From: Lancelot SIX Date: Thu, 8 Jun 2023 14:24:55 +0000 (+0100) Subject: gdb/testsuite: use proper int size for gdb.dwarf2/symbol_needs_eval*.exp X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=bdde90c4cea27aacb0b98b4b1fa7289cb1a96168;p=binutils-gdb.git gdb/testsuite: use proper int size for gdb.dwarf2/symbol_needs_eval*.exp We recently realized that symbol_needs_eval_fail.exp and symbol_needs_eval_timeout.exp invalidly dereference an int (4 bytes on x86_64) by reading 8 bytes (the size of a pointer). Here how it goes: In gdb/testsuite/gdb.dwarf2/symbol_needs_eval.c a global variable is defined: int exec_mask = 1; and later both tests build some DWARF using the assembler doing: set exec_mask_var [gdb_target_symbol exec_mask] ... DW_TAG_variable { {DW_AT_name a} {DW_AT_type :$int_type_label} {DW_AT_location { DW_OP_addr $exec_mask_var DW_OP_deref ... } } The definition of the DW_OP_deref (from Dwarf5 2.5.1.3 Stack Operations) says that "The size of the data retrieved from the dereferenced address is the size of an address on the target machine." On x86_64, the size of an int is 4 while the size of an address is 8. The result is that when evaluating this expression, the debugger reads outside of the `a` variable. Fix this by using `DW_OP_deref_size $int_size` instead. To achieve this, this patch adds the necessary steps so we can figure out what `sizeof(int)` evaluates to for the current target. While at it, also change the definition of the int type in the assembled DWARF information so we use the actual target's size for an int instead of the literal 4. Tested on x86_64 Linux. Approved-By: Tom Tromey --- diff --git a/gdb/testsuite/gdb.dwarf2/symbol_needs_eval_fail.exp b/gdb/testsuite/gdb.dwarf2/symbol_needs_eval_fail.exp index cb9826e90df..ec3913cb75f 100644 --- a/gdb/testsuite/gdb.dwarf2/symbol_needs_eval_fail.exp +++ b/gdb/testsuite/gdb.dwarf2/symbol_needs_eval_fail.exp @@ -47,11 +47,17 @@ if { [is_aarch64_target] } { standard_testfile symbol_needs_eval.c ${gdb_test_file_name}-dw.S +if [prepare_for_testing "failed to prepare" $testfile $srcfile {debug}] { + return +} + +set int_size [get_sizeof "int" -1] + # Make some DWARF for the test. set asm_file [standard_output_file $srcfile2] Dwarf::assemble $asm_file { - global dwarf_regnum regname + global dwarf_regnum regname int_size set exec_mask_var [gdb_target_symbol exec_mask] @@ -66,7 +72,7 @@ Dwarf::assemble $asm_file { int_type_label: DW_TAG_base_type { {DW_AT_name "int"} {DW_AT_encoding @DW_ATE_signed} - {DW_AT_byte_size 4 DW_FORM_sdata} + {DW_AT_byte_size $int_size DW_FORM_sdata} } # define artificial variable a @@ -75,7 +81,7 @@ Dwarf::assemble $asm_file { {DW_AT_type :$int_type_label} {DW_AT_location { DW_OP_addr $exec_mask_var - DW_OP_deref + DW_OP_deref_size $int_size # conditional jump to DW_OP_bregx DW_OP_bra 4 diff --git a/gdb/testsuite/gdb.dwarf2/symbol_needs_eval_timeout.exp b/gdb/testsuite/gdb.dwarf2/symbol_needs_eval_timeout.exp index 0e6ee6112ea..fc0b59175b1 100644 --- a/gdb/testsuite/gdb.dwarf2/symbol_needs_eval_timeout.exp +++ b/gdb/testsuite/gdb.dwarf2/symbol_needs_eval_timeout.exp @@ -47,11 +47,17 @@ if { [is_aarch64_target] } { standard_testfile symbol_needs_eval.c ${gdb_test_file_name}-dw.S +if [prepare_for_testing "failed to prepare" $testfile $srcfile {debug}] { + return +} + +set int_size [get_sizeof "int" -1] + # Make some DWARF for the test. set asm_file [standard_output_file $srcfile2] Dwarf::assemble $asm_file { - global dwarf_regnum regname + global dwarf_regnum regname int_size set exec_mask_var [gdb_target_symbol exec_mask] @@ -66,7 +72,7 @@ Dwarf::assemble $asm_file { int_type_label: DW_TAG_base_type { {DW_AT_name "int"} {DW_AT_encoding @DW_ATE_signed} - {DW_AT_byte_size 4 DW_FORM_sdata} + {DW_AT_byte_size $int_size DW_FORM_sdata} } # add info for variable exec_mask @@ -93,7 +99,7 @@ Dwarf::assemble $asm_file { {DW_AT_location { DW_OP_lit1 DW_OP_addr $exec_mask_var - DW_OP_deref + DW_OP_deref_size $int_size # jump to DW_OP_fbreg DW_OP_skip 4