https://sourceware.org/bugzilla/show_bug.cgi?id=25561
Bug ID: 25561
Summary: Incorrect DWARF info emitted when variable removed due
to --gc-sections
Product: binutils
Version: 2.33
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: ld
Assignee: unassigned at sourceware dot org
Reporter: chrisc112358 at gmail dot com
Target Milestone: ---
Overview:
The "DW_AT_location" of a variable which is discarded as a result of using the
CFLAG -fdata-sections and the LDFLAG --gc-sections will be incorrectly set to
0.
This will lead to incorrect results when trying to display the information for
the variable from a debugger.
Steps to Reproduce:
============================================================
// main.c
int *const g_const_variable = ((int *)0x40010000UL);
void main(void) {
*g_const_variable = 10;
}
============================================================
$ mkdir -p build
$ arm-none-eabi-gcc -g -fdata-sections -nostdlib
-Wl,--gc-sections,-Map,build/example.map -Wl,-emain -Wl,-Ttext=0x0 main.c -o
build/example.elf
$ arm-none-eabi-readelf --debug-dump build/example.elf > build/dwarf_dump.txt
============================================================
Inspecting build/dwarf_dump.txt I see an incorrect "DW_AT_location" of 0 for
g_const_variable:
<1><25>: Abbrev Number: 2 (DW_TAG_variable)
<26> DW_AT_name : (indirect string, offset: 0x0): g_const_variable
<2a> DW_AT_decl_file : 1
<2b> DW_AT_decl_line : 1
<2c> DW_AT_decl_column : 12
<2d> DW_AT_type : <0x3d>
<31> DW_AT_external : 1
<31> DW_AT_location : 5 byte block: 3 0 0 0 0 (DW_OP_addr: 0)
============================================================
In gdb you will see incorrect data displayed when viewing the variable
(gdb) p g_const_variable
$2 = (volatile int * const) 0xe52db004
(gdb) p &g_const_variable
$3 = (volatile int * const *) 0x0 <main>
============================================================
Expected Results:
I would expect the g_const_variable debug info to either be stripped from the
DWARF section or the DW_AT_location to be encoded such that the pointer address
can be recovered by a debugger.
Additional Notes:
I was using the latest release of the ARM gcc toolchain for testing
(https://developer.arm.com/tools-and-software/open-source-software/developer-tools/gnu-toolchain/gnu-rm/downloads)
but I tried some other gcc/ld versions as well and was seeing the same result
--
You are receiving this mail because:
You are on the CC list for the bug.
