Re: [RFC] gdb/riscv: Add target description support

[Andrew Burgess <andrew dot burgess at embecosm dot com>]

* Jim Wilson <jimw@sifive.com> [2018-11-08 13:57:10 -0800]:

> On Thu, Nov 8, 2018 at 8:07 AM Andrew Burgess
> <andrew.burgess@embecosm.com> wrote:
> >  riscv*-*-linux*)
> >         # Target: Linux/RISC-V
> > -       gdb_target_obs="riscv-linux-tdep.o riscv-tdep.o glibc-tdep.o \
> > +       gdb_target_obs="riscv-linux-tdep.oglibc-tdep.o \
> >                         linux-tdep.o solib-svr4.o symfile-mem.o linux-record.o"
> 
> There is a typo here, a space is missing between riscv-linux-tdep.o
> and glibc-tdep.o.

Good spot.  I'll fix this.

> 
> Otherwise this looks great.  I tested it on riscv64-linux and saw no
> new problems.  I also tested with riscv-tests/debug against openocd,
> and almost everything that failed before is now passing.  There are
> two tests that fail because riscv-tests/debug doesn't know how to
> parse the new {float...double...} string output for FP registers.
> There are two tests that fail for a 32-bit target with
>     (gdb) p/x fox = "This little piggy went to the market."
>     p/x fox = "This little piggy went to the market."^M
>     /scratch/jimw/openocd/fsf-gdb/riscv-gnu-toolchain/build/../riscv-gdb/gdb/regcac\
>     he.c:298: internal-error: void
> regcache::restore(readonly_detached_regcache*): \
>     Assertion `src != NULL' failed.^M

I have a fix for this issue, but it's not really related to target
register descriptions.  I hope to get this posted soon.

> And the same tests fail for a 64-bit target with a timeout.  That may
> only be one problem.  This should not prevent the XML register set
> support from going in now, and we can worry about debugging these
> problems later.  Even with the failures this is a major improvement
> over what we had before.

Great.  I just need to review John's feedback and then we can move
this forward.

Thanks again,

Andrew