Possibly dumb signal mapping question

Possibly dumb signal mapping question

[Kris Warkentin]

Say I'm building a gdb on Linux that remotely targets Neutrino, what is 
the acceptable way to map our OS signals onto TARGET_SIGNAL_*?  I'm 
looking at the target_signal_[to/from]_host functions but they look like 
they depend on the various SIG*s having been defined.  Wouldn't those 
defines collide with host defines?  Perhaps I'm getting confused for no 
good reason....  If I _were_ going to define them, what would be a good 
place?

I accidentally dropped the signal mapping code from our old port but in 
looking at it, it looks too hacky to live.  We've got some ugly 
target_signal_[to/from]_qnx functions that we just use to convert any 
signals sent to and from the remote target.  Is there a better way?

cheers,

Kris

Re: Possibly dumb signal mapping question

[Daniel Jacobowitz]

On Thu, Mar 17, 2005 at 04:14:16PM -0500, Kris Warkentin wrote:
> Say I'm building a gdb on Linux that remotely targets Neutrino, what is 
> the acceptable way to map our OS signals onto TARGET_SIGNAL_*?  I'm 
> looking at the target_signal_[to/from]_host functions but they look like 
> they depend on the various SIG*s having been defined.  Wouldn't those 
> defines collide with host defines?  Perhaps I'm getting confused for no 
> good reason....  If I _were_ going to define them, what would be a good 
> place?
> 
> I accidentally dropped the signal mapping code from our old port but in 
> looking at it, it looks too hacky to live.  We've got some ugly 
> target_signal_[to/from]_qnx functions that we just use to convert any 
> signals sent to and from the remote target.  Is there a better way?

Your stub should be doing this.  The TARGET_SIGNAL_* constants are the
on-the-wire values.

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: Possibly dumb signal mapping question

[Kris Warkentin]

Daniel Jacobowitz wrote:

>Your stub should be doing this.  The TARGET_SIGNAL_* constants are the
>on-the-wire values.
>

Thanks Daniel.  I guess I've got to do it the hard way.  I had just 
asked because I thought there might be some magic mapping routine like 
for the i386 registers.

cheers,

Kris

Re: Possibly dumb signal mapping question

[Daniel Jacobowitz]

On Fri, Mar 18, 2005 at 09:47:11AM -0500, Kris Warkentin wrote:
> Daniel Jacobowitz wrote:
> 
> >Your stub should be doing this.  The TARGET_SIGNAL_* constants are the
> >on-the-wire values.
> >
> 
> Thanks Daniel.  I guess I've got to do it the hard way.  I had just 
> asked because I thought there might be some magic mapping routine like 
> for the i386 registers.

No, there isn't - you could do it in your remote protocol target (don't you
use a different protocol?)

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: Possibly dumb signal mapping question

[Kris Warkentin]

Daniel Jacobowitz wrote:

>No, there isn't - you could do it in your remote protocol target (don't
>you
>use a different protocol?)
>

Yeah.  I just created some mapping functions to apply whenever the 
signal is sent or received on the wire.  It's in the remote part of our 
port that I haven't submitted yet.  I'm waiting for some time where I 
won't keep getting pre-empted.  It's annoying when I submit something to 
you guys and then can't devote my full attention to following up.

cheers,

Kris