200 lines
9.4 KiB
Plaintext
200 lines
9.4 KiB
Plaintext
# Copyright 2002-2018 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 <http://www.gnu.org/licenses/>.
|
|
|
|
# This file is part of the gdb testsuite
|
|
|
|
# Looking up methods by name, in programs with multiple compilation units.
|
|
|
|
# ====== PLEASE BE VERY CAREFUL WHEN CHANGING THIS TEST. =====
|
|
#
|
|
# The bug we're testing for (circa October 2002) is very sensitive to
|
|
# various conditions that are hard to control directly in the test
|
|
# suite. If you change the test, please revert this change, and make
|
|
# sure the test still fails:
|
|
#
|
|
# 2002-08-29 Jim Blandy <jimb@redhat.com>
|
|
#
|
|
# * symtab.c (lookup_symbol_aux): In the cases where we find a
|
|
# minimal symbol of an appropriate name and use its address to
|
|
# select a symtab to read and search, use `name' (as passed to us)
|
|
# as the demangled name when searching the symtab's global and
|
|
# static blocks, not the minsym's name.
|
|
#
|
|
# The original bug was that you'd try to set a breakpoint on a method
|
|
# (e.g., `break s::method1'), and you'd get an error, but if you
|
|
# repeated the command, it would work the second time:
|
|
#
|
|
# (gdb) break s::method1
|
|
# the class s does not have any method named method1
|
|
# Hint: try 's::method1<TAB> or 's::method1<ESC-?>
|
|
# (Note leading single quote.)
|
|
# (gdb) break s::method1
|
|
# Breakpoint 1 at 0x804841b: file psmang1.cc, line 13.
|
|
# (gdb)
|
|
#
|
|
# We observed this bug first using Stabs, and then using Dwarf 2.
|
|
#
|
|
# The problem was in lookup_symbol_aux: when looking up s::method1, it
|
|
# would fail to find it in any symtabs, find the minsym with the
|
|
# corresponding mangled name (say, `_ZN1S7method1Ev'), pass the
|
|
# minsym's address to find_pc_sect_symtab to look up the symtab
|
|
# (causing the compilation unit's full symbols to be read in), and
|
|
# then look up the symbol in that symtab's global block. All that is
|
|
# correct. However, it would pass the minsym's name as the NAME
|
|
# argument to lookup_block_symbol; a minsym's name is mangled, whereas
|
|
# lookup_block_symbol's NAME argument should be demangled.
|
|
#
|
|
# This is a pretty simple bug, but it turns out to be a bear to
|
|
# construct a test for. That's why this test case is so delicate. If
|
|
# you can see how to make it less so, please contribute a patch.
|
|
#
|
|
# Here are the twists:
|
|
#
|
|
# The bug only manifests itself when we call lookup_symbol to look up
|
|
# a method name (like "s::method1" or "s::method2"), and that method's
|
|
# definition is in a compilation unit for which we have read partial
|
|
# symbols, but not full symbols. The partial->full conversion must be
|
|
# caused by that specific lookup. (If we already have full symbols
|
|
# for the method's compilation unit, we won't need to look up the
|
|
# minsym, find the symtab for the minsym's address, and then call
|
|
# lookup_block_symbol; it's that last call where things go awry.)
|
|
#
|
|
# Now, when asked to set a breakpoint at `s::method1', GDB will first
|
|
# look up `s' to see if that is, in fact, the name of a class, and
|
|
# then look up 's::method1'. So we have to make sure that looking up
|
|
# `s' doesn't cause full symbols to be read for the compilation unit
|
|
# containing the definition of `s::method1'.
|
|
#
|
|
# The partial symbol tables for `psmang1.cc' and `psmang2.cc' will
|
|
# both have entries for `s'; GDB will read full symbols for whichever
|
|
# compilation unit's partial symbol table appears first in the
|
|
# objfile's list. The order in which compilation units appear in the
|
|
# partial symbol table list depends on how the program is linked, and
|
|
# how the debug info reader does the partial symbol scan. Ideally,
|
|
# the test shouldn't rely on them appearing in any particular order.
|
|
#
|
|
# So, since we don't know which compilation unit's full symbols are
|
|
# going to get read, we simply try looking up one method from each of
|
|
# the two compilation units. One of them has to come after the other
|
|
# in the partial symbol table list, so whichever comes later will
|
|
# still need its partial symbols read by the time we go to look up
|
|
# 's::methodX'.
|
|
#
|
|
# Second twist: don't move the common definition of `struct s' into a
|
|
# header file. If the compiler emits identical stabs for the
|
|
# #inclusion of that header file into psmang1.cc and into psmang2.cc,
|
|
# then the linker will do stabs compression, and replace one of the
|
|
# BINCL/EINCL regions with an EXCL stab, pointing to the other
|
|
# BINCL/EINCL region. GDB will read this, and record that the
|
|
# compilation unit that got the EXCL depends on the compilation unit
|
|
# that kept the BINCL/EINCL. Then, when it decides it needs to read
|
|
# full symbols for the former, it'll also read full symbols for the
|
|
# latter. Now, if it just so happens that the compilation unit that
|
|
# got the EXCL is also the first one with a definition of `s' in the
|
|
# partial symbol table list, then that first probe for `s' will cause
|
|
# both compilation units' full symbols to be read --- again defeating
|
|
# the test.
|
|
#
|
|
# We could work around this by having three compilation units, or by
|
|
# ensuring that the header file produces different stabs each time
|
|
# it's #included, but it seems simplest just to avoid compilation unit
|
|
# dependencies altogether, drop the header file, and duplicate the
|
|
# (pretty trivial) struct definition.
|
|
#
|
|
# Note that #including any header file at all into both compilation
|
|
# units --- say, <stdio.h> --- could create this sort of dependency.
|
|
#
|
|
# This is the aspect of the test which the debug format is most likely
|
|
# to affect, I think. The different formats create different kinds of
|
|
# inter-CU dependencies, which could mask the bug. It might be
|
|
# possible for the test to check that at least one of the partial
|
|
# symtabs remains unread, and fail otherwise --- the failure
|
|
# indicating that the test itself isn't going to catch the bug it was
|
|
# meant to, not that GDB is misbehaving.
|
|
#
|
|
# Third twist: given the way lookup_block_symbol is written, it's
|
|
# possible to find the symbol even when it gets passed a mangled name
|
|
# for its NAME parameter. There are three ways lookup_block_symbol
|
|
# might search a block, depending on how it was constructed:
|
|
#
|
|
# linear search. In this case, this bug will never manifest itself,
|
|
# since we check every symbol against NAME using SYMBOL_MATCHES_NAME.
|
|
# Since that macro checks its second argument (NAME) against both the
|
|
# mangled and demangled names of the symbol, this will always find the
|
|
# symbol successfully, so, no bug.
|
|
#
|
|
# hash table. If both the mangled and demangled names hash to the
|
|
# same bucket, then you'll again find the symbol "by accident", since
|
|
# we search the entire bucket using SYMBOL_SOURCE_NAME. Since GDB
|
|
# chooses the number of buckets based on the number of symbols, small
|
|
# compilation units may have only one hash bucket; in this case, the
|
|
# search always succeeds, even though we hashed on the wrong name.
|
|
# This test works around that by having a lot of dummy variables,
|
|
# making it less likely that the mangled and demangled names fall in
|
|
# the same bucket.
|
|
#
|
|
# binary search. (GDB 5.2 produced these sorts of blocks, and this
|
|
# test tries to detect the bug there, but subsequent versions of GDB
|
|
# almost never build them, and they may soon be removed entirely.) In
|
|
# this case, the symbols in the block are sorted by their
|
|
# SYMBOL_SOURCE_NAME (whose behavior depends on the current demangling
|
|
# setting, so that's wrong, but let's try to stay focussed).
|
|
# lookup_block_symbol does a binary search comparing NAME with
|
|
# SYMBOL_SOURCE_NAME until the range has been narrowed down to only a
|
|
# few symbols; then it starts a linear search forward from the lower
|
|
# end of that range, until it reaches a symbol whose
|
|
# SYMBOL_SOURCE_NAME follows NAME in lexicographic order. This means
|
|
# that, if you're doing a binary search for a mangled name in a block
|
|
# sorted by SYMBOL_SOURCE_NAME, you might find the symbol `by
|
|
# accident' if the mangled and demangled names happen to fall near
|
|
# each other in the ordering. The initial version of this patch used
|
|
# a class called `S'; all the other symbols in the compilation unit
|
|
# started with lower-case letters, so the demangled name `S::method1'
|
|
# sorted at the same place as the mangled name `_ZN1S7method1Ev': at
|
|
# the very beginning. Using a lower-case 's' as the name ensures that
|
|
# the demangled name falls after all the dummy symbols introduced for
|
|
# the hash table, as described above.
|
|
#
|
|
# This is all so tortured, someone will probably come up with still
|
|
# other ways this test could fail to do its job. If you need to make
|
|
# revisions, please be very careful.
|
|
|
|
#
|
|
# test running programs
|
|
#
|
|
|
|
|
|
if { [skip_cplus_tests] } { continue }
|
|
|
|
standard_testfile psmang1.cc psmang2.cc
|
|
|
|
if [get_compiler_info "c++"] {
|
|
return -1
|
|
}
|
|
|
|
if {[prepare_for_testing "failed to prepare" $testfile \
|
|
[list $srcfile $srcfile2] {debug c++}]} {
|
|
return -1
|
|
}
|
|
|
|
gdb_test "break s::method1" "Breakpoint .* at .*: file .*psmang1.cc.*"
|
|
|
|
# We have to exit and restart GDB here, to make sure that all the
|
|
# compilation units are psymtabs again.
|
|
|
|
clean_restart ${binfile}
|
|
|
|
gdb_test "break s::method2" "Breakpoint .* at .*: file .*psmang2.cc.*"
|