https://sourceware.org/bugzilla/show_bug.cgi?id=27246
--- Comment #11 from Khem Raj <raj.khem at gmail dot com> ---
(In reply to Khem Raj from comment #10)
> further reduced test is
>
> #define overalloc_at(f, l) \
> do { \
> fatal_exit("ERROR: attempting to allocate too large block at %s:%d",
> f, l);\
> } while (0)
>
> #define overalloc() overalloc_at(__FILE__, __LINE__)
>
> void foo()
> {
> overalloc();
> }
>
> % gcc-11.0.1 /tmp/a.c -c -gdwarf-5
> % /tmp/obj/gold/ld-new a.o
> /tmp/obj/gold/ld-new: internal error in format_file_lineno, at
> /home/kraj/work/binutils-gdb/gold/dwarf_reader.cc:2278
>
> ICE not seen when -gdwarf-4 is used
I compared the assembly output and difference ( dwarf4 vs dwarf5 ) that matters
is
that when dwarf5 is enabled then it emits this .file directive into .debug_line
section which gold does not like
@@ -289,38 +290,40 @@
.quad 0
.section .debug_line,"",@progbits
.Ldebug_line0:
+ .file 0
"/mnt/b/yoe/master/build/tmp/work/core2-64-yoe-linux/links/2.21-r0/links-2.21"
"/tmp/a.c"
.section .debug_str,"MS",@progbits,1
--
You are receiving this mail because:
You are on the CC list for the bug.
