu 20.04. >> >> [1] Documentation/arm64/memory-tagging-extension.rst (Core Dump Support) >> --- >> gdb/Makefile.in | 1 + >> gdb/NEWS | 10 ++ >> gdb/aarch64-linux-tdep.c | 167 +++++++++++++++++++ >> gdb/arch/aarch64-mte-linux.c | 56 +++++++ >> gdb/arch/aarch64-mte-linux.h | 10 ++ >> gdb/corelow.c | 62 +++++++ >> gdb/defs.h | 3 +- >> gdb/doc/gdb.texinfo | 19 +++ >> gdb/gcore.c | 83 ++++++++- >> gdb/gdbarch-components.py | 35 ++++ >> gdb/gdbarch-gen.h | 26 +++ >> gdb/gdbarch.c | 96 +++++++++++ >> gdb/linux-tdep.c | 39 ++++- >> gdb/memtag.c | 61 +++++++ >> gdb/memtag.h | 50 ++++++ >> gdb/testsuite/gdb.arch/aarch64-mte-gcore.c | 93 +++++++++++ >> gdb/testsuite/gdb.arch/aarch64-mte-gcore.exp | 107 ++++++++++++ >> 17 files changed, 910 insertions(+), 8 deletions(-) >> create mode 100644 gdb/memtag.c >> create mode 100644 gdb/memtag.h >> create mode 100644 gdb/testsuite/gdb.arch/aarch64-mte-gcore.c >> create mode 100644 gdb/testsuite/gdb.arch/aarch64-mte-gcore.exp >> >> diff --git a/gdb/Makefile.in b/gdb/Makefile.in >> index 418094775a5..fac9364bea4 100644 >> --- a/gdb/Makefile.in >> +++ b/gdb/Makefile.in >> @@ -1120,6 +1120,7 @@ COMMON_SFILES = \ >> memattr.c \ >> memory-map.c \ >> memrange.c \ >> + memtag.c \ >> minidebug.c \ >> minsyms.c \ >> mipsread.c \ >> diff --git a/gdb/NEWS b/gdb/NEWS >> index 982f4a1a18c..3d925dc3663 100644 >> --- a/gdb/NEWS >> +++ b/gdb/NEWS >> @@ -3,6 +3,16 @@ >> >> *** Changes since GDB 12 >> >> +* GDB now supports dumping memory tag data for AArch64 MTE. It also supports >> + reading memory tag data for AArch64 MTE from core files generated by >> + the gcore command or the Linux kernel. >> + >> + When a process uses memory-mapped pages protected by memory tags (for >> + example, AArch64 MTE), this additional information will be recorded in >> + the core file in the event of a crash or if GDB generates a core file >> + from the current process state. GDB will show this additional information >> + automatically, or through one of the memory-tag subcommands. >> + >> * GDB now supports hardware watchpoints on FreeBSD/Aarch64. >> >> * Remove support for building against Python 2, it is now only possible to >> diff --git a/gdb/aarch64-linux-tdep.c b/gdb/aarch64-linux-tdep.c >> index 55094b3d88b..12d98e71796 100644 >> --- a/gdb/aarch64-linux-tdep.c >> +++ b/gdb/aarch64-linux-tdep.c >> @@ -53,6 +53,9 @@ >> >> #include "gdbsupport/selftest.h" >> >> +#include "elf/common.h" >> +#include "elf/aarch64.h" >> + >> /* Signal frame handling. >> >> +------------+ ^ >> @@ -1781,6 +1784,155 @@ aarch64_linux_report_signal_info (struct gdbarch *gdbarch, >> } >> } >> >> +/* AArch64 Linux implementation of the gdbarch_create_memtag_section hook. */ >> + >> +static asection * >> +aarch64_linux_create_memtag_section (struct gdbarch *gdbarch, bfd *obfd, >> + CORE_ADDR address, size_t size) >> +{ >> + gdb_assert (obfd != nullptr); >> + gdb_assert (size > 0); >> + >> + /* Create the section and associated program header. */ >> + asection *mte_section = bfd_make_section_anyway (obfd, "memtag"); >> + >> + if (mte_section == nullptr) >> + return nullptr; >> + >> + bfd_set_section_vma (mte_section, address); >> + /* The size of the memory range covered by the memory tags. We reuse the >> + section's rawsize field for this purpose. */ >> + mte_section->rawsize = size; >> + /* Tags are stored packed as 2 tags per byte. */ >> + bfd_set_section_size (mte_section, (size / AARCH64_MTE_GRANULE_SIZE) / 2); >> + /* Make sure the section's flags has SEC_HAS_CONTENTS, otherwise BFD will >> + refuse to write data to this section. */ >> + bfd_set_section_flags (mte_section, SEC_HAS_CONTENTS); >> + >> + /* Store program header information. */ >> + bfd_record_phdr (obfd, PT_AARCH64_MEMTAG_MTE, 1, 0, 0, 0, 0, 0, 1, >> + &mte_section); >> + >> + return mte_section; >> +} >> + >> +/* Maximum number of tags to request. */ >> +#define MAX_TAGS_TO_TRANSFER 1024 >> + >> +/* AArch64 Linux implementation of the gdbarch_fill_memtag_section hook. */ >> + >> +static bool >> +aarch64_linux_fill_memtag_section (struct gdbarch *gdbarch, asection *osec) >> +{ >> + /* We only handle MTE tags for now. */ >> + >> + size_t segment_size = osec->rawsize; >> + CORE_ADDR start_address = bfd_section_vma (osec); >> + CORE_ADDR end_address = start_address + segment_size; >> + >> + /* Figure out how many tags we need to store in this memory range. */ >> + size_t granules = aarch64_mte_get_tag_granules (start_address, segment_size, >> + AARCH64_MTE_GRANULE_SIZE); >> + >> + /* If there are no tag granules to fetch, just return. */ >> + if (granules == 0) >> + return true; >> + >> + CORE_ADDR address = start_address; >> + >> + /* Vector of tags. */ >> + gdb::byte_vector tags; >> + >> + while (granules > 0) >> + { >> + /* Transfer tags in chunks. */ >> + gdb::byte_vector tags_read; >> + size_t xfer_len >> + = (granules >= MAX_TAGS_TO_TRANSFER)? >> + MAX_TAGS_TO_TRANSFER * AARCH64_MTE_GRANULE_SIZE : >> + granules * AARCH64_MTE_GRANULE_SIZE; >> + >> + if (!target_fetch_memtags (address, xfer_len, tags_read, >> + static_cast (memtag_type::allocation))) >> + { >> + warning (_("Failed to read MTE tags from memory range [%s,%s)."), >> + phex_nz (start_address, sizeof (start_address)), >> + phex_nz (end_address, sizeof (end_address))); >> + return false; >> + } >> + >> + /* Transfer over the tags that have been read. */ >> + tags.insert (tags.end (), tags_read.begin (), tags_read.end ()); >> + >> + /* Adjust the remaining granules and starting address. */ >> + granules -= tags_read.size (); >> + address += tags_read.size () * AARCH64_MTE_GRANULE_SIZE; >> + } >> + >> + /* Pack the MTE tag bits. */ >> + aarch64_mte_pack_tags (tags); >> + >> + if (!bfd_set_section_contents (osec->owner, osec, tags.data (), >> + 0, tags.size ())) >> + { >> + warning (_("Failed to write %s bytes of corefile memory " >> + "tag content (%s)."), >> + pulongest (tags.size ()), >> + bfd_errmsg (bfd_get_error ())); >> + } >> + return true; >> +} >> + >> +/* AArch64 Linux implementation of the gdbarch_decode_memtag_section >> + hook. Decode a memory tag section and return the requested tags. >> + >> + The section is guaranteed to cover the [ADDRESS, ADDRESS + length) >> + range. */ >> + >> +static gdb::byte_vector >> +aarch64_linux_decode_memtag_section (struct gdbarch *gdbarch, >> + bfd_section *section, >> + int type, >> + CORE_ADDR address, size_t length) >> +{ >> + gdb_assert (section != nullptr); >> + >> + /* The requested address must not be less than section->vma. */ >> + gdb_assert (section->vma <= address); >> + >> + /* Figure out how many tags we need to fetch in this memory range. */ >> + size_t granules = aarch64_mte_get_tag_granules (address, length, >> + AARCH64_MTE_GRANULE_SIZE); >> + /* Sanity check. */ >> + gdb_assert (granules > 0); >> + >> + /* Fetch the total number of tags in the range [VMA, address + length). */ >> + size_t granules_from_vma >> + = aarch64_mte_get_tag_granules (section->vma, >> + address - section->vma + length, >> + AARCH64_MTE_GRANULE_SIZE); >> + >> + /* Adjust the tags vector to contain the exact number of packed bytes. */ >> + gdb::byte_vector tags (((granules - 1) >> 1) + 1); >> + >> + /* Figure out the starting offset into the packed tags data. */ >> + file_ptr offset = ((granules_from_vma - granules) >> 1); >> + >> + if (!bfd_get_section_contents (section->owner, section, tags.data (), >> + offset, tags.size ())) >> + error (_("Couldn't read contents from memtag section.")); >> + >> + /* At this point, the tags are packed 2 per byte. Unpack them before >> + returning. */ >> + bool skip_first = ((granules_from_vma - granules) % 2) != 0; >> + aarch64_mte_unpack_tags (tags, skip_first); >> + >> + /* Resize to the exact number of tags that was requested. */ >> + tags.resize (granules); >> + >> + return tags; >> +} >> + >> static void >> aarch64_linux_init_abi (struct gdbarch_info info, struct gdbarch *gdbarch) >> { >> @@ -1864,6 +2016,21 @@ aarch64_linux_init_abi (struct gdbarch_info info, struct gdbarch *gdbarch) >> >> set_gdbarch_report_signal_info (gdbarch, >> aarch64_linux_report_signal_info); >> + >> + /* Core file helpers. */ >> + >> + /* Core file helper to create a memory tag section for a particular >> + PT_LOAD segment. */ >> + set_gdbarch_create_memtag_section >> + (gdbarch, aarch64_linux_create_memtag_section); >> + >> + /* Core file helper to fill a memory tag section with tag data. */ >> + set_gdbarch_fill_memtag_section >> + (gdbarch, aarch64_linux_fill_memtag_section); >> + >> + /* Core file helper to decode a memory tag section. */ >> + set_gdbarch_decode_memtag_section (gdbarch, >> + aarch64_linux_decode_memtag_section); >> } >> >> /* Initialize the aarch64_linux_record_tdep. */ >> diff --git a/gdb/arch/aarch64-mte-linux.c b/gdb/arch/aarch64-mte-linux.c >> index fc7a8cc00f7..3af6f364e91 100644 >> --- a/gdb/arch/aarch64-mte-linux.c >> +++ b/gdb/arch/aarch64-mte-linux.c >> @@ -21,6 +21,62 @@ >> >> /* See arch/aarch64-mte-linux.h */ >> >> +void >> +aarch64_mte_pack_tags (gdb::byte_vector &tags) >> +{ >> + /* Nothing to pack? */ >> + if (tags.empty ()) >> + return; >> + >> + /* If the tags vector has an odd number of elements, add another >> + zeroed-out element to make it even. This facilitates packing. */ >> + if ((tags.size () % 2) != 0) >> + tags.emplace_back (0); >> + >> + for (int unpacked = 0, packed = 0; unpacked < tags.size (); >> + unpacked += 2, packed++) >> + tags[packed] = (tags[unpacked + 1] << 4) | tags[unpacked]; >> + >> + /* Now we have half the size. */ >> + tags.resize (tags.size () / 2); >> +} >> + >> +/* See arch/aarch64-mte-linux.h */ >> + >> +void >> +aarch64_mte_unpack_tags (gdb::byte_vector &tags, bool skip_first) >> +{ >> + /* Nothing to unpack? */ >> + if (tags.empty ()) >> + return; >> + >> + /* An unpacked MTE tags vector will have twice the number of elements >> + compared to an unpacked one. */ >> + gdb::byte_vector unpacked_tags (tags.size () * 2); >> + >> + int unpacked = 0, packed = 0; >> + >> + if (skip_first) >> + { >> + /* We are not interested in the first unpacked element, just discard >> + it. */ >> + unpacked_tags[unpacked] = (tags[packed] >> 4) & 0xf; >> + unpacked++; >> + packed++; >> + } >> + >> + for (; packed < tags.size (); unpacked += 2, packed++) >> + { >> + unpacked_tags[unpacked] = tags[packed] & 0xf; >> + unpacked_tags[unpacked + 1] = (tags[packed] >> 4) & 0xf; >> + } >> + >> + /* Update the original tags vector. */ >> + tags = std::move (unpacked_tags); >> +} >> + >> +/* See arch/aarch64-mte-linux.h */ >> + >> size_t >> aarch64_mte_get_tag_granules (CORE_ADDR addr, size_t len, size_t granule_size) >> { >> diff --git a/gdb/arch/aarch64-mte-linux.h b/gdb/arch/aarch64-mte-linux.h >> index d158926feff..8a145b447aa 100644 >> --- a/gdb/arch/aarch64-mte-linux.h >> +++ b/gdb/arch/aarch64-mte-linux.h >> @@ -32,6 +32,7 @@ >> >> /* We have one tag per 16 bytes of memory. */ >> #define AARCH64_MTE_GRANULE_SIZE 16 >> +#define AARCH64_MTE_TAG_BIT_SIZE 4 >> #define AARCH64_MTE_LOGICAL_TAG_START_BIT 56 >> #define AARCH64_MTE_LOGICAL_MAX_VALUE 0xf >> >> @@ -71,4 +72,13 @@ extern CORE_ADDR aarch64_mte_set_ltag (CORE_ADDR address, CORE_ADDR tag); >> It is always possible to get the logical tag. */ >> extern CORE_ADDR aarch64_mte_get_ltag (CORE_ADDR address); >> >> +/* Given a TAGS vector containing 1 MTE tag per byte, pack the data as >> + 2 tags per byte and resize the vector. */ >> +void aarch64_mte_pack_tags (gdb::byte_vector &tags); >> + >> +/* Given a TAGS vector containing 2 MTE tags per byte, unpack the data as >> + 1 tag per byte and resize the vector. If SKIP_FIRST is TRUE, skip the >> + first unpacked element. Otherwise leave it in the unpacked vector. */ >> +void aarch64_mte_unpack_tags (gdb::byte_vector &tags, bool skip_first); >> + >> #endif /* ARCH_AARCH64_LINUX_H */ >> diff --git a/gdb/corelow.c b/gdb/corelow.c >> index 8c33fb7ebb2..8b8994f80db 100644 >> --- a/gdb/corelow.c >> +++ b/gdb/corelow.c >> @@ -52,6 +52,7 @@ >> #include >> #include "gdbcmd.h" >> #include "xml-tdesc.h" >> +#include "memtag.h" >> >> #ifndef O_LARGEFILE >> #define O_LARGEFILE 0 >> @@ -101,6 +102,13 @@ class core_target final : public process_stratum_target >> >> bool info_proc (const char *, enum info_proc_what) override; >> >> + bool supports_memory_tagging () override; >> + >> + /* Core file implementation of fetch_memtags. Fetch the memory tags from >> + core file notes. */ >> + bool fetch_memtags (CORE_ADDR address, size_t len, >> + gdb::byte_vector &tags, int type) override; >> + >> /* A few helpers. */ >> >> /* Getter, see variable definition. */ >> @@ -1162,6 +1170,60 @@ core_target::info_proc (const char *args, enum info_proc_what request) >> return true; >> } >> >> +/* Implementation of the "supports_memory_tagging" target_ops method. */ >> + >> +bool >> +core_target::supports_memory_tagging () >> +{ >> + /* Look for memory tag sections. If they exist, that means this core file >> + supports memory tagging. */ >> + >> + return (bfd_get_section_by_name (core_bfd, "memtag") != nullptr); >> +} >> + >> +/* Implementation of the "fetch_memtags" target_ops method. */ >> + >> +bool >> +core_target::fetch_memtags (CORE_ADDR address, size_t len, >> + gdb::byte_vector &tags, int type) >> +{ >> + struct gdbarch *gdbarch = target_gdbarch (); >> + >> + /* Make sure we have a way to decode the memory tag notes. */ >> + if (!gdbarch_decode_memtag_section_p (gdbarch)) >> + error (_("gdbarch_decode_memtag_section not implemented for this " >> + "architecture.")); >> + >> + memtag_section_info info; >> + info.memtag_section = nullptr; >> + >> + while (get_next_core_memtag_section (core_bfd, info.memtag_section, >> + address, info)) >> + { >> + size_t adjusted_length >> + = (address + len < info.end_address)? len : (info.end_address - address); >> + >> + /* Decode the memory tag note and return the tags. */ >> + gdb::byte_vector tags_read >> + = gdbarch_decode_memtag_section (gdbarch, info.memtag_section, type, >> + address, adjusted_length); >> + >> + /* Transfer over the tags that have been read. */ >> + tags.insert (tags.end (), tags_read.begin (), tags_read.end ()); >> + >> + /* ADDRESS + LEN may cross the boundaries of a particular memory tag >> + segment. Check if we need to fetch tags from a different section. */ >> + if (!tags_read.empty () && (address + len) < info.end_address) >> + return true; >> + >> + /* There are more tags to fetch. Update ADDRESS and LEN. */ >> + len -= (info.end_address - address); >> + address = info.end_address; >> + } >> + >> + return false; >> +} >> + >> /* Get a pointer to the current core target. If not connected to a >> core target, return NULL. */ >> >> diff --git a/gdb/defs.h b/gdb/defs.h >> index 99bfdd526ff..51a7576a56a 100644 >> --- a/gdb/defs.h >> +++ b/gdb/defs.h >> @@ -344,7 +344,8 @@ extern const char *pc_prefix (CORE_ADDR); >> >> typedef int (*find_memory_region_ftype) (CORE_ADDR addr, unsigned long size, >> int read, int write, int exec, >> - int modified, void *data); >> + int modified, bool memory_tagged, >> + void *data); >> >> /* * Possible lvalue types. Like enum language, this should be in >> value.h, but needs to be here for the same reason. */ >> diff --git a/gdb/doc/gdb.texinfo b/gdb/doc/gdb.texinfo >> index 38ad2ac32b0..36f10f20cfb 100644 >> --- a/gdb/doc/gdb.texinfo >> +++ b/gdb/doc/gdb.texinfo >> @@ -25555,6 +25555,25 @@ options that can be controlled at runtime and emulates the @code{prctl} >> option @code{PR_SET_TAGGED_ADDR_CTRL}. For further information, see the >> documentation in the Linux kernel. >> >> +@value{GDBN} supports dumping memory tag data to core files through the >> +@command{gcore} command and reading memory tag data from core files generated >> +by the @command{gcore} command or the Linux kernel. >> + >> +When a process uses memory-mapped pages protected by memory tags (for >> +example, AArch64 MTE), this additional information will be recorded in >> +the core file in the event of a crash or if @value{GDBN} generates a core file >> +from the current process state. >> + >> +The memory tag data will be used so developers can display the memory >> +tags from a particular memory region (using the @samp{m} modifier to the >> +@command{x} command, using the @command{print} command or using the various >> +@command{memory-tag} subcommands. >> + >> +In the case of a crash, @value{GDBN} will attempt to retrieve the memory tag >> +information automatically from the core file, and will show one of the above >> +messages depending on whether the synchronous or asynchronous mode is selected. >> +@xref{Memory Tagging}. @xref{Memory}. >> + >> @node i386 >> @subsection x86 Architecture-specific Issues >> >> diff --git a/gdb/gcore.c b/gdb/gcore.c >> index fdb22b72a07..b81ef81ab84 100644 >> --- a/gdb/gcore.c >> +++ b/gdb/gcore.c >> @@ -349,6 +349,12 @@ make_output_phdrs (bfd *obfd, asection *osec) >> int p_flags = 0; >> int p_type = 0; >> >> + /* Memory tag segments have already been handled by the architecture, as >> + those contain arch-specific information. If we have one of those, just >> + return. */ >> + if (startswith (bfd_section_name (osec), "memtag")) >> + return; >> + >> /* FIXME: these constants may only be applicable for ELF. */ >> if (startswith (bfd_section_name (osec), "load")) >> p_type = PT_LOAD; >> @@ -371,7 +377,8 @@ make_output_phdrs (bfd *obfd, asection *osec) >> >> static int >> gcore_create_callback (CORE_ADDR vaddr, unsigned long size, int read, >> - int write, int exec, int modified, void *data) >> + int write, int exec, int modified, bool memory_tagged, >> + void *data) >> { >> bfd *obfd = (bfd *) data; >> asection *osec; >> @@ -454,6 +461,45 @@ gcore_create_callback (CORE_ADDR vaddr, unsigned long size, int read, >> return 0; >> } >> >> +/* gdbarch_find_memory_region callback for creating a memory tag section. >> + DATA is 'bfd *' for the core file GDB is creating. */ >> + >> +static int >> +gcore_create_memtag_section_callback (CORE_ADDR vaddr, unsigned long size, >> + int read, int write, int exec, >> + int modified, bool memory_tagged, >> + void *data) >> +{ >> + /* Are there memory tags in this particular memory map entry? */ >> + if (!memory_tagged) >> + return 0; >> + >> + bfd *obfd = (bfd *) data; >> + >> + /* Ask the architecture to create a memory tag section for this particular >> + memory map entry. It will be populated with contents later, as we can't >> + start writing the contents before we have all the sections sorted out. */ >> + asection *memtag_section >> + = gdbarch_create_memtag_section (target_gdbarch (), obfd, vaddr, size); >> + >> + if (memtag_section == nullptr) >> + { >> + warning (_("Couldn't make gcore memory tag segment: %s"), >> + bfd_errmsg (bfd_get_error ())); >> + return 1; >> + } >> + >> + if (info_verbose) >> + { >> + gdb_printf (gdb_stdout, "Saved memory tag segment, %s bytes " >> + "at %s\n", >> + plongest (bfd_section_size (memtag_section)), >> + paddress (target_gdbarch (), vaddr)); >> + } >> + >> + return 0; >> +} >> + >> int >> objfile_find_memory_regions (struct target_ops *self, >> find_memory_region_ftype func, void *obfd) >> @@ -483,6 +529,7 @@ objfile_find_memory_regions (struct target_ops *self, >> (flags & SEC_READONLY) == 0, /* Writable. */ >> (flags & SEC_CODE) != 0, /* Executable. */ >> 1, /* MODIFIED is unknown, pass it as true. */ >> + false, /* No memory tags in the object file. */ >> obfd); >> if (ret != 0) >> return ret; >> @@ -496,6 +543,7 @@ objfile_find_memory_regions (struct target_ops *self, >> 1, /* Stack section will be writable. */ >> 0, /* Stack section will not be executable. */ >> 1, /* Stack section will be modified. */ >> + false, /* No memory tags in the object file. */ >> obfd); >> >> /* Make a heap segment. */ >> @@ -506,6 +554,7 @@ objfile_find_memory_regions (struct target_ops *self, >> 1, /* Heap section will be writable. */ >> 0, /* Heap section will not be executable. */ >> 1, /* Heap section will be modified. */ >> + false, /* No memory tags in the object file. */ >> obfd); >> >> return 0; >> @@ -555,6 +604,20 @@ gcore_copy_callback (bfd *obfd, asection *osec) >> } >> } >> >> +/* Callback to copy contents to a particular memory tag section. */ >> + >> +static void >> +gcore_copy_memtag_section_callback (bfd *obfd, asection *osec) >> +{ >> + /* We are only interested in "memtag" sections. */ >> + if (!startswith (bfd_section_name (osec), "memtag")) >> + return; >> + >> + /* Fill the section with memory tag contents. */ >> + if (!gdbarch_fill_memtag_section (target_gdbarch (), osec)) >> + error (_("Failed to fill memory tag section for core file.")); >> +} >> + >> static int >> gcore_memory_sections (bfd *obfd) >> { >> @@ -567,13 +630,27 @@ gcore_memory_sections (bfd *obfd) >> return 0; /* FIXME: error return/msg? */ >> } >> >> + /* Take care of dumping memory tags, if there are any. */ >> + if (!gdbarch_find_memory_regions_p (target_gdbarch ()) >> + || gdbarch_find_memory_regions (target_gdbarch (), >> + gcore_create_memtag_section_callback, >> + obfd) != 0) >> + { >> + if (target_find_memory_regions (gcore_create_memtag_section_callback, >> + obfd) != 0) >> + return 0; >> + } >> + >> /* Record phdrs for section-to-segment mapping. */ >> for (asection *sect : gdb_bfd_sections (obfd)) >> make_output_phdrs (obfd, sect); >> >> - /* Copy memory region contents. */ >> + /* Copy memory region and memory tag contents. */ >> for (asection *sect : gdb_bfd_sections (obfd)) >> - gcore_copy_callback (obfd, sect); >> + { >> + gcore_copy_callback (obfd, sect); >> + gcore_copy_memtag_section_callback (obfd, sect); >> + } >> >> return 1; >> } >> diff --git a/gdb/gdbarch-components.py b/gdb/gdbarch-components.py >> index e8f20c83ff0..6fa1b7591db 100644 >> --- a/gdb/gdbarch-components.py >> +++ b/gdb/gdbarch-components.py >> @@ -1522,6 +1522,41 @@ Find core file memory regions >> invalid=True, >> ) >> >> +Method( >> + comment=""" >> +Given a bfd OBFD, segment ADDRESS and SIZE, create a memory tag section to be dumped to a core file >> +""", >> + type="asection *", >> + name="create_memtag_section", >> + params=[("bfd *", "obfd"), ("CORE_ADDR", "address"), ("size_t", "size")], >> + predicate=True, >> + invalid=True, >> +) >> + >> +Method( >> + comment=""" >> +Given a memory tag section OSEC, fill OSEC's contents with the appropriate tag data >> +""", >> + type="bool", >> + name="fill_memtag_section", >> + params=[("asection *", "osec")], >> + predicate=True, >> + invalid=True, >> +) >> + >> +Method( >> + comment=""" >> +Decode a memory tag SECTION and return the tags of type TYPE contained in >> +the memory range [ADDRESS, ADDRESS + LENGTH). >> +If no tags were found, return an empty vector. >> +""", >> + type="gdb::byte_vector", >> + name="decode_memtag_section", >> + params=[("bfd_section *", "section"), ("int", "type"), ("CORE_ADDR", "address"), ("size_t", "length")], >> + predicate=True, >> + invalid=True, >> +) >> + >> Method( >> comment=""" >> Read offset OFFSET of TARGET_OBJECT_LIBRARIES formatted shared libraries list from >> diff --git a/gdb/gdbarch-gen.h b/gdb/gdbarch-gen.h >> index 882b9057b1a..1d19f51f21d 100644 >> --- a/gdb/gdbarch-gen.h >> +++ b/gdb/gdbarch-gen.h >> @@ -874,6 +874,32 @@ typedef int (gdbarch_find_memory_regions_ftype) (struct gdbarch *gdbarch, find_m >> extern int gdbarch_find_memory_regions (struct gdbarch *gdbarch, find_memory_region_ftype func, void *data); >> extern void set_gdbarch_find_memory_regions (struct gdbarch *gdbarch, gdbarch_find_memory_regions_ftype *find_memory_regions); >> >> +/* Given a bfd OBFD, segment ADDRESS and SIZE, create a memory tag section to be dumped to a core file */ >> + >> +extern bool gdbarch_create_memtag_section_p (struct gdbarch *gdbarch); >> + >> +typedef asection * (gdbarch_create_memtag_section_ftype) (struct gdbarch *gdbarch, bfd *obfd, CORE_ADDR address, size_t size); >> +extern asection * gdbarch_create_memtag_section (struct gdbarch *gdbarch, bfd *obfd, CORE_ADDR address, size_t size); >> +extern void set_gdbarch_create_memtag_section (struct gdbarch *gdbarch, gdbarch_create_memtag_section_ftype *create_memtag_section); >> + >> +/* Given a memory tag section OSEC, fill OSEC's contents with the appropriate tag data */ >> + >> +extern bool gdbarch_fill_memtag_section_p (struct gdbarch *gdbarch); >> + >> +typedef bool (gdbarch_fill_memtag_section_ftype) (struct gdbarch *gdbarch, asection *osec); >> +extern bool gdbarch_fill_memtag_section (struct gdbarch *gdbarch, asection *osec); >> +extern void set_gdbarch_fill_memtag_section (struct gdbarch *gdbarch, gdbarch_fill_memtag_section_ftype *fill_memtag_section); >> + >> +/* Decode a memory tag SECTION and return the tags of type TYPE contained in >> + the memory range [ADDRESS, ADDRESS + LENGTH). >> + If no tags were found, return an empty vector. */ >> + >> +extern bool gdbarch_decode_memtag_section_p (struct gdbarch *gdbarch); >> + >> +typedef gdb::byte_vector (gdbarch_decode_memtag_section_ftype) (struct gdbarch *gdbarch, bfd_section *section, int type, CORE_ADDR address, size_t length); >> +extern gdb::byte_vector gdbarch_decode_memtag_section (struct gdbarch *gdbarch, bfd_section *section, int type, CORE_ADDR address, size_t length); >> +extern void set_gdbarch_decode_memtag_section (struct gdbarch *gdbarch, gdbarch_decode_memtag_section_ftype *decode_memtag_section); >> + >> /* Read offset OFFSET of TARGET_OBJECT_LIBRARIES formatted shared libraries list from >> core file into buffer READBUF with length LEN. Return the number of bytes read >> (zero indicates failure). >> diff --git a/gdb/gdbarch.c b/gdb/gdbarch.c >> index a588bdef61a..f5dbacb14e7 100644 >> --- a/gdb/gdbarch.c >> +++ b/gdb/gdbarch.c >> @@ -171,6 +171,9 @@ struct gdbarch >> gdbarch_iterate_over_regset_sections_ftype *iterate_over_regset_sections; >> gdbarch_make_corefile_notes_ftype *make_corefile_notes; >> gdbarch_find_memory_regions_ftype *find_memory_regions; >> + gdbarch_create_memtag_section_ftype *create_memtag_section; >> + gdbarch_fill_memtag_section_ftype *fill_memtag_section; >> + gdbarch_decode_memtag_section_ftype *decode_memtag_section; >> gdbarch_core_xfer_shared_libraries_ftype *core_xfer_shared_libraries; >> gdbarch_core_xfer_shared_libraries_aix_ftype *core_xfer_shared_libraries_aix; >> gdbarch_core_pid_to_str_ftype *core_pid_to_str; >> @@ -527,6 +530,9 @@ verify_gdbarch (struct gdbarch *gdbarch) >> /* Skip verify of iterate_over_regset_sections, has predicate. */ >> /* Skip verify of make_corefile_notes, has predicate. */ >> /* Skip verify of find_memory_regions, has predicate. */ >> + /* Skip verify of create_memtag_section, has predicate. */ >> + /* Skip verify of fill_memtag_section, has predicate. */ >> + /* Skip verify of decode_memtag_section, has predicate. */ >> /* Skip verify of core_xfer_shared_libraries, has predicate. */ >> /* Skip verify of core_xfer_shared_libraries_aix, has predicate. */ >> /* Skip verify of core_pid_to_str, has predicate. */ >> @@ -1096,6 +1102,24 @@ gdbarch_dump (struct gdbarch *gdbarch, struct ui_file *file) >> gdb_printf (file, >> "gdbarch_dump: find_memory_regions = <%s>\n", >> host_address_to_string (gdbarch->find_memory_regions)); >> + gdb_printf (file, >> + "gdbarch_dump: gdbarch_create_memtag_section_p() = %d\n", >> + gdbarch_create_memtag_section_p (gdbarch)); >> + gdb_printf (file, >> + "gdbarch_dump: create_memtag_section = <%s>\n", >> + host_address_to_string (gdbarch->create_memtag_section)); >> + gdb_printf (file, >> + "gdbarch_dump: gdbarch_fill_memtag_section_p() = %d\n", >> + gdbarch_fill_memtag_section_p (gdbarch)); >> + gdb_printf (file, >> + "gdbarch_dump: fill_memtag_section = <%s>\n", >> + host_address_to_string (gdbarch->fill_memtag_section)); >> + gdb_printf (file, >> + "gdbarch_dump: gdbarch_decode_memtag_section_p() = %d\n", >> + gdbarch_decode_memtag_section_p (gdbarch)); >> + gdb_printf (file, >> + "gdbarch_dump: decode_memtag_section = <%s>\n", >> + host_address_to_string (gdbarch->decode_memtag_section)); >> gdb_printf (file, >> "gdbarch_dump: gdbarch_core_xfer_shared_libraries_p() = %d\n", >> gdbarch_core_xfer_shared_libraries_p (gdbarch)); >> @@ -3744,6 +3768,78 @@ set_gdbarch_find_memory_regions (struct gdbarch *gdbarch, >> gdbarch->find_memory_regions = find_memory_regions; >> } >> >> +bool >> +gdbarch_create_memtag_section_p (struct gdbarch *gdbarch) >> +{ >> + gdb_assert (gdbarch != NULL); >> + return gdbarch->create_memtag_section != NULL; >> +} >> + >> +asection * >> +gdbarch_create_memtag_section (struct gdbarch *gdbarch, bfd *obfd, CORE_ADDR address, size_t size) >> +{ >> + gdb_assert (gdbarch != NULL); >> + gdb_assert (gdbarch->create_memtag_section != NULL); >> + if (gdbarch_debug >= 2) >> + gdb_printf (gdb_stdlog, "gdbarch_create_memtag_section called\n"); >> + return gdbarch->create_memtag_section (gdbarch, obfd, address, size); >> +} >> + >> +void >> +set_gdbarch_create_memtag_section (struct gdbarch *gdbarch, >> + gdbarch_create_memtag_section_ftype create_memtag_section) >> +{ >> + gdbarch->create_memtag_section = create_memtag_section; >> +} >> + >> +bool >> +gdbarch_fill_memtag_section_p (struct gdbarch *gdbarch) >> +{ >> + gdb_assert (gdbarch != NULL); >> + return gdbarch->fill_memtag_section != NULL; >> +} >> + >> +bool >> +gdbarch_fill_memtag_section (struct gdbarch *gdbarch, asection *osec) >> +{ >> + gdb_assert (gdbarch != NULL); >> + gdb_assert (gdbarch->fill_memtag_section != NULL); >> + if (gdbarch_debug >= 2) >> + gdb_printf (gdb_stdlog, "gdbarch_fill_memtag_section called\n"); >> + return gdbarch->fill_memtag_section (gdbarch, osec); >> +} >> + >> +void >> +set_gdbarch_fill_memtag_section (struct gdbarch *gdbarch, >> + gdbarch_fill_memtag_section_ftype fill_memtag_section) >> +{ >> + gdbarch->fill_memtag_section = fill_memtag_section; >> +} >> + >> +bool >> +gdbarch_decode_memtag_section_p (struct gdbarch *gdbarch) >> +{ >> + gdb_assert (gdbarch != NULL); >> + return gdbarch->decode_memtag_section != NULL; >> +} >> + >> +gdb::byte_vector >> +gdbarch_decode_memtag_section (struct gdbarch *gdbarch, bfd_section *section, int type, CORE_ADDR address, size_t length) >> +{ >> + gdb_assert (gdbarch != NULL); >> + gdb_assert (gdbarch->decode_memtag_section != NULL); >> + if (gdbarch_debug >= 2) >> + gdb_printf (gdb_stdlog, "gdbarch_decode_memtag_section called\n"); >> + return gdbarch->decode_memtag_section (gdbarch, section, type, address, length); >> +} >> + >> +void >> +set_gdbarch_decode_memtag_section (struct gdbarch *gdbarch, >> + gdbarch_decode_memtag_section_ftype decode_memtag_section) >> +{ >> + gdbarch->decode_memtag_section = decode_memtag_section; >> +} >> + >> bool >> gdbarch_core_xfer_shared_libraries_p (struct gdbarch *gdbarch) >> { >> diff --git a/gdb/linux-tdep.c b/gdb/linux-tdep.c >> index 4e728a06e7e..8a83ed320cf 100644 >> --- a/gdb/linux-tdep.c >> +++ b/gdb/linux-tdep.c >> @@ -42,6 +42,7 @@ >> #include "gcore.h" >> #include "gcore-elf.h" >> #include "solib-svr4.h" >> +#include "memtag.h" >> >> #include >> #include >> @@ -1320,6 +1321,7 @@ typedef int linux_find_memory_region_ftype (ULONGEST vaddr, ULONGEST size, >> ULONGEST offset, ULONGEST inode, >> int read, int write, >> int exec, int modified, >> + bool memory_tagged, >> const char *filename, >> void *data); >> >> @@ -1470,10 +1472,11 @@ parse_smaps_data (const char *data, >> return smaps; >> } >> >> -/* See linux-tdep.h. */ >> +/* Helper that checks if an address is in a memory tag page for a live >> + process. */ >> >> -bool >> -linux_address_in_memtag_page (CORE_ADDR address) >> +static bool >> +linux_process_address_in_memtag_page (CORE_ADDR address) >> { >> if (current_inferior ()->fake_pid_p) >> return false; >> @@ -1505,6 +1508,30 @@ linux_address_in_memtag_page (CORE_ADDR address) >> return false; >> } >> >> +/* Helper that checks if an address is in a memory tag page for a core file >> + process. */ >> + >> +static bool >> +linux_core_file_address_in_memtag_page (CORE_ADDR address) >> +{ >> + if (core_bfd == nullptr) >> + return false; >> + >> + memtag_section_info info; >> + return get_next_core_memtag_section (core_bfd, nullptr, address, info); >> +} >> + >> +/* See linux-tdep.h. */ >> + >> +bool >> +linux_address_in_memtag_page (CORE_ADDR address) >> +{ >> + if (!target_has_execution ()) >> + return linux_core_file_address_in_memtag_page (address); >> + >> + return linux_process_address_in_memtag_page (address); >> +} >> + >> /* List memory regions in the inferior for a corefile. */ >> >> static int >> @@ -1593,6 +1620,7 @@ linux_find_memory_regions_full (struct gdbarch *gdbarch, >> map.offset, map.inode, map.read, map.write, map.exec, >> 1, /* MODIFIED is true because we want to dump >> the mapping. */ >> + map.vmflags.memory_tagging != 0, >> map.filename.c_str (), obfd); >> } >> } >> @@ -1621,12 +1649,14 @@ static int >> linux_find_memory_regions_thunk (ULONGEST vaddr, ULONGEST size, >> ULONGEST offset, ULONGEST inode, >> int read, int write, int exec, int modified, >> + bool memory_tagged, >> const char *filename, void *arg) >> { >> struct linux_find_memory_regions_data *data >> = (struct linux_find_memory_regions_data *) arg; >> >> - return data->func (vaddr, size, read, write, exec, modified, data->obfd); >> + return data->func (vaddr, size, read, write, exec, modified, memory_tagged, >> + data->obfd); >> } >> >> /* A variant of linux_find_memory_regions_full that is suitable as the >> @@ -1675,6 +1705,7 @@ static int >> linux_make_mappings_callback (ULONGEST vaddr, ULONGEST size, >> ULONGEST offset, ULONGEST inode, >> int read, int write, int exec, int modified, >> + bool memory_tagged, >> const char *filename, void *data) >> { >> struct linux_make_mappings_data *map_data >> diff --git a/gdb/memtag.c b/gdb/memtag.c >> new file mode 100644 >> index 00000000000..af86137c49d >> --- /dev/null >> +++ b/gdb/memtag.c >> @@ -0,0 +1,61 @@ >> +/* GDB generic memory tagging functions. >> + >> + Copyright (C) 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 . */ >> + >> +#include "defs.h" >> +#include "memtag.h" >> +#include "bfd.h" >> + >> +/* See memtag.h */ >> + >> +bool >> +get_next_core_memtag_section (bfd *abfd, asection *section, >> + CORE_ADDR address, memtag_section_info &info) >> +{ >> + /* If the caller provided no SECTION to start from, search from the >> + beginning. */ >> + if (section == nullptr) >> + section = bfd_get_section_by_name (abfd, "memtag"); >> + >> + /* Go through all the memtag sections and figure out if ADDRESS >> + falls within one of the memory ranges that contain tags. */ >> + while (section != nullptr) >> + { >> + size_t memtag_range_size = section->rawsize; >> + size_t tags_size = bfd_section_size (section); >> + >> + /* Empty memory range and empty tag dump should not happen. */ >> + gdb_assert (memtag_range_size != 0); >> + gdb_assert (tags_size != 0); >> + >> + CORE_ADDR start_address = bfd_section_vma (section); >> + CORE_ADDR end_address = start_address + memtag_range_size; >> + >> + /* Is the address within [start_address, end_address)? */ >> + if (address >= start_address >> + && address < end_address) >> + { >> + info.start_address = start_address; >> + info.end_address = end_address; >> + info.memtag_section = section; >> + return true; >> + } >> + section = bfd_get_next_section_by_name (abfd, section); >> + } >> + return false; >> +} >> diff --git a/gdb/memtag.h b/gdb/memtag.h >> new file mode 100644 >> index 00000000000..fe908c1e5e3 >> --- /dev/null >> +++ b/gdb/memtag.h >> @@ -0,0 +1,50 @@ >> +/* GDB generic memory tagging definitions. >> + Copyright (C) 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 . */ >> + >> +#ifndef MEMTAG_H >> +#define MEMTAG_H >> + >> +#include "bfd.h" >> + >> +struct memtag_section_info >> +{ >> + /* The start address of the tagged memory range. */ >> + CORE_ADDR start_address; >> + /* The final address of the tagged memory range. */ >> + CORE_ADDR end_address; >> + /* The section containing tags for the memory range >> + [start_address, end_address). */ >> + asection *memtag_section; >> +}; >> + >> +/* Helper function to walk through memory tag sections in a core file. >> + >> + Return TRUE if there is a "memtag" section containing ADDRESS. Return FALSE >> + otherwise. >> + >> + If SECTION is provided, search from that section onwards. If SECTION is >> + nullptr, then start a new search. >> + >> + If a "memtag" section containing ADDRESS is found, fill INFO with data >> + about such section. Otherwise leave it unchanged. */ >> + >> +bool get_next_core_memtag_section (bfd *abfd, asection *section, >> + CORE_ADDR address, >> + memtag_section_info &info); >> + >> +#endif /* MEMTAG_H */ >> diff --git a/gdb/testsuite/gdb.arch/aarch64-mte-gcore.c b/gdb/testsuite/gdb.arch/aarch64-mte-gcore.c >> new file mode 100644 >> index 00000000000..b20ebcff424 >> --- /dev/null >> +++ b/gdb/testsuite/gdb.arch/aarch64-mte-gcore.c >> @@ -0,0 +1,93 @@ >> +/* This test program 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 . */ >> + >> +/* Exercise AArch64's Memory Tagging Extension with tagged pointers. */ >> + >> +/* This test was based on the documentation for the AArch64 Memory Tagging >> + Extension from the Linux Kernel, found in the sources in >> + Documentation/arm64/memory-tagging-extension.rst. */ >> + >> +#include >> +#include >> +#include >> +#include >> +#include >> +#include >> +#include >> + >> +/* From arch/arm64/include/uapi/asm/hwcap.h */ >> +#define HWCAP2_MTE (1 << 18) >> + >> +/* From arch/arm64/include/uapi/asm/mman.h */ >> +#define PROT_MTE 0x20 >> + >> +/* From include/uapi/linux/prctl.h */ >> +#define PR_SET_TAGGED_ADDR_CTRL 55 >> +#define PR_GET_TAGGED_ADDR_CTRL 56 >> +#define PR_TAGGED_ADDR_ENABLE (1UL << 0) >> +#define PR_MTE_TCF_SHIFT 1 >> +#define PR_MTE_TCF_SYNC (1UL << PR_MTE_TCF_SHIFT) >> +#define PR_MTE_TAG_SHIFT 3 >> + >> +void >> +access_memory (unsigned char *tagged_ptr) >> +{ >> + tagged_ptr[0] = 'a'; >> +} >> + >> +int >> +main (int argc, char **argv) >> +{ >> + unsigned char *tagged_ptr; >> + unsigned long page_sz = sysconf (_SC_PAGESIZE); >> + unsigned long hwcap2 = getauxval(AT_HWCAP2); >> + >> + /* Bail out if MTE is not supported. */ >> + if (!(hwcap2 & HWCAP2_MTE)) >> + return 1; >> + >> + /* Enable the tagged address ABI, synchronous MTE tag check faults and >> + allow all non-zero tags in the randomly generated set. */ >> + if (prctl (PR_SET_TAGGED_ADDR_CTRL, >> + PR_TAGGED_ADDR_ENABLE | PR_MTE_TCF_SYNC >> + | (0xfffe << PR_MTE_TAG_SHIFT), >> + 0, 0, 0)) >> + { >> + perror ("prctl () failed"); >> + return 1; >> + } >> + >> + /* Create a mapping that will have PROT_MTE set. */ >> + tagged_ptr = mmap (0, page_sz, PROT_READ | PROT_WRITE, >> + MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); >> + if (tagged_ptr == MAP_FAILED) >> + { >> + perror ("mmap () failed"); >> + return 1; >> + } >> + >> + /* Enable MTE on the above anonymous mmap. */ >> + if (mprotect (tagged_ptr, page_sz, PROT_READ | PROT_WRITE | PROT_MTE)) >> + { >> + perror ("mprotect () failed"); >> + return 1; >> + } >> + >> + access_memory (tagged_ptr); >> + >> + return 0; >> +} >> diff --git a/gdb/testsuite/gdb.arch/aarch64-mte-gcore.exp b/gdb/testsuite/gdb.arch/aarch64-mte-gcore.exp >> new file mode 100644 >> index 00000000000..8a19c4b449e >> --- /dev/null >> +++ b/gdb/testsuite/gdb.arch/aarch64-mte-gcore.exp >> @@ -0,0 +1,107 @@ >> +# Copyright (C) 2018-2021 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 generating and reading a core file with MTE memory tags. >> + >> +if {![is_aarch64_target]} { >> + verbose "Skipping ${gdb_test_file_name}." >> + return >> +} >> + >> +standard_testfile >> +if { [prepare_for_testing "failed to prepare" ${testfile} ${srcfile}] } { >> + return -1 >> +} >> + >> +if ![runto_main] { >> + untested "could not run to main" >> + return -1 >> +} >> + >> +# Targets that don't support memory tagging should not execute the >> +# runtime memory tagging tests. >> +if {![supports_memtag]} { >> + unsupported "memory tagging unsupported" >> + return -1 >> +} >> + >> +gdb_breakpoint "access_memory" >> + >> +if [gdb_continue "access_memory"] { >> + return -1 >> +} >> + >> +# Set each tag granule to a different tag value, from 0x0 to 0xf. >> +set atag_msg "Allocation tag\\(s\\) updated successfully\." >> +for {set i 15} {$i >= 0} {incr i -1} { >> + set index [expr [expr 15 - $i] * 16] >> + set tag [format "%02x" $i] >> + gdb_test "memory-tag set-allocation-tag &tagged_ptr\[$index\] 1 $tag" \ >> + $atag_msg \ >> + "set memory tag of &tagged_ptr\[$index\] to $tag" >> +} >> + >> +# Run until a crash and confirm GDB displays memory tag violation >> +# information. >> +gdb_test "continue" \ >> + [multi_line \ >> + "Program received signal SIGSEGV, Segmentation fault" \ >> + "Memory tag violation while accessing address $hex" \ >> + "Allocation tag $hex" \ >> + "Logical tag $hex\." \ >> + "$hex in access_memory \\(.*\\) at .*" \ >> + ".*tagged_ptr\\\[0\\\] = 'a';"] \ >> + "display tag violation information for live process" >> + >> +# Generate the core file. >> +set core_filename [standard_output_file "$testfile.core"] >> +set core_generated [gdb_gcore_cmd "$core_filename" "generate core file"] >> + >> +if { !$core_generated } { >> + return -1 >> +} >> + >> +clean_restart $binfile >> + >> +# Load the core file and make sure we see the tag violation fault >> +# information. >> +gdb_test "core $core_filename" \ >> + [multi_line \ >> + "Core was generated by.*\." \ >> + "Program terminated with signal SIGSEGV, Segmentation fault" \ >> + "Memory tag violation while accessing address $hex" \ >> + "Allocation tag 0xf" \ >> + "Logical tag 0x0\." \ >> + "#0.*$hex in access_memory \\(.*\\) at .*" \ >> + ".*tagged_ptr\\\[0\\\] = 'a';"] \ >> + "core file shows tag violation information" >> + >> +# Make sure we have the tag_ctl register. >> +gdb_test "info register tag_ctl" \ >> + "tag_ctl.*$hex.*${::decimal}" \ >> + "tag_ctl is available" >> + >> +# Check if the tag granules have the expected values. If they do, that >> +# means the core file saved the tags properly and GDB has read them >> +# correctly. >> +for {set i 15} {$i >= 0} {incr i -1} { >> + set index [expr [expr 15 - $i] * 16] >> + set tag [format "%x" $i] >> + gdb_test "memory-tag print-allocation-tag &tagged_ptr\[$index\]" \ >> + "= 0x$tag" \ >> + "memory tag of &tagged_ptr\[$index\] is correct" >> +} > -- John Baldwin