Re: Cygwin 3.4.3 and 3.5.0... hangs in make, top, procps, ls /proc/PID/...

[Takashi Yano <takashi.yano@nifty.ne.jp>]

On Tue, 17 Jan 2023 21:46:20 +0100
Corinna Vinschen wrote:

> Hi Takashi,
> 
> On Jan 16 22:42, Corinna Vinschen via Cygwin wrote:
> > On Jan 16 16:18, Corinna Vinschen via Cygwin wrote:
> > > Actually, I' running your testcase on two machines in parallel now for
> > > quite some time, which only one hunk of 60675f1a7eb2 reverted, i.e.
> > > 
> > > diff --git a/winsup/cygwin/mm/shared.cc b/winsup/cygwin/mm/shared.cc
> > > index 893b20d289b4..6d6d2940b6d4 100644
> > > --- a/winsup/cygwin/mm/shared.cc
> > > +++ b/winsup/cygwin/mm/shared.cc
> > > @@ -140,7 +140,11 @@ open_shared (const WCHAR *name, int n, HANDLE& shared_h, DWORD size,
> > >    if (*m == SH_JUSTCREATE || *m == SH_JUSTOPEN)
> > >      addr = NULL;
> > >    else
> > > -    addr = (void *) region_address[*m];
> > > +    {
> > > +      addr = (void *) region_address[*m];
> > > +      VirtualFree (addr, 0, MEM_RELEASE);
> > > +    }
> > > +
> > >  
> > >    WCHAR map_buf[MAX_PATH];
> > >    WCHAR *mapname = NULL;
> > > 
> > > So far (and knocking on wood madly while writing this) the mapping
> > > problem didn't show up once.  Maybe you'd like to try the same?
> > 
> > Never mind, I encountered another error.  After a bit more debugging
> > I think I understand the problem now, and  I'm just in the process of
> > reworking open_shared.  This may take a day or two.  Stay tuned.
> 
> I pushed some patches to fix this issue.  Excessive debugging indicated
> that the reason cygcheck fails in this way is:
> 
> - It's a non-Cygwin process which
> 
> - is built with high-entropy ASLR and
> 
> - tries to load the Cygwin DLL dynamically and
> 
> - therefore suffers from the fact that recent Cygwin code doesn't
>   expect that certain memory regions are used by Windows itself.
>   Which they are, due to the high-entropy stuff.
> 
> The patches are supposed to make the code less rigid in terms of the
> addresses of certain memory regions, as well as dropping the
> high-entropy VA flag from builds of strace and cygcheck, both of which
> are loading the Cygwin DLL dynamically as part of their job.
> 
> The test release 3.5.0-0.116.g8d318bf142f7 contains the patches, for
> everybody to try.

Thank you very much for working on this problem. It seems that
it was unexpectedly large-scale modification.

I confirmed that the problem has been fixed with these patches.
The test case has been running for 11 hours but the problem does
not happen so far.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>