Re: GDB 10.1 release -- 2020-09-26 Update

[Simon Marchi <simark@simark.ca>]

On 2020-09-29 8:14 a.m., Strasuns, Mihails wrote:
> Hello,
>
> FYI: I am currently investigating a regression caused by 7188ed02d2a7e3fce00a0214e70457c5ef56df6b (Replace dwarf2_per_cu_data::cu backlink with per-objfile map).
> With my program it causes GDB to crash with the following backtrace:
>
> #0  __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
> #1  0x00007fe4e871c8b1 in __GI_abort () at abort.c:79
> #2  0x000055e6b5a17333 in handle_sigsegv (sig=11) at /users/mstrasun/sources/gdb/gdb/event-top.c:885
> #3  <signal handler called>
> #4  0x000055e6b5f08f9e in htab_find_with_hash (htab=0x0, element=0x7ffcd1f58440, hash=3394)
>     at /users/mstrasun/sources/gdb/libiberty/hashtab.c:591
> #5  0x000055e6b59d0b9b in follow_die_offset (sect_off=(unknown: 3394), offset_in_dwz=0,
>     ref_cu=0x7ffcd1f58528) at /users/mstrasun/sources/gdb/gdb/dwarf2/read.c:22273
> #6  0x000055e6b59d0c2e in follow_die_ref (src_die=0x0, attr=0x7ffcd1f58540, ref_cu=0x7ffcd1f58528)
>     at /users/mstrasun/sources/gdb/gdb/dwarf2/read.c:22288
> #7  0x000055e6b59ac51a in partial_die_full_name (pdi=0x55e6b87f02a0, cu=0x55e6b7811510)
>     at /users/mstrasun/sources/gdb/gdb/dwarf2/read.c:8296
> #8  0x000055e6b59ac65c in add_partial_symbol (pdi=0x55e6b87f02a0, cu=0x55e6b7811510)
>     at /users/mstrasun/sources/gdb/gdb/dwarf2/read.c:8323
> #9  0x000055e6b59ad040 in add_partial_subprogram (pdi=0x55e6b87f02a0, lowpc=0x7ffcd1f58760,
>     highpc=0x7ffcd1f58768, set_addrmap=0, cu=0x55e6b7811510)
>     at /users/mstrasun/sources/gdb/gdb/dwarf2/read.c:8591
> <...>
>
> After this patch die_hash ends up being null in this context.

If you can send a reproducer, I can take a look.

Simon