From: "Vivekanandan, Balasubramani" <balasubramani.vivekanandan@intel.com>
To: Matt Roper <matthew.d.roper@intel.com>,
<intel-gfx@lists.freedesktop.org>,
<intel-xe@lists.freedesktop.org>
Subject: Re: [PATCH 3/4] drm/xe: Drop usage of TIMESTAMP_OVERRIDE
Date: Mon, 24 Feb 2025 19:43:08 +0530 [thread overview]
Message-ID: <Z7x-dFvANs689sOV@bvivekan-mobl1> (raw)
In-Reply-To: <20250221003843.443559-9-matthew.d.roper@intel.com>
On 20.02.2025 16:38, Matt Roper wrote:
> On pre-Xe2 platforms, one of the approaches to initialize the GT command
> streamer frequency is to use the display reference clock. That's no
> longer relevant from Xe2 onward (i.e., all of the platforms where Xe is
> officially supported). Furthermore, use of TIMESTAMP_OVERRIDE to obtain
> the display reference clock is unnecessarily roundabout; the display
> driver already has a more reliable approach to obtain this value. Let's
> use the display driver's existing logic to determine the proper display
> reference clock in the rare cases where the hardware indicates we should
> do this.
>
> The one problem that arises here is if the Xe driver is run on an
> unsupported platform (i.e., pre-Xe2), CONFIG_DRM_XE_DISPLAY disabled
> (meaning we're not expecting to touch display hardware at all), and the
> platform has the rare CTC_MODE[1] setting indicating that display
> reference clock should be used as the GT CS refclk. If all of these
> conditions are true, the hardcoded 38.4 MHz value will still be correct
> for DG2 and MTL platforms, but there's a chance that we might not have
> the right value on the older Xe_LP platforms if we're trying not to
> touch the display hardware at all.
>
> Bspec: 62395
> Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Add my RB after addressing the warning from kernel test robot.
Reviewed-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Regards,
Bala
> ---
> drivers/gpu/drm/xe/display/xe_display.c | 6 ++++++
> drivers/gpu/drm/xe/display/xe_display.h | 4 ++++
> drivers/gpu/drm/xe/xe_gt_clock.c | 28 ++++++++-----------------
> 3 files changed, 19 insertions(+), 19 deletions(-)
>
> diff --git a/drivers/gpu/drm/xe/display/xe_display.c b/drivers/gpu/drm/xe/display/xe_display.c
> index 02a413a07382..e35d58079f0d 100644
> --- a/drivers/gpu/drm/xe/display/xe_display.c
> +++ b/drivers/gpu/drm/xe/display/xe_display.c
> @@ -17,6 +17,7 @@
> #include "intel_acpi.h"
> #include "intel_audio.h"
> #include "intel_bw.h"
> +#include "intel_cdclk.h"
> #include "intel_display.h"
> #include "intel_display_driver.h"
> #include "intel_display_irq.h"
> @@ -548,3 +549,8 @@ int xe_display_probe(struct xe_device *xe)
> unset_display_features(xe);
> return 0;
> }
> +
> +u32 xe_display_get_refclk(struct xe_device *xe)
> +{
> + return intel_display_get_refclk(&xe->display);
> +}
> diff --git a/drivers/gpu/drm/xe/display/xe_display.h b/drivers/gpu/drm/xe/display/xe_display.h
> index 685dc74402fb..b918f5d6b53a 100644
> --- a/drivers/gpu/drm/xe/display/xe_display.h
> +++ b/drivers/gpu/drm/xe/display/xe_display.h
> @@ -41,6 +41,8 @@ void xe_display_pm_runtime_suspend(struct xe_device *xe);
> void xe_display_pm_runtime_suspend_late(struct xe_device *xe);
> void xe_display_pm_runtime_resume(struct xe_device *xe);
>
> +u32 xe_display_get_refclk(struct xe_device *xe);
> +
> #else
>
> static inline int xe_display_driver_probe_defer(struct pci_dev *pdev) { return 0; }
> @@ -72,5 +74,7 @@ static inline void xe_display_pm_runtime_suspend(struct xe_device *xe) {}
> static inline void xe_display_pm_runtime_suspend_late(struct xe_device *xe) {}
> static inline void xe_display_pm_runtime_resume(struct xe_device *xe) {}
>
> +static u32 xe_display_get_refclk(struct xe_device *xe) { return 38400; }
> +
> #endif /* CONFIG_DRM_XE_DISPLAY */
> #endif /* _XE_DISPLAY_H_ */
> diff --git a/drivers/gpu/drm/xe/xe_gt_clock.c b/drivers/gpu/drm/xe/xe_gt_clock.c
> index cc2ae159298e..b61f944a7b03 100644
> --- a/drivers/gpu/drm/xe/xe_gt_clock.c
> +++ b/drivers/gpu/drm/xe/xe_gt_clock.c
> @@ -7,6 +7,7 @@
>
> #include "xe_gt_clock.h"
>
> +#include "display/xe_display.h"
> #include "regs/xe_gt_regs.h"
> #include "regs/xe_regs.h"
> #include "xe_assert.h"
> @@ -15,22 +16,6 @@
> #include "xe_macros.h"
> #include "xe_mmio.h"
>
> -static u32 read_reference_ts_freq(struct xe_gt *gt)
> -{
> - u32 ts_override = xe_mmio_read32(>->mmio, TIMESTAMP_OVERRIDE);
> - u32 base_freq, frac_freq;
> -
> - base_freq = REG_FIELD_GET(TIMESTAMP_OVERRIDE_US_COUNTER_DIVIDER_MASK,
> - ts_override) + 1;
> - base_freq *= 1000000;
> -
> - frac_freq = REG_FIELD_GET(TIMESTAMP_OVERRIDE_US_COUNTER_DENOMINATOR_MASK,
> - ts_override);
> - frac_freq = 1000000 / (frac_freq + 1);
> -
> - return base_freq + frac_freq;
> -}
> -
> static u32 get_crystal_clock_freq(u32 rpm_config_reg)
> {
> const u32 f19_2_mhz = 19200000;
> @@ -57,14 +42,19 @@ static u32 get_crystal_clock_freq(u32 rpm_config_reg)
>
> int xe_gt_clock_init(struct xe_gt *gt)
> {
> + struct xe_device *xe = gt_to_xe(gt);
> u32 ctc_reg = xe_mmio_read32(>->mmio, CTC_MODE);
> u32 freq = 0;
>
> /* Assuming gen11+ so assert this assumption is correct */
> - xe_gt_assert(gt, GRAPHICS_VER(gt_to_xe(gt)) >= 11);
> + xe_gt_assert(gt, GRAPHICS_VER(xe) >= 11);
>
> - if (ctc_reg & CTC_SOURCE_DIVIDE_LOGIC) {
> - freq = read_reference_ts_freq(gt);
> + /*
> + * Use of the display reference clock to determine GT CS frequency
> + * is only relevant to pre-Xe2 platforms.
> + */
> + if (GRAPHICS_VER(xe) < 20 && ctc_reg & CTC_SOURCE_DIVIDE_LOGIC) {
> + freq = xe_display_get_refclk(xe);
> } else {
> u32 c0 = xe_mmio_read32(>->mmio, RPM_CONFIG0);
>
> --
> 2.48.1
>
next prev parent reply other threads:[~2025-02-24 14:13 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-02-21 0:38 [PATCH 0/4] Stop accessing display TIMESTAMP_OVERRIDE in GT code Matt Roper
2025-02-21 0:38 ` [PATCH 1/4] drm/i915/display: Make refclk fetching logic reusable Matt Roper
2025-02-24 9:13 ` Jani Nikula
2025-02-24 13:31 ` Vivekanandan, Balasubramani
2025-02-21 0:38 ` [PATCH 2/4] drm/i915/gt: Replace TIMESTAMP_OVERRIDE readout Matt Roper
2025-02-24 13:55 ` Vivekanandan, Balasubramani
2025-02-21 0:38 ` [PATCH 3/4] drm/xe: Drop usage of TIMESTAMP_OVERRIDE Matt Roper
2025-02-21 3:35 ` Lucas De Marchi
2025-02-21 4:25 ` kernel test robot
2025-02-24 14:13 ` Vivekanandan, Balasubramani [this message]
2025-02-21 0:38 ` [PATCH 4/4] drm/xe/sriov: Drop TIMESTAMP_OVERRIDE from Xe2 runtime regs Matt Roper
2025-02-21 0:44 ` ✓ CI.Patch_applied: success for Stop accessing display TIMESTAMP_OVERRIDE in GT code Patchwork
2025-02-21 0:44 ` ✓ CI.checkpatch: " Patchwork
2025-02-21 0:46 ` ✓ CI.KUnit: " Patchwork
2025-02-21 1:02 ` ✓ CI.Build: " Patchwork
2025-02-21 1:05 ` ✓ CI.Hooks: " Patchwork
2025-02-21 1:06 ` ✗ CI.checksparse: warning " Patchwork
2025-02-21 1:26 ` ✓ Xe.CI.BAT: success " Patchwork
2025-02-21 17:01 ` ✗ Xe.CI.Full: failure " 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=Z7x-dFvANs689sOV@bvivekan-mobl1 \
--to=balasubramani.vivekanandan@intel.com \
--cc=intel-gfx@lists.freedesktop.org \
--cc=intel-xe@lists.freedesktop.org \
--cc=matthew.d.roper@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