gdb/testsuite/gdb.cp/psmang.exp

# Copyright 2002-2021 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 <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.*"