Re: [PATCH v11 0/3] gdbstub and TCG plugin improvements

[Pierrick Bouvier]

On 2/5/24 13:31, Alex Bennée wrote:
> Akihiko Odaki <akihiko.odaki@daynix.com> writes:
>
> > On 2024/02/03 22:58, Alex Bennée wrote:
> > > Akihiko Odaki <akihiko.odaki@daynix.com> writes:
> > >
> > > > On 2024/02/03 20:08, Alex Bennée wrote:
> > > > > Akihiko Odaki <akihiko.odaki@daynix.com> writes:
> > > > >
> > > > > > This series extracts fixes and refactorings that can be applied
> > > > > > independently from "[PATCH v9 00/23] plugins: Allow to read registers".
> > > > > >
> > > > > > The patch "target/riscv: Move MISA limits to class" was replaced with
> > > > > > patch "target/riscv: Move misa_mxl_max to class" since I found instances
> > > > > > may have different misa_ext_mask.
> > > > > As this is re-based on Alistair's riscv-to-apply.next tree I'll wait
> > > > > for
> > > > > this to go through the RiscV trees and then re-base the plugin patches
> > > > > and dropping the merged riscv patches from my tree.
> > > > > In the meantime feel free to review:
> > > > >      Message-Id: <20240122145610.413836-1-alex.bennee@linaro.org>
> > > > >      Date: Mon, 22 Jan 2024 14:55:49 +0000
> > > > >      Subject: [PATCH v3 00/21] plugin updates (register access) for 9.0 
> > > > > (pre-PR?)
> > > > >      From: =?UTF-8?q?Alex=20Benn=C3=A9e?= <alex.bennee@linaro.org>
> > > > > For:
> > > > >      contrib/plugins: extend execlog to track register changes
> > > > >      gdbstub: expose api to find registers
> > > > > So I can add this to my maintainer omnibus series for the next PR I
> > > > > send.
> > > >
> > > > I added one trivial comment to: "gdbstub: expose api to find registers"
> > > >
> > > > "contrib/plugins: extend execlog to track register changes" depends on
> > > > "plugins: add an API to read registers". The comments for the patch in
> > > > the following email are not addressed yet:
> > > > https://lore.kernel.org/all/4b2156ed-688d-4617-b52d-200413f01156@daynix.com/
> > > I don't think we need to serialise with the BQL as the structures
> > > are
> > > per-CPU (and created on vCPU creation).
> >
> > qemu_plugin_get_registers() has vcpu parameter, which can refer to a
> > different vcpu the caller is on (or the caller may not be in a vcpu
> > context at all).
>
> It should only be called from the current cpu context. We can either
> assert that or make it implicit like qemu_plugin_insn_disas does.
> However we will need to ensure current_cpu is set before the vcpu_init
> callback.
>
> Pierrick has had to move these initialisations around for the scoreboard
> work so they are now run with safe work once the thread starts.

As a complement, in the series I'll post, the work is run 
asynchronously, but not "safe_async", which means it's not under an 
exclusive section.

If you need this guarantee for registers API, it's better to add this.

> > > As far as the restructuring we can move it into gdbstub later if
> > > there
> > > is a need to. At the moment the structure is just housekeeping for
> > > plugins.
> >
> > Certainly we can move it later, but adding the code in the plugin
> > infrastructure now won't help in that case.