From: Tom de Vries Date: Mon, 23 Aug 2021 10:08:25 +0000 (+0200) Subject: [gdb/testsuite] Add dummy start and end CUs in dwarf assembly X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=5ef670d81fd;p=binutils-gdb.git [gdb/testsuite] Add dummy start and end CUs in dwarf assembly Say one compiles a hello.c: ... $ gcc -g hello.c ... On openSUSE Leap 15.2 and Tumbleweed, the CU for hello.c is typically not the first in .debug_info, nor the last, due to presence of debug information in objects for sources like: - ../sysdeps/x86_64/start.S - init.c - ../sysdeps/x86_64/crti.S - elf-init.c - ../sysdeps/x86_64/crtn.S. On other systems, say ubuntu 18.04.5, the CU for hello.c is typically the first and the last in .debug_info. This difference has caused me to find some errors in the dwarf assembly using openSUSE, that didn't show up on other platforms. Force the same situation on other platforms by adding a dummy start and end CU. Tested on x86_64-linux. gdb/testsuite/ChangeLog: 2021-08-22 Tom de Vries PR testsuite/28235 * lib/dwarf.exp (Dwarf::dummy_cu): New proc. (Dwarf::assemble): Add dummy start and end CU. --- diff --git a/gdb/testsuite/lib/dwarf.exp b/gdb/testsuite/lib/dwarf.exp index 52886d0701b..ce7b5983ab9 100644 --- a/gdb/testsuite/lib/dwarf.exp +++ b/gdb/testsuite/lib/dwarf.exp @@ -2252,6 +2252,13 @@ namespace eval Dwarf { } } + # Emit a dummy CU. + proc dummy_cu {} { + # Generate a CU with default options and empty body. + cu {} { + } + } + # The top-level interface to the DWARF assembler. # FILENAME is the name of the file where the generated assembly # code is written. @@ -2291,7 +2298,7 @@ namespace eval Dwarf { } set _output_file [open $filename w] - set _cu_count 0 + set _cu_count -1 _empty_array _deferred_output set _defer "" set _label_num 0 @@ -2304,11 +2311,19 @@ namespace eval Dwarf { set _debug_addr_index 0 + # Dummy CU at the start to ensure that the first CU in $body is not + # the first in .debug_info. + dummy_cu + # Not "uplevel" here, because we want to evaluate in this # namespace. This is somewhat bad because it means we can't # readily refer to outer variables. eval $body + # Dummy CU at the end to ensure that the last CU in $body is not + # the last in .debug_info. + dummy_cu + _write_deferred_output catch {close $_output_file}