Re: [PATCH v3 1/2] riscv: Introduce support for hardware break/watchpoints

[Himanshu Chauhan]

Hi,

On Tue, Apr 7, 2026 at 3:59 PM Ilya Mamay <mmamayka01@xxxxxxxxx> wrote:
>
> Hi, Himanshu
>
> Thanks for your reply. However, could you please explain the actual purpose of
> this ecall_lock? I still don't understand why it is needed.

The arch_install/uinstall calls can be called as part of ptrace system
calls. This path can be preempted and leave the in-kernel state of the
hardware breakpoints inconsistent. I am using these for this purpose.

Regards
Himanshu

>
> On Mon, Apr 06, 2026 at 10:21:06AM +0530, Himanshu Chauhan wrote:
> > Hi,
> >
> > On Wed, Mar 18, 2026 at 5:30 PM Ilya Mamay <mmamayka01@xxxxxxxxx> wrote:
> > >
> > > Hi,
> > >
> > > Thank you for this patch - it;s a valuable addition to the RISC-V HBP support.
> > > I have one question regarding the locking:
> > >
> > > On Mon, Feb 23, 2026 at 10:19:17AM +0530, Himanshu Chauhan wrote:
> > > > +int arch_install_hw_breakpoint(struct perf_event *event)
> > > > +{
> > > > +     struct arch_hw_breakpoint *bp = counter_arch_bp(event);
> > > > +     union sbi_dbtr_shmem_entry *shmem = this_cpu_ptr(sbi_dbtr_shmem);
> > > > +     struct sbi_dbtr_data_msg *xmit;
> > > > +     struct sbi_dbtr_id_msg *recv;
> > > > +     struct perf_event **slot;
> > > > +     unsigned long idx;
> > > > +     struct sbiret ret;
> > > > +     int err = 0;
> > > > +
> > > > +     raw_spin_lock_irqsave(this_cpu_ptr(&ecall_lock),
> > > > +                           *this_cpu_ptr(&ecall_lock_flags));
> > >
> > > Could you clarify the purpose of the ecall_lock in arch_install_hw_breakpoint and
> > > arch_update_hw_breakpoint? Why is it used only in two functions mentioned above,
> > > but not in other PMU callbacks like arch_uninstall_hw_breakpoint?
> > >
> > I probably missed it in arch_uninstall_hw_breakpoint. I will take care of it.
> >
> > Thanks & Regards
> > Himanshu
> >
> > > Regards
> > > Ilya
> > >
> > > > _______________________________________________
> > > > linux-riscv mailing list
> > > > linux-riscv@xxxxxxxxxxxxxxxxxxx
> > > > http://lists.infradead.org/mailman/listinfo/linux-riscv