This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Fix hardware watchpoints on PowerPC servers
- From: "Ulrich Weigand" <uweigand at de dot ibm dot com>
- To: emachado at linux dot vnet dot ibm dot com (Edjunior Barbosa Machado)
- Cc: gdb-patches at sourceware dot org
- Date: Mon, 13 May 2013 15:19:20 +0200 (CEST)
- Subject: Re: [PATCH] Fix hardware watchpoints on PowerPC servers
Edjunior Machado wrote:
> initially developed only for bookE (embedded processors), the new ptrace
> interface is now available for Power server processors too [1].
>
> This patch fixes the insertion of hardware watchpoints on Power servers
> using this new interface available on kernel 3.7 and newer (currently, the
> region should be aligned to 8 bytes, which is also the max length). It reduces
> the number of unexpected failures in gdb testsuite in 40 failures (ppc64
> running kernel 3.8).
So the kernel now reports it supports PPC_DEBUG_FEATURE_DATA_BP_RANGE,
but then rejects any attempt to set a range BP of more than 8 bytes?
And there's no way to check for this ahead of time except for checking
the BookE hwcaps bits? That seems unfortunate ...
But I guess if that's the way the kernel interface is now, we'll have to
add that check.
> * ppc-linux-nat.c (ppc_linux_region_ok_for_hw_watchpoint): Check if the
> region is ok for a hardware watchpoint using the new ptrace interface
> on Power servers.
This is OK.
As noted by Luis, it would be good to rename the "booke" terminology,
maybe using something like "have_ptrace_hwdebug" instead of
have_ptrace_booke_interface. But that should be a separate patch.
Thanks,
Ulrich
--
Dr. Ulrich Weigand
GNU Toolchain for Linux on System z and Cell BE
Ulrich.Weigand@de.ibm.com