This is the mail archive of the gdb-patches@sources.redhat.com mailing list for the GDB project.


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

Re: S390 gdb patches




Hi Andrew,

As you can appreciate I've spent a lot of time getting this code into a fit
state to be accepted in the repository
( probably 3 to 4 weeks ).

What is your phone number so we can negociate here.


What is the logic behind not using TRUE or FALSE definitions ?

>s390-tdep.c includes <netinet/in.h>
>
>    I can't see any reason for this (well except
>    to get at the htonl() macros and you shouldn't
>    be using them).

This probably can be worked around, have you any better ideas ?


>tm-s390.h
>
>    If I'm reading the code right, much of this
>    file is conditional on WANT_S390_TGT_DEFS.
>    That code should be deleted.

We already did talk about this I said I wanted to keep the
old definitions e.g. FRAME_CHAIN etc.
As I believe if the multiarch was mature it could generate this
code from the macros & this is the only stuff that is currently
documented.

To be honest it will not fit for s390, I've only implemented
this stuff to get it accepted e.g. when running 31 bit programs
on a 64 bit machine we will produce 31 bit core dumps.
I also presume sparc has this problem whatever the claims are
unless they came up with a generic core dump format ( unlikely ).
This would also require multiple regset definitions.
You mentioned on a web page the idea behind multi arch
was to run on multiple different processors within the same
box & gave playstation 2 as an example, it is  short of this goal
& I cannot see why we have this interface rammed down our throat if
it simply isn't mature.





> I'm not clear on the copyright status of the files:
>         s390-gdbregs.
>         s390-regs-common.h
>         sigcontext.h
>    which were lifted from the Linux kernel.
>
>    Does IBM currently control the copyright on those
>    files?
I'm no legal genius,
Well they are under GPL so I think Richard Stallman
& Linus can fight about ownership & may the best loyar win.
The files are required to.
1) To keep the kernels register formats compatible with gdb.
2) Remote debugging.
3) Backchaining out of signal handlers.



>    __u8 et.al. are used liberally.  Within GDB
>    I believe that ``bfd_byte'' is used.

I like these definitions what happens if I want
a signed or unsigned bfd_byte. I'll change this ..... grudgingly.


>gdbarch.sh:
>    I'll figure out what macros you needed and multi
>    arch them my self (avoid assignment :-)

I cannot see any point in rewriting my gdbarch.sh macros unless ,
1) You find something wrong with them or they require improvment, fair
enough.
2)  Don't rewrite them just to claim "ownership" of them.


D.J. Barrow Linux for S/390 kernel developer
eMail: djbarrow@de.ibm.com,barrow_dj@yahoo.com
Phone: +49-(0)7031-16-2583
IBM Germany Lab, Schönaicherstr. 220, 71032 Böblingen



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