From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) by sourceware.org (Postfix) with ESMTP id C02253858D39 for ; Fri, 25 Oct 2024 03:35:15 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org C02253858D39 Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=redhat.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=redhat.com ARC-Filter: OpenARC Filter v1.0.0 sourceware.org C02253858D39 Authentication-Results: server2.sourceware.org; arc=none smtp.remote-ip=170.10.133.124 ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1729827326; cv=none; b=cutvC7FfOTlWGATaVlQt2OZ5EDbp6HeH+lTWfI1X6qqbJtd6sPNnyks/zSYawfuIyWmwHm7iXIhkZWJkX5IZbvmUCY8Qu1LddbAHbQtT+0sUSUd+ZFdHa5gJXcL5jAt27u0UOiE32HsHrhLo/IZ1CiX6Fq6LUbuusnIfc2aOigk= ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1729827326; c=relaxed/simple; bh=3zNcUMD7chDJsNWef7nLJ1nWvJEVxHuHTAQVb5B2e4k=; h=DKIM-Signature:From:To:Subject:Date:Message-ID:MIME-Version; b=Lu0Qe2yz9L2m0zMR3C99csdQAWz9yEalKAF1P1D0tF5Kh92T95cxkDa329P8KgjfJZMxEOTJkuPGzYSxL0wL3u/7DAmNL9MlVtGhlmhMtQ+YCy7tG/8tKEweV3qExSAzusznEl0RYvTx4a5mPDmLK0mbQ+7SurQtfr5LV15UHJA= ARC-Authentication-Results: i=1; server2.sourceware.org DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1729827315; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=G1qOjXhN6MVrHNXAPw8XEMh5mU4qG+qGot/jjb9gdDk=; b=Z7HcdGquCejBIgDemUem1AKz4Tb27gUsUiPLpKyq+eCm+5SeW2KyIjDDNy4mUPK/QSeOjk +j6ih9kUHak8FLNdPCWM2VCJ0cgt6JPiOGsTnIpeCpURflUP9miBttx6yBN98L+II0lbZJ 5NH+52/kbP769WTtLslOlDUiFNRALEk= Received: from mx-prod-mc-05.mail-002.prod.us-west-2.aws.redhat.com (ec2-54-186-198-63.us-west-2.compute.amazonaws.com [54.186.198.63]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-100-Gcm_rIknPTmCqrL8bRcCHA-1; Thu, 24 Oct 2024 23:35:04 -0400 X-MC-Unique: Gcm_rIknPTmCqrL8bRcCHA-1 Received: from mx-prod-int-01.mail-002.prod.us-west-2.aws.redhat.com (mx-prod-int-01.mail-002.prod.us-west-2.aws.redhat.com [10.30.177.4]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) by mx-prod-mc-05.mail-002.prod.us-west-2.aws.redhat.com (Postfix) with ESMTPS id C16D91955F41 for ; Fri, 25 Oct 2024 03:35:02 +0000 (UTC) Received: from f41-1.lan (unknown [10.22.64.38]) by mx-prod-int-01.mail-002.prod.us-west-2.aws.redhat.com (Postfix) with ESMTP id E92B83000198; Fri, 25 Oct 2024 03:35:01 +0000 (UTC) From: Kevin Buettner To: gdb-patches@sourceware.org Cc: Kevin Buettner Subject: [PATCH v3 03/11] Track and fetch TLS module ids for MUSL and GLIBC Date: Thu, 24 Oct 2024 19:26:21 -0700 Message-ID: <20241025033431.36274-4-kevinb@redhat.com> In-Reply-To: <20241025033431.36274-1-kevinb@redhat.com> References: <20241025033431.36274-1-kevinb@redhat.com> MIME-Version: 1.0 X-Scanned-By: MIMEDefang 3.4.1 on 10.30.177.4 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset="US-ASCII"; x-default=true X-Spam-Status: No, score=-11.0 required=5.0 tests=BAYES_00,DKIMWL_WL_HIGH,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,DKIM_VALID_EF,GIT_PATCH_0,RCVD_IN_DNSWL_NONE,RCVD_IN_MSPIKE_H3,RCVD_IN_MSPIKE_WL,SPF_HELO_NONE,SPF_NONE,TXREP autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org List-Id: This commit adds, to solib-svr4.h and solib-svr4.c, functions glibc_link_map_to_tls_module_id and musl_link_map_to_tls_module_id for use with callers in linux-tdep.c (which is not in this commit). It adds a number of helper functions for implementing link map to module id support. It also renames existing function 'find_program_interpreter' to 'svr4_find_program_interpreter' and makes it visible to other source files within GDB. It will be used in the libc sniffing code in linux-tdep.c in a later commit in this series. The libc sniffer is needed in order to know which link map to module id function to call as the method for determining module ids differs between libc / dynamic linker implementations. These details are discussed in comments in the patch. Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=24548 Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=31563 --- gdb/solib-svr4.c | 203 ++++++++++++++++++++++++++++++++++++++++++++++- gdb/solib-svr4.h | 12 +++ 2 files changed, 211 insertions(+), 4 deletions(-) diff --git a/gdb/solib-svr4.c b/gdb/solib-svr4.c index 7999a8e6c7e..b331aac6122 100644 --- a/gdb/solib-svr4.c +++ b/gdb/solib-svr4.c @@ -405,6 +405,9 @@ struct svr4_info The special entry zero is reserved for a linear list to support gdbstubs that do not support namespaces. */ std::map> solib_lists; + + bool glibc_tls_slots_inited = false; + std::vector glibc_tls_slots; }; /* Per-program-space data key. */ @@ -586,10 +589,10 @@ read_program_header (int type, int *p_arch_size, CORE_ADDR *base_addr) return buf; } +/* See solib-svr4.h. */ -/* Return program interpreter string. */ -static std::optional -find_program_interpreter (void) +std::optional +svr4_find_program_interpreter () { /* If we have a current exec_bfd, use its section table. */ if (current_program_space->exec_bfd () @@ -1516,6 +1519,194 @@ svr4_fetch_objfile_link_map (struct objfile *objfile) return 0; } +/* Return true if bfd section BFD_SECT is a thread local section + (i.e. either named ".tdata" or ".tbss"), and false otherwise. */ + +static bool +is_thread_local_section (struct bfd_section *bfd_sect) +{ + return ((strcmp (bfd_sect->name, ".tdata") == 0 + || strcmp (bfd_sect->name, ".tbss") == 0) + && bfd_sect->size != 0); +} + +/* Return true if objfile OBJF contains a thread local section, and + false otherwise. */ + +static bool +has_thread_local_section (const objfile *objf) +{ + for (obj_section *objsec : objf->sections ()) + if (is_thread_local_section (objsec->the_bfd_section)) + return true; + return false; +} + +/* Return true if solib SO contains a thread local section, and false + otherwise. */ + +static bool +has_thread_local_section (const solib &so) +{ + for (const target_section &p : so.sections) + if (is_thread_local_section (p.the_bfd_section)) + return true; + return false; +} + +/* For the MUSL C library, given link map address LM_ADDR, return the + corresponding TLS module id, or 0 if not found. + + Background: Unlike the mechanism used by glibc (see below), the + scheme used by the MUSL C library is pretty simple. If the + executable contains TLS variables it gets module id 1. Otherwise, + the first shared object loaded which contains TLS variables is + assigned to module id 1. TLS-containing shared objects are then + assigned consecutive module ids, based on the order that they are + loaded. When unloaded via dlclose, module ids are reassigned as if + that module had never been loaded. */ + +int +musl_link_map_to_tls_module_id (CORE_ADDR lm_addr) +{ + /* When lm_addr is zero, the program is statically linked. Any TLS + variables will be in module id 1. */ + if (lm_addr == 0) + return 1; + + int mod_id = 0; + if (has_thread_local_section (current_program_space->symfile_object_file)) + mod_id++; + + struct svr4_info *info = get_svr4_info (current_program_space); + + /* Cause svr4_current_sos() to be run if it hasn't been already. */ + if (info->main_lm_addr == 0) + solib_add (NULL, 0, auto_solib_add); + + /* Handle case where lm_addr corresponds to the main program. + Return value is either 0, when there are no TLS variables, or 1, + when there are. */ + if (lm_addr == info->main_lm_addr) + return mod_id; + + /* Iterate through the shared objects, possibly incrementing the + module id, and returning mod_id should a match be found. */ + for (const solib &so : current_program_space->solibs ()) + { + if (has_thread_local_section (so)) + mod_id++; + + auto *li = gdb::checked_static_cast (so.lm_info.get ()); + if (li->lm_addr == lm_addr) + return mod_id; + } + return 0; +} + +/* For GLIBC, given link map address LM_ADDR, return the corresponding TLS + module id, or 0 if not found. */ + +int +glibc_link_map_to_tls_module_id (CORE_ADDR lm_addr) +{ + /* When lm_addr is zero, the program is statically linked. Any TLS + variables will be in module id 1. */ + if (lm_addr == 0) + return 1; + + /* Look up lm_addr in the TLS slot data structure. */ + struct svr4_info *info = get_svr4_info (current_program_space); + auto it = std::find (info->glibc_tls_slots.begin (), + info->glibc_tls_slots.end (), + lm_addr); + if (it == info->glibc_tls_slots.end ()) + return 0; + else + return 1 + it - info->glibc_tls_slots.begin (); +} + +/* Conditionally, based on whether the shared object, SO, contains TLS + variables, assign a link map address to a TLS module id slot. This + code is GLIBC-specific and may only work for specific GLIBC + versions. That said, it is known to work for (at least) GLIBC + versions 2.27 thru 2.40. + + Background: In order to implement internal TLS address lookup + code, it is necessary to find the module id that has been + associated with a specific link map address. In GLIBC, the TLS + module id is stored in struct link_map, in the member + 'l_tls_modid'. While the first several members of struct link_map + are part of the SVR4 ABI, the offset to l_tls_modid definitely is + not. Therefore, since we don't know the offset to l_tls_modid, we + cannot simply look it up - which is a shame, because things would + be so much more easy and obviously accurate, if we could access + l_tls_modid. + + GLIBC has a concept of TLS module id slots. These slots are + allocated consecutively as shared objects containing TLS variables + are loaded. When unloaded (e.g. via dlclose()), the corresponding + slot is marked as unused, but may be used again when later loading + a shared object. + + The functions tls_maybe_fill_slot and tls_maybe_erase_slot are + associated with the observers 'solib_loaded' and 'solib_unloaded'. + They (attempt to) track use of TLS module id slots in the same way + that GLIBC does, which will hopefully provide an accurate module id + when asked to provide it via glibc_link_map_to_tls_module_id(), + above. */ + +static void +tls_maybe_fill_slot (solib &so) +{ + struct svr4_info *info = get_svr4_info (current_program_space); + if (!info->glibc_tls_slots_inited) + { + /* Cause svr4_current_sos() to be run if it hasn't been already. */ + if (info->main_lm_addr == 0) + svr4_current_sos_direct (info); + + /* Quit early when main_lm_addr is still 0. */ + if (info->main_lm_addr == 0) + return; + + /* Also quit early when symfile_object_file is not yet known. */ + if (current_program_space->symfile_object_file == nullptr) + return; + + if (has_thread_local_section (current_program_space->symfile_object_file)) + info->glibc_tls_slots.push_back (info->main_lm_addr); + info->glibc_tls_slots_inited = true; + } + + if (has_thread_local_section (so)) + { + auto it = std::find (info->glibc_tls_slots.begin (), + info->glibc_tls_slots.end (), + 0); + auto *li = gdb::checked_static_cast (so.lm_info.get ()); + if (it == info->glibc_tls_slots.end ()) + info->glibc_tls_slots.push_back (li->lm_addr); + else + *it = li->lm_addr; + } +} + +/* Remove a link map address from the TLS module slot data structure. + As noted above, this code is GLIBC-specific. */ + +static void +tls_maybe_erase_slot (program_space *pspace, const solib &so) +{ + struct svr4_info *info = get_svr4_info (pspace); + auto *li = gdb::checked_static_cast (so.lm_info.get ()); + auto it = std::find (info->glibc_tls_slots.begin (), + info->glibc_tls_slots.end (), + li->lm_addr); + if (it != info->glibc_tls_slots.end ()) + *it = 0; +} + /* On some systems, the only way to recognize the link map entry for the main executable file is by looking at its name. Return non-zero iff SONAME matches one of the known main executable names. */ @@ -2305,7 +2496,7 @@ enable_break (struct svr4_info *info, int from_tty) /* Find the program interpreter; if not found, warn the user and drop into the old breakpoint at symbol code. */ std::optional interp_name_holder - = find_program_interpreter (); + = svr4_find_program_interpreter (); if (interp_name_holder) { const char *interp_name = (const char *) interp_name_holder->data (); @@ -3382,4 +3573,8 @@ _initialize_svr4_solib () { gdb::observers::free_objfile.attach (svr4_free_objfile_observer, "solib-svr4"); + + /* Set up observers for tracking GLIBC TLS module id slots. */ + gdb::observers::solib_loaded.attach (tls_maybe_fill_slot, "solib-svr4"); + gdb::observers::solib_unloaded.attach (tls_maybe_erase_slot, "solib-svr4"); } diff --git a/gdb/solib-svr4.h b/gdb/solib-svr4.h index 579fe6d9843..6e25441f221 100644 --- a/gdb/solib-svr4.h +++ b/gdb/solib-svr4.h @@ -112,4 +112,16 @@ extern struct link_map_offsets *svr4_lp64_fetch_link_map_offsets (void); SVR4 run time loader. */ int svr4_in_dynsym_resolve_code (CORE_ADDR pc); +/* For the MUSL C library, given link map address LM_ADDR, return the + corresponding TLS module id, or 0 if not found. */ +int musl_link_map_to_tls_module_id (CORE_ADDR lm_addr); + +/* For GLIBC, given link map address LM_ADDR, return the corresponding TLS + module id, or 0 if not found. */ +int glibc_link_map_to_tls_module_id (CORE_ADDR lm_addr); + +/* Return program interpreter string. */ + +std::optional svr4_find_program_interpreter (); + #endif /* solib-svr4.h */ -- 2.46.2