> Hi Joel.  Sorry for how long it took to get to this...

No problem - it must have taken me just as long to answer...

I followed all your suggestions. I'm still a little unsure about:

> Make sure you don't need to handle EINTR here.  You may want to take
> a look at linux-low.c:my_waitpid and see whether you could do something
> of the sorts, to be able to get rid of the usleep.  

I was going to say that I haven't faced a situation where I received
an EINTR, but I looked at the man page, and indeed, they mention it
explicitly.

I wonder if I could just copy/paste the implementation for Linux
with support for clone removed.... I will do that next, but I suggest
that we treat this as a separate patch? That way, we can get the bulk
out of the way, and focus on enhancing that part of the patch only.

> > +  if (status->kind == TARGET_WAITKIND_STOPPED
> > +      && status->value.integer == SIGTRAP)
> 
> `enum target_signal' vs host signal mixup continued.

Can you spot me on this? I made the changes that I think needed
to be made, but can you verify that I made the right ones?

I also made a change to the patch that imports vasprintf: I added
a -DHAVE_CONFIG_H to the compile command, to avoid 'function implicitly
defined' gcc warnings. The file looks like this:

    #ifdef HAVE_CONFIG_H
    #include "config.h"
    #endif
    [...]
    #ifdef HAVE_STRING_H
    #include <string.h>
    #endif

I added the define to build this file with gdbserver's config.h,
so that system headers whose inclusion is conditionalized get
included when appropriate.

The modified hunk is:

    +# We build vasprintf with -DHAVE_CONFIG_H because we want that unit to
    +# include our config.h file.  Otherwise, some system headers do not get
    +# included, and the compiler emits a warning about implicitly defined
    +# functions (missing declaration).
    +vasprintf.o: $(srcdir)/../../libiberty/vasprintf.c
    + $(CC) -o vasprintf.o -c $(CPPFLAGS) $(INTERNAL_CFLAGS) -DHAVE_CONFIG_H $<

Tested on lynxos 4.0. Does this look OK?

-- 
Joel