This is the mail archive of the
mailing list for the Cygwin project.
Re: About cygwin1.dll
- From: Brian Dessent <brian at dessent dot net>
- To: ÊÀÑå Íõ <wsycqyz at hotmail dot com>, cygwin at cygwin dot com
- Date: Sat, 18 Mar 2006 23:30:53 -0800
- Subject: Re: About cygwin1.dll
- References: <BAY113-F9CCC4DB92E517A0036CFDA3DA0@phx.gbl>
- Reply-to: cygwin at cygwin dot com
Please post to the mailing list instead of sending email to me
directly. By doing this you help everyone because there may be others
that will benefit, and the thread will be archived.
"ÊÀÑå Íõ" wrote:
> I've read your article"[patch] fix spurious SIGSEGV faults under Cygwin",
> it's so helpful for me to understand the problem "Program received signal
> SIGSEGV, Segmentation fault.0x610af8b8 in pthread_key_create..."
> But I am new for linux, and I don't how to use your attached patch in the
You shouldn't use my patch. It's obsolete, and the problem is already
fixed in a better way by changes made by Christopher Faylor. These
fixes are already checked in to CVS so just build from current sources
and you will get the fix.
> Could you please send me the problem-solved version of cygwin1.dll
> directly, or send me some link about how to use your patch?
The problem is not in cygwin1.dll, it is in a lack of communication
between the DLL and gdb. You need to build both from CVS in order to
get the fix. For the cygwin side you can just use a daily snapshot.
For gdb you will have to build it from CVS or wait for an updated
Please remember that these SIGSEGVs you are seeing are not real
segmentation violations, they are just a consequence of how error
checking is done in the DLL. They are expected to happen, and when they
occur a fault handler in the cygwin DLL is in place to catch them and
handle the situation cleanly. Consequently you can always just type "c"
or "continue" at the gdb prompt and everything will resume as usual.
That is all that the above mentioned changes do -- make gdb ignore the
fault and continue.
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html