# Copyright 1997, 1998, 1999, 2007, 2008, 2009, 2010, 2011 # 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 was written by Fred Fish. (fnf@cygnus.com) # These tests are the same as those in callfuncs.exp, except that the # test program here does not call malloc. # # "What in the world does malloc have to do with calling functions in # the inferior?" Well, nothing. GDB's ability to invoke a function # in the inferior program works just fine in programs that have no # malloc function available. It doesn't rely on the inferior's # malloc, directly or indirectly. It just uses the inferior's stack # space. # # "Then what's the point of this test file?" Well, it just so happens # that this file, in addition to testing inferior function calls, also # tests GDB's ability to evaluate string literals (like "string 1" and # "string 2" in the tests below). Evaluating *those* sorts of # expressions does require malloc. # # (As an extension to C, GDB also has a syntax for literal arrays of # anything, not just characters. For example, the expression # {2,3,4,5} (which appears in the tests below) evaluates to an array # of four ints. So rather than talking just about string literals, # we'll use the broader term "array literals".) # # Now, in this file, we only evaluate array literals when we're about # to pass them to a function, but don't be confused --- this is a red # herring. You can evaluate "abcdef" even if you're not about to pass # that to a function, and doing so requires malloc even if you're just # going to store a pointer to it in a variable, like this: # # (gdb) ptype s # type = char * # (gdb) set variable s = "abcdef" # # According to C's rules for evaluating expressions, arrays are # converted into pointers to their first element. This means that, in # order to evaluate an expression like "abcdef", GDB needs to actually # find some memory in the inferior we can plop the characters into; # then we use that memory's address as the address of our array # literal. GDB finds this memory by calling the inferior's malloc # function, if it has one. So, evaluating an array literal depends on # performing an inferior function call, but not vice versa. (GDB # can't just allocate the space on the stack; the pointer may remain # live long after the current frame has been popped.) # # "But, if evaluating array literals requires malloc, what's the point # of testing that GDB can do so in a program that doesn't have malloc? # It can't work!" On most systems, that's right, but HP-UX has some # sort of dynamic linking magic that ensures that *every* program has # malloc. So on HP-UX, GDB can evaluate array literals even in # inferior programs that don't use malloc. That's why this test is in # gdb.hp. # # This file has, for some reason, led to well more than its fair share # of misunderstandings about the relationship between array literal # expressions and inferior function calls. Folks talk as if you can # only evaluate array literals when you're about to pass them to a # function. I think they're assuming that, since GDB is constructing # a new frame on the inferior's stack (correct), it's going to use # that space for the array literals (incorrect). Remember that those # array literals may need to be live long after the inferior function # call returns; GDB can't tell. # # What makes the confusion worse is that there *is* a relationship # between array literals and inferior function calls --- GDB uses # inferior function calls to evaluate array literals. But many people # jump to other, incorrect conclusions about this. if $tracelevel then { strace $tracelevel } if { [skip_hp_tests] } then { continue } set testfile "callfwmall" set srcfile ${testfile}.c set binfile ${objdir}/${subdir}/${testfile} if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug}] != "" } { untested callfwmall.exp return -1 } # Create and source the file that provides information about the compiler # used to compile the test case. if [get_compiler_info ${binfile}] { return -1; } if {$hp_aCC_compiler} { set prototypes 1 } else { set prototypes 0 } # Some targets can't call functions, so don't even bother with this # test. if [target_info exists gdb,cannot_call_functions] { setup_xfail "*-*-*" 2416 fail "This target can not call functions" continue } # Set the current language to C. This counts as a test. If it # fails, then we skip the other tests. proc set_lang_c {} { global gdb_prompt send_gdb "set language c\n" gdb_expect { -re ".*$gdb_prompt $" {} timeout { fail "set language c (timeout)" ; return 0 } } send_gdb "show language\n" gdb_expect { -re ".* source language is \"c\".*$gdb_prompt $" { pass "set language to \"c\"" return 1 } -re ".*$gdb_prompt $" { fail "setting language to \"c\"" return 0 } timeout { fail "can't show language (timeout)" return 0 } } } # FIXME: Before calling this proc, we should probably verify that # we can call inferior functions and get a valid integral value # returned. # Note that it is OK to check for 0 or 1 as the returned values, because C # specifies that the numeric value of a relational or logical expression # (computed in the inferior) is 1 for true and 0 for false. proc do_function_calls {} { global prototypes global gcc_compiled global gdb_prompt # We need to up this because this can be really slow on some boards. set timeout 60; gdb_test "p t_char_values(0,0)" " = 0" gdb_test "p t_char_values('a','b')" " = 1" gdb_test "p t_char_values(char_val1,char_val2)" " = 1" gdb_test "p t_char_values('a',char_val2)" " = 1" gdb_test "p t_char_values(char_val1,'b')" " = 1" gdb_test "p t_short_values(0,0)" " = 0" gdb_test "p t_short_values(10,-23)" " = 1" gdb_test "p t_short_values(short_val1,short_val2)" " = 1" gdb_test "p t_short_values(10,short_val2)" " = 1" gdb_test "p t_short_values(short_val1,-23)" " = 1" gdb_test "p t_int_values(0,0)" " = 0" gdb_test "p t_int_values(87,-26)" " = 1" gdb_test "p t_int_values(int_val1,int_val2)" " = 1" gdb_test "p t_int_values(87,int_val2)" " = 1" gdb_test "p t_int_values(int_val1,-26)" " = 1" gdb_test "p t_long_values(0,0)" " = 0" gdb_test "p t_long_values(789,-321)" " = 1" gdb_test "p t_long_values(long_val1,long_val2)" " = 1" gdb_test "p t_long_values(789,long_val2)" " = 1" gdb_test "p t_long_values(long_val1,-321)" " = 1" if ![target_info exists gdb,skip_float_tests] { gdb_test "p t_float_values(0.0,0.0)" " = 0" # These next four tests fail on the mn10300. # The first value is passed in regs, the other in memory. # Gcc emits different stabs for the two parameters; the first is # claimed to be a float, the second a double. # dbxout.c in gcc claims this is the desired behavior. setup_xfail "mn10300-*-*" gdb_test "p t_float_values(3.14159,-2.3765)" " = 1" setup_xfail "mn10300-*-*" gdb_test "p t_float_values(float_val1,float_val2)" " = 1" setup_xfail "mn10300-*-*" gdb_test "p t_float_values(3.14159,float_val2)" " = 1" setup_xfail "mn10300-*-*" gdb_test "p t_float_values(float_val1,-2.3765)" " = 1" # Test passing of arguments which might not be widened. gdb_test "p t_float_values2(0.0,0.0)" " = 0" # Although PR 5318 mentions SunOS specifically, this seems # to be a generic problem on quite a few platforms. if $prototypes then { setup_xfail "sparc-*-*" "mips*-*-*" 5318 if {!$gcc_compiled} then { setup_xfail "alpha-dec-osf2*" "i*86-*-sysv4*" 5318 } } gdb_test "p t_float_values2(3.14159,float_val2)" " = 1" gdb_test "p t_small_values(1,2,3,4,5,6,7,8,9,10)" " = 55" gdb_test "p t_double_values(0.0,0.0)" " = 0" gdb_test "p t_double_values(45.654,-67.66)" " = 1" gdb_test "p t_double_values(double_val1,double_val2)" " = 1" gdb_test "p t_double_values(45.654,double_val2)" " = 1" gdb_test "p t_double_values(double_val1,-67.66)" " = 1" } gdb_test "p t_string_values(string_val2,string_val1)" " = 0" gdb_test "p t_string_values(string_val1,string_val2)" " = 1" gdb_test "p t_string_values(\"string 1\",\"string 2\")" " = 1" gdb_test "p t_string_values(\"string 1\",string_val2)" " = 1" gdb_test "p t_string_values(string_val1,\"string 2\")" " = 1" gdb_test "p t_char_array_values(char_array_val2,char_array_val1)" " = 0" gdb_test "p t_char_array_values(char_array_val1,char_array_val2)" " = 1" gdb_test "p t_char_array_values(\"carray 1\",\"carray 2\")" " = 1" gdb_test "p t_char_array_values(\"carray 1\",char_array_val2)" " = 1" gdb_test "p t_char_array_values(char_array_val1,\"carray 2\")" " = 1" gdb_test "p doubleit(4)" " = 8" gdb_test "p add(4,5)" " = 9" gdb_test "p t_func_values(func_val2,func_val1)" " = 0" gdb_test "p t_func_values(func_val1,func_val2)" " = 1" # On the rs6000, we need to pass the address of the trampoline routine, # not the address of add itself. I don't know how to go from add to # the address of the trampoline. Similar problems exist on the HPPA, # and in fact can present an unsolvable problem as the stubs may not # even exist in the user's program. We've slightly recoded t_func_values # to avoid such problems in the common case. This may or may not help # the RS6000. setup_xfail "rs6000*-*-*" if {![istarget hppa*-*-hpux*]} then { gdb_test "p t_func_values(add,func_val2)" " = 1" } setup_xfail "rs6000*-*-*" if {![istarget hppa*-*-hpux*]} then { gdb_test "p t_func_values(func_val1,doubleit)" " = 1" } gdb_test "p t_call_add(func_val1,3,4)" " = 7" setup_xfail "rs6000*-*-*" if {![istarget hppa*-*-hpux*]} then { gdb_test "p t_call_add(add,3,4)" " = 7" } gdb_test "p t_enum_value1(enumval1)" " = 1" gdb_test "p t_enum_value1(enum_val1)" " = 1" gdb_test "p t_enum_value1(enum_val2)" " = 0" gdb_test "p t_enum_value2(enumval2)" " = 1" gdb_test "p t_enum_value2(enum_val2)" " = 1" gdb_test "p t_enum_value2(enum_val1)" " = 0" gdb_test "p sum_args(1,{2})" " = 2" gdb_test "p sum_args(2,{2,3})" " = 5" gdb_test "p sum_args(3,{2,3,4})" " = 9" gdb_test "p sum_args(4,{2,3,4,5})" " = 14" gdb_test "p sum10 (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)" " = 55" gdb_test "p t_structs_c(struct_val1)" "= 120 'x'" \ "call inferior func with struct - returns char" gdb_test "p t_structs_s(struct_val1)" "= 87" \ "call inferior func with struct - returns short" gdb_test "p t_structs_i(struct_val1)" "= 76" \ "call inferior func with struct - returns int" gdb_test "p t_structs_l(struct_val1)" "= 51" \ "call inferior func with struct - returns long" gdb_test "p t_structs_f(struct_val1)" "= 2.12.*" \ "call inferior func with struct - returns float" gdb_test "p t_structs_d(struct_val1)" "= 9.87.*" \ "call inferior func with struct - returns double" gdb_test "p t_structs_a(struct_val1)" "= (.unsigned char .. )?\"foo\"" \ "call inferior func with struct - returns char *" } # Start with a fresh gdb. gdb_exit gdb_start gdb_reinitialize_dir $srcdir/$subdir gdb_load ${binfile} gdb_test "set print sevenbit-strings" "" gdb_test "set print address off" "" gdb_test "set width 0" "" if { $hp_aCC_compiler } { # Do not set language explicitly to 'C'. This will cause aCC # tests to fail because promotion rules are different. Just let # the language be set to the default. if { ![runto_main] } { gdb_suppress_tests; } gdb_test "set overload-resolution 0" ".*" } else { if { ![set_lang_c] } { gdb_suppress_tests; } else { if { ![runto_main] } { gdb_suppress_tests; } } } gdb_test "next" ".*" do_function_calls return 0