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: Heads up: *possible* bug in cygwin

Christopher Faylor wrote:
On Mon, Dec 30, 2002 at 11:18:55PM -0500, Charles Wilson wrote:

If somebody with a debuggable cygwin kernel could look into this, I'd appreciate it. I'll try to follow up on my own, but it takes FOREVER to do a 'cvs update' on the cygwin source tree over a 28.8k modem...
Sigh. Finally built a debuggable cygwin from current CVS. Here's the stacktrace from the SIGSEGV.

Problems "in the bowels" of malloc are invariably caused by memory
corruption, like double freeing of a pointer, overrunning of a buffer,
ignoring of OOM conditions, etc.  Given that the malloc in cygwin (to
say nothing of Doug Lea's malloc) goes through a fairly heavy workout
every day, I'd suspect the application before I'd suspect cygwin.
I would too -- but I can't see where any of the arguments to the functions, or any of the operations within the glib subroutines along the way, do any of that. It all =seems= okay, but I'll probably have to pull out pencil and paper and keep track of all the pointers by hand. Or can I just turn on -DDEBUG when compiling

Plus, glib-2.0.7 is at least as well tested as cygwin (and might be close to as well tested as dlmalloc). It's hard to imagine that a buffer overrun or double-free was overlooked in glib's own testsuite, given that folks on non-cygwin platforms can use stuff like purify and electricfence when they test.

And finally, that doesn't explain why the *same code*, *unchanged*, worked on May 1, 2002, but doesn't now. I wish I had the cygwin dll and importlib from then, so I could eliminate that variable...

Hmmm...glib does a lot of testing against NULL to determine whether a pointer has been initialized -- but declares these pointers as
gpointer foo;
gpointer foo = NULL;
Did gcc (pre 3.2) automatically initialize data to 0, while gcc-3.2 does not? Hmmm...waitaminute, I do have gcc2 installed...


Unsubscribe info:
Bug reporting:

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