Re: [PATCH 6.6] membarrier: riscv: Add full memory barrier in switch_mm()

[Alexandre Ghiti]

On 10/09/2024 15:25, Greg KH wrote:
> On Tue, Sep 10, 2024 at 02:35:06PM +0200, Alexandre Ghiti wrote:
> > On 10/09/2024 13:58, Greg KH wrote:
> > > On Tue, Sep 10, 2024 at 01:31:04PM +0200, Alexandre Ghiti wrote:
> > > > Hi Greg,
> > > >
> > > > On 10/09/2024 09:32, Greg KH wrote:
> > > > > On Mon, Sep 09, 2024 at 10:57:01AM +0800, WangYuli wrote:
> > > > > > From: Andrea Parri <parri.andrea@xxxxxxxxx>
> > > > > >
> > > > > > [ Upstream commit d6cfd1770f20392d7009ae1fdb04733794514fa9 ]
> > > > > >
> > > > > > The membarrier system call requires a full memory barrier after storing
> > > > > > to rq->curr, before going back to user-space.  The barrier is only
> > > > > > needed when switching between processes: the barrier is implied by
> > > > > > mmdrop() when switching from kernel to userspace, and it's not needed
> > > > > > when switching from userspace to kernel.
> > > > > >
> > > > > > Rely on the feature/mechanism ARCH_HAS_MEMBARRIER_CALLBACKS and on the
> > > > > > primitive membarrier_arch_switch_mm(), already adopted by the PowerPC
> > > > > > architecture, to insert the required barrier.
> > > > > >
> > > > > > Fixes: fab957c11efe2f ("RISC-V: Atomic and Locking Code")
> > > > > > Signed-off-by: Andrea Parri <parri.andrea@xxxxxxxxx>
> > > > > > Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
> > > > > > Link: https://lore.kernel.org/r/20240131144936.29190-2-parri.andrea@xxxxxxxxx
> > > > > > Signed-off-by: Palmer Dabbelt <palmer@xxxxxxxxxxxx>
> > > > > > Signed-off-by: WangYuli <wangyuli@xxxxxxxxxxxxx>
> > > > > > ---
> > > > > >     MAINTAINERS                         |  2 +-
> > > > > >     arch/riscv/Kconfig                  |  1 +
> > > > > >     arch/riscv/include/asm/membarrier.h | 31 +++++++++++++++++++++++++++++
> > > > > >     arch/riscv/mm/context.c             |  2 ++
> > > > > >     kernel/sched/core.c                 |  5 +++--
> > > > > >     5 files changed, 38 insertions(+), 3 deletions(-)
> > > > > >     create mode 100644 arch/riscv/include/asm/membarrier.h
> > > > > Now queued up, thanks.
> > > > The original patch was merged in 6.9 and the Fixes tag points to a commit
> > > > introduced in v4.15. So IIUC, this patch should have been backported
> > > > "automatically" to the releases < 6.9 right? As stated in the documentation
> > > > (process/stable-kernel-rules.html):
> > > >
> > > > "Note, such tagging is unnecessary if the stable team can derive the
> > > > appropriate versions from Fixes: tags."
> > > >
> > > > Or did we miss something?
> > > Yes, you didn't tag cc: stable at all in this commit, which is why we
> > > did not see it.  The documentation says that :)
> >
> > Ok, some patches seem to make it to stable without the cc: stable tag (like
> > the one below for example), so I thought it was not necessary.
> Yes, they do, because many maintainers and developers don't follow the
> normal rules, so we sweep the tree when we have a chance and do a "best
> effort" backport at times.
>
> But again, it is NOT guaranteed that this will happen, and even if it
> does, you will NOT get emails saying the attempt-at-backporting fails,
> if it fails, as obviously the developers involved weren't explicitly
> asking for it to be backported.
>
> So in short, just properly tag things for stable, and all will be fine.


Great, thank you, I'll make sure we tag the fixes to backport properly!

Thanks again,

Alex


> thanks,
>
> greg k-h