From: sashiko-bot@kernel.org
To: "Cody Kang" <codykang.hk@gmail.com>
Cc: dri-devel@lists.freedesktop.org, devicetree@vger.kernel.org,
linux-phy@lists.infradead.org, linux-clk@vger.kernel.org,
robh@kernel.org, conor+dt@kernel.org, vkoul@kernel.org,
neil.armstrong@linaro.org, olteanv@gmail.com
Subject: Re: [PATCH v3 09/17] drm/spacemit: add Saturn DPU hardware backend
Date: Sat, 21 Nov 2026 08:01:22 +0000 [thread overview]
Message-ID: <sashiko-outbox-226811@kernel.org> (raw)
In-Reply-To: <20261121-k3-display-v3-9-ec419e8e96b8@gmail.com>
Thank you for your contribution! Sashiko AI review found 4 potential
issues to consider:
- [High] drm/spacemit: vfp underflow in saturn_init_tmg() when front
porch < 7
- [High] drm/spacemit: unmasked IRQ dispatch and premature vblank event
in ISR
- [Medium] drm/spacemit: stale underrun IRQ and ur_mask_armed race on
CFG_RDY
- [Medium] drm/spacemit: race on cfg_rdy_timer and INT_CFG_RDY in
spacemit_dpu_run()
--
commit feb91f776850387e06b93738a2a9ed3971674069
Author: Cody Kang <codykang@vsit.ai>
drm/spacemit: add Saturn DPU hardware backend
This commit adds the Saturn DPU hardware backend, covering clock setup,
scene control and timing generator programming, RDMA channel and composer
configuration, line-buffer sizing, and interrupt handling.
> diff --git a/drivers/gpu/drm/spacemit/dpu/dpu_saturn.c b/drivers/gpu/drm/spacemit/dpu/dpu_saturn.c
> new file mode 100644
> index 0000000000000..704e64fb9ec7e
> --- /dev/null
> +++ b/drivers/gpu/drm/spacemit/dpu/dpu_saturn.c
> @@ -0,0 +1,353 @@
[ ... ]
> +/* Complete a commit whose lists the engine will not confirm any more. */
> +void spacemit_dpu_complete_event(struct spacemit_crtc *a_crtc)
> +{
> + struct drm_crtc *crtc = &a_crtc->crtc;
> + struct drm_pending_vblank_event *event;
> + unsigned long flags;
> +
> + spin_lock_irqsave(&crtc->dev->event_lock, flags);
> + event = a_crtc->event;
> + a_crtc->event = NULL;
> + if (event)
> + drm_crtc_send_vblank_event(crtc, event);
> + spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
> +
> + if (event)
> + drm_crtc_vblank_put(crtc);
> +}
> +
> +void spacemit_cfg_rdy_timer_handler(struct timer_list *t)
> +{
> + struct spacemit_crtc *a_crtc = timer_container_of(a_crtc, t,
> + cfg_rdy_timer);
> +
> + drm_err_ratelimited(a_crtc->crtc.dev, "CFG_RDY handshake timeout\n");
> + /* A CFG_RDY that still comes belongs to no commit any more. */
> + saturn_hee_enable_irq(a_crtc, INT_CFG_RDY, false);
> + spacemit_dpu_complete_event(a_crtc);
> +}
[ ... ]
> +/* Returns false when none of the pending status bits was this CRTC's. */
> +bool spacemit_dpu_online_isr(struct spacemit_crtc *a_crtc)
> +{
> + u32 irq_raw, irq_bit, irq_ur_bit, handled;
> + struct spacemit_drm_private *priv = a_crtc->crtc.dev->dev_private;
> + struct spacemit_hw_device *hwdev = priv->hwdev;
> +
> + irq_raw = saturn_hee_get_int_sts(hwdev);
> + irq_bit = saturn_hee_get_irq_bit(INT_UNDERRUN);
> + irq_ur_bit = irq_raw & irq_bit;
> + handled = irq_ur_bit;
> + if (irq_ur_bit && !a_crtc->ur_reported) {
> + saturn_hee_clr_int_sts(a_crtc, irq_ur_bit);
> + drm_err_ratelimited(a_crtc->crtc.dev, "underrun\n");
> + a_crtc->ur_reported = true;
> + }
> + irq_bit = saturn_hee_get_irq_bit(INT_CFG_RDY);
> + if (irq_raw & irq_bit) {
> + saturn_hee_clr_int_sts(a_crtc, irq_bit);
> + timer_delete(&a_crtc->cfg_rdy_timer);
> + a_crtc->ur_reported = false;
> + saturn_hee_enable_irq(a_crtc, INT_CFG_RDY, false);
> + handled |= irq_bit;
> +
> + /* The lists are taken: the next vblank completes the commit. */
> + spin_lock(&a_crtc->crtc.dev->event_lock);
> + if (a_crtc->event) {
> + drm_crtc_arm_vblank_event(&a_crtc->crtc, a_crtc->event);
> + a_crtc->event = NULL;
> + }
> + spin_unlock(&a_crtc->crtc.dev->event_lock);
> + }
> + irq_bit = saturn_hee_get_irq_bit(INT_VSYNC);
> + if (irq_raw & irq_bit) {
> + saturn_hee_clr_int_sts(a_crtc, irq_bit);
> + drm_crtc_handle_vblank(&a_crtc->crtc);
> + handled |= irq_bit;
> + }
> + irq_bit = saturn_hee_get_irq_bit(INT_REST) | irq_ur_bit;
> + if (irq_raw & irq_bit) {
> + saturn_hee_clr_int_sts(a_crtc, irq_bit);
> + handled |= irq_raw & irq_bit;
> + }
> +
> + return handled;
> +}
[Severity: High]
Can spacemit_dpu_online_isr() complete a commit's vblank event a frame
early when INT_CFG_RDY and INT_VSYNC are both set in irq_raw?
As noted in saturn_hee_irq_enable() and spacemit_dpu_run(),
DPU_ONLINE_IRQ_STS latches status bits even while their mask bits in
DPU_ONLINE_IRQ_MSK are cleared, yet saturn_hee_enable_irq() unmasks
interrupts without clearing DPU_ONLINE_IRQ_STS first and
spacemit_dpu_online_isr() does not mask irq_raw against
DPU_ONLINE_IRQ_MSK. When INT_CFG_RDY fires while INT_VSYNC is also set
(either from a frame boundary or latched while vblank was masked), the
INT_CFG_RDY branch calls drm_crtc_arm_vblank_event() and the following
INT_VSYNC branch immediately calls drm_crtc_handle_vblank() in the same
ISR invocation. That delivers the flip event before the next vblank and
unblocks drm_atomic_helper_wait_for_flip_done() early, allowing the next
commit to overwrite the active command list slot.
> +
> +void spacemit_dpu_run(struct drm_crtc *crtc, struct drm_atomic_commit *state)
> +{
> + struct spacemit_crtc *a_crtc = to_spacemit_crtc(crtc);
> + struct spacemit_drm_private *priv = a_crtc->crtc.dev->dev_private;
> + struct spacemit_hw_device *hwdev = priv->hwdev;
> +
> + saturn_hee_conf_dpuctrl(crtc, state);
> +
> + /* flush all config register writes before CFG_RDY is raised */
> + mb();
> +
> + /* Ack first: one left by a commit that timed out is not this one's. */
> + saturn_hee_clr_int_sts(a_crtc, saturn_hee_get_irq_bit(INT_CFG_RDY));
> + saturn_hee_enable_irq(a_crtc, INT_CFG_RDY, true);
> +
> + mod_timer(&a_crtc->cfg_rdy_timer,
> + jiffies + msecs_to_jiffies(SATURN_CFG_RDY_TIMEOUT_MS));
> + saturn_hee_cfg_ready(a_crtc, hwdev);
> +
> + if (unlikely(a_crtc->is_1st_f)) {
> + drm_dbg(a_crtc->crtc.dev, "DPU start\n");
> + a_crtc->is_1st_f = false;
> + saturn_hee_sw_start(a_crtc, hwdev);
> + }
> +}
[Severity: Medium]
Can spacemit_cfg_rdy_timer_handler() race with spacemit_dpu_online_isr()
and spacemit_dpu_run() to complete the next commit's vblank event early or
leave INT_CFG_RDY masked?
CPU0 (<softirq>) CPU1
spacemit_cfg_rdy_timer_handler()
drm_err_ratelimited(...)
<hardirq>
spacemit_dpu_online_isr()
timer_delete(&a_crtc->cfg_rdy_timer)
drm_crtc_arm_vblank_event(...)
drm_crtc_handle_vblank(...)
spacemit_crtc_atomic_flush()
a_crtc->event = new_event
spacemit_dpu_run()
saturn_hee_enable_irq(..., true)
saturn_hee_enable_irq(..., false)
spacemit_dpu_complete_event()
drm_crtc_send_vblank_event() <-- completes new commit early
timer_delete() in spacemit_dpu_online_isr() does not wait for an in-flight
spacemit_cfg_rdy_timer_handler(), so the timer callback can mask the next
commit's INT_CFG_RDY and complete its a_crtc->event before the hardware
consumes its command lists.
In addition, spacemit_dpu_run() unmasks INT_CFG_RDY before mod_timer() and
saturn_hee_cfg_ready(). If a late CFG_RDY from a previously timed-out
commit arrives right after saturn_hee_enable_irq(a_crtc, INT_CFG_RDY,
true), spacemit_dpu_online_isr() runs timer_delete() before the timer is
armed, masks INT_CFG_RDY, and arms a_crtc->event before both_cfg_rdy is
written for the new commit.
> diff --git a/drivers/gpu/drm/spacemit/dpu/dpu_saturn_hee.c b/drivers/gpu/drm/spacemit/dpu/dpu_saturn_hee.c
> new file mode 100644
> index 0000000000000..1a194fe7cc7ba
> --- /dev/null
> +++ b/drivers/gpu/drm/spacemit/dpu/dpu_saturn_hee.c
> @@ -0,0 +1,670 @@
[ ... ]
> +void saturn_hee_enable_irq(struct spacemit_crtc *a_crtc,
> + enum spacemit_dpu_irq irq_id, bool enable)
> +{
> + saturn_enable_irq_mask(a_crtc, enable,
> + DPU_INT_BASE_ADDR + DPU_ONLINE_IRQ_MSK,
> + saturn_hee_get_irq_bit(irq_id));
> +}
> +
> +void saturn_hee_cfg_ready(struct spacemit_crtc *a_crtc,
> + struct spacemit_hw_device *hwdev)
> +{
> + u32 base = DPU_SCENE_CTL_ADDR(COMPOSER1);
> +
> + saturn_hee_enable_irq(a_crtc, INT_UNDERRUN, true);
> + a_crtc->ur_mask_armed = true;
> + dpu_write(hwdev, DPU_CTL_REG, base, both_cfg_rdy, 1);
> +}
[Severity: Medium]
Does enabling INT_UNDERRUN in saturn_hee_cfg_ready() before setting
a_crtc->ur_mask_armed and without clearing DPU_ONLINE_IRQ_STS race with
spacemit_dpu_online_isr()?
saturn_hee_cfg_ready()
saturn_hee_enable_irq(a_crtc, INT_UNDERRUN, true)
<hardirq>
spacemit_dpu_online_isr()
saturn_hee_clr_int_sts(a_crtc, ...)
if (... && a_crtc->ur_mask_armed) <-- still false; stays unmasked
a_crtc->ur_mask_armed = true
After an underrun, INT_UNDERRUN is masked for the rest of the frame while
the hardware continues latching DPU_INT_FRM_TIMING_UNFLOW in
DPU_ONLINE_IRQ_STS. When the next commit calls saturn_hee_cfg_ready(),
unmasking INT_UNDERRUN without clearing DPU_ONLINE_IRQ_STS immediately
triggers spacemit_dpu_online_isr(). If the interrupt arrives before
a_crtc->ur_mask_armed = true (which also happens in
saturn_hee_irq_enable()), saturn_hee_clr_int_sts() sees ur_mask_armed ==
false and leaves INT_UNDERRUN unmasked; if it arrives after
ur_mask_armed = true but before INT_CFG_RDY clears ur_reported, the bottom
saturn_hee_clr_int_sts() call in spacemit_dpu_online_isr() immediately
masks INT_UNDERRUN again before the new commit takes effect.
> +
> +void saturn_hee_sw_start(struct spacemit_crtc *a_crtc,
> + struct spacemit_hw_device *hwdev)
> +{
> + u32 base;
> +
> + base = DPU_SCENE_CTL_ADDR(COMPOSER1);
> + dpu_write_reg(hwdev, DPU_CTL_REG, base, sw_start, 1);
> +}
> +
> +static void saturn_init_tmg(struct spacemit_crtc *a_crtc,
> + const struct drm_display_mode *mode)
> +{
> + u32 base = TMG0_BASE_ADDR;
> + struct spacemit_drm_private *priv = a_crtc->crtc.dev->dev_private;
> + struct spacemit_hw_device *hwdev = priv->hwdev;
> + u16 vfp, vbp, vsync, hfp, hbp, hsync;
> + void __iomem *tmg_addr;
> + u32 value;
> +
> + tmg_addr = hwdev->base + TMG0_BASE_ADDR;
> +
> + hsync = mode->hsync_end - mode->hsync_start;
> + hbp = mode->htotal - mode->hsync_end;
> + hfp = mode->hsync_start - mode->hdisplay;
> + vsync = mode->vsync_end - mode->vsync_start;
> + vbp = mode->vtotal - mode->vsync_end;
> + vfp = mode->vsync_start - mode->vdisplay;
> + drm_dbg(a_crtc->crtc.dev, "timing " DRM_MODE_FMT "\n", DRM_MODE_ARG(mode));
> +
> + dpu_write(hwdev, TMG_REG, base, disp_ready_man_en, 0);
> + dpu_write(hwdev, TMG_REG, base, background_r, 0);
> + dpu_write(hwdev, TMG_REG, base, background_g, 0x0);
> + dpu_write(hwdev, TMG_REG, base, background_b, 0xff);
> + dpu_write(hwdev, TMG_REG, base, eof_1st_ln_dly_num, vfp - 7);
> + dpu_write(hwdev, TMG_REG, base, eof_2nd_ln_dly_num, vfp - 6);
> + dpu_write(hwdev, TMG_REG, base, split_en, 0);
> + dpu_write(hwdev, TMG_REG, base, cmd_screen, 0);
> + dpu_write(hwdev, TMG_REG, base, cmd_wait_en, 0);
> + dpu_write(hwdev, TMG_REG, base, cmd_wait_te, 0);
> + dpu_write(hwdev, TMG_REG, base, sof_pre_ln_num, 0);
> + dpu_write(hwdev, TMG_REG, base, hfp, hfp);
> + dpu_write(hwdev, TMG_REG, base, hbp, hbp);
> + dpu_write(hwdev, TMG_REG, base, vfp, vfp);
> + dpu_write(hwdev, TMG_REG, base, vbp, vbp);
> + dpu_write(hwdev, TMG_REG, base, hsync_width, hsync);
> + dpu_write(hwdev, TMG_REG, base, vsync_width, vsync);
> + dpu_write(hwdev, TMG_REG, base, hsp, 1);
> + dpu_write(hwdev, TMG_REG, base, vsp, 1);
> + dpu_write(hwdev, TMG_REG, base, h_active, mode->hdisplay);
> + dpu_write(hwdev, TMG_REG, base, v_active, mode->vdisplay);
> + dpu_write(hwdev, TMG_REG, base, fm_timing_en, 1);
> + /* The pipeline always hands RGB888 to the DP/eDP encoder. */
> + dpu_write(hwdev, TMG_REG, base, user, OUTFMT_RGB888);
> +
> + value = readl(tmg_addr + 4 * dpu_field_word(TMG_REG, vfp));
> + value &= ~dpu_field_mask(TMG_REG, vfp);
> + value |= vfp;
> + writel(value, tmg_addr + 4 * dpu_field_word(TMG_REG, vfp));
[Severity: High]
Does vfp - 7 and vfp - 6 underflow when the display mode's vertical front
porch is smaller than 7 lines?
Standard modes such as 1920x1080@60Hz (vdisplay = 1080, vsync_start =
1084, so vfp = 4) have vfp < 7, which wraps the 16-bit eof_1st_ln_dly_num
and eof_2nd_ln_dly_num fields to 0xfff9..0xffff, exceeding vtotal so the
EOF line-delay counters cannot fire within the frame. Also, the raw
readl()/writel() sequence at the end of saturn_init_tmg() redundantly
rewrites the same vfp field that dpu_write(hwdev, TMG_REG, base, vfp, vfp)
just programmed above.
--
Sashiko AI review · https://sashiko.dev/#/patchset/20261121-k3-display-v3-0-ec419e8e96b8@gmail.com?part=9
WARNING: multiple messages have this Message-ID (diff)
From: sashiko-bot@kernel.org
To: "Cody Kang" <codykang.hk@gmail.com>
Cc: dri-devel@lists.freedesktop.org, devicetree@vger.kernel.org,
linux-phy@lists.infradead.org, linux-clk@vger.kernel.org,
robh@kernel.org, conor+dt@kernel.org, vkoul@kernel.org,
neil.armstrong@linaro.org, olteanv@gmail.com
Subject: Re: [PATCH v3 09/17] drm/spacemit: add Saturn DPU hardware backend
Date: Sat, 21 Nov 2026 08:01:22 +0000 [thread overview]
Message-ID: <sashiko-outbox-226811@kernel.org> (raw)
In-Reply-To: <20261121-k3-display-v3-9-ec419e8e96b8@gmail.com>
Thank you for your contribution! Sashiko AI review found 4 potential
issues to consider:
- [High] drm/spacemit: vfp underflow in saturn_init_tmg() when front
porch < 7
- [High] drm/spacemit: unmasked IRQ dispatch and premature vblank event
in ISR
- [Medium] drm/spacemit: stale underrun IRQ and ur_mask_armed race on
CFG_RDY
- [Medium] drm/spacemit: race on cfg_rdy_timer and INT_CFG_RDY in
spacemit_dpu_run()
--
commit feb91f776850387e06b93738a2a9ed3971674069
Author: Cody Kang <codykang@vsit.ai>
drm/spacemit: add Saturn DPU hardware backend
This commit adds the Saturn DPU hardware backend, covering clock setup,
scene control and timing generator programming, RDMA channel and composer
configuration, line-buffer sizing, and interrupt handling.
> diff --git a/drivers/gpu/drm/spacemit/dpu/dpu_saturn.c b/drivers/gpu/drm/spacemit/dpu/dpu_saturn.c
> new file mode 100644
> index 0000000000000..704e64fb9ec7e
> --- /dev/null
> +++ b/drivers/gpu/drm/spacemit/dpu/dpu_saturn.c
> @@ -0,0 +1,353 @@
[ ... ]
> +/* Complete a commit whose lists the engine will not confirm any more. */
> +void spacemit_dpu_complete_event(struct spacemit_crtc *a_crtc)
> +{
> + struct drm_crtc *crtc = &a_crtc->crtc;
> + struct drm_pending_vblank_event *event;
> + unsigned long flags;
> +
> + spin_lock_irqsave(&crtc->dev->event_lock, flags);
> + event = a_crtc->event;
> + a_crtc->event = NULL;
> + if (event)
> + drm_crtc_send_vblank_event(crtc, event);
> + spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
> +
> + if (event)
> + drm_crtc_vblank_put(crtc);
> +}
> +
> +void spacemit_cfg_rdy_timer_handler(struct timer_list *t)
> +{
> + struct spacemit_crtc *a_crtc = timer_container_of(a_crtc, t,
> + cfg_rdy_timer);
> +
> + drm_err_ratelimited(a_crtc->crtc.dev, "CFG_RDY handshake timeout\n");
> + /* A CFG_RDY that still comes belongs to no commit any more. */
> + saturn_hee_enable_irq(a_crtc, INT_CFG_RDY, false);
> + spacemit_dpu_complete_event(a_crtc);
> +}
[ ... ]
> +/* Returns false when none of the pending status bits was this CRTC's. */
> +bool spacemit_dpu_online_isr(struct spacemit_crtc *a_crtc)
> +{
> + u32 irq_raw, irq_bit, irq_ur_bit, handled;
> + struct spacemit_drm_private *priv = a_crtc->crtc.dev->dev_private;
> + struct spacemit_hw_device *hwdev = priv->hwdev;
> +
> + irq_raw = saturn_hee_get_int_sts(hwdev);
> + irq_bit = saturn_hee_get_irq_bit(INT_UNDERRUN);
> + irq_ur_bit = irq_raw & irq_bit;
> + handled = irq_ur_bit;
> + if (irq_ur_bit && !a_crtc->ur_reported) {
> + saturn_hee_clr_int_sts(a_crtc, irq_ur_bit);
> + drm_err_ratelimited(a_crtc->crtc.dev, "underrun\n");
> + a_crtc->ur_reported = true;
> + }
> + irq_bit = saturn_hee_get_irq_bit(INT_CFG_RDY);
> + if (irq_raw & irq_bit) {
> + saturn_hee_clr_int_sts(a_crtc, irq_bit);
> + timer_delete(&a_crtc->cfg_rdy_timer);
> + a_crtc->ur_reported = false;
> + saturn_hee_enable_irq(a_crtc, INT_CFG_RDY, false);
> + handled |= irq_bit;
> +
> + /* The lists are taken: the next vblank completes the commit. */
> + spin_lock(&a_crtc->crtc.dev->event_lock);
> + if (a_crtc->event) {
> + drm_crtc_arm_vblank_event(&a_crtc->crtc, a_crtc->event);
> + a_crtc->event = NULL;
> + }
> + spin_unlock(&a_crtc->crtc.dev->event_lock);
> + }
> + irq_bit = saturn_hee_get_irq_bit(INT_VSYNC);
> + if (irq_raw & irq_bit) {
> + saturn_hee_clr_int_sts(a_crtc, irq_bit);
> + drm_crtc_handle_vblank(&a_crtc->crtc);
> + handled |= irq_bit;
> + }
> + irq_bit = saturn_hee_get_irq_bit(INT_REST) | irq_ur_bit;
> + if (irq_raw & irq_bit) {
> + saturn_hee_clr_int_sts(a_crtc, irq_bit);
> + handled |= irq_raw & irq_bit;
> + }
> +
> + return handled;
> +}
[Severity: High]
Can spacemit_dpu_online_isr() complete a commit's vblank event a frame
early when INT_CFG_RDY and INT_VSYNC are both set in irq_raw?
As noted in saturn_hee_irq_enable() and spacemit_dpu_run(),
DPU_ONLINE_IRQ_STS latches status bits even while their mask bits in
DPU_ONLINE_IRQ_MSK are cleared, yet saturn_hee_enable_irq() unmasks
interrupts without clearing DPU_ONLINE_IRQ_STS first and
spacemit_dpu_online_isr() does not mask irq_raw against
DPU_ONLINE_IRQ_MSK. When INT_CFG_RDY fires while INT_VSYNC is also set
(either from a frame boundary or latched while vblank was masked), the
INT_CFG_RDY branch calls drm_crtc_arm_vblank_event() and the following
INT_VSYNC branch immediately calls drm_crtc_handle_vblank() in the same
ISR invocation. That delivers the flip event before the next vblank and
unblocks drm_atomic_helper_wait_for_flip_done() early, allowing the next
commit to overwrite the active command list slot.
> +
> +void spacemit_dpu_run(struct drm_crtc *crtc, struct drm_atomic_commit *state)
> +{
> + struct spacemit_crtc *a_crtc = to_spacemit_crtc(crtc);
> + struct spacemit_drm_private *priv = a_crtc->crtc.dev->dev_private;
> + struct spacemit_hw_device *hwdev = priv->hwdev;
> +
> + saturn_hee_conf_dpuctrl(crtc, state);
> +
> + /* flush all config register writes before CFG_RDY is raised */
> + mb();
> +
> + /* Ack first: one left by a commit that timed out is not this one's. */
> + saturn_hee_clr_int_sts(a_crtc, saturn_hee_get_irq_bit(INT_CFG_RDY));
> + saturn_hee_enable_irq(a_crtc, INT_CFG_RDY, true);
> +
> + mod_timer(&a_crtc->cfg_rdy_timer,
> + jiffies + msecs_to_jiffies(SATURN_CFG_RDY_TIMEOUT_MS));
> + saturn_hee_cfg_ready(a_crtc, hwdev);
> +
> + if (unlikely(a_crtc->is_1st_f)) {
> + drm_dbg(a_crtc->crtc.dev, "DPU start\n");
> + a_crtc->is_1st_f = false;
> + saturn_hee_sw_start(a_crtc, hwdev);
> + }
> +}
[Severity: Medium]
Can spacemit_cfg_rdy_timer_handler() race with spacemit_dpu_online_isr()
and spacemit_dpu_run() to complete the next commit's vblank event early or
leave INT_CFG_RDY masked?
CPU0 (<softirq>) CPU1
spacemit_cfg_rdy_timer_handler()
drm_err_ratelimited(...)
<hardirq>
spacemit_dpu_online_isr()
timer_delete(&a_crtc->cfg_rdy_timer)
drm_crtc_arm_vblank_event(...)
drm_crtc_handle_vblank(...)
spacemit_crtc_atomic_flush()
a_crtc->event = new_event
spacemit_dpu_run()
saturn_hee_enable_irq(..., true)
saturn_hee_enable_irq(..., false)
spacemit_dpu_complete_event()
drm_crtc_send_vblank_event() <-- completes new commit early
timer_delete() in spacemit_dpu_online_isr() does not wait for an in-flight
spacemit_cfg_rdy_timer_handler(), so the timer callback can mask the next
commit's INT_CFG_RDY and complete its a_crtc->event before the hardware
consumes its command lists.
In addition, spacemit_dpu_run() unmasks INT_CFG_RDY before mod_timer() and
saturn_hee_cfg_ready(). If a late CFG_RDY from a previously timed-out
commit arrives right after saturn_hee_enable_irq(a_crtc, INT_CFG_RDY,
true), spacemit_dpu_online_isr() runs timer_delete() before the timer is
armed, masks INT_CFG_RDY, and arms a_crtc->event before both_cfg_rdy is
written for the new commit.
> diff --git a/drivers/gpu/drm/spacemit/dpu/dpu_saturn_hee.c b/drivers/gpu/drm/spacemit/dpu/dpu_saturn_hee.c
> new file mode 100644
> index 0000000000000..1a194fe7cc7ba
> --- /dev/null
> +++ b/drivers/gpu/drm/spacemit/dpu/dpu_saturn_hee.c
> @@ -0,0 +1,670 @@
[ ... ]
> +void saturn_hee_enable_irq(struct spacemit_crtc *a_crtc,
> + enum spacemit_dpu_irq irq_id, bool enable)
> +{
> + saturn_enable_irq_mask(a_crtc, enable,
> + DPU_INT_BASE_ADDR + DPU_ONLINE_IRQ_MSK,
> + saturn_hee_get_irq_bit(irq_id));
> +}
> +
> +void saturn_hee_cfg_ready(struct spacemit_crtc *a_crtc,
> + struct spacemit_hw_device *hwdev)
> +{
> + u32 base = DPU_SCENE_CTL_ADDR(COMPOSER1);
> +
> + saturn_hee_enable_irq(a_crtc, INT_UNDERRUN, true);
> + a_crtc->ur_mask_armed = true;
> + dpu_write(hwdev, DPU_CTL_REG, base, both_cfg_rdy, 1);
> +}
[Severity: Medium]
Does enabling INT_UNDERRUN in saturn_hee_cfg_ready() before setting
a_crtc->ur_mask_armed and without clearing DPU_ONLINE_IRQ_STS race with
spacemit_dpu_online_isr()?
saturn_hee_cfg_ready()
saturn_hee_enable_irq(a_crtc, INT_UNDERRUN, true)
<hardirq>
spacemit_dpu_online_isr()
saturn_hee_clr_int_sts(a_crtc, ...)
if (... && a_crtc->ur_mask_armed) <-- still false; stays unmasked
a_crtc->ur_mask_armed = true
After an underrun, INT_UNDERRUN is masked for the rest of the frame while
the hardware continues latching DPU_INT_FRM_TIMING_UNFLOW in
DPU_ONLINE_IRQ_STS. When the next commit calls saturn_hee_cfg_ready(),
unmasking INT_UNDERRUN without clearing DPU_ONLINE_IRQ_STS immediately
triggers spacemit_dpu_online_isr(). If the interrupt arrives before
a_crtc->ur_mask_armed = true (which also happens in
saturn_hee_irq_enable()), saturn_hee_clr_int_sts() sees ur_mask_armed ==
false and leaves INT_UNDERRUN unmasked; if it arrives after
ur_mask_armed = true but before INT_CFG_RDY clears ur_reported, the bottom
saturn_hee_clr_int_sts() call in spacemit_dpu_online_isr() immediately
masks INT_UNDERRUN again before the new commit takes effect.
> +
> +void saturn_hee_sw_start(struct spacemit_crtc *a_crtc,
> + struct spacemit_hw_device *hwdev)
> +{
> + u32 base;
> +
> + base = DPU_SCENE_CTL_ADDR(COMPOSER1);
> + dpu_write_reg(hwdev, DPU_CTL_REG, base, sw_start, 1);
> +}
> +
> +static void saturn_init_tmg(struct spacemit_crtc *a_crtc,
> + const struct drm_display_mode *mode)
> +{
> + u32 base = TMG0_BASE_ADDR;
> + struct spacemit_drm_private *priv = a_crtc->crtc.dev->dev_private;
> + struct spacemit_hw_device *hwdev = priv->hwdev;
> + u16 vfp, vbp, vsync, hfp, hbp, hsync;
> + void __iomem *tmg_addr;
> + u32 value;
> +
> + tmg_addr = hwdev->base + TMG0_BASE_ADDR;
> +
> + hsync = mode->hsync_end - mode->hsync_start;
> + hbp = mode->htotal - mode->hsync_end;
> + hfp = mode->hsync_start - mode->hdisplay;
> + vsync = mode->vsync_end - mode->vsync_start;
> + vbp = mode->vtotal - mode->vsync_end;
> + vfp = mode->vsync_start - mode->vdisplay;
> + drm_dbg(a_crtc->crtc.dev, "timing " DRM_MODE_FMT "\n", DRM_MODE_ARG(mode));
> +
> + dpu_write(hwdev, TMG_REG, base, disp_ready_man_en, 0);
> + dpu_write(hwdev, TMG_REG, base, background_r, 0);
> + dpu_write(hwdev, TMG_REG, base, background_g, 0x0);
> + dpu_write(hwdev, TMG_REG, base, background_b, 0xff);
> + dpu_write(hwdev, TMG_REG, base, eof_1st_ln_dly_num, vfp - 7);
> + dpu_write(hwdev, TMG_REG, base, eof_2nd_ln_dly_num, vfp - 6);
> + dpu_write(hwdev, TMG_REG, base, split_en, 0);
> + dpu_write(hwdev, TMG_REG, base, cmd_screen, 0);
> + dpu_write(hwdev, TMG_REG, base, cmd_wait_en, 0);
> + dpu_write(hwdev, TMG_REG, base, cmd_wait_te, 0);
> + dpu_write(hwdev, TMG_REG, base, sof_pre_ln_num, 0);
> + dpu_write(hwdev, TMG_REG, base, hfp, hfp);
> + dpu_write(hwdev, TMG_REG, base, hbp, hbp);
> + dpu_write(hwdev, TMG_REG, base, vfp, vfp);
> + dpu_write(hwdev, TMG_REG, base, vbp, vbp);
> + dpu_write(hwdev, TMG_REG, base, hsync_width, hsync);
> + dpu_write(hwdev, TMG_REG, base, vsync_width, vsync);
> + dpu_write(hwdev, TMG_REG, base, hsp, 1);
> + dpu_write(hwdev, TMG_REG, base, vsp, 1);
> + dpu_write(hwdev, TMG_REG, base, h_active, mode->hdisplay);
> + dpu_write(hwdev, TMG_REG, base, v_active, mode->vdisplay);
> + dpu_write(hwdev, TMG_REG, base, fm_timing_en, 1);
> + /* The pipeline always hands RGB888 to the DP/eDP encoder. */
> + dpu_write(hwdev, TMG_REG, base, user, OUTFMT_RGB888);
> +
> + value = readl(tmg_addr + 4 * dpu_field_word(TMG_REG, vfp));
> + value &= ~dpu_field_mask(TMG_REG, vfp);
> + value |= vfp;
> + writel(value, tmg_addr + 4 * dpu_field_word(TMG_REG, vfp));
[Severity: High]
Does vfp - 7 and vfp - 6 underflow when the display mode's vertical front
porch is smaller than 7 lines?
Standard modes such as 1920x1080@60Hz (vdisplay = 1080, vsync_start =
1084, so vfp = 4) have vfp < 7, which wraps the 16-bit eof_1st_ln_dly_num
and eof_2nd_ln_dly_num fields to 0xfff9..0xffff, exceeding vtotal so the
EOF line-delay counters cannot fire within the frame. Also, the raw
readl()/writel() sequence at the end of saturn_init_tmg() redundantly
rewrites the same vfp field that dpu_write(hwdev, TMG_REG, base, vfp, vfp)
just programmed above.
--
Sashiko AI review · https://sashiko.dev/#/patchset/20261121-k3-display-v3-0-ec419e8e96b8@gmail.com?part=9
--
linux-phy mailing list
linux-phy@lists.infradead.org
https://lists.infradead.org/mailman/listinfo/linux-phy
next prev parent reply other threads:[~2026-11-21 8:01 UTC|newest]
Thread overview: 82+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-21 7:40 [PATCH v3 00/17] drm/spacemit: add SpacemiT K3 display support Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 01/17] dt-bindings: display: spacemit: add K3 Saturn DPU controller Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:44 ` sashiko-bot
2026-11-21 7:44 ` sashiko-bot
2026-11-21 7:40 ` [PATCH v3 02/17] dt-bindings: phy: add SpacemiT K3 Innosilicon DP PHY Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 03/17] dt-bindings: display: spacemit: add K3 Innosilicon DP/eDP controller Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:45 ` sashiko-bot
2026-11-21 7:45 ` sashiko-bot
2026-11-21 7:40 ` [PATCH v3 04/17] dt-bindings: soc: spacemit: allow eDP/DP PHY PLL pixel clocks on K3 APMU Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 05/17] phy: spacemit: add Innosilicon DP TX PHY driver Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 06/17] clk: spacemit: k3: parent eDP/DP pixel clock to the PHY PLL Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 07/17] drm/spacemit: add Saturn DPU register model Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 08/17] drm/spacemit: add Saturn DPU core types, cmdlist and display MMU Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 09/17] drm/spacemit: add Saturn DPU hardware backend Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 8:01 ` sashiko-bot [this message]
2026-11-21 8:01 ` sashiko-bot
2026-11-21 7:40 ` [PATCH v3 10/17] drm/spacemit: add Saturn DPU KMS pipeline Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 8:06 ` sashiko-bot
2026-11-21 8:06 ` sashiko-bot
2026-11-21 7:40 ` [PATCH v3 11/17] drm/spacemit: add Saturn DPU DRM device driver Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 12/17] drm/spacemit: add Innosilicon DP/eDP controller bridge driver Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 8:08 ` sashiko-bot
2026-11-21 8:08 ` sashiko-bot
2026-11-21 7:40 ` [PATCH v3 13/17] MAINTAINERS: add SpacemiT K3 display driver entry Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 14/17] riscv: dts: spacemit: k3: add display nodes Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 15/17] riscv: dts: spacemit: k3-pico-itx: enable the DisplayPort output Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 16/17] riscv: dts: spacemit: k3-com260-ifx: " Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` [PATCH v3 17/17] riscv: defconfig: spacemit: k3: enable display driver Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang via B4 Relay
2026-11-21 7:40 ` Cody Kang
2026-11-21 7:40 ` Cody Kang via B4 Relay
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=sashiko-outbox-226811@kernel.org \
--to=sashiko-bot@kernel.org \
--cc=codykang.hk@gmail.com \
--cc=conor+dt@kernel.org \
--cc=devicetree@vger.kernel.org \
--cc=dri-devel@lists.freedesktop.org \
--cc=linux-clk@vger.kernel.org \
--cc=linux-phy@lists.infradead.org \
--cc=neil.armstrong@linaro.org \
--cc=olteanv@gmail.com \
--cc=robh@kernel.org \
--cc=sashiko-reviews@lists.linux.dev \
--cc=vkoul@kernel.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.