This is the mail archive of the mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re[2]: cygwin1.dll [1.3.11(CVS)]: Changes since 06/19 snapshot cause page fault in kernel32.dll

Hello Christopher,

Thursday, June 20, 2002, 7:16:28 PM, you wrote:

CF> On Thu, Jun 20, 2002 at 05:57:58PM +0200, Pavel Tsekov wrote:
>>With CYGWIN_SLEEP I was able to attach before the crash occures. Here
>>is a backtrace from gdb:
>>Program received signal SIGSEGV, Segmentation fault.
>>0x61093419 in MTinterface::Init (this=0x610f8000, forked=0)
>>    at ../../../../cygwin-snapshot/winsup/cygwin/
>>296       mainthread.win32_obj_id = myself->hProcess;
>>The problem is that operator -> (myself->) returns NULL.

CF> Hmm.  myself is never supposed to be NULL.  Or at least it shouldn't be
CF> NULL once the constructors are run.  This function is called after the
CF> constructors are run, so...

It's not 'myself' that is NULL - it is the return value of the operator -> applied
on myself i.e.

This method of the pinfo class returned NULL becase procinfo member
was NULL.
_pinfo *operator -> () const {return procinfo;}

CF> Out of curiousity, does the latest cygwin snapshot die the same way?

Yes! It works fine - I mean the precompiled 20020619 dll. I'm just downloading the
snapshot source now and will compile it and see if it works. But this
is strange - I update my source tree with the diff from the snapshot
page when a new snapshot is released - it has worked always till now :(

If this matters:

When building, I always completly remove the contents of my build
directory. Then I configure with:
CFLAGS=-g CXXFLAGS=-g ../cygwin-snapshot/configure --enable-debugging

Unsubscribe info:
Bug reporting:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]