As pre-approved by Alan in
https://sourceware.org/pipermail/binutils/2021-September/118019.html
and I believe people have run into getting testsuite failures for
test-environments with "long" directory names, at least once more
since that time. Enough. I grepped the gas, binutils and ld
testsuites for "CU:" to catch target-specific occurrences, but I
noticed none. I chose to remove "CU:" on the objdump tests instead of
changing options to get the wide format, so as to keep the name of the
test consistent with actual options; but added it to the readelf
options for the gas test as I believe the "CU:" format is preferable.
Tested for cris-elf and native x86_64-pc-linux-gnu.
binutils:
* dwarf.c (display_debug_lines_decoded): Don't check the
string length of the directory, instead emit the "CU: dir/name"
format only if wide output is requested.
* testsuite/binutils-all/dw5.W, testsuite/binutils-all/objdump.WL:
Adjust accordingly.
gas:
* testsuite/gas/elf/dwarf-5-loc0.d: Add -W to readelf options.
else
directory = (char *) directory_table[ix - 1];
- if (do_wide || strlen (directory) < 76)
+ if (do_wide)
printf (_("CU: %s/%s:\n"), directory, file_table[0].name);
else
printf ("%s:\n", file_table[0].name);
Contents of the .debug_line section:
-CU: ./main.c:
+main.c:
File name Line number Starting address View Stmt
main.c 6 0x1234 x
main.c 6 0x12346 x
Contents of the \.debug_line section:
-CU: \./dw2-decodedline\.c:
+dw2-decodedline\.c:
File name Line number Starting address View Stmt
directory/file1\.c:
#as: -gdwarf-5
-#readelf: -wL
+#readelf: -W -wL
#name: DWARF5 .loc 0
# The am33 cr16 crx ft32 mn10 msp430 nds32 and rl78 targets do not evaluate the subtraction of symbols at assembly time.
# The riscv targets do not support the subtraction of symbols.