This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH, v2] Expect SI_KERNEL or TRAP_BRKPT si_code values for MIPS breakpoint traps
- From: Pedro Alves <palves at redhat dot com>
- To: "Maciej W. Rozycki" <macro at imgtec dot com>
- Cc: Luis Machado <lgustavo at codesourcery dot com>, gdb-patches at sourceware dot org
- Date: Wed, 24 Feb 2016 11:57:42 +0000
- Subject: Re: [PATCH, v2] Expect SI_KERNEL or TRAP_BRKPT si_code values for MIPS breakpoint traps
- Authentication-results: sourceware.org; auth=none
- References: <1456145829-5052-1-git-send-email-lgustavo at codesourcery dot com> <56CB0A42 dot 70507 at redhat dot com> <alpine dot DEB dot 2 dot 00 dot 1602221612230 dot 15885 at tp dot orcam dot me dot uk>
On 02/22/2016 04:19 PM, Maciej W. Rozycki wrote:
> On Mon, 22 Feb 2016, Pedro Alves wrote:
>
>> I'd like to have a short paragraph in linux-nat.h about MIPS's behavior,
>> below the ppc comment. If you look at the x86 table above, which cases get
>> SI_KERNEL, and which don't? Single-step is not applicable, since the
>> MIPS doesn't handle single-stepping itself (IIRC), but do we also get
>> SI_KERNEL hardware bps and watchpoints, for example?
>
> As I previously noted the MIPS Linux port currently does not support
> hardware breakpoints even though hardware may have them implemented
> (they're optional). Only hardware watchpoints are supported right now.
Ah.
> You are right in that there's no OS single-stepping support defined in
> the MIPS architecture and consequently Linux cannot support it on the MIPS
> target. Only JTAG single-stepping is possible with MIPS hardware.
Yeah. I see I missed a word, sorry. I meant
"since the MIPS _kernel_ doesn't handle single-stepping itself".
I was thinking about e.g., uprobes, and imagining that since the kernel
needs to advance past uprobes-set breakpoints somehow, that the kernel
could have learned to software single-step, and that that might have been
hooked to PTRACE_SINGLESTEP.
Thanks,
Pedro Alves