[Bug 44492] New: winedbg reports unknown mode, fails to backtrace etc. in protected-mode

https://bugs.winehq.org/show_bug.cgi?id=44492

Bug ID: 44492
           Summary: winedbg reports unknown mode, fails to backtrace etc.
                    in protected-mode
           Product: Wine
           Version: unspecified
          Hardware: x86
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: -unknown
          Assignee: [e-mail]
          Reporter: [e-mail]
      Distribution: ---

While trying to debug bug #42554 I found that winedbg persistently refused to
give backtraces, both in the crash dialog and when started via `wine winedbg
MSBSOLAR.EXE`. The following output is given:

------
Unhandled exception: page fault on read access to 0x00004540 bad address.
In (null) mode.
Register dump:
 CS:12af SS:11d7 DS:12c7 ES:1237 FS:0063 GS:006bBad segment (4567)
Stack dump:
000c: sel=0067 base=00000000 limit=00000000 32-bit rw-
Backtrace:
=>0 Unknown mode ffffffff (Unknown mode ffffffff
)
Unknown mode ffffffff: -- no code accessible --
------

Tracing this back to its source, I find that ptrace seems to be returning
incorrect output:

0039: get_selector_entry( handle=002c, entry=597 )
0030: *signal* signal=19
0039: get_selector_entry() = 0 { base=00000000, limit=00000000, flags=52 }

Obviously the base should not be NULL, nor the limit 0, and the flags should be
9b (WINE_LDT_FLAGS_ALLOCATED|WINE_LDT_FLAGS_CODE). If I add traces to
libs/wine/ldt.c, nothing is modifying the entry past the point at which it is
set, so it seems like the fault lies in ptrace... an audacious claim to be
sure.

I'm almost sure this used to work correctly, but I can't figure out when. I
checked wine-2.10 and 3.0, and it was still broken then. I also checked on Arch
(well, Manjaro) just in case this was another Ubuntu-related ptrace breakage,
but it shows the same behaviour there.