[PATCH v5 2/3] reset: tenstorrent: Add reset controller for Atlantis

Anirudh Srinivasan asrinivasan at oss.tenstorrent.com
Fri Feb 6 20:10:20 PST 2026


On Thu, Feb 5, 2026 at 12:47 PM Krzysztof Kozlowski <krzk at kernel.org> wrote:
>
> On 05/02/2026 15:18, Anirudh Srinivasan wrote:
> > Hello Krzysztof,
> >
> > On Thu, Feb 5, 2026 at 3:27 AM Krzysztof Kozlowski <krzk at kernel.org> wrote:
> >>
> >> On Mon, Feb 02, 2026 at 01:46:50PM -0600, Anirudh Srinivasan wrote:
> >>> diff --git a/include/soc/tenstorrent/atlantis-prcm.h b/include/soc/tenstorrent/atlantis-prcm.h
> >>> new file mode 100644
> >>> index 000000000000..841516cbefd9
> >>> --- /dev/null
> >>> +++ b/include/soc/tenstorrent/atlantis-prcm.h
> >>> @@ -0,0 +1,31 @@
> >>> +/* SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) */
> >>> +/*
> >>> + * Shared definitions for Atlantis PRCM Clock and Reset Drivers
> >>> + *
> >>> + * Copyright (c) 2026 Tenstorrent
> >>> + */
> >>> +#ifndef __SOC_ATLANTIS_PRCM_H__
> >>> +#define __SOC_ATLANTIS_PRCM_H__
> >>> +
> >>> +#include <linux/bits.h>
> >>> +#include <linux/types.h>
> >>> +
> >>> +struct atlantis_prcm_adev {
> >>> +     struct auxiliary_device adev;
> >>> +     struct regmap *regmap;
> >>> +};
> >>
> >> Drop, there is no driver-wide user of this.
> >
> >>
> >>> +
> >>> +static inline struct atlantis_prcm_adev *
> >>> +to_atlantis_prcm_adev(struct auxiliary_device *adev)
> >>
> >> Same here.
> >>
> >> Do not add symbols to headers which nothing uses. These are private to
> >> the driver so they must stay ONLY in the driver.
> >
> > The definitions in these files are shared between the clock and reset
> > driver (which are in drivers/clk/tenstorrent/ and drivers/reset/),
> > hence they have been put here.
>
>
> Ah, but then I wonder why your clock driver needs to access reset driver
> structures. Or vice versa. You are only expected to get regmap from the
> parent, no?

Yes, it seems to be doable without these structs. I will remove them,
and move the register definitions into the specific drivers.

>
>
> Best regards,
> Krzysztof



More information about the linux-riscv mailing list