[PATCH] server: Fix setting context flags in get_thread_context.

[Jacek Caban]

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=49011
Signed-off-by: Jacek Caban <jacek(a)codeweavers.com>
---
  dlls/kernel32/tests/thread.c | 15 +++++++++++++++
  server/thread.c              |  2 +-
  2 files changed, 16 insertions(+), 1 deletion(-)

[Alexandre Julliard]

Jacek Caban <jacek(a)codeweavers.com> writes:
> Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=49011
> Signed-off-by: Jacek Caban <jacek(a)codeweavers.com>
> ---
>  dlls/kernel32/tests/thread.c | 15 +++++++++++++++
>  server/thread.c              |  2 +-
>  2 files changed, 16 insertions(+), 1 deletion(-)

This fixes the server crash, but the tests are still failing here:

../../../tools/runtest -q -P wine -T ../../.. -M ntdll.dll -p ntdll_test.exe exception && touch exception.ok
exception.c:951: Test failed: NtGetContextThread failed with c0000022
exception.c:952: Test failed: dr0 340000
exception.c:953: Test failed: dr7 3
exception.c:2745: Test failed: NtGetContextThread failed with c0000022
exception.c:2746: Test failed: expected 0, got ffffffff
exception.c:2747: Test failed: expected 0, got ffffffff
exception.c:2748: Test failed: expected 0, got ffffffff
exception.c:2749: Test failed: expected 0, got ffffffff
exception.c:2750: Test failed: expected 0, got ffffffff
exception.c:2751: Test failed: expected 0, got 400
make: *** [Makefile:331: exception.ok] Error 10

-- 
Alexandre Julliard
julliard(a)winehq.org

[Jacek Caban]

On 27.04.2020 17:38, Alexandre Julliard wrote:
> Jacek Caban <jacek(a)codeweavers.com> writes:
> > Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=49011
> > Signed-off-by: Jacek Caban <jacek(a)codeweavers.com>
> > ---
> >   dlls/kernel32/tests/thread.c | 15 +++++++++++++++
> >   server/thread.c              |  2 +-
> >   2 files changed, 16 insertions(+), 1 deletion(-)
> > This fixes the server crash, but the tests are still failing here:

The attached patch should fix it. It does the right thing, but doing the 
same for setting system registers is more tricky. I had an 
implementation that tracked more closely valid registers in context 
object, which allowed moving more parts to select request. This would 
solve the problem. I ended up removing that part late in clean up 
process, because I wrongly considered that an uninteresting 
optimization. I will revisit it.


Thanks,

Jacek