From: "Ville Syrjälä" <ville.syrjala@linux.intel.com>
To: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org, intel-xe@lists.freedesktop.org,
Rodrigo Vivi <rodrigo.vivi@intel.com>
Subject: Re: [PATCH 4/5] drm/i915/gmbus: Fix i2c bit banging for PTL+
Date: Wed, 14 Oct 2026 18:56:53 +0300 [thread overview]
Message-ID: <as-mReJ7rAGinlgY@intel.com> (raw)
In-Reply-To: <179199289730.125250.2949692993509316966@jlahtine-mobl>
On Wed, Oct 14, 2026 at 06:48:17PM +0300, Joonas Lahtinen wrote:
> Hi,
>
> This didn't cherry pick cleanly into drm-intel-fixes.
>
> Do we still want to send this into v7.3 given how fresh and
> relatively non-trivial it is and we're past -rc7 already?
Should be fine to let this sit in -next for a while. The code
tries the gmbus path first (which should usually work), and bit
banging is just a fallback.
>
> If so, we'd need a version that applies on drm-intel-fixes and I can
> then send an extra PR to see if it will get pulled in time still.
>
> Regards, Joonas
>
> Quoting Ville Syrjala (2026-10-08 01:23:13)
> > From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> >
> > The current implementation of w/a 16025573575 does not actually work.
> > The GPIO register mask bits are write only, and thus he read-modify-write
> > cycles we do will not preserve them. Furthermore the implementation makes
> > no effort to preserve the actual direction/output value bits, which is
> > a must if all the mask bits were actually set all the time.
> >
> > Rewrite the whole thing so that we always set all the mask bits,
> > and correctly preserve the previous state of the direction/output value
> > bits. And do that on all platforms so that we only have a single codepath
> > to worry about. As a bonus this approach will also make the code
> > work on upcoming platforms where the mask bits no longer even exist.
> >
> > Smoke tested on alm,cl,tgl,lnl,ptl.
> >
> > Fixes: 7565fd5dcb0b ("drm/i915/gmbus: Add Wa_16025573575 for PTL/WCL for bit-bashing")
> > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > ---
> > drivers/gpu/drm/i915/display/intel_gmbus.c | 98 ++++++++++++++--------
> > 1 file changed, 61 insertions(+), 37 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_gmbus.c b/drivers/gpu/drm/i915/display/intel_gmbus.c
> > index 854f439ac5ba..b6c63a466233 100644
> > --- a/drivers/gpu/drm/i915/display/intel_gmbus.c
> > +++ b/drivers/gpu/drm/i915/display/intel_gmbus.c
> > @@ -246,25 +246,45 @@ static void bxt_gmbus_clock_gating(struct intel_display *display,
> > !enable ? BXT_GMBUS_GATING_DIS : 0);
> > }
> >
> > -static u32 get_reserved(struct intel_gmbus *bus)
> > +static u32 gpio_mask_bits(struct intel_display *display)
> > {
> > - struct intel_display *display = bus->display;
> > - u32 preserve_bits = 0;
> > + return GPIO_CLOCK_DIR_MASK | GPIO_CLOCK_VAL_MASK |
> > + GPIO_DATA_DIR_MASK | GPIO_DATA_VAL_MASK;
> > +}
> > +
> > +static u32 gpio_preserved_bits(struct intel_display *display)
> > +{
> > + u32 val = GPIO_CLOCK_DIR_OUT | GPIO_CLOCK_VAL_OUT |
> > + GPIO_DATA_DIR_OUT | GPIO_DATA_VAL_OUT;
> >
> > /*
> > * On gen2/3 these are either internal pullup disables, or
> > * documented as reserved bits that need to be preserved.
> > */
> > if (DISPLAY_VER(display) < 4)
> > - preserve_bits |= GPIO_DATA_PULLUP_DISABLE |
> > + val |= GPIO_DATA_PULLUP_DISABLE |
> > GPIO_CLOCK_PULLUP_DISABLE;
> >
> > - /* Wa_16025573575: the masks bits need to be preserved through out */
> > - if (intel_display_wa(display, INTEL_DISPLAY_WA_16025573575))
> > - preserve_bits |= GPIO_CLOCK_DIR_MASK | GPIO_CLOCK_VAL_MASK |
> > - GPIO_DATA_DIR_MASK | GPIO_DATA_VAL_MASK;
> > + return val;
> > +}
> >
> > - return intel_de_read_fw(display, bus->gpio_reg) & preserve_bits;
> > +static u32 gpio_read_reg(struct intel_gmbus *bus)
> > +{
> > + struct intel_display *display = bus->display;
> > + u32 val;
> > +
> > + val = intel_de_read_fw(display, bus->gpio_reg);
> > +
> > + val &= gpio_preserved_bits(display);
> > +
> > + /*
> > + * Wa_16025573575:
> > + * The (write only!) masks bits need to be kept set through out
> > + * on PTL. Do it always to keep the code uniform for all platforms.
> > + */
> > + val |= gpio_mask_bits(display);
> > +
> > + return val;
> > }
> >
> > static int get_clock(void *data)
> > @@ -287,16 +307,16 @@ static void set_clock(void *data, int state_high)
> > {
> > struct intel_gmbus *bus = data;
> > struct intel_display *display = bus->display;
> > - u32 reserved = get_reserved(bus);
> > - u32 clock_bits;
> > + u32 val;
> >
> > - if (state_high)
> > - clock_bits = GPIO_CLOCK_DIR_IN | GPIO_CLOCK_DIR_MASK;
> > - else
> > - clock_bits = GPIO_CLOCK_DIR_OUT | GPIO_CLOCK_DIR_MASK |
> > - GPIO_CLOCK_VAL_MASK;
> > + val = gpio_read_reg(bus);
> >
> > - intel_de_write_fw(display, bus->gpio_reg, reserved | clock_bits);
> > + val &= ~(GPIO_CLOCK_DIR_OUT | GPIO_CLOCK_VAL_OUT);
> > +
> > + if (!state_high)
> > + val |= GPIO_CLOCK_DIR_OUT;
> > +
> > + intel_de_write_fw(display, bus->gpio_reg, val);
> > intel_de_posting_read_fw(display, bus->gpio_reg);
> > }
> >
> > @@ -304,32 +324,36 @@ static void set_data(void *data, int state_high)
> > {
> > struct intel_gmbus *bus = data;
> > struct intel_display *display = bus->display;
> > - u32 reserved = get_reserved(bus);
> > - u32 data_bits;
> > + u32 val;
> >
> > - if (state_high)
> > - data_bits = GPIO_DATA_DIR_IN | GPIO_DATA_DIR_MASK;
> > - else
> > - data_bits = GPIO_DATA_DIR_OUT | GPIO_DATA_DIR_MASK |
> > - GPIO_DATA_VAL_MASK;
> > + val = gpio_read_reg(bus);
> >
> > - intel_de_write_fw(display, bus->gpio_reg, reserved | data_bits);
> > + val &= ~(GPIO_DATA_DIR_OUT | GPIO_DATA_VAL_OUT);
> > +
> > + if (!state_high)
> > + val |= GPIO_DATA_DIR_OUT;
> > +
> > + intel_de_write_fw(display, bus->gpio_reg, val);
> > intel_de_posting_read_fw(display, bus->gpio_reg);
> > }
> >
> > static void
> > -ptl_handle_mask_bits(struct intel_gmbus *bus, bool set)
> > +gpio_set_mask(struct intel_gmbus *bus, bool enable)
> > {
> > struct intel_display *display = bus->display;
> > - u32 reg_val = intel_de_read_fw(display, bus->gpio_reg);
> > - u32 mask_bits = GPIO_CLOCK_DIR_MASK | GPIO_CLOCK_VAL_MASK |
> > - GPIO_DATA_DIR_MASK | GPIO_DATA_VAL_MASK;
> > - if (set)
> > - reg_val |= mask_bits;
> > + u32 val;
> > +
> > + val = gpio_read_reg(bus);
> > +
> > + val &= ~(GPIO_DATA_DIR_OUT | GPIO_CLOCK_DIR_OUT |
> > + GPIO_DATA_VAL_OUT | GPIO_CLOCK_VAL_OUT);
> > +
> > + if (enable)
> > + val |= gpio_mask_bits(display);
> > else
> > - reg_val &= ~mask_bits;
> > + val &= ~gpio_mask_bits(display);
> >
> > - intel_de_write_fw(display, bus->gpio_reg, reg_val);
> > + intel_de_write_fw(display, bus->gpio_reg, val);
> > intel_de_posting_read_fw(display, bus->gpio_reg);
> > }
> >
> > @@ -344,8 +368,8 @@ intel_gpio_pre_xfer(struct i2c_adapter *adapter)
> >
> > intel_gmbus_reset(display);
> >
> > - if (intel_display_wa(display, INTEL_DISPLAY_WA_16025573575))
> > - ptl_handle_mask_bits(bus, true);
> > + if (gpio_mask_bits(display))
> > + gpio_set_mask(bus, true);
> >
> > set_data(bus, 1);
> > set_clock(bus, 1);
> > @@ -362,8 +386,8 @@ intel_gpio_post_xfer(struct i2c_adapter *adapter)
> > set_data(bus, 1);
> > set_clock(bus, 1);
> >
> > - if (intel_display_wa(display, INTEL_DISPLAY_WA_16025573575))
> > - ptl_handle_mask_bits(bus, false);
> > + if (gpio_mask_bits(display))
> > + gpio_set_mask(bus, false);
> >
> > if (display->platform.pineview)
> > pnv_gmbus_clock_gating(display, true);
> > --
> > 2.55.0
> >
--
Ville Syrjälä
Intel
next prev parent reply other threads:[~2026-10-14 15:57 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-07 22:23 [PATCH 0/5] drm/i915/gmbus: Fix PTL+ i2c bit banging Ville Syrjala
2026-10-07 22:23 ` [PATCH 1/5] drm/i915/gmbus: Stop frobbing with pin direction in .getsda()/.getscl() Ville Syrjala
2026-10-08 9:34 ` Sebastian Brzezinka
2026-10-07 22:23 ` [PATCH 2/5] drm/i915/gmbus: Handle the internal pullup disable bits more sensibly Ville Syrjala
2026-10-12 14:28 ` Nautiyal, Ankit K
2026-10-07 22:23 ` [PATCH 3/5] drm/i915/gmbus: Reorder the PNV clock gating GPIO w/a Ville Syrjala
2026-10-08 9:49 ` Sebastian Brzezinka
2026-10-07 22:23 ` [PATCH 4/5] drm/i915/gmbus: Fix i2c bit banging for PTL+ Ville Syrjala
2026-10-08 9:48 ` Sebastian Brzezinka
2026-10-12 14:25 ` Nautiyal, Ankit K
2026-10-14 15:48 ` Joonas Lahtinen
2026-10-14 15:56 ` Ville Syrjälä [this message]
2026-10-15 5:56 ` Joonas Lahtinen
2026-10-07 22:23 ` [PATCH 5/5] drm/i915/gmbus: Remove the INTEL_DISPLAY_WA_16025573575 leftovers Ville Syrjala
2026-10-08 9:48 ` Sebastian Brzezinka
2026-10-07 22:33 ` ✓ CI.KUnit: success for drm/i915/gmbus: Fix PTL+ i2c bit banging Patchwork
2026-10-07 23:54 ` ✓ Xe.CI.BAT: " Patchwork
2026-10-08 3:36 ` ✓ Xe.CI.FULL: " Patchwork
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=as-mReJ7rAGinlgY@intel.com \
--to=ville.syrjala@linux.intel.com \
--cc=intel-gfx@lists.freedesktop.org \
--cc=intel-xe@lists.freedesktop.org \
--cc=joonas.lahtinen@linux.intel.com \
--cc=rodrigo.vivi@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox