From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from simark.ca by simark.ca with LMTP id 2P5QN6TY419QJgAAWB0awg (envelope-from ) for ; Wed, 23 Dec 2020 18:54:12 -0500 Received: by simark.ca (Postfix, from userid 112) id D6FF61F0AA; Wed, 23 Dec 2020 18:54:12 -0500 (EST) X-Spam-Checker-Version: SpamAssassin 3.4.2 (2018-09-13) on simark.ca X-Spam-Level: X-Spam-Status: No, score=0.4 required=5.0 tests=DKIM_SIGNED,MAILING_LIST_MULTI, RDNS_NONE,T_DKIM_INVALID,URIBL_BLOCKED autolearn=no autolearn_force=no version=3.4.2 Received: from sourceware.org (unknown [8.43.85.97]) (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 simark.ca (Postfix) with ESMTPS id D57401E590 for ; Wed, 23 Dec 2020 18:54:11 -0500 (EST) Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id 9D8D73865C2A; Wed, 23 Dec 2020 23:54:11 +0000 (GMT) Received: from mail-wm1-x32e.google.com (mail-wm1-x32e.google.com [IPv6:2a00:1450:4864:20::32e]) by sourceware.org (Postfix) with ESMTPS id CE2783857004 for ; Wed, 23 Dec 2020 23:54:07 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.3.2 sourceware.org CE2783857004 Authentication-Results: sourceware.org; dmarc=none (p=none dis=none) header.from=embecosm.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=andrew.burgess@embecosm.com Received: by mail-wm1-x32e.google.com with SMTP id c133so278579wme.4 for ; Wed, 23 Dec 2020 15:54:07 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=embecosm.com; s=google; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=IuRwl5c5TZZxve4ltt56bqltBiQuc3NPQL1XQoxTyR0=; b=aN7LSmGHaeVlaLeGnZ7bK82sO1g3NK7UygomDit60UlNBvzDrc2hJV+tVLICBblDUy ExsPxhVEfECg0i4fUVDK/H3o4jF5yeHDzTZdf5xBuVQb1XGGYvJxZEebcc4ILmFu2RZY NPBs93nyxuDVUfyyUboICGmjVaY++XJo361thMqGFZxVzetKOViiRlDbtZh2oq2ANGML L9hOYVgHKf2dyTfuU9c6SDQnhK9sAT04jbj7JZNr3Ilel/RKqzeeA6VLYcHVPQrPJg93 WvWuhisgDLbaeKp6+m0feYadfRzaoaDnXuYi3vbg6JAZZEPZhXMGvN98U/uAMd4L1iuT aC8w== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=IuRwl5c5TZZxve4ltt56bqltBiQuc3NPQL1XQoxTyR0=; b=ltjqB98WRP2rwIJoq2Tj+1BaFveciwjPDr4IodZwf8usV4vr1F+SAM9My/Q+pujOE0 Ig9RcISznkhUvajjQZF9ApgMpmBmHI6FqMs6hIFV47HOZZ97x+2i5ANUCfSAfuE5TRPU fYddfIUmBFw2zAhe9llByM91FyOxgC3Z0rxVKkAiGluWtwXdir74pKJ9wF5IaY8QmD6d fzWu3NzgXs9RxpV1Fu0Bv01EoKqMtzPVTwx0Q1Ak2yyEpfjdzb+dxHrKL2x9UaO4d0nl JT8pWp2PgQDhntl1ImLPgD+0ReKUlN/gAbYQJE/sCyJseGBGAQdDZ/IvaskVeUGgiGiB rkbQ== X-Gm-Message-State: AOAM530Bm+kwiTbqz1sBAM+Nq85Q2ESH4tIi3MCPRFlkTE1kEOfMSWTV YKhOjIlyy5BFPUqmsO7U0JDfMJTI1AOHqQ== X-Google-Smtp-Source: ABdhPJwkiI5gvN25d+xqWbpEDxjYgUScHF3IZAC0Uy0gA3l0jIFc55dbgb/XcgP1USdSf7YI/6suDQ== X-Received: by 2002:a05:600c:1483:: with SMTP id c3mr1757184wmh.87.1608767646499; Wed, 23 Dec 2020 15:54:06 -0800 (PST) Received: from localhost (host109-154-20-128.range109-154.btcentralplus.com. [109.154.20.128]) by smtp.gmail.com with ESMTPSA id k10sm35060832wrq.38.2020.12.23.15.54.05 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 23 Dec 2020 15:54:05 -0800 (PST) From: Andrew Burgess To: gdb-patches@sourceware.org Subject: [PATCH 2/2] gdb: avoid resolving dynamic properties for non-allocated arrays Date: Wed, 23 Dec 2020 23:53:55 +0000 Message-Id: X-Mailer: git-send-email 2.25.4 In-Reply-To: References: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit 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: , Errors-To: gdb-patches-bounces@sourceware.org Sender: "Gdb-patches" In PR gdb/27059 an issue was discovered where GDB would sometimes trigger undefined behaviour in the form of signed integer overflow. The problem here is that GDB was reading random garbage from the inferior memory space, assuming this data was valid, and performing arithmetic on it. This bug raises an interesting general problem with GDB's DWARF expression evaluator, which is this: We currently assume that the DWARF expressions being evaluated are well formed, and well behaving. As an example, this is the expression that the bug was running into problems on, this was used as the expression for a DW_AT_byte_stride of a DW_TAG_subrange_type: DW_OP_push_object_address; DW_OP_plus_uconst: 88; DW_OP_deref; DW_OP_push_object_address; DW_OP_plus_uconst: 32; DW_OP_deref; DW_OP_mul Two values are read from the inferior and multiplied together. GDB should not assume that any value read from the inferior is in any way sane, as such the implementation of DW_OP_mul should be guarding against overflow and doing something semi-sane here. However, it turns out that the original bug PR gdb/27059, is hitting a more specific case, which doesn't require changes to the DWARF expression evaluator, so I'm going to leave the above issue for another day. In the test mentioned in the bug GDB is actually trying to resolve the dynamic type of a Fortran array that is NOT allocated. A non-allocated Fortran array is one that does not have any data allocated for it yet, and even the upper and lower bounds of the array are not yet known. It turns out that, at least for gfortran compiled code, the data fields that describe the byte-stride are not initialised until the array is allocated. This leads me to the following conclusion: GDB should not try to resolve the bounds, or stride information for an array that is not allocated (or not associated, a similar, but slightly different Fortran feature). Instead, each of these properties should be set to undefined if the array is not allocated (or associated). That is what this commit does. There's a new flag that is passed around during the dynamic array resolution. When this flag is true the dynamic properties are resolved using the DWARF expressions as they currently are, but when this flag is false the expressions are not evaluated, and instead the properties are set to undefined. gdb/ChangeLog: PR gdb/27059 * eval.c (evaluate_subexp_for_sizeof): Handle not allocated and not associated arrays. * f-lang.c (fortran_adjust_dynamic_array_base_address_hack): Don't adjust arrays that are not allocated/associated. * gdbtypes.c (resolve_dynamic_range): Update header comment. Add new parameter which is used to sometimes set dynamic properties to undefined. (resolve_dynamic_array_or_string): Update header comment. Add new parameter which is used to guard evaluating dynamic properties. Resolve allocated/associated properties first. gdb/testsuite/ChangeLog: PR gdb/27059 * gdb.dwarf2/dyn-type-unallocated.c: New file. * gdb.dwarf2/dyn-type-unallocated.exp: New file. --- gdb/ChangeLog | 14 ++ gdb/eval.c | 12 +- gdb/f-lang.c | 4 + gdb/gdbtypes.c | 69 +++++++--- gdb/testsuite/ChangeLog | 6 + .../gdb.dwarf2/dyn-type-unallocated.c | 40 ++++++ .../gdb.dwarf2/dyn-type-unallocated.exp | 122 ++++++++++++++++++ 7 files changed, 243 insertions(+), 24 deletions(-) create mode 100644 gdb/testsuite/gdb.dwarf2/dyn-type-unallocated.c create mode 100644 gdb/testsuite/gdb.dwarf2/dyn-type-unallocated.exp diff --git a/gdb/eval.c b/gdb/eval.c index c781fde0614..dadadbb8353 100644 --- a/gdb/eval.c +++ b/gdb/eval.c @@ -2961,10 +2961,14 @@ evaluate_subexp_for_sizeof (struct expression *exp, int *pos, { val = evaluate_subexp (nullptr, exp, pos, EVAL_NORMAL); type = value_type (val); - if (type->code () == TYPE_CODE_ARRAY - && is_dynamic_type (type->index_type ()) - && type->bounds ()->high.kind () == PROP_UNDEFINED) - return allocate_optimized_out_value (size_type); + if (type->code () == TYPE_CODE_ARRAY) + { + if (type_not_allocated (type) || type_not_associated (type)) + return value_zero (size_type, not_lval); + else if (is_dynamic_type (type->index_type ()) + && type->bounds ()->high.kind () == PROP_UNDEFINED) + return allocate_optimized_out_value (size_type); + } } else (*pos) += 4; diff --git a/gdb/f-lang.c b/gdb/f-lang.c index 28a66fdde94..e06bbb42956 100644 --- a/gdb/f-lang.c +++ b/gdb/f-lang.c @@ -1391,6 +1391,10 @@ fortran_adjust_dynamic_array_base_address_hack (struct type *type, { gdb_assert (type->code () == TYPE_CODE_ARRAY); + /* We can't adjust the base address for arrays that have no content. */ + if (type_not_allocated (type) || type_not_associated (type)) + return address; + int ndimensions = calc_f77_array_dims (type); LONGEST total_offset = 0; diff --git a/gdb/gdbtypes.c b/gdb/gdbtypes.c index 8e90c4b108b..8237eedaec8 100644 --- a/gdb/gdbtypes.c +++ b/gdb/gdbtypes.c @@ -2177,11 +2177,20 @@ static struct type *resolve_dynamic_type_internal /* Given a dynamic range type (dyn_range_type) and a stack of struct property_addr_info elements, return a static version - of that type. */ + of that type. + + When RESOLVE_P is true then the returned static range is created by + actually evaluating any dynamic properties within the range type, while + when RESOLVE_P is false the returned static range has all of the bounds + and stride information set to undefined. The RESOLVE_P set to false + case will be used when evaluating a dynamic array that is not + allocated, or not associated, i.e. the bounds information might not be + initialized yet. */ static struct type * resolve_dynamic_range (struct type *dyn_range_type, - struct property_addr_info *addr_stack) + struct property_addr_info *addr_stack, + bool resolve_p = true) { CORE_ADDR value; struct type *static_range_type, *static_target_type; @@ -2190,13 +2199,13 @@ resolve_dynamic_range (struct type *dyn_range_type, gdb_assert (dyn_range_type->code () == TYPE_CODE_RANGE); const struct dynamic_prop *prop = &dyn_range_type->bounds ()->low; - if (dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) + if (resolve_p && dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) low_bound.set_const_val (value); else low_bound.set_undefined (); prop = &dyn_range_type->bounds ()->high; - if (dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) + if (resolve_p && dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) { high_bound.set_const_val (value); @@ -2209,7 +2218,7 @@ resolve_dynamic_range (struct type *dyn_range_type, bool byte_stride_p = dyn_range_type->bounds ()->flag_is_byte_stride; prop = &dyn_range_type->bounds ()->stride; - if (dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) + if (resolve_p && dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) { stride.set_const_val (value); @@ -2242,11 +2251,16 @@ resolve_dynamic_range (struct type *dyn_range_type, /* Resolves dynamic bound values of an array or string type TYPE to static ones. ADDR_STACK is a stack of struct property_addr_info to be used if - needed during the dynamic resolution. */ + needed during the dynamic resolution. + + When RESOLVE_P is true then the dynamic properties of TYPE are + evaluated, otherwise the dynamic properties of TYPE are not evaluated, + instead we assume the array is not allocated/associated yet. */ static struct type * resolve_dynamic_array_or_string (struct type *type, - struct property_addr_info *addr_stack) + struct property_addr_info *addr_stack, + bool resolve_p = true) { CORE_ADDR value; struct type *elt_type; @@ -2262,29 +2276,44 @@ resolve_dynamic_array_or_string (struct type *type, type = copy_type (type); - elt_type = type; - range_type = check_typedef (elt_type->index_type ()); - range_type = resolve_dynamic_range (range_type, addr_stack); - - /* Resolve allocated/associated here before creating a new array type, which - will update the length of the array accordingly. */ + /* Resolve the allocated and associated properties before doing anything + else. If an array is not allocated or not associated then (at least + for Fortran) there is no guarantee that the data to define the upper + bound, lower bound, or stride will be correct. If RESOLVE_P is + already false at this point then this is not the first dimension of + the array and a more outer dimension has already marked this array as + not allocated/associated, as such we just ignore this property. This + is fine as GDB only checks the allocated/associated on the outer most + dimension of the array. */ prop = TYPE_ALLOCATED_PROP (type); - if (prop != NULL && dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) - prop->set_const_val (value); + if (prop != NULL && resolve_p + && dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) + { + prop->set_const_val (value); + if (value == 0) + resolve_p = false; + } prop = TYPE_ASSOCIATED_PROP (type); - if (prop != NULL && dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) - prop->set_const_val (value); + if (prop != NULL && resolve_p + && dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) + { + prop->set_const_val (value); + if (value == 0) + resolve_p = false; + } - ary_dim = check_typedef (TYPE_TARGET_TYPE (elt_type)); + range_type = check_typedef (type->index_type ()); + range_type = resolve_dynamic_range (range_type, addr_stack, resolve_p); + ary_dim = check_typedef (TYPE_TARGET_TYPE (type)); if (ary_dim != NULL && ary_dim->code () == TYPE_CODE_ARRAY) - elt_type = resolve_dynamic_array_or_string (ary_dim, addr_stack); + elt_type = resolve_dynamic_array_or_string (ary_dim, addr_stack, resolve_p); else elt_type = TYPE_TARGET_TYPE (type); prop = type->dyn_prop (DYN_PROP_BYTE_STRIDE); - if (prop != NULL) + if (prop != NULL && resolve_p) { if (dwarf2_evaluate_property (prop, NULL, addr_stack, &value)) { diff --git a/gdb/testsuite/gdb.dwarf2/dyn-type-unallocated.c b/gdb/testsuite/gdb.dwarf2/dyn-type-unallocated.c new file mode 100644 index 00000000000..453c54d626b --- /dev/null +++ b/gdb/testsuite/gdb.dwarf2/dyn-type-unallocated.c @@ -0,0 +1,40 @@ +/* This testcase is part of GDB, the GNU debugger. + + Copyright 2020 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 . */ + +#include "../lib/attributes.h" + +/* Our fake dynamic object. */ +void *dyn_object; + +void __attribute__((noinline)) ATTRIBUTE_NOCLONE +marker () +{ /* Nothing. */ } + +int +main () +{ + asm ("main_label: .globl main_label"); + + /* Initialise the dynamic object. */ + dyn_object = 0; + + asm ("marker_label: .globl marker_label"); + marker (); /* Break here. */ + + return 0; +} + diff --git a/gdb/testsuite/gdb.dwarf2/dyn-type-unallocated.exp b/gdb/testsuite/gdb.dwarf2/dyn-type-unallocated.exp new file mode 100644 index 00000000000..60cf8abc899 --- /dev/null +++ b/gdb/testsuite/gdb.dwarf2/dyn-type-unallocated.exp @@ -0,0 +1,122 @@ +# Copyright 2020 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 . + +# Test for issue PR gdb/27059. The problem was that when resolving a +# dynamic type that was not-allocated GDB would still try to execute +# the DWARF expressions for the upper, lower, and byte-stride values. +# +# The problem is that, at least in some gfortran compiled programs, +# these values are undefined until the array is allocated. +# +# As a result, executing the dwarf expressions was triggering integer +# overflow in some cases. +# +# This test aims to make the sometimes occurring integer overflow a +# more noticeable error by creating an array that is always marked as +# not-allocated. +# +# The dwarf expressions for the various attributes then contains an +# infinite loop. If GDB ever tries to execute these expressions we +# will get a test timeout. With this issue fixed the expressions are +# never executed and the test completes as we'd expect. + +load_lib dwarf.exp + +if {![dwarf2_support]} { + return 0 +} + +standard_testfile .c -dw.S + +if { [prepare_for_testing "failed to prepare" ${testfile} ${srcfile}] } { + return -1 +} + +set asm_file [standard_output_file $srcfile2] +Dwarf::assemble $asm_file { + cu {} { + global srcfile + + compile_unit { + {producer "gcc" } + {language @DW_LANG_Fortran90} + {name ${srcfile}} + {low_pc 0 addr} + } { + declare_labels array_type_label integer_type_label + + set int_size [get_sizeof "int" "UNKNOWN"] + set voidp_size [get_sizeof "void *" "UNKNOWN"] + + integer_type_label: DW_TAG_base_type { + {DW_AT_byte_size $int_size DW_FORM_sdata} + {DW_AT_encoding @DW_ATE_signed} + {DW_AT_name integer} + } + + array_type_label: DW_TAG_array_type { + {DW_AT_type :$integer_type_label} + {DW_AT_data_location { + DW_OP_push_object_address + DW_OP_deref + } SPECIAL_expr} + {DW_AT_allocated { + DW_OP_push_object_address + DW_OP_deref_size ${voidp_size} + DW_OP_lit0 + DW_OP_ne + } SPECIAL_expr} + } { + DW_TAG_subrange_type { + {DW_AT_type :$integer_type_label} + {DW_AT_lower_bound { + DW_OP_skip -3 + } SPECIAL_expr} + {DW_AT_upper_bound { + DW_OP_skip -3 + } SPECIAL_expr} + {DW_AT_byte_stride { + DW_OP_skip -3 + } SPECIAL_expr} + } + } + + DW_TAG_variable { + {DW_AT_location { + DW_OP_addr [gdb_target_symbol dyn_object] + } SPECIAL_expr} + {name "dyn_object"} + {type :$array_type_label} + } + subprogram { + {external 1 flag} + {MACRO_AT_func {main}} + } + } + } +} + +if { [prepare_for_testing "failed to prepare" "${testfile}" \ + [list $srcfile $asm_file] {nodebug}] } { + return -1 +} + +if ![runto_main] { + return -1 +} + +gdb_breakpoint "marker_label" +gdb_continue_to_breakpoint "stop at marker_label" +gdb_test "ptype dyn_object" "type = integer, allocatable \\(:\\)" -- 2.25.4