Re: [PR 31727] Recognize -2 as a tombstone value in .debug_line

[Tom Tromey <tom@tromey.com>]

>>>>> "Dmitry" == Dmitry Neverov <dmitry.neverov@jetbrains.com> writes:

Dmitry> Commit a8caed5d7faa639a1e6769eba551d15d8ddd9510 handled the tombstone
Dmitry> value -1 used by lld (https://reviews.llvm.org/D81784).  The
Dmitry> referenced lld commit also uses the tombstone value -2 for
Dmitry> pre-DWARF-v5
Dmitry> (https://github.com/llvm/llvm-project/commit/e618ccbf431f6730edb6d1467a127c3a52fd57f7).

Dmitry> If not handled, -2 breaks the pc step range calculation and triggers
Dmitry> the assertion:

Dmitry>   gdb/infrun.c:2794: internal-error: resume_1: Assertion
Dmitry>   `pc_in_thread_step_range (pc, tp)' failed.

Dmitry> This commit adds -2 tombstone value and handles it in the same way as -1.

Dmitry> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=31727

I wonder if it's possible to write a test case for this using the DWARF
assembler.

If that looks too hard, though, I wouldn't bother, considering that the
-2 approach is gone from lld now.  In that case the patch is OK:
Approved-By: Tom Tromey <tom@tromey.com>

Tom