This is the mail archive of the cygwin 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: MS or cygwin dll debug tools/ was "sys/sockio.h" etc.

Mike Marchywka wrote:
> > .rdata relocs it is then.
> This is why I lurk as I now remember something about this and, sure, I can find it on google.
> Of course, it would still be nice to have a fully instrumented load ( so where do I find the
> unreolcatable data?) .
> I relinked with -strip and used dumpbin to find all the symbols in rdata but there are several hundred.
> I thought I could grep the code for suspicious consts but no luck so far.

The simplest method is just to tweak the linker script to make .rdata
writable (or rather, to map both .data and .rdata input segments into
the same .data output section.)

There is no method that I'm aware of to find the name of the offending
symbol.  Actually, that's not quite true, see
<>.  As I said
in that message though, those might just be coindicental and certainly
not guaranteed across OS versions or whatever.

> Isn't there some way to find the offending relocation attempt? What does strace know
> about the stituation or does it just echo stuff from Windoze?

strace under Cygwin is not like strace on other operating systems, it is
really just a glorified viewer for a bunch of debug_printf statements
that exist in the Cygwin code.  The error occurs before the process even
begins execution, during the period where NTLDR loads libraries into its
address space.  So, strace will be of no use.


Unsubscribe info:
Problem reports:

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