drm: bridge: dw_hdmi: Misc enable/disable, CEC and EDID cleanup
Subsystems (12)
Jonas Karlman
reviewerv7 cover drm: bridge: dw_hdmi: Misc enable/disable, CEC and EDID cleanup
This is a revival of an old dw-hdmi series and is the first series part of a new effort to upstream old LibreELEC HDMI 2.0 patches for Rockchip RK33xx devices.
This series ensure poweron/poweroff and CEC phys addr invalidation is happening during normal DRM funcs, ensures EDID and CEC phys addr is updated in detect() similar to how the bridge connector works with a HDMI bridge attached, and also changes to debounce hotplug processing to prevent a full disable/enable cycle during a HPD low voltage pulse.
After this series HPD, EDID and CEC handling should work very similar regardless is the dw-hdmi connector or the bridge connector is used. It should also help ensure a smoother transition when dw-hdmi is fully converted into a HDMI bridge in a future series.
These changes have mainly been tested on Rockchip RK3328, RK3399 and RK3568 devices using both the dw-hdmi connector and also using a basic convert to use a bridge connector. The changes has also been tested on Amlogic S905X, S905Y2 and A311D devices that uses the bridge connector.
Testing with a Rock Pi 4 (RK3399) using a Reaspberry Pi Monitor with Linux kms client console using drm.debug=0xe should log something like following:
Power cycle monitor using the power button:
[CONNECTOR:68:HDMI-A-1] CEA VCDB 0x4a
[CONNECTOR:68:HDMI-A-1] HDMI: DVI dual 0, max TMDS clock 0 kHz [CONNECTOR:68:HDMI-A-1] ELD monitor RPI MON156
[CONNECTOR:68:HDMI-A-1] HDMI: latency present 0 0, video latency 0 0, audio latency 0 0 [CONNECTOR:68:HDMI-A-1] ELD size 36, SAD count 1
[CONNECTOR:68:HDMI-A-1] Same epoch counter 10
Cable unplugged:
[CONNECTOR:68:HDMI-A-1] EDID changed, epoch counter 11 [CONNECTOR:68:HDMI-A-1] status updated from connected to disconnected [CONNECTOR:68:HDMI-A-1] Changed epoch counter 10 => 12 [CONNECTOR:68:HDMI-A-1] generating connector hotplug event [CONNECTOR:68:HDMI-A-1] Sent hotplug event
Cable connected:
[CONNECTOR:68:HDMI-A-1] CEA VCDB 0x4a
[CONNECTOR:68:HDMI-A-1] HDMI: DVI dual 0, max TMDS clock 0 kHz [CONNECTOR:68:HDMI-A-1] ELD monitor RPI MON156
[CONNECTOR:68:HDMI-A-1] HDMI: latency present 0 0, video latency 0 0, audio latency 0 0 [CONNECTOR:68:HDMI-A-1] ELD size 36, SAD count 1
[CONNECTOR:68:HDMI-A-1] status updated from disconnected to connected [CONNECTOR:68:HDMI-A-1] Changed epoch counter 12 => 13 [CONNECTOR:68:HDMI-A-1] generating connector hotplug event [CONNECTOR:68:HDMI-A-1] Sent hotplug event
This series has evolved into an initial part of a larger multi series
effort to:
- drm: bridge: dw_hdmi: Misc enable/disable, CEC and EDID cleanup [v7]
- drm/bridge: dw-hdmi: Improve input/output bus format handling
- drm/bridge: dw-hdmi: Convert to a HDMI bridge and use of bridge connector
- drm/bridge: dw-hdmi: Add and use tmds_char_rate_valid() plat data ops
- drm/meson: hdmi: Misc cleanup and use CEC notifier helpers [v1]
- phy: rockchip: inno-hdmi: Change TMDS rate handling to configure() ops [v4]
- drm/rockchip: dw_hdmi: Misc cleanup and propagate bus format [v2]
- drm/rockchip: dw_hdmi: Enable YCbCr and Deep Color modes
Link to snapshot: https://github.com/Kwiboo/linux-rockchip/commits/next-20260518-rk-hdmi-v5/
Changes in v7:
- Add patch to fix use-after-free when CEC adapter is unregistered
- Only declare CEC notifier support when CEC device register succeeds
- Re-order patches to drop drm_edid_raw() use before exposing more usage
- Change to free irq before mute and clear using IH regs
- Rebased on next-20260518
Link to v6: https://lore.kernel.org/dri-devel/[email protected]/
Changes in v6:
- Update EDID and CEC phys addr in the bridge detect() func
- Add CEC notifier bridge op for the bridge connector
- Change back to disable_delayed_work_sync() in hpd disable ops,
a possible deadlock is avoided by not using drm_bridge_hpd_notify()
- Drop use of a suspend helper now that hpd disable ops use sync() calls
- Ensure HPD interrupt is masked and IRQ handler is disabled early
in dw_hdmi_remove() to prevent any irq re-arming of delayed work
- Update a few commit messages and cover letter
- Collect t-b tags
Link to v5: https://lore.kernel.org/dri-devel/[email protected]/
Changes in v5:
- Add patch that holds a bridge ref until connector cleanup, to fix
a use-after-free issue during connector cleanup
- Add patch that unregister CEC notifier during connector cleanup
- Add patch that adds a common suspend helper
- Add patch that drops call to drm_bridge_hpd_notify()
- Collect r-b tag
- Rebased on next-20260508
Link to v4: https://lore.kernel.org/dri-devel/[email protected]/
Changes in v4:
- Change to use generic CEC notifier helpers
- Disable/mask hpd_work until enable_hpd()/hpd_enable()
- Read connector status directly from HW regs in hpd_work
- Continued rework of HDP and RXSENSE interrupt handling
- Collect r-b tags
- Rebased on next-20260430
Link to v3: https://lore.kernel.org/dri-devel/[email protected]/
Changes in v3:
- Rework EDID refresh handling to closer match bridge connector
- Use delayed work to debounce HPD processing
- Update commit messages
- Collect r-b tags
- Rebased on next-20260401
Link to v2: https://lore.kernel.org/dri-devel/[email protected]/
Changes in v2:
- Add patch to disable scrambler feature when not supported
- Add patch to only notify connected status on HPD interrupt
- Update commit messages
- Collect r-b tags
- Rebased on next-20240906
Link to v1: https://lore.kernel.org/dri-devel/[email protected]/
Jonas Karlman (23):
drm: bridge: dw_hdmi: Disable scrambler feature when not supported
drm: bridge: dw_hdmi: Only notify connected status on HPD interrupt
drm: bridge: dw_hdmi: Free IRQ before CEC adapter is unregistered
drm: bridge: dw_hdmi: Hold bridge ref until connector cleanup
drm: bridge: dw_hdmi: Call poweron/poweroff from atomic enable/disable
drm: bridge: dw_hdmi: Use passed mode instead of stored previous_mode
drm: bridge: dw_hdmi: Fold poweron and setup functions
drm: bridge: dw_hdmi: Remove previous_mode and mode_set
drm: bridge: dw_hdmi: Unregister CEC notifier during connector cleanup
drm: bridge: dw_hdmi: Invalidate CEC phys addr from connector detect
drm: bridge: dw_hdmi: Remove cec_notifier_mutex
drm: bridge: dw_hdmi: Extract dw_hdmi_connector_status_update()
drm: bridge: dw_hdmi: Use dw_hdmi_connector_status_update()
drm: bridge: dw_hdmi: Use display_info is_hdmi and has_audio
drm: bridge: dw_hdmi: Use generic CEC notifier helpers
drm: bridge: dw_hdmi: Update EDID and CEC phys addr in bridge detect()
drm: bridge: dw_hdmi: Declare bridge CEC notifier support
drm: bridge: dw_hdmi: Drop call to drm_bridge_hpd_notify()
drm: bridge: dw_hdmi: Use delayed_work to debounce hotplug event
drm: bridge: dw_hdmi: Rework HDP and RXSENSE interrupt handling
drm: bridge: dw_hdmi: Remove the empty dw_hdmi_setup_rx_sense()
drm: bridge: dw_hdmi: Remove the empty dw_hdmi_phy_update_hpd()
drm: bridge: dw_hdmi: Merge top and bottom half IRQ handlers
drivers/gpu/drm/bridge/imx/imx8mp-hdmi-tx.c | 1 - drivers/gpu/drm/bridge/synopsys/Kconfig | 1 + drivers/gpu/drm/bridge/synopsys/dw-hdmi-cec.c | 1 + drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 502 +++++++----------- drivers/gpu/drm/meson/meson_dw_hdmi.c | 3 - drivers/gpu/drm/rockchip/dw_hdmi-rockchip.c | 2 - drivers/gpu/drm/sun4i/sun8i_hdmi_phy.c | 2 - include/drm/bridge/dw_hdmi.h | 6 - 8 files changed, 192 insertions(+), 326 deletions(-)
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 1/23 drm: bridge: dw_hdmi: Disable scrambler feature when not supported
The scrambler feature can be left enabled when hotplugging from a sink and mode that require scrambling to a sink that does not support SCDC or scrambling.
Typically a blank screen or 'no signal' message can be observed after using a HDMI 2.0 4K@60Hz mode and then hotplugging to a sink that only support HDMI 1.4.
Fix this by disabling the scrambler feature when SCDC is not supported.
---
v7: No change
v6: Collect t-b tag
v5: No change
v4: No change
v3: Collect r-b tag
v2: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 2 ++ 1 file changed, 2 insertions(+)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +2 -0
@@ -2135,6 +2135,8 @@ static void hdmi_av_composer(struct dw_hdmi *hdmi, HDMI_MC_SWRSTZ);drm_scdc_set_scrambling (hdmi->curr_conn,0 ); } + }else if (hdmi->version >=0x200a ) { +hdmi_writeb (hdmi,0 ,HDMI_FC_SCRAMBLER_CTRL ); }/* Set up horizontal active pixel width */
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 2/23 drm: bridge: dw_hdmi: Only notify connected status on HPD interrupt
In reply to Jonas Karlmandrm_helper_hpd_irq_event() and drm_bridge_hpd_notify() may incorrectly be called with a connected status when HPD is high and RX sense is changed. This typically happens when the HDMI cable is unplugged, shortly before the HPD is changed to low.
The original intent of commit da09daf88108 ("drm: bridge: dw_hdmi: only trigger hotplug event on link change") was to signal hotplug event at correct interrupt states.
Based on the commit message the intent was to trigger hotplug event:
- when HPD goes high (plugin)
- when both HPD and RX sense has gone low (plugout)
However, following interrupt state changes can typically be observed when the HDMI cable is unplugged:
- RX interrupt: HPD=high RX=low -> triggers a connected event
- HPD interrupt: HPD=low RX=low -> triggers a disconnected event
Fix this by only notify connected status on the HPD interrupt when HPD is going high, not on the RX sense interrupt when RX sense is changed.
After this a connected event should be triggered when HPD=high at HPD interrupt, and a disconnected event should be triggered when both HPD=low and RX=low at either HPD or RX interrupt.
---
v7: No change
v6: Collect t-b tag
v5: No change
v4: Collect r-b tag
v3: Update commit message
v2: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +2 -1
@@ -3157,7 +3157,8 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id)mutex_unlock (&hdmi->cec_notifier_mutex); } -if (phy_stat &HDMI_PHY_HPD ) +if ((intr_stat &HDMI_IH_PHY_STAT0_HPD ) && + (phy_stat &HDMI_PHY_HPD )) status = connector_status_connected;if (!(phy_stat & (HDMI_PHY_HPD |HDMI_PHY_RX_SENSE )))
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 3/23 drm: bridge: dw_hdmi: Free IRQ before CEC adapter is unregistered
In reply to Jonas KarlmanThe interrupt allocated with devm_request_threaded_irq() can be use-after-free when the devres release action try to free_irq().
KASAN report a slab-use-after-free in dw_hdmi_cec_hardirq during unbind:
Call trace: [...] dw_hdmi_cec_hardirq+0x4cc/0x560 free_irq+0x48c/0x7e4 devm_irq_release+0x54/0x90 dr_node_release+0x38/0x5c release_nodes+0xac/0x130 devres_release_all+0xf4/0x1b0 device_unbind_cleanup+0x28/0x1f8 device_release_driver_internal+0x358/0x470 device_release_driver+0x18/0x24 bus_remove_device+0x33c/0x4f0 device_del+0x2d8/0x790 platform_device_del+0x34/0x1e0 platform_device_unregister+0x14/0x3c dw_hdmi_remove+0x74/0x180 [...]
Freed by:
[...]
kfree+0x1dc/0x5dc cec_delete_adapter+0xd4/0x118 cec_devnode_release+0xa4/0xe0 device_release+0xa0/0x200 kobject_put+0x14c/0x26c put_device+0x14/0x30 cec_unregister_adapter+0x20c/0x280 dw_hdmi_cec_remove+0x8c/0xd0 [...]
Explicitly devm_free_irq() before the CEC adapter is unregistered to fix this possible use-after-free issue.
---
v7: New patch
KASAN report a slab-use-after-free in dw_hdmi_cec_hardirq when,
echo fe0a0000.hdmi > /sys/bus/platform/drivers/dwhdmi-rockchip/unbind
on a Rockchip RK3566 device prior to this fix.
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi-cec.c | 1 + 1 file changed, 1 insertion(+)
drivers/gpu/drm/bridge/synopsys/dw-hdmi-cec.c +1 -0
@@ -309,6 +309,7 @@ static void dw_hdmi_cec_remove(struct platform_device *pdev)struct dw_hdmi_cec *cec =platform_get_drvdata (pdev);cec_notifier_cec_adap_unregister (cec->notify, cec->adap); +devm_free_irq (&pdev->dev, cec->irq, cec->adap);cec_unregister_adapter (cec->adap); }
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v7 03/23] drm: bridge: dw_hdmi: Free IRQ before CEC adapter is unregistered
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] Freeing the shared IRQ before masking the hardware interrupt source creates a race condition that…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Freeing the shared IRQ before masking the hardware interrupt source creates a race condition that can lead to an unhandled interrupt storm, permanently disabling the shared IRQ line. --
commit 860a0518c5583dda606c670b609794b60e6bc07b
Author: Jonas Karlman <[email protected]>
drm: bridge: dw_hdmi: Free IRQ before CEC adapter is unregistered
This commit explicitly frees the IRQ before the CEC adapter is unregistered to prevent a use-after-free issue. This addresses a KASAN slab-use-after-free report triggered during device unbind when the devres release action frees the IRQ.
drivers/gpu/drm/bridge/synopsys/dw-hdmi-cec.c static void dw_hdmi_cec_remove(struct platform_device *pdev) · line 312
Quoted 9 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi-cec.c +1
-0@@ -309,6 +309,7 @@ static void dw_hdmi_cec_remove(struct platform_device *pdev)struct dw_hdmi_cec *cec =platform_get_drvdata (pdev);cec_notifier_cec_adap_unregister (cec->notify, cec->adap); +devm_free_irq (&pdev->dev, cec->irq, cec->adap);
Does this sequence create a race condition that could result in an unhandled interrupt storm?
The dw-hdmi-cec driver shares this IRQ line with the main dw-hdmi driver. By freeing the IRQ handler before calling cec_unregister_adapter(), the hardware CEC interrupts might still be unmasked and active.
If the CEC hardware asserts an interrupt during this window, the shared IRQ line will go high. The remaining handlers on that line won't clear the CEC interrupt, returning IRQ_NONE.
This could lead to an interrupt storm, causing the kernel to permanently disable the shared IRQ line and break the main dw-hdmi driver's functionality.
Should the CEC hardware interrupts be explicitly masked before devm_free_irq() is called?
Quoted 2 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi-cec.c +0
-0cec_unregister_adapter (cec->adap); }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=3
Re: [PATCH v7 03/23] drm: bridge: dw_hdmi: Free IRQ before CEC adapter is unregistered
In reply to Jonas KarlmanQuoted 35 lines · Jonas Karlman
The interrupt allocated with
devm_request_threaded_irq()can be use-after-free when the devres release action try tofree_irq().KASAN report a slab-use-after-free in
dw_hdmi_cec_hardirqduring unbind:Call trace: [...] dw_hdmi_cec_hardirq+0x4cc/0x560 free_irq+0x48c/0x7e4 devm_irq_release+0x54/0x90 dr_node_release+0x38/0x5c release_nodes+0xac/0x130 devres_release_all+0xf4/0x1b0 device_unbind_cleanup+0x28/0x1f8 device_release_driver_internal+0x358/0x470 device_release_driver+0x18/0x24 bus_remove_device+0x33c/0x4f0 device_del+0x2d8/0x790 platform_device_del+0x34/0x1e0 platform_device_unregister+0x14/0x3c dw_hdmi_remove+0x74/0x180 [...]Freed by:
[...]kfree+0x1dc/0x5dc cec_delete_adapter+0xd4/0x118 cec_devnode_release+0xa4/0xe0 device_release+0xa0/0x200 kobject_put+0x14c/0x26c put_device+0x14/0x30 cec_unregister_adapter+0x20c/0x280 dw_hdmi_cec_remove+0x8c/0xd0 [...]
Quoted 5 lines
Explicitly
devm_free_irq()before the CEC adapter is unregistered to fix this possible use-after-free issue.Fixes a616e63c56ef ("drm/bridge: dw-hdmi: add cec driver")Signed-off-by Jonas Karlman <[email protected]>
Regards,
Hans
Quoted 22 lines
---
v7: New patchKASAN report a slab-use-after-free in
dw_hdmi_cec_hardirqwhen,
echo fe0a0000.hdmi >/sys/bus/platform/drivers/dwhdmi-rockchip/unbind
on a Rockchip RK3566 device prior to this fix.--- drivers/gpu/drm/bridge/synopsys/dw-hdmi-cec.c | 1 + 1 file changed, 1 insertion(+)drivers/gpu/drm/bridge/synopsys/dw-hdmi-cec.c +1
-0@@ -309,6 +309,7 @@ static void dw_hdmi_cec_remove(struct platform_device *pdev)struct dw_hdmi_cec *cec =platform_get_drvdata (pdev);cec_notifier_cec_adap_unregister (cec->notify, cec->adap); +devm_free_irq (&pdev->dev, cec->irq, cec->adap);cec_unregister_adapter (cec->adap); }
Jonas Karlman
reviewerPATCH v7 4/23 drm: bridge: dw_hdmi: Hold bridge ref until connector cleanup
In reply to Jonas Karlmandrmres connector cleanup typically run after devres has released the last dw-hdmi bridge reference. Since struct dw_hdmi, where the connector lives, is freed when the last bridge reference is released, connector cleanup can end up accessing freed memory.
Call trace without a bridge reference held until connector cleanup:
- dw_hdmi_bridge_detach()
- dw_hdmi_bridge_destroy() <<-- struct dw_hdmi is free()
- [drm:drm_managed_release] drmres release begin
- [drm:drm_managed_release] REL (...) drm_mode_config_init_release (0 bytes)
- dw_hdmi_connector_destroy()
- drm_connector_cleanup() <<-- drm_connector is use-after-free
[...]
- [drm:drm_managed_release] drmres release end
Hold a bridge reference for as long as the connector exists and drop it after drm_connector_cleanup() has completed to keep struct dw_hdmi alive until connector teardown is finished and avoids the use-after-free.
Call trace with a bridge reference held until connector cleanup:
- dw_hdmi_bridge_detach()
- [drm:drm_managed_release] drmres release begin
- [drm:drm_managed_release] REL (...) drm_mode_config_init_release (0 bytes)
- dw_hdmi_connector_destroy()
- drm_connector_cleanup() <<-- drm_connector is destroy()
- drm_bridge_put()
- dw_hdmi_bridge_destroy() <<-- struct dw_hdmi is free()
[...]
- [drm:drm_managed_release] drmres release end
---
v7: Add fixes tag, re-order patch
v6: Collect t-b tag
v5: New patch
This use-after-free issue likely existed before commit ed6987b67418 when devm_kzalloc() was used instead of devm_drm_bridge_alloc(). However, v6.16-rc1 first introduced bridge refcount and drm_bridge_put(), parts that are used to help fix the use-after-free issue.
KASAN report a slab-use-after-free in __refcount_add_not_zero when,
echo fe0a0000.hdmi > /sys/bus/platform/drivers/dwhdmi-rockchip/unbind
on a Rockchip RK3566 device prior to this fix.
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 23 ++++++++++++++++++----- 1 file changed, 18 insertions(+), 5 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +18 -5
@@ -2568,10 +2568,18 @@ static void dw_hdmi_connector_force(struct drm_connector *connector)mutex_unlock (&hdmi->mutex); } +static void dw_hdmi_connector_destroy (struct drm_connector *connector) +{ + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); + +drm_connector_cleanup (connector); +drm_bridge_put (&hdmi->bridge); +} +static const struct drm_connector_funcs dw_hdmi_connector_funcs = { .fill_modes = drm_helper_probe_single_connector_modes, .detect = dw_hdmi_connector_detect, - .destroy = drm_connector_cleanup, + .destroy = dw_hdmi_connector_destroy, .force = dw_hdmi_connector_force, .reset = drm_atomic_helper_connector_reset, .atomic_duplicate_state = drm_atomic_helper_connector_duplicate_state,
@@ -2588,6 +2596,7 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)struct drm_connector *connector = &hdmi->connector;struct cec_connector_info conn_info;struct cec_notifier *notifier; +int ret;if (hdmi->version >=0x200a ) connector->ycbcr_420_allowed =
@@ -2600,10 +2609,14 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)drm_connector_helper_add (connector, &dw_hdmi_connector_helper_funcs); -drm_connector_init_with_ddc (hdmi->bridge.dev, connector, - &dw_hdmi_connector_funcs, -DRM_MODE_CONNECTOR_HDMIA , - hdmi->ddc); + ret =drm_connector_init_with_ddc (hdmi->bridge.dev, connector, + &dw_hdmi_connector_funcs, +DRM_MODE_CONNECTOR_HDMIA , + hdmi->ddc); +if (ret) +return ret; + +drm_bridge_get (&hdmi->bridge);/* * drm_connector_attach_max_bpc_property() requires the
Signature · 1 line
2.54.0
Luca Ceresoli
reviewerRe: [PATCH v7 04/23] drm: bridge: dw_hdmi: Hold bridge ref until connector cleanup
On Mon, 18 May 2026 18:01:40 +0000, Jonas Karlman <[email protected]> wrote:
Hello Jonas,
Quoted 11 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-2@@ -2600,10 +2609,14 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)drm_connector_helper_add (connector, &dw_hdmi_connector_helper_funcs); - drm_connector_init_with_ddc(hdmi->bridge.dev, connector, - &dw_hdmi_connector_funcs,
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static int dw_hdmi_connector_create(struct dw_hdmi *hdmi) · line 2619
Quoted 10 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +8
-2-DRM_MODE_CONNECTOR_HDMIA , - hdmi->ddc); + ret =drm_connector_init_with_ddc (hdmi->bridge.dev, connector, + &dw_hdmi_connector_funcs, +DRM_MODE_CONNECTOR_HDMIA , + hdmi->ddc); +if (ret) +return ret; + +drm_bridge_get (&hdmi->bridge);
I'm not fully following the code paths, but both the report and the fix make sense to me. Only I'd move the drm_bridge_get() before drm_connector_init_with_ddc(), to avoid a short window where no reference is held and the bridge might be destroyed before drm_bridge_get() is called. I'm not sure this can happen, but it's better to write the code in a way that clearly makes it impossible.
Otherwise looks good.
Luca
Signature · 3 lines
Luca Ceresoli, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
Jonas Karlman
reviewerRe: [PATCH v7 04/23] drm: bridge: dw_hdmi: Hold bridge ref until connector cleanup
Hello Luca,
Quoted 24 lines · Luca Ceresoli
On Mon, 18 May 2026 18:01:40 +0000, Jonas Karlman <[email protected]> wrote:
Hello Jonas,
Quoted 20 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +7
-4@@ -2600,10 +2609,14 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)drm_connector_helper_add (connector, &dw_hdmi_connector_helper_funcs); -drm_connector_init_with_ddc (hdmi->bridge.dev, connector, - &dw_hdmi_connector_funcs, -DRM_MODE_CONNECTOR_HDMIA , - hdmi->ddc); + ret =drm_connector_init_with_ddc (hdmi->bridge.dev, connector, + &dw_hdmi_connector_funcs, +DRM_MODE_CONNECTOR_HDMIA , + hdmi->ddc); +if (ret) +return ret; +
Quoted 8 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static int dw_hdmi_connector_create(struct dw_hdmi *hdmi) · line 2619
Quoted 1 line
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +1
-0+drm_bridge_get (&hdmi->bridge);I'm not fully following the code paths, but both the report and the fix make sense to me. Only I'd move the
drm_bridge_get()beforedrm_connector_init_with_ddc(), to avoid a short window where no reference is held and the bridge might be destroyed beforedrm_bridge_get()is called. I'm not sure this can happen, but it's better to write the code in a way that clearly makes it impossible.
dw_hdmi_connector_create() is only called from dw_hdmi_bridge_attach() so the bridge should already have a ref for the lifetime of this call.
I explicitly chose the placement after drm_connector_init_with_ddc() to ensure ref count is correctly balanced without having to add a drm_bridge_put() call in any error path. I.e. connector destroy() is only called when drm_connector_init_with_ddc() succeeds.
This code/call is also planned to be removed in a future series, so I do not think moving drm_bridge_get() is necessary unless you think we need to protect against possible bad behavior from DRM core?
Regards,
Jonas
Quoted 5 lines
Otherwise looks good.
Luca
Re: [PATCH v7 04/23] drm: bridge: dw_hdmi: Hold bridge ref until connector cleanup
Hello Jonas,
Quoted 36 lines · Jonas Karlman
Hello Luca,
Quoted 24 lines · Luca Ceresoli
On Mon, 18 May 2026 18:01:40 +0000, Jonas Karlman <[email protected]> wrote:
Hello Jonas,
Quoted 20 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +7
-4@@ -2600,10 +2609,14 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)drm_connector_helper_add (connector, &dw_hdmi_connector_helper_funcs); -drm_connector_init_with_ddc (hdmi->bridge.dev, connector, - &dw_hdmi_connector_funcs, -DRM_MODE_CONNECTOR_HDMIA , - hdmi->ddc); + ret =drm_connector_init_with_ddc (hdmi->bridge.dev, connector, + &dw_hdmi_connector_funcs, +DRM_MODE_CONNECTOR_HDMIA , + hdmi->ddc); +if (ret) +return ret; +Quoted 8 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static int dw_hdmi_connector_create(struct dw_hdmi *hdmi) · line 2619
Quoted 1 line
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +1
-0+drm_bridge_get (&hdmi->bridge);I'm not fully following the code paths, but both the report and the fix make sense to me. Only I'd move the
drm_bridge_get()beforedrm_connector_init_with_ddc(), to avoid a short window where no reference is held and the bridge might be destroyed beforedrm_bridge_get()is called. I'm not sure this can happen, but it's better to write the code in a way that clearly makes it impossible.
Quoted 2 lines
dw_hdmi_connector_create()is only called fromdw_hdmi_bridge_attach()so the bridge should already have a ref for the lifetime of this call.
Ah, that's true. So the patch is correct.
In case you send a new iteration, please add this extra explanation to the commit message, similar to the above paragraph.
Quoted 6 lines
I explicitly chose the placement after
drm_connector_init_with_ddc()to ensure ref count is correctly balanced without having to add adrm_bridge_put()call in any error path. I.e. connectordestroy()is only called whendrm_connector_init_with_ddc()succeeds.This code/call is also planned to be removed in a future series,
In order to remove the !DRM_BRIDGE_ATTACH_NO_CONNECTOR case? That would be welcome!
Luca
Jonas Karlman
reviewerRe: [PATCH v7 04/23] drm: bridge: dw_hdmi: Hold bridge ref until connector cleanup
Hello Luca,
Quoted 39 lines · Luca Ceresoli
Hello Jonas,
Quoted 36 lines · Jonas Karlman
Hello Luca,
Quoted 32 lines · Luca Ceresoli
On Mon, 18 May 2026 18:01:40 +0000, Jonas Karlman <[email protected]> wrote:
Hello Jonas,
Quoted 11 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-2@@ -2600,10 +2609,14 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)drm_connector_helper_add (connector, &dw_hdmi_connector_helper_funcs); - drm_connector_init_with_ddc(hdmi->bridge.dev, connector, - &dw_hdmi_connector_funcs,drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static int dw_hdmi_connector_create(struct dw_hdmi *hdmi) · line 2619
Quoted 10 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +8
-2-DRM_MODE_CONNECTOR_HDMIA , - hdmi->ddc); + ret =drm_connector_init_with_ddc (hdmi->bridge.dev, connector, + &dw_hdmi_connector_funcs, +DRM_MODE_CONNECTOR_HDMIA , + hdmi->ddc); +if (ret) +return ret; + +drm_bridge_get (&hdmi->bridge);I'm not fully following the code paths, but both the report and the fix make sense to me. Only I'd move the
drm_bridge_get()beforedrm_connector_init_with_ddc(), to avoid a short window where no reference is held and the bridge might be destroyed beforedrm_bridge_get()is called. I'm not sure this can happen, but it's better to write the code in a way that clearly makes it impossible.
Quoted 6 lines
Quoted 2 lines
dw_hdmi_connector_create()is only called fromdw_hdmi_bridge_attach()so the bridge should already have a ref for the lifetime of this call.Ah, that's true. So the patch is correct.
Reviewed-by Luca Ceresoli <[email protected]>
Thanks.
Quoted 2 lines
In case you send a new iteration, please add this extra explanation to the commit message, similar to the above paragraph.
Sure, I will include a note about this if/when I need to re-spin.
Quoted 9 lines
Quoted 6 lines
I explicitly chose the placement after
drm_connector_init_with_ddc()to ensure ref count is correctly balanced without having to add adrm_bridge_put()call in any error path. I.e. connectordestroy()is only called whendrm_connector_init_with_ddc()succeeds.This code/call is also planned to be removed in a future series,
In order to remove the !DRM_BRIDGE_ATTACH_NO_CONNECTOR case? That would be welcome!
That would be an end goal, however initial plan/step was to just change to use drm_bridge_connector_init() inside this driver [1] or possible move that to the consuming driver (imx6, rockchip and sun8i), in an unpolished future series [2].
Fully change to use ATTACH_NO_CONNECTOR for those affected drivers may possibly be pushed to a follow-up future series.
Main end goal of my current effort is to enable support for Deep Color and YCbCr output modes on Rockchip RK32xx/RK33xx/RK356x devices [3].
[1] https://github.com/Kwiboo/linux-rockchip/commit/813b55961e5a8fa864ea157e2793e76ca4967bac
[2] https://github.com/Kwiboo/linux-rockchip/compare/7e9084cc75011ce28b1ceafec804091438eed1ff...3b5507aa260eb8306554c34a0c362e514ea41c3b
[3] https://github.com/Kwiboo/linux-rockchip/commits/next-20260518-rk-hdmi-v5/
Regards,
Jonas
Quoted 3 lines
Luca
Jonas Karlman
reviewerPATCH v7 5/23 drm: bridge: dw_hdmi: Call poweron/poweroff from atomic enable/disable
In reply to Jonas KarlmanChange to only call poweron/poweroff from atomic_enable/atomic_disable funcs instead of trying to be clever by keeping a bridge_is_on state and poweron/off in the hotplug IRQ handler.
The bridge is already enabled/disabled depending on connection status with the call to drm_helper_hpd_irq_event() in hotplug IRQ handler.
---
v7: No change
v6: Update commit message,
Collect t-b tag
v5: No change
v4: No change
v3: Collect r-b tag
v2: Update commit message
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 33 ++--------------------- 1 file changed, 2 insertions(+), 31 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +2 -31
@@ -171,7 +171,6 @@ struct dw_hdmi {enum drm_connector_force force;/* mutex-protected force state */ struct drm_connector *curr_conn;/* current connector (only valid when !disabled) */ bool disabled;/* DRM has disabled our bridge */ -bool bridge_is_on;/* indicates the bridge is on */ bool rxsense;/* rxsense state */ u8 phy_mask;/* desired phy int mask settings */ u8 mc_clkdis;/* clock disable register */
@@ -2400,8 +2399,6 @@ static void initialize_hdmi_ih_mutes(struct dw_hdmi *hdmi)static void dw_hdmi_poweron (struct dw_hdmi *hdmi) { - hdmi->bridge_is_on = true; -/* * The curr_conn field is guaranteed to be valid here, as this function * is only be called when !hdmi->disabled.
@@ -2415,30 +2412,6 @@ static void dw_hdmi_poweroff(struct dw_hdmi *hdmi) hdmi->phy.ops->disable (hdmi, hdmi->phy.data); hdmi->phy.enabled = false; } - - hdmi->bridge_is_on = false; -} - -static void dw_hdmi_update_power (struct dw_hdmi *hdmi) -{ -int force = hdmi->force; - -if (hdmi->disabled) { - force =DRM_FORCE_OFF ; - }else if (force ==DRM_FORCE_UNSPECIFIED ) { -if (hdmi->rxsense) - force =DRM_FORCE_ON ; -else - force =DRM_FORCE_OFF ; - } - -if (force ==DRM_FORCE_OFF ) { -if (hdmi->bridge_is_on) -dw_hdmi_poweroff (hdmi); - }else { -if (!hdmi->bridge_is_on) -dw_hdmi_poweron (hdmi); - } }/*
@@ -2563,7 +2536,6 @@ static void dw_hdmi_connector_force(struct drm_connector *connector)mutex_lock (&hdmi->mutex); hdmi->force = connector->force; -dw_hdmi_update_power (hdmi);dw_hdmi_update_phy_mask (hdmi);mutex_unlock (&hdmi->mutex); }
@@ -3001,7 +2973,7 @@ static void dw_hdmi_bridge_atomic_disable(struct drm_bridge *bridge,mutex_lock (&hdmi->mutex); hdmi->disabled = true; hdmi->curr_conn =NULL ; -dw_hdmi_update_power (hdmi); +dw_hdmi_poweroff (hdmi);dw_hdmi_update_phy_mask (hdmi);handle_plugged_change (hdmi, false);mutex_unlock (&hdmi->mutex);
@@ -3019,7 +2991,7 @@ static void dw_hdmi_bridge_atomic_enable(struct drm_bridge *bridge,mutex_lock (&hdmi->mutex); hdmi->disabled = false; hdmi->curr_conn = connector; -dw_hdmi_update_power (hdmi); +dw_hdmi_poweron (hdmi);dw_hdmi_update_phy_mask (hdmi);handle_plugged_change (hdmi, true);mutex_unlock (&hdmi->mutex);
@@ -3119,7 +3091,6 @@ void dw_hdmi_setup_rx_sense(struct dw_hdmi *hdmi, bool hpd, bool rx_sense)if (hpd) hdmi->rxsense = true; -dw_hdmi_update_power (hdmi);dw_hdmi_update_phy_mask (hdmi); }mutex_unlock (&hdmi->mutex);
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 6/23 drm: bridge: dw_hdmi: Use passed mode instead of stored previous_mode
In reply to Jonas KarlmanUse the passed mode instead of mixing use of passed mode and the stored previous_mode in dw_hdmi_setup(). The passed mode is currenly always the previous_mode.
Also fix a small typo and add a variable to help shorten a code line.
---
v7: No change
v6: Collect t-b tag
v5: No change
v4: No change
v3: Collect r-b tag
v2: Update commit message, s/type/typo/
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +4 -5
@@ -2258,6 +2258,7 @@ static int dw_hdmi_setup(struct dw_hdmi *hdmi,const struct drm_connector *connector, const struct drm_display_mode *mode) { + const struct drm_display_info *display = &connector->display_info;int ret;hdmi_disable_overflow_interrupts (hdmi);
@@ -2303,12 +2304,10 @@ static int dw_hdmi_setup(struct dw_hdmi *hdmi, hdmi->hdmi_data.video_mode.mdataenablepolarity = true;/* HDMI Initialization Step B.1 */ -hdmi_av_composer (hdmi, &connector->display_info, mode); +hdmi_av_composer (hdmi, display, mode); -/* HDMI Initializateion Step B.2 */ - ret = hdmi->phy.ops->init (hdmi, hdmi->phy.data, - &connector->display_info, - &hdmi->previous_mode); +/* HDMI Initialization Step B.2 */ + ret = hdmi->phy.ops->init (hdmi, hdmi->phy.data, display, mode);if (ret)return ret; hdmi->phy.enabled = true;
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 7/23 drm: bridge: dw_hdmi: Fold poweron and setup functions
In reply to Jonas KarlmanFold the poweron and setup functions into one function and use the adjusted_mode directly from the new crtc_state to remove the need of storing previous_mode.
---
v7: No change
v6: Collect t-b tag
v5: No change, rebase on next-20260508
v4: No change
v3: Collect r-b tag
v2: No change
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 21 ++++++++------------- 1 file changed, 8 insertions(+), 13 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +8 -13
@@ -2254,9 +2254,9 @@ static void hdmi_disable_overflow_interrupts(struct dw_hdmi *hdmi) HDMI_IH_MUTE_FC_STAT2); } -static int dw_hdmi_setup (struct dw_hdmi *hdmi, -const struct drm_connector *connector, -const struct drm_display_mode *mode) +static int dw_hdmi_poweron (struct dw_hdmi *hdmi, +const struct drm_connector *connector, +const struct drm_display_mode *mode) {const struct drm_display_info *display = &connector->display_info;int ret;
@@ -2396,15 +2396,6 @@ static void initialize_hdmi_ih_mutes(struct dw_hdmi *hdmi)hdmi_writeb (hdmi, ih_mute,HDMI_IH_MUTE ); } -static void dw_hdmi_poweron (struct dw_hdmi *hdmi) -{ -/* -* The curr_conn field is guaranteed to be valid here, as this function -* is only be called when !hdmi->disabled. -*/ -dw_hdmi_setup (hdmi, hdmi->curr_conn, &hdmi->previous_mode); -} -static void dw_hdmi_poweroff (struct dw_hdmi *hdmi) {if (hdmi->phy.enabled) {
@@ -2982,15 +2973,19 @@ static void dw_hdmi_bridge_atomic_enable(struct drm_bridge *bridge, struct drm_atomic_commit *state) {struct dw_hdmi *hdmi = bridge->driver_private; +const struct drm_display_mode *mode;struct drm_connector *connector; +struct drm_crtc *crtc; connector =drm_atomic_get_new_connector_for_encoder (state, bridge->encoder); + crtc =drm_atomic_get_new_connector_state (state, connector)->crtc; + mode = &drm_atomic_get_new_crtc_state (state, crtc)->adjusted_mode;mutex_lock (&hdmi->mutex); hdmi->disabled = false; hdmi->curr_conn = connector; -dw_hdmi_poweron (hdmi); +dw_hdmi_poweron (hdmi, connector, mode);dw_hdmi_update_phy_mask (hdmi);handle_plugged_change (hdmi, true);mutex_unlock (&hdmi->mutex);
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 8/23 drm: bridge: dw_hdmi: Remove previous_mode and mode_set
In reply to Jonas KarlmanWith the use of adjusted_mode directly from the crtc_state there is no longer a need to store a copy in previous_mode, remove it and the now unneeded mode_set ops.
---
v7: No change
v6: Collect t-b tag
v5: No change, rebase on next-20260508
v4: No change
v3: Collect r-b tag
v2: No change
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 19 +------------------ 1 file changed, 1 insertion(+), 18 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +1 -18
@@ -156,8 +156,6 @@ struct dw_hdmi {bool enabled; } phy; -struct drm_display_mode previous_mode; -struct i2c_adapter *ddc;void __iomem *regs;bool sink_is_hdmi;
@@ -167,7 +165,7 @@ struct dw_hdmi {struct pinctrl_state *default_state;struct pinctrl_state *unwedge_state; -struct mutex mutex;/* for state below and previous_mode */ +struct mutex mutex;/* for state below */ enum drm_connector_force force;/* mutex-protected force state */ struct drm_connector *curr_conn;/* current connector (only valid when !disabled) */ bool disabled;/* DRM has disabled our bridge */
@@ -2941,20 +2939,6 @@ dw_hdmi_bridge_mode_valid(struct drm_bridge *bridge,return mode_status; } -static void dw_hdmi_bridge_mode_set (struct drm_bridge *bridge, -const struct drm_display_mode *orig_mode, -const struct drm_display_mode *mode) -{ -struct dw_hdmi *hdmi = bridge->driver_private; - -mutex_lock (&hdmi->mutex); - -/* Store the display mode for plugin/DKMS poweron events */ -drm_mode_copy (&hdmi->previous_mode, mode); - -mutex_unlock (&hdmi->mutex); -} -static void dw_hdmi_bridge_atomic_disable (struct drm_bridge *bridge,struct drm_atomic_commit *state) {
@@ -3018,7 +3002,6 @@ static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .atomic_get_input_bus_fmts = dw_hdmi_bridge_atomic_get_input_bus_fmts, .atomic_enable = dw_hdmi_bridge_atomic_enable, .atomic_disable = dw_hdmi_bridge_atomic_disable, - .mode_set = dw_hdmi_bridge_mode_set, .mode_valid = dw_hdmi_bridge_mode_valid, .detect = dw_hdmi_bridge_detect, .edid_read = dw_hdmi_bridge_edid_read,
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 9/23 drm: bridge: dw_hdmi: Unregister CEC notifier during connector cleanup
In reply to Jonas KarlmanThe CEC notifier is being unregistered when the bridge detach, something that happens earlier than normal connector cleanup.
Change to unregister the CEC notifier at connector cleanup, in the connector .destroy() func, to align the lifetime of the connector and the CEC notifier and closer match a drmres handled generic CEC notifier.
---
v7: No change
v6: Collect t-b tag
v5: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +5 -11
@@ -2532,6 +2532,11 @@ static void dw_hdmi_connector_destroy(struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); +mutex_lock (&hdmi->cec_notifier_mutex); +cec_notifier_conn_unregister (hdmi->cec_notifier); + hdmi->cec_notifier =NULL ; +mutex_unlock (&hdmi->cec_notifier_mutex); +drm_connector_cleanup (connector);drm_bridge_put (&hdmi->bridge); }
@@ -2909,16 +2914,6 @@ static int dw_hdmi_bridge_attach(struct drm_bridge *bridge,return dw_hdmi_connector_create (hdmi); } -static void dw_hdmi_bridge_detach (struct drm_bridge *bridge) -{ -struct dw_hdmi *hdmi = bridge->driver_private; - -mutex_lock (&hdmi->cec_notifier_mutex); -cec_notifier_conn_unregister (hdmi->cec_notifier); - hdmi->cec_notifier =NULL ; -mutex_unlock (&hdmi->cec_notifier_mutex); -} - static enum drm_mode_status dw_hdmi_bridge_mode_valid(struct drm_bridge *bridge, const struct drm_display_info *info,
@@ -2996,7 +2991,6 @@ static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .atomic_destroy_state = drm_atomic_helper_bridge_destroy_state, .atomic_reset = drm_atomic_helper_bridge_reset, .attach = dw_hdmi_bridge_attach, - .detach = dw_hdmi_bridge_detach, .atomic_check = dw_hdmi_bridge_atomic_check, .atomic_get_output_bus_fmts = dw_hdmi_bridge_atomic_get_output_bus_fmts, .atomic_get_input_bus_fmts = dw_hdmi_bridge_atomic_get_input_bus_fmts,
Signature · 1 line
2.54.0
Re: [PATCH v7 09/23] drm: bridge: dw_hdmi: Unregister CEC notifier during connector cleanup
Quoted 9 lines · Jonas Karlman
The CEC notifier is being unregistered when the bridge detach, something that happens earlier than normal connector cleanup.
Change to unregister the CEC notifier at connector cleanup, in the connector .destroy() func, to align the lifetime of the connector and the CEC notifier and closer match a drmres handled generic CEC notifier.
Regards,
Hans
Quoted 49 lines
---
v7: No change
v6: Collect t-b tag
v5: New patch--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +5
-11@@ -2532,6 +2532,11 @@ static void dw_hdmi_connector_destroy(struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); +mutex_lock (&hdmi->cec_notifier_mutex); +cec_notifier_conn_unregister (hdmi->cec_notifier); + hdmi->cec_notifier =NULL ; +mutex_unlock (&hdmi->cec_notifier_mutex); +drm_connector_cleanup (connector);drm_bridge_put (&hdmi->bridge); }@@ -2909,16 +2914,6 @@ static int dw_hdmi_bridge_attach(struct drm_bridge *bridge,return dw_hdmi_connector_create (hdmi); } -static void dw_hdmi_bridge_detach (struct drm_bridge *bridge) -{ -struct dw_hdmi *hdmi = bridge->driver_private; - -mutex_lock (&hdmi->cec_notifier_mutex); -cec_notifier_conn_unregister (hdmi->cec_notifier); - hdmi->cec_notifier =NULL ; -mutex_unlock (&hdmi->cec_notifier_mutex); -} - static enum drm_mode_status dw_hdmi_bridge_mode_valid(struct drm_bridge *bridge, const struct drm_display_info *info,@@ -2996,7 +2991,6 @@ static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .atomic_destroy_state = drm_atomic_helper_bridge_destroy_state, .atomic_reset = drm_atomic_helper_bridge_reset, .attach = dw_hdmi_bridge_attach, - .detach = dw_hdmi_bridge_detach, .atomic_check = dw_hdmi_bridge_atomic_check, .atomic_get_output_bus_fmts = dw_hdmi_bridge_atomic_get_output_bus_fmts, .atomic_get_input_bus_fmts = dw_hdmi_bridge_atomic_get_input_bus_fmts,
Re: [PATCH v7 09/23] drm: bridge: dw_hdmi: Unregister CEC notifier during connector cleanup
In reply to Jonas KarlmanQuoted 8 lines · Jonas Karlman
The CEC notifier is being unregistered when the bridge detach, something that happens earlier than normal connector cleanup.
Change to unregister the CEC notifier at connector cleanup, in the connector .destroy() func, to align the lifetime of the connector and the CEC notifier and closer match a drmres handled generic CEC notifier.
[...]
Signature · 3 lines
Luca Ceresoli, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
Jonas Karlman
reviewerPATCH v7 10/23 drm: bridge: dw_hdmi: Invalidate CEC phys addr from connector detect
In reply to Jonas KarlmanWait until the connector detect ops is called to invalidate CEC phys addr instead of doing it directly from the irq handler.
The CEC notifier is only registered for the dw-hdmi connector so this has no impact when the bridge connector is used.
---
v7: Update commit message
v6: Collect t-b tag
v5: No change
v4: No change
v3: No change
v2: Collect r-b tag
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +11 -7
@@ -2472,7 +2472,17 @@ dw_hdmi_connector_detect(struct drm_connector *connector, bool force) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); -return dw_hdmi_detect (hdmi); +enum drm_connector_status status; + + status =dw_hdmi_detect (hdmi); + +if (status == connector_status_disconnected) { +mutex_lock (&hdmi->cec_notifier_mutex); +cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); +mutex_unlock (&hdmi->cec_notifier_mutex); + } + +return status; }static int dw_hdmi_connector_get_modes (struct drm_connector *connector)
@@ -3106,12 +3116,6 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) phy_stat &HDMI_PHY_HPD , phy_stat & HDMI_PHY_RX_SENSE); -if ((phy_stat & (HDMI_PHY_RX_SENSE |HDMI_PHY_HPD )) ==0 ) { -mutex_lock (&hdmi->cec_notifier_mutex); -cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); -mutex_unlock (&hdmi->cec_notifier_mutex); - } -if ((intr_stat &HDMI_IH_PHY_STAT0_HPD ) && (phy_stat &HDMI_PHY_HPD )) status = connector_status_connected;
Signature · 1 line
2.54.0
Re: [PATCH v7 10/23] drm: bridge: dw_hdmi: Invalidate CEC phys addr from connector detect
Quoted 9 lines · Jonas Karlman
Wait until the connector detect ops is called to invalidate CEC phys addr instead of doing it directly from the irq handler.
The CEC notifier is only registered for the dw-hdmi connector so this has no impact when the bridge connector is used.
Reviewed-by Neil Armstrong <[email protected]>Signed-off-by Jonas Karlman <[email protected]>
Regards,
Hans
Quoted 47 lines
---
v7: Update commit message
v6: Collect t-b tag
v5: No change
v4: No change
v3: No change
v2: Collect r-b tag--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +11
-7@@ -2472,7 +2472,17 @@ dw_hdmi_connector_detect(struct drm_connector *connector, bool force) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); -return dw_hdmi_detect (hdmi); +enum drm_connector_status status; + + status =dw_hdmi_detect (hdmi); + +if (status == connector_status_disconnected) { +mutex_lock (&hdmi->cec_notifier_mutex); +cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); +mutex_unlock (&hdmi->cec_notifier_mutex); + } + +return status; }static int dw_hdmi_connector_get_modes (struct drm_connector *connector)@@ -3106,12 +3116,6 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) phy_stat &HDMI_PHY_HPD , phy_stat & HDMI_PHY_RX_SENSE); -if ((phy_stat & (HDMI_PHY_RX_SENSE |HDMI_PHY_HPD )) ==0 ) { -mutex_lock (&hdmi->cec_notifier_mutex); -cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); -mutex_unlock (&hdmi->cec_notifier_mutex); - } -if ((intr_stat &HDMI_IH_PHY_STAT0_HPD ) && (phy_stat &HDMI_PHY_HPD )) status = connector_status_connected;
Jonas Karlman
reviewerPATCH v7 11/23 drm: bridge: dw_hdmi: Remove cec_notifier_mutex
In reply to Jonas KarlmanWith CEC phys addr invalidation moved away from the irq handler there is no longer a need for cec_notifier_mutex, remove it.
---
v7: No change
v6: Collect t-b tag
v5: No change, cec_notifier_conn_unregister() call moved
v4: No change
v3: No change
v2: Collect r-b tag
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 11 +---------- 1 file changed, 1 insertion(+), 10 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +1 -10
@@ -189,7 +189,6 @@ struct dw_hdmi {void (*enable_audio)(struct dw_hdmi *hdmi);void (*disable_audio)(struct dw_hdmi *hdmi); -struct mutex cec_notifier_mutex;struct cec_notifier *cec_notifier;hdmi_codec_plugged_cb plugged_cb;
@@ -2476,11 +2475,8 @@ dw_hdmi_connector_detect(struct drm_connector *connector, bool force) status =dw_hdmi_detect (hdmi); -if (status == connector_status_disconnected) { -mutex_lock (&hdmi->cec_notifier_mutex); +if (status == connector_status_disconnected)cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); -mutex_unlock (&hdmi->cec_notifier_mutex); - }return status; }
@@ -2542,10 +2538,8 @@ static void dw_hdmi_connector_destroy(struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); -mutex_lock (&hdmi->cec_notifier_mutex);cec_notifier_conn_unregister (hdmi->cec_notifier); hdmi->cec_notifier =NULL ; -mutex_unlock (&hdmi->cec_notifier_mutex);drm_connector_cleanup (connector);drm_bridge_put (&hdmi->bridge);
@@ -2612,9 +2606,7 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)if (!notifier)return -ENOMEM ; -mutex_lock (&hdmi->cec_notifier_mutex); hdmi->cec_notifier = notifier; -mutex_unlock (&hdmi->cec_notifier_mutex);return 0 ; }
@@ -3323,7 +3315,6 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev,mutex_init (&hdmi->mutex);mutex_init (&hdmi->audio_mutex); -mutex_init (&hdmi->cec_notifier_mutex);spin_lock_init (&hdmi->audio_lock); ddc_node =of_parse_phandle (np,"ddc-i2c-bus" ,0 );
Signature · 1 line
2.54.0
Re: [PATCH v7 11/23] drm: bridge: dw_hdmi: Remove cec_notifier_mutex
Quoted 6 lines · Jonas Karlman
With CEC phys addr invalidation moved away from the irq handler there is no longer a need for
cec_notifier_mutex, remove it.Reviewed-by Neil Armstrong <[email protected]>Signed-off-by Jonas Karlman <[email protected]>
Nice, I wondered why that mutex was there at all.
Regards,
Hans
Quoted 65 lines
---
v7: No change
v6: Collect t-b tag
v5: No change,cec_notifier_conn_unregister()call moved
v4: No change
v3: No change
v2: Collect r-b tag--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 11 +---------- 1 file changed, 1 insertion(+), 10 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +1
-10@@ -189,7 +189,6 @@ struct dw_hdmi {void (*enable_audio)(struct dw_hdmi *hdmi);void (*disable_audio)(struct dw_hdmi *hdmi); -struct mutex cec_notifier_mutex;struct cec_notifier *cec_notifier;hdmi_codec_plugged_cb plugged_cb;@@ -2476,11 +2475,8 @@ dw_hdmi_connector_detect(struct drm_connector *connector, bool force) status =dw_hdmi_detect (hdmi); -if (status == connector_status_disconnected) { -mutex_lock (&hdmi->cec_notifier_mutex); +if (status == connector_status_disconnected)cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); -mutex_unlock (&hdmi->cec_notifier_mutex); - }return status; }@@ -2542,10 +2538,8 @@ static void dw_hdmi_connector_destroy(struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); -mutex_lock (&hdmi->cec_notifier_mutex);cec_notifier_conn_unregister (hdmi->cec_notifier); hdmi->cec_notifier =NULL ; -mutex_unlock (&hdmi->cec_notifier_mutex);drm_connector_cleanup (connector);drm_bridge_put (&hdmi->bridge);@@ -2612,9 +2606,7 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)if (!notifier)return -ENOMEM ; -mutex_lock (&hdmi->cec_notifier_mutex); hdmi->cec_notifier = notifier; -mutex_unlock (&hdmi->cec_notifier_mutex);return 0 ; }@@ -3323,7 +3315,6 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev,mutex_init (&hdmi->mutex);mutex_init (&hdmi->audio_mutex); -mutex_init (&hdmi->cec_notifier_mutex);spin_lock_init (&hdmi->audio_lock); ddc_node =of_parse_phandle (np,"ddc-i2c-bus" ,0 );
Jonas Karlman
reviewerPATCH v7 12/23 drm: bridge: dw_hdmi: Extract dw_hdmi_connector_status_update()
In reply to Jonas KarlmanMove connector EDID update and CEC phys addr handling to a helper function as a preparation before moving EDID refresh from get_modes funcs to detect/force funcs.
---
v7: No change
v6: Pass struct dw_hdmi as a parameter, to allow calls from bridge funcs,
Collect t-b tag
v5: No change
v4: Collect r-b tag
v3: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 27 ++++++++++++++--------- 1 file changed, 17 insertions(+), 10 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +17 -10
@@ -2466,6 +2466,21 @@ static const struct drm_edid *dw_hdmi_edid_read(struct dw_hdmi *hdmi,* DRM Connector Operations */ +static void +dw_hdmi_connector_status_update (struct dw_hdmi *hdmi, +struct drm_connector *connector, +enum drm_connector_status status) +{ +const struct drm_edid *drm_edid; + + drm_edid =dw_hdmi_edid_read (hdmi, connector); +drm_edid_connector_update (connector, drm_edid); +drm_edid_free (drm_edid); + +cec_notifier_set_phys_addr (hdmi->cec_notifier, + connector->display_info.source_physical_address); +} +static enum drm_connector_status dw_hdmi_connector_detect (struct drm_connector *connector,bool force) {
@@ -2485,18 +2500,10 @@ static int dw_hdmi_connector_get_modes(struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); -const struct drm_edid *drm_edid; -int ret; - drm_edid =dw_hdmi_edid_read (hdmi, connector); +dw_hdmi_connector_status_update (hdmi, connector, connector->status); -drm_edid_connector_update (connector, drm_edid); -cec_notifier_set_phys_addr (hdmi->cec_notifier, - connector->display_info.source_physical_address); - ret =drm_edid_connector_add_modes (connector); -drm_edid_free (drm_edid); - -return ret; +return drm_edid_connector_add_modes (connector); } static int dw_hdmi_connector_atomic_check(struct drm_connector *connector,
Signature · 1 line
2.54.0
Re: [PATCH v7 12/23] drm: bridge: dw_hdmi: Extract dw_hdmi_connector_status_update()
Quoted 7 lines · Jonas Karlman
Move connector EDID update and CEC phys addr handling to a helper function as a preparation before moving EDID refresh from
get_modesfuncs to detect/force funcs.Reviewed-by Nicolas Frattaroli <[email protected]>Signed-off-by Jonas Karlman <[email protected]>
Regards,
Hans
Quoted 58 lines
---
v7: No change
v6: Pass structdw_hdmias a parameter, to allow calls from bridge funcs,
Collect t-b tag
v5: No change
v4: Collect r-b tag
v3: New patch--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 27 ++++++++++++++--------- 1 file changed, 17 insertions(+), 10 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +17
-10@@ -2466,6 +2466,21 @@ static const struct drm_edid *dw_hdmi_edid_read(struct dw_hdmi *hdmi,* DRM Connector Operations */ +static void +dw_hdmi_connector_status_update (struct dw_hdmi *hdmi, +struct drm_connector *connector, +enum drm_connector_status status) +{ +const struct drm_edid *drm_edid; + + drm_edid =dw_hdmi_edid_read (hdmi, connector); +drm_edid_connector_update (connector, drm_edid); +drm_edid_free (drm_edid); + +cec_notifier_set_phys_addr (hdmi->cec_notifier, + connector->display_info.source_physical_address); +} +static enum drm_connector_status dw_hdmi_connector_detect (struct drm_connector *connector,bool force) {@@ -2485,18 +2500,10 @@ static int dw_hdmi_connector_get_modes(struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); -const struct drm_edid *drm_edid; -int ret; - drm_edid =dw_hdmi_edid_read (hdmi, connector); +dw_hdmi_connector_status_update (hdmi, connector, connector->status); -drm_edid_connector_update (connector, drm_edid); -cec_notifier_set_phys_addr (hdmi->cec_notifier, - connector->display_info.source_physical_address); - ret =drm_edid_connector_add_modes (connector); -drm_edid_free (drm_edid); - -return ret; +return drm_edid_connector_add_modes (connector); } static int dw_hdmi_connector_atomic_check(struct drm_connector *connector,
Jonas Karlman
reviewerPATCH v7 13/23 drm: bridge: dw_hdmi: Use dw_hdmi_connector_status_update()
In reply to Jonas KarlmanUpdate connector EDID and CEC phys addr from detect and force funcs to ensure that userspace always have access to latest read EDID after a sink use a HPD low voltage pulse to indicate that EDID has changed.
With EDID being updated in detect and force funcs, there should no longer be a need to re-read EDID in get_modes funcs, so drop it.
This change make the dw-hdmi connector work more closely like the bridge connector does with a hdmi bridge.
---
v7: No change
v6: Pass struct dw_hdmi as a parameter,
Collect t-b tag
v5: No change
v4: Move last_connector_result assign in force ops to this patch,
Collect r-b tag
v3: Reworked 'Update EDID during hotplug processing' patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 28 ++++++++++++----------- 1 file changed, 15 insertions(+), 13 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +15 -13
@@ -2473,36 +2473,36 @@ dw_hdmi_connector_status_update(struct dw_hdmi *hdmi, {const struct drm_edid *drm_edid; +if (status == connector_status_disconnected) { +drm_edid_connector_update (connector,NULL ); +cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); +return ; + } + drm_edid =dw_hdmi_edid_read (hdmi, connector);drm_edid_connector_update (connector, drm_edid);drm_edid_free (drm_edid); -cec_notifier_set_phys_addr (hdmi->cec_notifier, - connector->display_info.source_physical_address); +if (status == connector_status_connected) +cec_notifier_set_phys_addr (hdmi->cec_notifier, + connector->display_info.source_physical_address); }static enum drm_connector_status dw_hdmi_connector_detect (struct drm_connector *connector,bool force) { - struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, - connector); + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);enum drm_connector_status status; status =dw_hdmi_detect (hdmi); -if (status == connector_status_disconnected) -cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); +dw_hdmi_connector_status_update (hdmi, connector, status);return status; }static int dw_hdmi_connector_get_modes (struct drm_connector *connector) { - struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, - connector); - -dw_hdmi_connector_status_update (hdmi, connector, connector->status); -return drm_edid_connector_add_modes (connector); }
@@ -2532,13 +2532,15 @@ static int dw_hdmi_connector_atomic_check(struct drm_connector *connector,static void dw_hdmi_connector_force (struct drm_connector *connector) { - struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, - connector); + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);mutex_lock (&hdmi->mutex); hdmi->force = connector->force; + hdmi->last_connector_result = connector->status;dw_hdmi_update_phy_mask (hdmi);mutex_unlock (&hdmi->mutex); + +dw_hdmi_connector_status_update (hdmi, connector, connector->status); }static void dw_hdmi_connector_destroy (struct drm_connector *connector)
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v7 13/23] drm: bridge: dw_hdmi: Use dw_hdmi_connector_status_update()
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] The CEC physical address is no longer set when the connector status is `connector_status_unknown`…
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The CEC physical address is no longer set when the connector status is connector_status_unknown, breaking CEC functionality for displays with unreliable HPD.
- [High] Unconditional EDID read over I2C in detect() ignores the force parameter, causing severe performance degradation and I2C spam. --
commit d5d270011a2b02f003218ae7b96f62fae2b447e9
Author: Jonas Karlman <[email protected]>
drm: bridge: dw_hdmi: Use dw_hdmi_connector_status_update()
This commit updates the dw-hdmi connector to fetch EDID and update CEC physical addresses during detect and force functions rather than in get_modes. It aims to ensure userspace always has access to the latest read EDID.
Quoted 3 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-0
[ ... ]
Quoted 1 line
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-0@@ -2473,36 +2473,36 @@ dw_hdmi_connector_status_update(struct dw_hdmi *hdmi,
[ ... ]
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c dw_hdmi_connector_status_update(struct dw_hdmi *hdmi, · line 2489
Quoted 10 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +3
-2drm_edid =dw_hdmi_edid_read (hdmi, connector);drm_edid_connector_update (connector, drm_edid);drm_edid_free (drm_edid); -cec_notifier_set_phys_addr (hdmi->cec_notifier, - connector->display_info.source_physical_address); +if (status == connector_status_connected) +cec_notifier_set_phys_addr (hdmi->cec_notifier, + connector->display_info.source_physical_address); }
Does this break CEC functionality for displays with unreliable HPD?
By guarding the cec_notifier_set_phys_addr() call with a check for connector_status_connected, the CEC physical address update is skipped if the status evaluates to connector_status_unknown.
Previously, this was called unconditionally after a successful EDID read, which allowed CEC to work even when HPD was unreliable.
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c dw_hdmi_connector_status_update(struct dw_hdmi *hdmi, · line 2499
Quoted 13 lines
static enum
drm_connector_statusdw_hdmi_connector_detect(struct drm_connector *connector, bool force){drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +2
-4- struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, - connector); + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);enum drm_connector_status status; status =dw_hdmi_detect (hdmi); -if (status == connector_status_disconnected) -cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); +dw_hdmi_connector_status_update (hdmi, connector, status);
Will this cause performance issues by performing unconditional I2C EDID reads on every detect call?
The dw_hdmi_connector_status_update() function reads the EDID over I2C unless the status is disconnected, but this completely ignores the force parameter passed to detect. If a system-wide hotplug event occurs on another port, drm_helper_hpd_irq_event() will call detect with force=false on all HPD-capable connectors.
Could this trigger an unnecessary 20-50ms I2C EDID read here, spamming the I2C bus and blocking the shared KMS system workqueue?
Quoted 3 lines
return status;
}
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=13
Re: [PATCH v7 13/23] drm: bridge: dw_hdmi: Use dw_hdmi_connector_status_update()
In reply to Jonas KarlmanQuoted 7 lines · Jonas Karlman
Update connector EDID and CEC phys addr from detect and force funcs to ensure that userspace always have access to latest read EDID after a sink use a HPD low voltage pulse to indicate that EDID has changed.
With EDID being updated in detect and force funcs, there should no longer be a need to re-read EDID in
get_modesfuncs, so drop it.
Quoted 6 lines
This change make the dw-hdmi connector work more closely like the bridge connector does with a hdmi bridge.
Reviewed-by Nicolas Frattaroli <[email protected]>Signed-off-by Jonas Karlman <[email protected]>
Regards,
Hans
Quoted 82 lines
---
v7: No change
v6: Pass structdw_hdmias a parameter,
Collect t-b tag
v5: No change
v4: Movelast_connector_resultassign in force ops to this patch,
Collect r-b tag
v3: Reworked 'Update EDID during hotplug processing' patch--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 28 ++++++++++++----------- 1 file changed, 15 insertions(+), 13 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +15
-13@@ -2473,36 +2473,36 @@ dw_hdmi_connector_status_update(struct dw_hdmi *hdmi, {const struct drm_edid *drm_edid; +if (status == connector_status_disconnected) { +drm_edid_connector_update (connector,NULL ); +cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); +return ; + } + drm_edid =dw_hdmi_edid_read (hdmi, connector);drm_edid_connector_update (connector, drm_edid);drm_edid_free (drm_edid); -cec_notifier_set_phys_addr (hdmi->cec_notifier, - connector->display_info.source_physical_address); +if (status == connector_status_connected) +cec_notifier_set_phys_addr (hdmi->cec_notifier, + connector->display_info.source_physical_address); }static enum drm_connector_status dw_hdmi_connector_detect (struct drm_connector *connector,bool force) { - struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, - connector); + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);enum drm_connector_status status; status =dw_hdmi_detect (hdmi); -if (status == connector_status_disconnected) -cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); +dw_hdmi_connector_status_update (hdmi, connector, status);return status; }static int dw_hdmi_connector_get_modes (struct drm_connector *connector) { - struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, - connector); - -dw_hdmi_connector_status_update (hdmi, connector, connector->status); -return drm_edid_connector_add_modes (connector); }@@ -2532,13 +2532,15 @@ static int dw_hdmi_connector_atomic_check(struct drm_connector *connector,static void dw_hdmi_connector_force (struct drm_connector *connector) { - struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, - connector); + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);mutex_lock (&hdmi->mutex); hdmi->force = connector->force; + hdmi->last_connector_result = connector->status;dw_hdmi_update_phy_mask (hdmi);mutex_unlock (&hdmi->mutex); + +dw_hdmi_connector_status_update (hdmi, connector, connector->status); }static void dw_hdmi_connector_destroy (struct drm_connector *connector)
Jonas Karlman
reviewerPATCH v7 14/23 drm: bridge: dw_hdmi: Use display_info is_hdmi and has_audio
In reply to Jonas Karlmandrm_edid_connector_update() is being called from bridge connector funcs and from detect and force funcs for dw-hdmi connector.
Change to use is_hdmi and has_audio from display_info directly instead of keeping our own state in sink_is_hdmi and sink_has_audio.
Also remove the old and unused edid struct member and related define.
---
v7: No change, re-order patch
v6: Collect t-b tag
v5: No change
v4: Collect r-b tag
v3: No change
v2: Collect r-b tag
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 32 ++++------------------- 1 file changed, 5 insertions(+), 27 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +5 -27
@@ -46,8 +46,6 @@#define DDC_CI_ADDR 0x37#define DDC_SEGMENT_ADDR 0x30 -#define HDMI_EDID_LEN 512 -/* DW-HDMI Controller >= 0x200a are at least compliant with SCDC version 1 */ #define SCDC_MIN_SOURCE_VERSION 0x1
@@ -147,8 +145,6 @@ struct dw_hdmi {int vic; -u8 edid[HDMI_EDID_LEN ]; -struct {const struct dw_hdmi_phy_ops *ops;const char *name;
@@ -158,8 +154,6 @@ struct dw_hdmi {struct i2c_adapter *ddc;void __iomem *regs; -bool sink_is_hdmi; -bool sink_has_audio;struct pinctrl *pinctrl;struct pinctrl_state *default_state;
@@ -2056,7 +2050,7 @@ static void hdmi_av_composer(struct dw_hdmi *hdmi, HDMI_FC_INVIDCONF_IN_I_P_INTERLACED :HDMI_FC_INVIDCONF_IN_I_P_PROGRESSIVE ; - inv_val |= hdmi->sink_is_hdmi ? + inv_val |= display->is_hdmi ?HDMI_FC_INVIDCONF_DVI_MODEZ_HDMI_MODE :HDMI_FC_INVIDCONF_DVI_MODEZ_DVI_MODE ;
@@ -2292,7 +2286,7 @@ static int dw_hdmi_poweron(struct dw_hdmi *hdmi,if (hdmi->hdmi_data.enc_out_bus_format ==MEDIA_BUS_FMT_FIXED ) hdmi->hdmi_data.enc_out_bus_format =MEDIA_BUS_FMT_RGB888_1X24 ; - hdmi->hdmi_data.rgb_limited_range = hdmi->sink_is_hdmi && + hdmi->hdmi_data.rgb_limited_range = display->is_hdmi &&drm_default_rgb_quant_range (mode) ==HDMI_QUANTIZATION_RANGE_LIMITED ;
@@ -2312,7 +2306,7 @@ static int dw_hdmi_poweron(struct dw_hdmi *hdmi,/* HDMI Initialization Step B.3 */ dw_hdmi_enable_video_path (hdmi); -if (hdmi->sink_has_audio) { +if (display->has_audio) {dev_dbg (hdmi->dev,"sink has audio support\n" );/* HDMI Initialization Step E - Configure audio */
@@ -2321,7 +2315,7 @@ static int dw_hdmi_poweron(struct dw_hdmi *hdmi, }/* not for DVI mode */ -if (hdmi->sink_is_hdmi) { +if (display->is_hdmi) {dev_dbg (hdmi->dev,"%s HDMI mode\n" , __func__);/* HDMI Initialization Step F - Configure AVI InfoFrame */
@@ -2435,29 +2429,13 @@ static const struct drm_edid *dw_hdmi_edid_read(struct dw_hdmi *hdmi, struct drm_connector *connector) {const struct drm_edid *drm_edid; -const struct edid *edid;if (!hdmi->ddc)return NULL ; drm_edid =drm_edid_read_ddc (connector, hdmi->ddc); -if (!drm_edid) { +if (!drm_edid)dev_dbg (hdmi->dev,"failed to get edid\n" ); -return NULL ; - } - -/* -* FIXME: This should use connector->display_info.is_hdmi and -* connector->display_info.has_audio from a path that has read the EDID -* and called drm_edid_connector_update(). -*/ - edid =drm_edid_raw (drm_edid); - -dev_dbg (hdmi->dev,"got edid: width[%d] x height[%d]\n" , - edid->width_cm, edid->height_cm); - - hdmi->sink_is_hdmi =drm_detect_hdmi_monitor (edid); - hdmi->sink_has_audio =drm_detect_monitor_audio (edid);return drm_edid; }
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 15/23 drm: bridge: dw_hdmi: Use generic CEC notifier helpers
In reply to Jonas KarlmanThe commit 8b1a8f8b2002 ("drm/display: add CEC helpers code") added generic CEC helpers to be used by HDMI drivers.
Replace the open-coded CEC notifier handling with use of the generic CEC notifier helpers. Ensure DRM_DISPLAY_HDMI_CEC_NOTIFIER_HELPER is also selected when DRM_DW_HDMI_CEC is enabled so that the CEC helpers is available.
The drmm release action for the generic CEC notifier should run just before dw_hdmi_connector_destroy(), closely matching the lifetime of the replaced CEC notifier and the connector.
---
v7: No change
v6: Update commit message,
Collect t-b tag
v5: Collect r-b tag
v4: New patch
--- drivers/gpu/drm/bridge/synopsys/Kconfig | 1 + drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 26 +++++------------------ 2 files changed, 6 insertions(+), 21 deletions(-)
drivers/gpu/drm/bridge/synopsys/Kconfig +1 -0
@@ -49,6 +49,7 @@ config DRM_DW_HDMI_CEC depends on DRM_DW_HDMI select CEC_CORE select CEC_NOTIFIER + select DRM_DISPLAY_HDMI_CEC_NOTIFIER_HELPER help Support the CE interface which is part of the Synopsys Designware HDMI block.
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +5 -21
@@ -23,12 +23,11 @@#include <linux/dma-mapping.h> #include <linux/spinlock.h> -#include <media/cec-notifier.h> -#include <linux/media-bus-format.h> #include <linux/videodev2.h> #include <drm/bridge/dw_hdmi.h> +#include <drm/display/drm_hdmi_cec_helper.h> #include <drm/display/drm_hdmi_helper.h> #include <drm/display/drm_scdc_helper.h> #include <drm/drm_atomic.h>
@@ -183,8 +182,6 @@ struct dw_hdmi {void (*enable_audio)(struct dw_hdmi *hdmi);void (*disable_audio)(struct dw_hdmi *hdmi); -struct cec_notifier *cec_notifier; -hdmi_codec_plugged_cb plugged_cb;struct device *codec_dev;enum drm_connector_status last_connector_result;
@@ -2453,7 +2450,7 @@ dw_hdmi_connector_status_update(struct dw_hdmi *hdmi,if (status == connector_status_disconnected) {drm_edid_connector_update (connector,NULL ); -cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); +drm_connector_cec_phys_addr_invalidate (connector);return ; }
@@ -2462,8 +2459,7 @@ dw_hdmi_connector_status_update(struct dw_hdmi *hdmi,drm_edid_free (drm_edid);if (status == connector_status_connected) -cec_notifier_set_phys_addr (hdmi->cec_notifier, - connector->display_info.source_physical_address); +drm_connector_cec_phys_addr_set (connector); } static enum drm_connector_status
@@ -2525,9 +2521,6 @@ static void dw_hdmi_connector_destroy(struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); -cec_notifier_conn_unregister (hdmi->cec_notifier); - hdmi->cec_notifier =NULL ; -drm_connector_cleanup (connector);drm_bridge_put (&hdmi->bridge); }
@@ -2550,8 +2543,6 @@ static const struct drm_connector_helper_funcs dw_hdmi_connector_helper_funcs =static int dw_hdmi_connector_create (struct dw_hdmi *hdmi) {struct drm_connector *connector = &hdmi->connector; -struct cec_connector_info conn_info; -struct cec_notifier *notifier;int ret;if (hdmi->version >=0x200a )
@@ -2587,15 +2578,8 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)drm_connector_attach_encoder (connector, hdmi->bridge.encoder); -cec_fill_conn_info_from_drm (&conn_info, connector); - - notifier =cec_notifier_conn_register (hdmi->dev,NULL , &conn_info); -if (!notifier) -return -ENOMEM ; - - hdmi->cec_notifier = notifier; - -return 0 ; +return drmm_connector_hdmi_cec_notifier_register (connector,NULL , + hdmi->dev); }/* -----------------------------------------------------------------------------
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v7 15/23] drm: bridge: dw_hdmi: Use generic CEC notifier helpers
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] Use-After-Free and memory leak due to drmm action lifetime mismatch with the connector --
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Use-After-Free and memory leak due to drmm action lifetime mismatch with the connector --
commit 7a3d2d4c9ae6107ffef88b867d73cc39dd7e8567
Author: Jonas Karlman <[email protected]>
drm: bridge: dw_hdmi: Use generic CEC notifier helpers
This commit replaces the open-coded CEC notifier handling with the generic DRM HDMI CEC notifier helpers.
Quoted 3 lines
The drmm release action for the generic CEC notifier should run just before
dw_hdmi_connector_destroy(), closely matching the lifetime of the replaced CEC notifier and the connector.
Is this assumption accurate for all drivers using dw_hdmi?
When master drivers like sun4i or imx are unbound, drm_mode_config_cleanup() is often called synchronously, which in turn calls dw_hdmi_connector_destroy().
If a userspace process holds the DRM file descriptor open, the drm_device outlives the unbind. Doesn't this mean the drmm release action will run much later than dw_hdmi_connector_destroy(), rather than just before it?
Quoted 4 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-0
[ ... ]
Quoted 6 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-2@@ -2587,15 +2578,8 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)drm_connector_attach_encoder (connector, hdmi->bridge.encoder); -cec_fill_conn_info_from_drm (&conn_info, connector); -
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static int dw_hdmi_connector_create(struct dw_hdmi *hdmi) · line 2583
Quoted 10 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +2
-7- notifier =cec_notifier_conn_register (hdmi->dev,NULL , &conn_info); -if (!notifier) -return -ENOMEM ; - - hdmi->cec_notifier = notifier; - -return 0 ; +return drmm_connector_hdmi_cec_notifier_register (connector,NULL , + hdmi->dev); }
Could this cause a use-after-free when the drm_device is finally released?
Looking at dw_hdmi_connector_destroy(), it cleans up the connector state:
dw_hdmi_connector_destroy()
drm_connector_cleanup(connector)
...
memset(connector, 0, sizeof(*connector));Since the connector is embedded in struct dw_hdmi which is managed by devres and the bridge refcount, the memory will be freed when the platform device unbind completes.
When the drmm action drm_connector_hdmi_cec_notifier_unregister() eventually executes during drm_dev_release(), won't it attempt to dereference the already freed connector pointer?
If the memory happens to still be accessible, won't it read the zeroed connector->cec.data as a NULL pointer, silently leaking the CEC notifier?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=15
Re: [PATCH v7 15/23] drm: bridge: dw_hdmi: Use generic CEC notifier helpers
In reply to Jonas KarlmanQuoted 8 lines · Jonas Karlman
The commit
8b1a8f8b2002("drm/display: add CEC helpers code") added generic CEC helpers to be used by HDMI drivers.Replace the open-coded CEC notifier handling with use of the generic CEC notifier helpers. Ensure
DRM_DISPLAY_HDMI_CEC_NOTIFIER_HELPERis also selected whenDRM_DW_HDMI_CECis enabled so that the CEC helpers is available.
Quoted 7 lines
The drmm release action for the generic CEC notifier should run just before
dw_hdmi_connector_destroy(), closely matching the lifetime of the replaced CEC notifier and the connector.Reviewed-by Neil Armstrong <[email protected]>Signed-off-by Jonas Karlman <[email protected]>
Regards,
Hans
Quoted 106 lines
---
v7: No change
v6: Update commit message,
Collect t-b tag
v5: Collect r-b tag
v4: New patch--- drivers/gpu/drm/bridge/synopsys/Kconfig | 1 + drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 26 +++++------------------ 2 files changed, 6 insertions(+), 21 deletions(-)drivers/gpu/drm/bridge/synopsys/Kconfig +1
-0@@ -49,6 +49,7 @@ config DRM_DW_HDMI_CEC depends on DRM_DW_HDMI select CEC_CORE select CEC_NOTIFIER + select DRM_DISPLAY_HDMI_CEC_NOTIFIER_HELPER help Support the CE interface which is part of the Synopsys Designware HDMI block.drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +5
-21@@ -23,12 +23,11 @@#include <linux/dma-mapping.h> #include <linux/spinlock.h> -#include <media/cec-notifier.h> -#include <linux/media-bus-format.h> #include <linux/videodev2.h> #include <drm/bridge/dw_hdmi.h> +#include <drm/display/drm_hdmi_cec_helper.h> #include <drm/display/drm_hdmi_helper.h> #include <drm/display/drm_scdc_helper.h> #include <drm/drm_atomic.h> @@ -183,8 +182,6 @@ struct dw_hdmi {void (*enable_audio)(struct dw_hdmi *hdmi);void (*disable_audio)(struct dw_hdmi *hdmi); -struct cec_notifier *cec_notifier; -hdmi_codec_plugged_cb plugged_cb;struct device *codec_dev;enum drm_connector_status last_connector_result;@@ -2453,7 +2450,7 @@ dw_hdmi_connector_status_update(struct dw_hdmi *hdmi,if (status == connector_status_disconnected) {drm_edid_connector_update (connector,NULL ); -cec_notifier_phys_addr_invalidate (hdmi->cec_notifier); +drm_connector_cec_phys_addr_invalidate (connector);return ; }@@ -2462,8 +2459,7 @@ dw_hdmi_connector_status_update(struct dw_hdmi *hdmi,drm_edid_free (drm_edid);if (status == connector_status_connected) -cec_notifier_set_phys_addr (hdmi->cec_notifier, - connector->display_info.source_physical_address); +drm_connector_cec_phys_addr_set (connector); } static enum drm_connector_status@@ -2525,9 +2521,6 @@ static void dw_hdmi_connector_destroy(struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); -cec_notifier_conn_unregister (hdmi->cec_notifier); - hdmi->cec_notifier =NULL ; -drm_connector_cleanup (connector);drm_bridge_put (&hdmi->bridge); }@@ -2550,8 +2543,6 @@ static const struct drm_connector_helper_funcs dw_hdmi_connector_helper_funcs =static int dw_hdmi_connector_create (struct dw_hdmi *hdmi) {struct drm_connector *connector = &hdmi->connector; -struct cec_connector_info conn_info; -struct cec_notifier *notifier;int ret;if (hdmi->version >=0x200a )@@ -2587,15 +2578,8 @@ static int dw_hdmi_connector_create(struct dw_hdmi *hdmi)drm_connector_attach_encoder (connector, hdmi->bridge.encoder); -cec_fill_conn_info_from_drm (&conn_info, connector); - - notifier =cec_notifier_conn_register (hdmi->dev,NULL , &conn_info); -if (!notifier) -return -ENOMEM ; - - hdmi->cec_notifier = notifier; - -return 0 ; +return drmm_connector_hdmi_cec_notifier_register (connector,NULL , + hdmi->dev); }/* -----------------------------------------------------------------------------
Jonas Karlman
reviewerPATCH v7 16/23 drm: bridge: dw_hdmi: Update EDID and CEC phys addr in bridge detect()
In reply to Jonas KarlmanUpdate EDID and CEC phys addr in the bridge detect() func to closely match the behavior of a bridge connector with a HDMI bridge attached and the dw-hdmi connector.
This change introduce a slight delay to the bridge connector detect() and get_modes() funcs due to multiple EDID reads. This is an acceptable added delay to help ensure EDID and CEC phys addr always are correct.
---
v7: Update commit message
v6: New patch
This is a temporary change until dw-hdmi is fully converted into a HDMI bridge in a future part of this multi-series effort.
The patch "drm/bridge-connector: Use cached connector status in .get_modes()" [1] can help remove one unnecessary EDID read until dw-hdmi is fully converted into a HDMI bridge.
[1] https://lore.kernel.org/dri-devel/[email protected]/
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +10 -1
@@ -2947,8 +2947,17 @@ static enum drm_connector_status dw_hdmi_bridge_detect(struct drm_bridge *bridge, struct drm_connector *connector) { struct dw_hdmi *hdmi = bridge->driver_private; +enum drm_connector_status status; -return dw_hdmi_detect (hdmi); + status =dw_hdmi_detect (hdmi); + +/* +* Update EDID and CEC phys addr to match the behavior of a bridge +* connector with a HDMI bridge attached and the dw-hdmi connector. +*/ +dw_hdmi_connector_status_update (hdmi, connector, status); + +return status; } static const struct drm_edid *dw_hdmi_bridge_edid_read(struct drm_bridge *bridge,
Signature · 1 line
2.54.0
Re: [PATCH v7 16/23] drm: bridge: dw_hdmi: Update EDID and CEC phys addr in bridge detect()
Quoted 38 lines · Jonas Karlman
Update EDID and CEC phys addr in the bridge
detect()func to closely match the behavior of a bridge connector with a HDMI bridge attached and the dw-hdmi connector.This change introduce a slight delay to the bridge connector
detect()andget_modes()funcs due to multiple EDID reads. This is an acceptable added delay to help ensure EDID and CEC phys addr always are correct.Signed-off-by Jonas Karlman <[email protected]>---
v7: Update commit message
v6: New patchThis is a temporary change until dw-hdmi is fully converted into a HDMI bridge in a future part of this multi-series effort.
The patch "drm/bridge-connector: Use cached connector status in .get_modes()" [1] can help remove one unnecessary EDID read until dw-hdmi is fully converted into a HDMI bridge.
[1] https://lore.kernel.org/dri-devel/[email protected]/
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +3
-1@@ -2947,8 +2947,17 @@ static enum drm_connector_status dw_hdmi_bridge_detect(struct drm_bridge *bridge, struct drm_connector *connector) { struct dw_hdmi *hdmi = bridge->driver_private; +enum drm_connector_status status; -return dw_hdmi_detect (hdmi); + status =dw_hdmi_detect (hdmi); +
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static enum drm_connector_status · line 2963
Quoted 10 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +7
-0+/* +* Update EDID and CEC phys addr to match the behavior of a bridge +* connector with a HDMI bridge attached and the dw-hdmi connector. +*/ +dw_hdmi_connector_status_update (hdmi, connector, status); + +return status; } static const struct drm_edid *dw_hdmi_bridge_edid_read(struct drm_bridge *bridge,
Thanks,
Neil
Jonas Karlman
reviewerPATCH v7 17/23 drm: bridge: dw_hdmi: Declare bridge CEC notifier support
In reply to Jonas KarlmanEDID and CEC phys addr is now being updated in bridge detect() func, making it possible to have CEC notifier support using the bridge connector.
Add the CEC notifier bridge op to instruct the bridge connector to make use of the generic CEC notifier helpers.
---
v7: Only declare CEC notifier support when CEC device register succeeds
v6: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +4 -0
@@ -3515,6 +3515,10 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev, pdevinfo.dma_mask =0 ; hdmi->cec =platform_device_register_full (&pdevinfo); +if (!IS_ERR (hdmi->cec)) { + hdmi->bridge.ops |=DRM_BRIDGE_OP_HDMI_CEC_NOTIFIER ; + hdmi->bridge.hdmi_cec_dev = hdmi->dev; + } }drm_bridge_add (&hdmi->bridge);
Signature · 1 line
2.54.0
Re: [PATCH v7 17/23] drm: bridge: dw_hdmi: Declare bridge CEC notifier support
Quoted 8 lines · Jonas Karlman
EDID and CEC phys addr is now being updated in bridge
detect()func, making it possible to have CEC notifier support using the bridge connector.Add the CEC notifier bridge op to instruct the bridge connector to make use of the generic CEC notifier helpers.
Signed-off-by Jonas Karlman <[email protected]>
Regards,
Hans
Quoted 22 lines
---
v7: Only declare CEC notifier support when CEC device register succeeds
v6: New patch--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 4 ++++ 1 file changed, 4 insertions(+)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +4
-0@@ -3515,6 +3515,10 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev, pdevinfo.dma_mask =0 ; hdmi->cec =platform_device_register_full (&pdevinfo); +if (!IS_ERR (hdmi->cec)) { + hdmi->bridge.ops |=DRM_BRIDGE_OP_HDMI_CEC_NOTIFIER ; + hdmi->bridge.hdmi_cec_dev = hdmi->dev; + } }drm_bridge_add (&hdmi->bridge);
Jonas Karlman
reviewerPATCH v7 18/23 drm: bridge: dw_hdmi: Drop call to drm_bridge_hpd_notify()
In reply to Jonas KarlmanThe use of calls to both drm_helper_hpd_irq_event() and drm_bridge_hpd_notify() from the HPD IRQ handler may cause multiple hotplug uevents and modesets when the bridge connector is used.
Use of drm_helper_hpd_irq_event() cause the internal DRM function check_connector_changed() to be called, which in turn calls the connector detect()/force() funcs to detect any connection status or epoch changes, and when changed trigger a hotplug uevent. This also help ensure that EDID and CEC phys addr is updated.
If only a call drm_bridge_hpd_notify() would be used, a custom connector status/EDID change detection logic needs to be implemented, to fully match what check_connector_changed() already provides.
The bridge connector detect() func also ensures that any hpd_notify() funcs are called for all bridges in the chain, so there is not really any need to have a call to drm_bridge_hpd_notify() here.
With both calls there is two hotplug uevents, two modesets and a total of four .hpd_notify() calls (using a bridge connector):
dw_hdmi_irq(): EVENT=plugout
drm_helper_hpd_irq_event():
dw_hdmi_bridge_hpd_notify(status=2)
[drm:check_connector_changed] [CONNECTOR:46:HDMI-A-1] status updated from connected to disconnected [drm:check_connector_changed] [CONNECTOR:46:HDMI-A-1] Changed epoch counter 1 => 2 [drm:drm_sysfs_connector_hotplug_event] [CONNECTOR:46:HDMI-A-1] generating connector hotplug event
drm_client_hotplug():
[drm:drm_fb_helper_hotplug_event]
[drm:drm_client_modeset_probe]
[drm:drm_helper_probe_single_connector_modes] [CONNECTOR:46:HDMI-A-1] dw_hdmi_bridge_hpd_notify(status=2)
[drm:drm_helper_probe_single_connector_modes] [CONNECTOR:46:HDMI-A-1] disconnected [drm:drm_edid_connector_update] [CONNECTOR:46:HDMI-A-1] EDID changed, epoch counter 3 [drm:drm_client_modeset_probe] No connectors reported connected with modes [drm:drm_client_modeset_probe] [CONNECTOR:46:HDMI-A-1] enabled? no [drm:drm_client_firmware_config.isra.0] Not using firmware configuration [drm:drm_client_modeset_probe] picking CRTCs for 3840x2160 config [drm:drm_client_hotplug] fbdev: ret=0
drm_bridge_hpd_notify():
dw_hdmi_bridge_hpd_notify(status=2)
[drm:drm_sysfs_connector_hotplug_event] [CONNECTOR:46:HDMI-A-1] generating connector hotplug event
drm_client_hotplug():
[drm:drm_fb_helper_hotplug_event]
[drm:drm_client_modeset_probe]
[drm:drm_helper_probe_single_connector_modes] [CONNECTOR:46:HDMI-A-1] dw_hdmi_bridge_hpd_notify(status=2)
[drm:drm_helper_probe_single_connector_modes] [CONNECTOR:46:HDMI-A-1] disconnected [drm:drm_client_modeset_probe] No connectors reported connected with modes [drm:drm_client_modeset_probe] [CONNECTOR:46:HDMI-A-1] enabled? no [drm:drm_client_firmware_config.isra.0] Not using firmware configuration [drm:drm_client_modeset_probe] picking CRTCs for 3840x2160 config [drm:drm_client_hotplug] fbdev: ret=0
Change to only call drm_helper_hpd_irq_event() from HPD IRQ handler to ensure that only one hotplug uevent is sent to userspace when connection status or EDID changes.
With only a call the drm_helper_hpd_irq_event() there is only a single hotplug uevent and only two .hpd_notify() calls:
dw_hdmi_irq(): EVENT=plugout
drm_helper_hpd_irq_event():
dw_hdmi_bridge_hpd_notify(status=2)
[drm:check_connector_changed] [CONNECTOR:46:HDMI-A-1] status updated from connected to disconnected [drm:check_connector_changed] [CONNECTOR:46:HDMI-A-1] Changed epoch counter 1 => 2 [drm:drm_sysfs_connector_hotplug_event] [CONNECTOR:46:HDMI-A-1] generating connector hotplug event
drm_client_hotplug():
[drm:drm_fb_helper_hotplug_event]
[drm:drm_client_modeset_probe]
[drm:drm_helper_probe_single_connector_modes] [CONNECTOR:46:HDMI-A-1] dw_hdmi_bridge_hpd_notify(status=2)
[drm:drm_helper_probe_single_connector_modes] [CONNECTOR:46:HDMI-A-1] disconnected [drm:drm_edid_connector_update] [CONNECTOR:46:HDMI-A-1] EDID changed, epoch counter 3 [drm:drm_client_modeset_probe] No connectors reported connected with modes [drm:drm_client_modeset_probe] [CONNECTOR:46:HDMI-A-1] enabled? no [drm:drm_client_firmware_config.isra.0] Not using firmware configuration [drm:drm_client_modeset_probe] picking CRTCs for 3840x2160 config [drm:drm_client_hotplug] fbdev: ret=0
---
v7: No change
v6: Drop the call from IRQ handler instead, prior to use of a HPD
delayed work to avoid a possible deadlock with use of sync() calls in the bridge hpd_disable() ops,
Update commit message,
Collect t-b tag
v5: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +1 -3
@@ -3101,10 +3101,8 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) status == connector_status_connected ? "plugin" : "plugout"); -if (hdmi->bridge.dev) { +if (hdmi->bridge.dev)drm_helper_hpd_irq_event (hdmi->bridge.dev); -drm_bridge_hpd_notify (&hdmi->bridge, status); - } }hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 );
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v7 18/23] drm: bridge: dw_hdmi: Drop call to drm_bridge_hpd_notify()
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] Removing `drm_bridge_hpd_notify()` violates the `DRM_BRIDGE_OP_HPD` contract and replaces targeted…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Removing drm_bridge_hpd_notify() violates the DRM_BRIDGE_OP_HPD contract and replaces targeted bridge HPD notification with an inefficient global device poll. --
commit f2225f0de66ea8205ddbd49531c0c402dab0c72b
Author: Jonas Karlman <[email protected]>
drm: bridge: dw_hdmi: Drop call to drm_bridge_hpd_notify()
This commit removes the call to drm_bridge_hpd_notify() from the dw_hdmi hotplug interrupt handler. It relies instead solely on drm_helper_hpd_irq_event() to trigger hotplug events in order to prevent multiple hotplug uevents and redundant modeset probes.
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) · line 3107
Quoted 13 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +1
-3@@ -3101,10 +3101,8 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) status == connector_status_connected ? "plugin" : "plugout"); -if (hdmi->bridge.dev) { +if (hdmi->bridge.dev)drm_helper_hpd_irq_event (hdmi->bridge.dev); -drm_bridge_hpd_notify (&hdmi->bridge, status); - }
Does removing the drm_bridge_hpd_notify() call violate the DRM_BRIDGE_OP_HPD API contract?
The dw_hdmi driver is registered with the DRM_BRIDGE_OP_HPD flag, which typically requires the driver to call drm_bridge_hpd_notify() on status changes so that registered callbacks like drm_bridge_connector_hpd_cb() are properly invoked.
By relying solely on drm_helper_hpd_irq_event(), doesn't this replace the targeted bridge HPD notification with a global device poll?
Looking at drm_helper_hpd_irq_event() in drm_probe_helper.c, it acquires dev->mode_config.mutex and iterates over all connectors on the DRM device, calling their detect() functions synchronously. In multi-display systems (for example, HDMI alongside eDP or DSI), will an HDMI hotplug interrupt now trigger slow DDC/I2C/AUX probes on those completely unrelated connectors?
While later patches or discussions might address using delayed_work to debounce hotplug events to avoid mutex deadlocks, could the driver resolve those locking issues without bypassing the bridge framework's targeted notification mechanism and forcing an inefficient global poll?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=18
Jonas Karlman
reviewerPATCH v7 19/23 drm: bridge: dw_hdmi: Use delayed_work to debounce hotplug event
In reply to Jonas KarlmanHDMI Specification Version 1.4b chapter 8.5 mentions:
An HDMI Sink shall not assert high voltage level on its Hot Plug Detect pin when the E-EDID is not available for reading.
A Source may use a high voltage level Hot Plug Detect signal to initiate the reading of E-EDID data.
An HDMI Sink shall indicate any change to the contents of the E-EDID by driving a low voltage level pulse on the Hot Plug Detect pin. This pulse shall be at least 100 msec.
Use a delayed work to debounce reacting on HPD events to improve handling of a HPD low voltage level pulse when a sink changes the EDID.
The delayed work is only enabled between enable_hpd()/hpd_enable() and disable_hpd()/hpd_disable() calls from core, i.e. enabled after attach/bind/resume and disabled before detach/unbind/suspend.
The 1100 msec hotplug debounce timeout was arbitrarily picked to match other drivers using same const, and testing using a Raspberry Pi Monitor seem to use a 200-300 msec pulse when going from standby to power on state.
---
v7: Change to free irq before mute and clear using IH regs, also include
clear of STAT0_RX_SENSE
v6: Change back to disable_delayed_work_sync() in hpd disable ops,
Ensure HPD interrupt is masked and IRQ handler is disabled early in dw_hdmi_remove() to prevent any irq re-arming of delayed work, Drop use of suspend helper
v5: Change to none-sync disable_delayed_work() in hpd disable ops,
Change to cancel_delayed_work_sync() in remove,
Add cancel_delayed_work_sync() to new suspend helper
v4: Disable/mask delayed_work until enable_hpd()/hpd_enable(),
Read connector status directly from HW regs in hpd_work
v3: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 80 +++++++++++++++++++++-- 1 file changed, 75 insertions(+), 5 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +75 -5
@@ -50,6 +50,8 @@#define HDMI14_MAX_TMDSCLK 340000000 +#define HOTPLUG_DEBOUNCE_MS 1100 +static const u16 csc_coeff_default[3 ][4 ] = { {0x2000 ,0x0000 ,0x0000 ,0x0000 }, {0x0000 ,0x2000 ,0x0000 ,0x0000 },
@@ -185,6 +187,7 @@ struct dw_hdmi {hdmi_codec_plugged_cb plugged_cb;struct device *codec_dev;enum drm_connector_status last_connector_result; +struct delayed_work hpd_work; };const struct dw_hdmi_plat_data *dw_hdmi_to_plat_data (struct dw_hdmi *hdmi)
@@ -2517,6 +2520,20 @@ static void dw_hdmi_connector_force(struct drm_connector *connector)dw_hdmi_connector_status_update (hdmi, connector, connector->status); } +static void dw_hdmi_connector_enable_hpd (struct drm_connector *connector) +{ + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); + +enable_delayed_work (&hdmi->hpd_work); +} + +static void dw_hdmi_connector_disable_hpd (struct drm_connector *connector) +{ + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); + +disable_delayed_work_sync (&hdmi->hpd_work); +} +static void dw_hdmi_connector_destroy (struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);
@@ -2538,6 +2555,8 @@ static const struct drm_connector_funcs dw_hdmi_connector_funcs = {static const struct drm_connector_helper_funcs dw_hdmi_connector_helper_funcs = { .get_modes = dw_hdmi_connector_get_modes, .atomic_check = dw_hdmi_connector_atomic_check, + .enable_hpd = dw_hdmi_connector_enable_hpd, + .disable_hpd = dw_hdmi_connector_disable_hpd, };static int dw_hdmi_connector_create (struct dw_hdmi *hdmi)
@@ -2968,6 +2987,20 @@ static const struct drm_edid *dw_hdmi_bridge_edid_read(struct drm_bridge *bridgereturn dw_hdmi_edid_read (hdmi, connector); } +static void dw_hdmi_bridge_hpd_enable (struct drm_bridge *bridge) +{ +struct dw_hdmi *hdmi = bridge->driver_private; + +enable_delayed_work (&hdmi->hpd_work); +} + +static void dw_hdmi_bridge_hpd_disable (struct drm_bridge *bridge) +{ +struct dw_hdmi *hdmi = bridge->driver_private; + +disable_delayed_work_sync (&hdmi->hpd_work); +} +static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .atomic_duplicate_state = drm_atomic_helper_bridge_duplicate_state, .atomic_destroy_state = drm_atomic_helper_bridge_destroy_state,
@@ -2981,6 +3014,8 @@ static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .mode_valid = dw_hdmi_bridge_mode_valid, .detect = dw_hdmi_bridge_detect, .edid_read = dw_hdmi_bridge_edid_read, + .hpd_enable = dw_hdmi_bridge_hpd_enable, + .hpd_disable = dw_hdmi_bridge_hpd_disable, };/* -----------------------------------------------------------------------------
@@ -3101,8 +3136,8 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) status == connector_status_connected ? "plugin" : "plugout"); -if (hdmi->bridge.dev) -drm_helper_hpd_irq_event (hdmi->bridge.dev); +mod_delayed_work (system_percpu_wq, &hdmi->hpd_work, +msecs_to_jiffies (HOTPLUG_DEBOUNCE_MS )); }hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 );
@@ -3112,6 +3147,29 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id)return IRQ_HANDLED ; } +static void dw_hdmi_hpd_work (struct work_struct *work) +{ + struct dw_hdmi *hdmi = container_of(work, struct dw_hdmi, hpd_work.work); +struct drm_device *dev = hdmi->bridge.dev; + +if (WARN_ON (!dev)) +return ; + +/* +* Notify the DRM core of the HPD event using drm_helper_hpd_irq_event() +* instead of drm_bridge_hpd_notify(). This will cause the DRM function +* check_connector_changed() to be called, which in turn calls the +* connector detect()/force() funcs to detect any connection status or +* epoch changes. The bridge connector detect() func also ensures that +* any hpd_notify() funcs are called for all bridges in the chain. +* +* drm_bridge_hpd_notify() shares a mutex with drm_bridge_hpd_disable(), +* and can result in a deadlock due to the disable_delayed_work_sync() +* call to wait on work to complete in dw_hdmi_bridge_hpd_disable(). +*/ +drm_helper_hpd_irq_event (dev); +} +static const struct dw_hdmi_phy_data dw_hdmi_phys[] = { { .type =DW_HDMI_PHY_DWC_HDMI_TX_PHY ,
@@ -3396,6 +3454,9 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev, goto err_res; } +INIT_DELAYED_WORK (&hdmi->hpd_work, dw_hdmi_hpd_work); +disable_delayed_work (&hdmi->hpd_work); + ret =devm_request_threaded_irq (dev, irq, dw_hdmi_hardirq, dw_hdmi_irq,IRQF_SHARED ,dev_name (dev), hdmi);
@@ -3532,6 +3593,18 @@ EXPORT_SYMBOL_GPL(dw_hdmi_probe);void dw_hdmi_remove (struct dw_hdmi *hdmi) { +struct platform_device *pdev =to_platform_device (hdmi->dev); +int irq =platform_get_irq (pdev,0 ); + +/* Free, mute and clear phy interrupts */ +devm_free_irq (hdmi->dev, irq, hdmi); +hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , +HDMI_IH_PHY_STAT0 ); + +/* Cancel any pending hot plug work */ +cancel_delayed_work_sync (&hdmi->hpd_work); +drm_bridge_remove (&hdmi->bridge);if (hdmi->audio && !IS_ERR (hdmi->audio))
@@ -3539,9 +3612,6 @@ void dw_hdmi_remove(struct dw_hdmi *hdmi)if (!IS_ERR (hdmi->cec))platform_device_unregister (hdmi->cec); -/* Disable all interrupts */ -hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); -if (hdmi->i2c)i2c_del_adapter (&hdmi->i2c->adap);else
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v7 19/23] drm: bridge: dw_hdmi: Use delayed_work to debounce hotplug event
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] The debounce logic completely filters out the 100ms EDID change pulse, causing the DRM core to…
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The debounce logic completely filters out the 100ms EDID change pulse, causing the DRM core to miss the hotplug event.
- [High] The shared IRQ handler is freed before hardware interrupts are masked, potentially causing an unhandled interrupt storm. --
commit 2b07e0d99061a1389b7eef1c604f98ccf0e7d115
Author: Jonas Karlman <[email protected]>
drm: bridge: dw_hdmi: Use delayed_work to debounce hotplug event
This commit introduces a delayed work queue to debounce HDMI hotplug detect (HPD) events. It aims to improve the handling of EDID change pulses from the sink by applying an 1100 msec timeout between the event and the DRM core notification.
Quoted 3 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-0
[ ... ]
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) · line 3141
Quoted 9 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +2
-2@@ -3101,8 +3136,8 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) status == connector_status_connected ? "plugin" : "plugout"); -if (hdmi->bridge.dev) -drm_helper_hpd_irq_event (hdmi->bridge.dev); +mod_delayed_work (system_percpu_wq, &hdmi->hpd_work, +msecs_to_jiffies (HOTPLUG_DEBOUNCE_MS )); }
Does this debounce logic unintentionally filter out the 100ms EDID change pulse entirely?
When the sink signals an EDID change by pulsing HPD low for 100ms and then high again, mod_delayed_work() will be called on both edges. The second call on the rising edge will reset the timer, pushing the work execution to 1100ms after the pin returns high.
When dw_hdmi_hpd_work() finally runs and calls drm_helper_hpd_irq_event(), the DRM connector detect function will see the current HPD status as connected. Since the previous status known to the DRM core was also connected (because the brief disconnect was never sampled), check_connector_changed() will conclude that nothing changed. The epoch counter will not increment, and the userspace uevent might be lost.
[ ... ]
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c EXPORT_SYMBOL_GPL(dw_hdmi_probe); · line 3603
Quoted 12 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +8
-0@@ -3532,6 +3593,18 @@ EXPORT_SYMBOL_GPL(dw_hdmi_probe);void dw_hdmi_remove (struct dw_hdmi *hdmi) { +struct platform_device *pdev =to_platform_device (hdmi->dev); +int irq =platform_get_irq (pdev,0 ); + +/* Free, mute and clear phy interrupts */ +devm_free_irq (hdmi->dev, irq, hdmi); +hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , +HDMI_IH_PHY_STAT0 );
Could freeing the shared IRQ handler before masking the hardware interrupts cause an unhandled interrupt storm?
If the HDMI device asserts a level-triggered interrupt exactly when devm_free_irq() is called, the interrupt line will stay asserted but the dw_hdmi handler will no longer be registered to clear it. Because the IRQ is shared (requested with IRQF_SHARED), the kernel will repeatedly invoke other handlers on this IRQ line.
Those other handlers will all return IRQ_NONE, which can cause the kernel to permanently disable the shared IRQ line.
Should the hardware interrupts be explicitly masked and cleared before the handler is freed?
Quoted 5 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +4
-0+ +/* Cancel any pending hot plug work */ +cancel_delayed_work_sync (&hdmi->hpd_work); +drm_bridge_remove (&hdmi->bridge);
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=19
Neil Armstrong
maintainerRe: [PATCH v7 19/23] drm: bridge: dw_hdmi: Use delayed_work to debounce hotplug event
In reply to Jonas KarlmanHi,
Quoted 15 lines · Jonas Karlman
HDMI Specification Version 1.4b chapter 8.5 mentions:
An HDMI Sink shall not assert high voltage level on its Hot Plug Detect pin when the E-EDID is not available for reading.
A Source may use a high voltage level Hot Plug Detect signal to initiate the reading of E-EDID data.
An HDMI Sink shall indicate any change to the contents of the E-EDID by driving a low voltage level pulse on the Hot Plug Detect pin. This pulse shall be at least 100 msec.
Use a delayed work to debounce reacting on HPD events to improve handling of a HPD low voltage level pulse when a sink changes the EDID.
Quoted 8 lines
The delayed work is only enabled between enable_hpd()/hpd_enable() and disable_hpd()/hpd_disable() calls from core, i.e. enabled after attach/bind/resume and disabled before detach/unbind/suspend.
The 1100 msec hotplug debounce timeout was arbitrarily picked to match other drivers using same const, and testing using a Raspberry Pi Monitor seem to use a 200-300 msec pulse when going from standby to power on state.
The logic looks ok, but I'm puzzled by the 1.1 sec debounce, which after plugging in a monitor will only send an irq event after 1.1s which is very long.
Since the spec says 100ms and the real worls values are more like 200-300ms, I would first reduce this to 500ms.
But as I understand the code right now, on the first HPD front the irq work is programmed to run after the debounce time, but if it's a pulse the irq would also trigger on the second HPD front and then delay again the work after the debounce time.
My understanding of a debounce was that we "ignore" the pulse by only generating a single irq event when the pulse is finished.
The current code does that, we will only have a single irq event and the HPD will return as connected state, good. But this delays the irq event 1.1s _after_ the end of the pulse, which I would expect the event to be send at tht debounce time after the start of the pulse.
Like, program the work at the beginning of the pulse, if somehow the pulse ends before the debounce time, send the irq event immediately, otherwise let the debounce work run after the debounce time which will trigger a disconnect event.
But the delay is too high, 1.1s could be a manual unplug/plug or bad connector with false contact on the hpd pin.
I would rather reduce this to something more realistic like 500ms or less and try to better handle the pulse somehow. But I don't have any idea if the scheme I described is doable.
Neil
Quoted 180 lines
Signed-off-by Jonas Karlman <[email protected]>---
v7: Change to free irq before mute and clear using IH regs, also include
clear ofSTAT0_RX_SENSE
v6: Change back todisable_delayed_work_sync()in hpd disable ops,
Ensure HPD interrupt is masked and IRQ handler is disabled early indw_hdmi_remove()to prevent any irq re-arming of delayed work, Drop use of suspend helper
v5: Change to none-syncdisable_delayed_work()in hpd disable ops,
Change tocancel_delayed_work_sync()in remove,
Addcancel_delayed_work_sync()to new suspend helper
v4: Disable/maskdelayed_workuntil enable_hpd()/hpd_enable(),
Read connector status directly from HW regs inhpd_work
v3: New patch--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 80 +++++++++++++++++++++-- 1 file changed, 75 insertions(+), 5 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +75
-5@@ -50,6 +50,8 @@#define HDMI14_MAX_TMDSCLK 340000000 +#define HOTPLUG_DEBOUNCE_MS 1100 +static const u16 csc_coeff_default[3 ][4 ] = { {0x2000 ,0x0000 ,0x0000 ,0x0000 }, {0x0000 ,0x2000 ,0x0000 ,0x0000 },@@ -185,6 +187,7 @@ struct dw_hdmi {hdmi_codec_plugged_cb plugged_cb;struct device *codec_dev;enum drm_connector_status last_connector_result; +struct delayed_work hpd_work; };const struct dw_hdmi_plat_data *dw_hdmi_to_plat_data (struct dw_hdmi *hdmi)@@ -2517,6 +2520,20 @@ static void dw_hdmi_connector_force(struct drm_connector *connector)dw_hdmi_connector_status_update (hdmi, connector, connector->status); } +static void dw_hdmi_connector_enable_hpd (struct drm_connector *connector) +{ + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); + +enable_delayed_work (&hdmi->hpd_work); +} + +static void dw_hdmi_connector_disable_hpd (struct drm_connector *connector) +{ + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); + +disable_delayed_work_sync (&hdmi->hpd_work); +} +static void dw_hdmi_connector_destroy (struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);@@ -2538,6 +2555,8 @@ static const struct drm_connector_funcs dw_hdmi_connector_funcs = {static const struct drm_connector_helper_funcs dw_hdmi_connector_helper_funcs = { .get_modes = dw_hdmi_connector_get_modes, .atomic_check = dw_hdmi_connector_atomic_check, + .enable_hpd = dw_hdmi_connector_enable_hpd, + .disable_hpd = dw_hdmi_connector_disable_hpd, };static int dw_hdmi_connector_create (struct dw_hdmi *hdmi)@@ -2968,6 +2987,20 @@ static const struct drm_edid *dw_hdmi_bridge_edid_read(struct drm_bridge *bridgereturn dw_hdmi_edid_read (hdmi, connector); } +static void dw_hdmi_bridge_hpd_enable (struct drm_bridge *bridge) +{ +struct dw_hdmi *hdmi = bridge->driver_private; + +enable_delayed_work (&hdmi->hpd_work); +} + +static void dw_hdmi_bridge_hpd_disable (struct drm_bridge *bridge) +{ +struct dw_hdmi *hdmi = bridge->driver_private; + +disable_delayed_work_sync (&hdmi->hpd_work); +} +static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .atomic_duplicate_state = drm_atomic_helper_bridge_duplicate_state, .atomic_destroy_state = drm_atomic_helper_bridge_destroy_state,@@ -2981,6 +3014,8 @@ static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .mode_valid = dw_hdmi_bridge_mode_valid, .detect = dw_hdmi_bridge_detect, .edid_read = dw_hdmi_bridge_edid_read, + .hpd_enable = dw_hdmi_bridge_hpd_enable, + .hpd_disable = dw_hdmi_bridge_hpd_disable, };/* ----------------------------------------------------------------------------- @@ -3101,8 +3136,8 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) status == connector_status_connected ? "plugin" : "plugout"); -if (hdmi->bridge.dev) -drm_helper_hpd_irq_event (hdmi->bridge.dev); +mod_delayed_work (system_percpu_wq, &hdmi->hpd_work, +msecs_to_jiffies (HOTPLUG_DEBOUNCE_MS )); }hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 );@@ -3112,6 +3147,29 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id)return IRQ_HANDLED ; } +static void dw_hdmi_hpd_work (struct work_struct *work) +{ + struct dw_hdmi *hdmi = container_of(work, struct dw_hdmi, hpd_work.work); +struct drm_device *dev = hdmi->bridge.dev; + +if (WARN_ON (!dev)) +return ; + +/* +* Notify the DRM core of the HPD event using drm_helper_hpd_irq_event() +* instead of drm_bridge_hpd_notify(). This will cause the DRM function +* check_connector_changed() to be called, which in turn calls the +* connector detect()/force() funcs to detect any connection status or +* epoch changes. The bridge connector detect() func also ensures that +* any hpd_notify() funcs are called for all bridges in the chain. +* +* drm_bridge_hpd_notify() shares a mutex with drm_bridge_hpd_disable(), +* and can result in a deadlock due to the disable_delayed_work_sync() +* call to wait on work to complete in dw_hdmi_bridge_hpd_disable(). +*/ +drm_helper_hpd_irq_event (dev); +} +static const struct dw_hdmi_phy_data dw_hdmi_phys[] = { { .type =DW_HDMI_PHY_DWC_HDMI_TX_PHY ,@@ -3396,6 +3454,9 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev, goto err_res; } +INIT_DELAYED_WORK (&hdmi->hpd_work, dw_hdmi_hpd_work); +disable_delayed_work (&hdmi->hpd_work); + ret =devm_request_threaded_irq (dev, irq, dw_hdmi_hardirq, dw_hdmi_irq,IRQF_SHARED ,dev_name (dev), hdmi);@@ -3532,6 +3593,18 @@ EXPORT_SYMBOL_GPL(dw_hdmi_probe);void dw_hdmi_remove (struct dw_hdmi *hdmi) { +struct platform_device *pdev =to_platform_device (hdmi->dev); +int irq =platform_get_irq (pdev,0 ); + +/* Free, mute and clear phy interrupts */ +devm_free_irq (hdmi->dev, irq, hdmi); +hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , +HDMI_IH_PHY_STAT0 ); + +/* Cancel any pending hot plug work */ +cancel_delayed_work_sync (&hdmi->hpd_work); +drm_bridge_remove (&hdmi->bridge);if (hdmi->audio && !IS_ERR (hdmi->audio))@@ -3539,9 +3612,6 @@ void dw_hdmi_remove(struct dw_hdmi *hdmi)if (!IS_ERR (hdmi->cec))platform_device_unregister (hdmi->cec); -/* Disable all interrupts */ -hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); -if (hdmi->i2c)i2c_del_adapter (&hdmi->i2c->adap);else
Jonas Karlman
reviewerRe: [PATCH v7 19/23] drm: bridge: dw_hdmi: Use delayed_work to debounce hotplug event
Hi Neil,
Quoted 27 lines · Neil Armstrong
Hi,
Quoted 15 lines · Jonas Karlman
HDMI Specification Version 1.4b chapter 8.5 mentions:
An HDMI Sink shall not assert high voltage level on its Hot Plug Detect pin when the E-EDID is not available for reading.
A Source may use a high voltage level Hot Plug Detect signal to initiate the reading of E-EDID data.
An HDMI Sink shall indicate any change to the contents of the E-EDID by driving a low voltage level pulse on the Hot Plug Detect pin. This pulse shall be at least 100 msec.
Use a delayed work to debounce reacting on HPD events to improve handling of a HPD low voltage level pulse when a sink changes the EDID.
Quoted 8 lines
The delayed work is only enabled between enable_hpd()/hpd_enable() and disable_hpd()/hpd_disable() calls from core, i.e. enabled after attach/bind/resume and disabled before detach/unbind/suspend.
The 1100 msec hotplug debounce timeout was arbitrarily picked to match other drivers using same const, and testing using a Raspberry Pi Monitor seem to use a 200-300 msec pulse when going from standby to power on state.
Quoted 5 lines
The logic looks ok, but I'm puzzled by the 1.1 sec debounce, which after plugging in a monitor will only send an irq event after 1.1s which is very long.
Since the spec says 100ms and the real worls values are more like 200-300ms, I would first reduce this to 500ms.
You are correct, this value was picked based on existing values used by other drivers:
exynos/exynos_hdmi.c:#define HOTPLUG_DEBOUNCE_MS 1100
bridge/ti-tfp410.c:#define HOTPLUG_DEBOUNCE_MS 1100
amd/display/amdgpu_dm/amdgpu_dm.h:#define AMDGPU_DM_MAX_HDMI_HPD_DEBOUNCE_MS 5000
rockchip/dw_hdmi_qp-rockchip.c:#define HOTPLUG_DEBOUNCE_MS 150
150 ms was too short for my test monitor (Raspberry Pi Monitor), it does a HPD low voltage level pulse when powering on of around 200+ ms.
[82.641903] dw_hdmi_hardirq: EVENT=plugout [82.841939] dw_hdmi_hardirq: EVENT=plugin
And on my LG OLED 4K TV there was typically a pulse of around 700-900 ms. So the 1.1 seconds used by other drivers seemed like a good candidate :-)
Quoted 4 lines
But as I understand the code right now, on the first HPD front the irq work is programmed to run after the debounce time, but if it's a pulse the irq would also trigger on the second HPD front and then delay again the work after the debounce time.
Your analysis is correct, any HPD event would keep debouncing adding 1.1 timout from each HPD irq, both plugout and plugin.
The intention was to allow for up to 1.1 seconds to pass between a plugout and a plugin, before we treated a plugout as a full disconnect event, and not to delay a possible connected event by 1.1 seconds.
Quoted 2 lines
My understanding of a debounce was that we "ignore" the pulse by only generating a single irq event when the pulse is finished.
Correct, as long as the pulse was less than 1.1 seconds.
Quoted 4 lines
The current code does that, we will only have a single irq event and the HPD will return as connected state, good. But this delays the irq event 1.1s _after_ the end of the pulse, which I would expect the event to be send at tht debounce time after the start of the pulse.
Good catch and I agree, we should delay/timeout the disconnected state longer than plugin and react on plugin almost immediately.
Quoted 10 lines
Like, program the work at the beginning of the pulse, if somehow the pulse ends before the debounce time, send the irq event immediately, otherwise let the debounce work run after the debounce time which will trigger a disconnect event.
But the delay is too high, 1.1s could be a manual unplug/plug or bad connector with false contact on the hpd pin.
I would rather reduce this to something more realistic like 500ms or less and try to better handle the pulse somehow. But I don't have any idea if the scheme I described is doable.
We can configure different delays for plugout and plugin, I am currently testing something like following:
#define HOTPLUG_CONNECTED_DEBOUNCE_MS 150 #define HOTPLUG_DISCONNECTED_DEBOUNCE_MS 500
delay = status == connector_status_connected ?
HOTPLUG_CONNECTED_DEBOUNCE_MS : HOTPLUG_DISCONNECTED_DEBOUNCE_MS; mod_delayed_work(system_percpu_wq, &hdmi->hpd_work, msecs_to_jiffies(delay));
That would mean:
- at plugout we (re-)start the timer to trigger in 500ms
- at plugin we (re-)start the timer to trigger in 150ms
- whenever the timer triggers the hpd_work is started
- the hpw_work trigger normal detect() handling to determin if connector status (or EDID) has changed
Example during power on of a Raspberry Pi Monitor:
[82.641903] dw_hdmi_hardirq: EVENT=plugout [82.648276] dw_hdmi_schedule_hpd_work(status=2) [82.841939] dw_hdmi_hardirq: EVENT=plugin [82.848211] dw_hdmi_schedule_hpd_work(status=1) [83.008573] dw_hdmi_hpd_work(): START [83.020358] dw_hdmi_bridge_detect() [83.034958] dw_hdmi_bridge_edid_read() [83.187102] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] CEA VCDB 0x4a [83.194471] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] HDMI: DVI dual 0, max TMDS clock 0 kHz [83.201755] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] ELD monitor RPI MON156 [83.208968] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] HDMI: latency present 0 0, video latency 0 0, audio latency 0 0 [83.216588] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] ELD size 36, SAD count 1 [83.224445] [drm:check_connector_changed] [CONNECTOR:55:HDMI-A-1] Same epoch counter 2 [83.231553] dw_hdmi_hpd_work(): END
Around 205ms between HPD=0 and HPD=1 and around 160ms from plugin until delayed hpd_work starts.
And at full plugout the debounce time is around 497 ms:
[95.125105] dw_hdmi_hardirq: EVENT=plugout [95.147586] dw_hdmi_schedule_hpd_work(status=2) [95.645277] dw_hdmi_hpd_work(): START [95.667741] dw_hdmi_bridge_detect() [95.691523] [drm:drm_edid_connector_update] [CONNECTOR:55:HDMI-A-1] EDID changed, epoch counter 3 [95.709861] [drm:check_connector_changed] [CONNECTOR:55:HDMI-A-1] status updated from connected to disconnected [95.722674] [drm:check_connector_changed] [CONNECTOR:55:HDMI-A-1] Changed epoch counter 2 => 4 [95.735173] [drm:drm_sysfs_connector_hotplug_event] [CONNECTOR:55:HDMI-A-1] generating connector hotplug event [95.746326] [drm:drm_fb_helper_hotplug_event] [95.754871] [drm:drm_client_modeset_probe] [95.762429] [drm:drm_helper_probe_single_connector_modes] [CONNECTOR:55:HDMI-A-1] [95.769758] dw_hdmi_bridge_detect() [95.776597] [drm:drm_helper_probe_single_connector_modes] [CONNECTOR:55:HDMI-A-1] disconnected [95.784389] [drm:drm_client_modeset_probe] No connectors reported connected with modes [95.791865] [drm:drm_client_modeset_probe] [CONNECTOR:55:HDMI-A-1] enabled? no [95.799426] [drm:drm_client_firmware_config.isra.0] Not using firmware configuration [95.807140] [drm:drm_client_modeset_probe] picking CRTCs for 1920x1080 config [95.818422] dw_hdmi_bridge_atomic_disable() [95.826412] [drm:vop2_plane_atomic_disable] Smart0-win0 disable [95.868341] [drm:drm_client_hotplug] fbdev: ret=0 [95.878206] dw_hdmi_hpd_work(): END
With a much quicker reaction on plugin event maybe the 1.1 seconds could stay to avoid having to do a full teardown, disable() + enable() cycle, at slightly longer HPD pulses.
Or maybe it is time to re-spin an old cec-adapter debounce series [1] that can allow for some additional time until the CEC phys addr is fully invalidated at a longer HPD low voltage level pulse.
Regards,
Jonas
Quoted 184 lines
Neil
Quoted 180 lines
Signed-off-by Jonas Karlman <[email protected]>---
v7: Change to free irq before mute and clear using IH regs, also include
clear ofSTAT0_RX_SENSE
v6: Change back todisable_delayed_work_sync()in hpd disable ops,
Ensure HPD interrupt is masked and IRQ handler is disabled early indw_hdmi_remove()to prevent any irq re-arming of delayed work, Drop use of suspend helper
v5: Change to none-syncdisable_delayed_work()in hpd disable ops,
Change tocancel_delayed_work_sync()in remove,
Addcancel_delayed_work_sync()to new suspend helper
v4: Disable/maskdelayed_workuntil enable_hpd()/hpd_enable(),
Read connector status directly from HW regs inhpd_work
v3: New patch--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 80 +++++++++++++++++++++-- 1 file changed, 75 insertions(+), 5 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +75
-5@@ -50,6 +50,8 @@#define HDMI14_MAX_TMDSCLK 340000000 +#define HOTPLUG_DEBOUNCE_MS 1100 +static const u16 csc_coeff_default[3 ][4 ] = { {0x2000 ,0x0000 ,0x0000 ,0x0000 }, {0x0000 ,0x2000 ,0x0000 ,0x0000 },@@ -185,6 +187,7 @@ struct dw_hdmi {hdmi_codec_plugged_cb plugged_cb;struct device *codec_dev;enum drm_connector_status last_connector_result; +struct delayed_work hpd_work; };const struct dw_hdmi_plat_data *dw_hdmi_to_plat_data (struct dw_hdmi *hdmi)@@ -2517,6 +2520,20 @@ static void dw_hdmi_connector_force(struct drm_connector *connector)dw_hdmi_connector_status_update (hdmi, connector, connector->status); } +static void dw_hdmi_connector_enable_hpd (struct drm_connector *connector) +{ + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); + +enable_delayed_work (&hdmi->hpd_work); +} + +static void dw_hdmi_connector_disable_hpd (struct drm_connector *connector) +{ + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); + +disable_delayed_work_sync (&hdmi->hpd_work); +} +static void dw_hdmi_connector_destroy (struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);@@ -2538,6 +2555,8 @@ static const struct drm_connector_funcs dw_hdmi_connector_funcs = {static const struct drm_connector_helper_funcs dw_hdmi_connector_helper_funcs = { .get_modes = dw_hdmi_connector_get_modes, .atomic_check = dw_hdmi_connector_atomic_check, + .enable_hpd = dw_hdmi_connector_enable_hpd, + .disable_hpd = dw_hdmi_connector_disable_hpd, };static int dw_hdmi_connector_create (struct dw_hdmi *hdmi)@@ -2968,6 +2987,20 @@ static const struct drm_edid *dw_hdmi_bridge_edid_read(struct drm_bridge *bridgereturn dw_hdmi_edid_read (hdmi, connector); } +static void dw_hdmi_bridge_hpd_enable (struct drm_bridge *bridge) +{ +struct dw_hdmi *hdmi = bridge->driver_private; + +enable_delayed_work (&hdmi->hpd_work); +} + +static void dw_hdmi_bridge_hpd_disable (struct drm_bridge *bridge) +{ +struct dw_hdmi *hdmi = bridge->driver_private; + +disable_delayed_work_sync (&hdmi->hpd_work); +} +static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .atomic_duplicate_state = drm_atomic_helper_bridge_duplicate_state, .atomic_destroy_state = drm_atomic_helper_bridge_destroy_state,@@ -2981,6 +3014,8 @@ static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .mode_valid = dw_hdmi_bridge_mode_valid, .detect = dw_hdmi_bridge_detect, .edid_read = dw_hdmi_bridge_edid_read, + .hpd_enable = dw_hdmi_bridge_hpd_enable, + .hpd_disable = dw_hdmi_bridge_hpd_disable, };/* ----------------------------------------------------------------------------- @@ -3101,8 +3136,8 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) status == connector_status_connected ? "plugin" : "plugout"); -if (hdmi->bridge.dev) -drm_helper_hpd_irq_event (hdmi->bridge.dev); +mod_delayed_work (system_percpu_wq, &hdmi->hpd_work, +msecs_to_jiffies (HOTPLUG_DEBOUNCE_MS )); }hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 );@@ -3112,6 +3147,29 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id)return IRQ_HANDLED ; } +static void dw_hdmi_hpd_work (struct work_struct *work) +{ + struct dw_hdmi *hdmi = container_of(work, struct dw_hdmi, hpd_work.work); +struct drm_device *dev = hdmi->bridge.dev; + +if (WARN_ON (!dev)) +return ; + +/* +* Notify the DRM core of the HPD event using drm_helper_hpd_irq_event() +* instead of drm_bridge_hpd_notify(). This will cause the DRM function +* check_connector_changed() to be called, which in turn calls the +* connector detect()/force() funcs to detect any connection status or +* epoch changes. The bridge connector detect() func also ensures that +* any hpd_notify() funcs are called for all bridges in the chain. +* +* drm_bridge_hpd_notify() shares a mutex with drm_bridge_hpd_disable(), +* and can result in a deadlock due to the disable_delayed_work_sync() +* call to wait on work to complete in dw_hdmi_bridge_hpd_disable(). +*/ +drm_helper_hpd_irq_event (dev); +} +static const struct dw_hdmi_phy_data dw_hdmi_phys[] = { { .type =DW_HDMI_PHY_DWC_HDMI_TX_PHY ,@@ -3396,6 +3454,9 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev, goto err_res; } +INIT_DELAYED_WORK (&hdmi->hpd_work, dw_hdmi_hpd_work); +disable_delayed_work (&hdmi->hpd_work); + ret =devm_request_threaded_irq (dev, irq, dw_hdmi_hardirq, dw_hdmi_irq,IRQF_SHARED ,dev_name (dev), hdmi);@@ -3532,6 +3593,18 @@ EXPORT_SYMBOL_GPL(dw_hdmi_probe);void dw_hdmi_remove (struct dw_hdmi *hdmi) { +struct platform_device *pdev =to_platform_device (hdmi->dev); +int irq =platform_get_irq (pdev,0 ); + +/* Free, mute and clear phy interrupts */ +devm_free_irq (hdmi->dev, irq, hdmi); +hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , +HDMI_IH_PHY_STAT0 ); + +/* Cancel any pending hot plug work */ +cancel_delayed_work_sync (&hdmi->hpd_work); +drm_bridge_remove (&hdmi->bridge);if (hdmi->audio && !IS_ERR (hdmi->audio))@@ -3539,9 +3612,6 @@ void dw_hdmi_remove(struct dw_hdmi *hdmi)if (!IS_ERR (hdmi->cec))platform_device_unregister (hdmi->cec); -/* Disable all interrupts */ -hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); -if (hdmi->i2c)i2c_del_adapter (&hdmi->i2c->adap);else
Neil Armstrong
maintainerRe: [PATCH v7 19/23] drm: bridge: dw_hdmi: Use delayed_work to debounce hotplug event
Quoted 152 lines · Jonas Karlman
Hi Neil,
Quoted 27 lines · Neil Armstrong
Hi,
Quoted 15 lines · Jonas Karlman
HDMI Specification Version 1.4b chapter 8.5 mentions:
An HDMI Sink shall not assert high voltage level on its Hot Plug Detect pin when the E-EDID is not available for reading.
A Source may use a high voltage level Hot Plug Detect signal to initiate the reading of E-EDID data.
An HDMI Sink shall indicate any change to the contents of the E-EDID by driving a low voltage level pulse on the Hot Plug Detect pin. This pulse shall be at least 100 msec.
Use a delayed work to debounce reacting on HPD events to improve handling of a HPD low voltage level pulse when a sink changes the EDID.
Quoted 8 lines
The delayed work is only enabled between enable_hpd()/hpd_enable() and disable_hpd()/hpd_disable() calls from core, i.e. enabled after attach/bind/resume and disabled before detach/unbind/suspend.
The 1100 msec hotplug debounce timeout was arbitrarily picked to match other drivers using same const, and testing using a Raspberry Pi Monitor seem to use a 200-300 msec pulse when going from standby to power on state.
Quoted 5 lines
The logic looks ok, but I'm puzzled by the 1.1 sec debounce, which after plugging in a monitor will only send an irq event after 1.1s which is very long.
Since the spec says 100ms and the real worls values are more like 200-300ms, I would first reduce this to 500ms.
You are correct, this value was picked based on existing values used by other drivers:
exynos/exynos_hdmi.c:#define
HOTPLUG_DEBOUNCE_MS1100
bridge/ti-tfp410.c:#defineHOTPLUG_DEBOUNCE_MS1100
amd/display/amdgpu_dm/amdgpu_dm.h:#defineAMDGPU_DM_MAX_HDMI_HPD_DEBOUNCE_MS5000
rockchip/dw_hdmi_qp-rockchip.c:#defineHOTPLUG_DEBOUNCE_MS150150 ms was too short for my test monitor (Raspberry Pi Monitor), it does a HPD low voltage level pulse when powering on of around 200+ ms.
[82.641903] dw_hdmi_hardirq: EVENT=plugout [82.841939] dw_hdmi_hardirq: EVENT=pluginAnd on my LG OLED 4K TV there was typically a pulse of around 700-900 ms. So the 1.1 seconds used by other drivers seemed like a good candidate :-)
Quoted 4 lines
But as I understand the code right now, on the first HPD front the irq work is programmed to run after the debounce time, but if it's a pulse the irq would also trigger on the second HPD front and then delay again the work after the debounce time.
Your analysis is correct, any HPD event would keep debouncing adding 1.1 timout from each HPD irq, both plugout and plugin.
The intention was to allow for up to 1.1 seconds to pass between a plugout and a plugin, before we treated a plugout as a full disconnect event, and not to delay a possible connected event by 1.1 seconds.
Quoted 2 lines
My understanding of a debounce was that we "ignore" the pulse by only generating a single irq event when the pulse is finished.
Correct, as long as the pulse was less than 1.1 seconds.
Quoted 4 lines
The current code does that, we will only have a single irq event and the HPD will return as connected state, good. But this delays the irq event 1.1s _after_ the end of the pulse, which I would expect the event to be send at tht debounce time after the start of the pulse.
Good catch and I agree, we should delay/timeout the disconnected state longer than plugin and react on plugin almost immediately.
Quoted 10 lines
Like, program the work at the beginning of the pulse, if somehow the pulse ends before the debounce time, send the irq event immediately, otherwise let the debounce work run after the debounce time which will trigger a disconnect event.
But the delay is too high, 1.1s could be a manual unplug/plug or bad connector with false contact on the hpd pin.
I would rather reduce this to something more realistic like 500ms or less and try to better handle the pulse somehow. But I don't have any idea if the scheme I described is doable.
We can configure different delays for plugout and plugin, I am currently testing something like following:
#define HOTPLUG_CONNECTED_DEBOUNCE_MS 150 #define HOTPLUG_DISCONNECTED_DEBOUNCE_MS 500delay = status == connector_status_connected ? HOTPLUG_CONNECTED_DEBOUNCE_MS : HOTPLUG_DISCONNECTED_DEBOUNCE_MS; mod_delayed_work(system_percpu_wq, &hdmi->hpd_work, msecs_to_jiffies(delay));That would mean:
- at plugout we (re-)start the timer to trigger in 500ms
- at plugin we (re-)start the timer to trigger in 150ms
- whenever the timer triggers thehpd_workis started
- thehpw_worktrigger normaldetect()handling to determin if connector status (or EDID) has changedExample during power on of a Raspberry Pi Monitor:
[82.641903] dw_hdmi_hardirq: EVENT=plugout [82.648276] dw_hdmi_schedule_hpd_work(status=2) [82.841939] dw_hdmi_hardirq: EVENT=plugin [82.848211] dw_hdmi_schedule_hpd_work(status=1) [83.008573] dw_hdmi_hpd_work(): START [83.020358] dw_hdmi_bridge_detect() [83.034958] dw_hdmi_bridge_edid_read() [83.187102] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] CEA VCDB 0x4a [83.194471] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] HDMI: DVI dual 0, max TMDS clock 0 kHz [83.201755] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] ELD monitor RPI MON156 [83.208968] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] HDMI: latency present 0 0, video latency 0 0, audio latency 0 0 [83.216588] [drm:update_display_info.part.0] [CONNECTOR:55:HDMI-A-1] ELD size 36, SAD count 1 [83.224445] [drm:check_connector_changed] [CONNECTOR:55:HDMI-A-1] Same epoch counter 2 [83.231553] dw_hdmi_hpd_work(): ENDAround 205ms between HPD=0 and HPD=1 and around 160ms from plugin until delayed
hpd_workstarts.And at full plugout the debounce time is around 497 ms:
[95.125105] dw_hdmi_hardirq: EVENT=plugout [95.147586] dw_hdmi_schedule_hpd_work(status=2) [95.645277] dw_hdmi_hpd_work(): START [95.667741] dw_hdmi_bridge_detect() [95.691523] [drm:drm_edid_connector_update] [CONNECTOR:55:HDMI-A-1] EDID changed, epoch counter 3 [95.709861] [drm:check_connector_changed] [CONNECTOR:55:HDMI-A-1] status updated from connected to disconnected [95.722674] [drm:check_connector_changed] [CONNECTOR:55:HDMI-A-1] Changed epoch counter 2 => 4 [95.735173] [drm:drm_sysfs_connector_hotplug_event] [CONNECTOR:55:HDMI-A-1] generating connector hotplug event [95.746326] [drm:drm_fb_helper_hotplug_event] [95.754871] [drm:drm_client_modeset_probe] [95.762429] [drm:drm_helper_probe_single_connector_modes] [CONNECTOR:55:HDMI-A-1] [95.769758] dw_hdmi_bridge_detect() [95.776597] [drm:drm_helper_probe_single_connector_modes] [CONNECTOR:55:HDMI-A-1] disconnected [95.784389] [drm:drm_client_modeset_probe] No connectors reported connected with modes [95.791865] [drm:drm_client_modeset_probe] [CONNECTOR:55:HDMI-A-1] enabled? no [95.799426] [drm:drm_client_firmware_config.isra.0] Not using firmware configuration [95.807140] [drm:drm_client_modeset_probe] picking CRTCs for 1920x1080 config [95.818422] dw_hdmi_bridge_atomic_disable() [95.826412] [drm:vop2_plane_atomic_disable] Smart0-win0 disable [95.868341] [drm:drm_client_hotplug] fbdev: ret=0 [95.878206] dw_hdmi_hpd_work(): END
Quoted 3 lines
With a much quicker reaction on plugin event maybe the 1.1 seconds could stay to avoid having to do a full teardown,
disable()+enable()cycle, at slightly longer HPD pulses.
Yes those traces a good
Quoted 4 lines
Or maybe it is time to re-spin an old cec-adapter debounce series [1] that can allow for some additional time until the CEC phys addr is fully invalidated at a longer HPD low voltage level pulse.
Either one are good, the cec-adapt one looks simpler but still effective.
Thanks,
Neil
Quoted 191 lines
Regards,
JonasQuoted 184 lines
Neil
Quoted 180 lines
Signed-off-by Jonas Karlman <[email protected]>---
v7: Change to free irq before mute and clear using IH regs, also include
clear ofSTAT0_RX_SENSE
v6: Change back todisable_delayed_work_sync()in hpd disable ops,
Ensure HPD interrupt is masked and IRQ handler is disabled early indw_hdmi_remove()to prevent any irq re-arming of delayed work, Drop use of suspend helper
v5: Change to none-syncdisable_delayed_work()in hpd disable ops,
Change tocancel_delayed_work_sync()in remove,
Addcancel_delayed_work_sync()to new suspend helper
v4: Disable/maskdelayed_workuntil enable_hpd()/hpd_enable(),
Read connector status directly from HW regs inhpd_work
v3: New patch--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 80 +++++++++++++++++++++-- 1 file changed, 75 insertions(+), 5 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +75
-5@@ -50,6 +50,8 @@#define HDMI14_MAX_TMDSCLK 340000000 +#define HOTPLUG_DEBOUNCE_MS 1100 +static const u16 csc_coeff_default[3 ][4 ] = { {0x2000 ,0x0000 ,0x0000 ,0x0000 }, {0x0000 ,0x2000 ,0x0000 ,0x0000 },@@ -185,6 +187,7 @@ struct dw_hdmi {hdmi_codec_plugged_cb plugged_cb;struct device *codec_dev;enum drm_connector_status last_connector_result; +struct delayed_work hpd_work; };const struct dw_hdmi_plat_data *dw_hdmi_to_plat_data (struct dw_hdmi *hdmi)@@ -2517,6 +2520,20 @@ static void dw_hdmi_connector_force(struct drm_connector *connector)dw_hdmi_connector_status_update (hdmi, connector, connector->status); } +static void dw_hdmi_connector_enable_hpd (struct drm_connector *connector) +{ + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); + +enable_delayed_work (&hdmi->hpd_work); +} + +static void dw_hdmi_connector_disable_hpd (struct drm_connector *connector) +{ + struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector); + +disable_delayed_work_sync (&hdmi->hpd_work); +} +static void dw_hdmi_connector_destroy (struct drm_connector *connector) { struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);@@ -2538,6 +2555,8 @@ static const struct drm_connector_funcs dw_hdmi_connector_funcs = {static const struct drm_connector_helper_funcs dw_hdmi_connector_helper_funcs = { .get_modes = dw_hdmi_connector_get_modes, .atomic_check = dw_hdmi_connector_atomic_check, + .enable_hpd = dw_hdmi_connector_enable_hpd, + .disable_hpd = dw_hdmi_connector_disable_hpd, };static int dw_hdmi_connector_create (struct dw_hdmi *hdmi)@@ -2968,6 +2987,20 @@ static const struct drm_edid *dw_hdmi_bridge_edid_read(struct drm_bridge *bridgereturn dw_hdmi_edid_read (hdmi, connector); } +static void dw_hdmi_bridge_hpd_enable (struct drm_bridge *bridge) +{ +struct dw_hdmi *hdmi = bridge->driver_private; + +enable_delayed_work (&hdmi->hpd_work); +} + +static void dw_hdmi_bridge_hpd_disable (struct drm_bridge *bridge) +{ +struct dw_hdmi *hdmi = bridge->driver_private; + +disable_delayed_work_sync (&hdmi->hpd_work); +} +static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .atomic_duplicate_state = drm_atomic_helper_bridge_duplicate_state, .atomic_destroy_state = drm_atomic_helper_bridge_destroy_state,@@ -2981,6 +3014,8 @@ static const struct drm_bridge_funcs dw_hdmi_bridge_funcs = { .mode_valid = dw_hdmi_bridge_mode_valid, .detect = dw_hdmi_bridge_detect, .edid_read = dw_hdmi_bridge_edid_read, + .hpd_enable = dw_hdmi_bridge_hpd_enable, + .hpd_disable = dw_hdmi_bridge_hpd_disable, };/* ----------------------------------------------------------------------------- @@ -3101,8 +3136,8 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) status == connector_status_connected ? "plugin" : "plugout"); -if (hdmi->bridge.dev) -drm_helper_hpd_irq_event (hdmi->bridge.dev); +mod_delayed_work (system_percpu_wq, &hdmi->hpd_work, +msecs_to_jiffies (HOTPLUG_DEBOUNCE_MS )); }hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 );@@ -3112,6 +3147,29 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id)return IRQ_HANDLED ; } +static void dw_hdmi_hpd_work (struct work_struct *work) +{ + struct dw_hdmi *hdmi = container_of(work, struct dw_hdmi, hpd_work.work); +struct drm_device *dev = hdmi->bridge.dev; + +if (WARN_ON (!dev)) +return ; + +/* +* Notify the DRM core of the HPD event using drm_helper_hpd_irq_event() +* instead of drm_bridge_hpd_notify(). This will cause the DRM function +* check_connector_changed() to be called, which in turn calls the +* connector detect()/force() funcs to detect any connection status or +* epoch changes. The bridge connector detect() func also ensures that +* any hpd_notify() funcs are called for all bridges in the chain. +* +* drm_bridge_hpd_notify() shares a mutex with drm_bridge_hpd_disable(), +* and can result in a deadlock due to the disable_delayed_work_sync() +* call to wait on work to complete in dw_hdmi_bridge_hpd_disable(). +*/ +drm_helper_hpd_irq_event (dev); +} +static const struct dw_hdmi_phy_data dw_hdmi_phys[] = { { .type =DW_HDMI_PHY_DWC_HDMI_TX_PHY ,@@ -3396,6 +3454,9 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev, goto err_res; } +INIT_DELAYED_WORK (&hdmi->hpd_work, dw_hdmi_hpd_work); +disable_delayed_work (&hdmi->hpd_work); + ret =devm_request_threaded_irq (dev, irq, dw_hdmi_hardirq, dw_hdmi_irq,IRQF_SHARED ,dev_name (dev), hdmi);@@ -3532,6 +3593,18 @@ EXPORT_SYMBOL_GPL(dw_hdmi_probe);void dw_hdmi_remove (struct dw_hdmi *hdmi) { +struct platform_device *pdev =to_platform_device (hdmi->dev); +int irq =platform_get_irq (pdev,0 ); + +/* Free, mute and clear phy interrupts */ +devm_free_irq (hdmi->dev, irq, hdmi); +hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , +HDMI_IH_PHY_STAT0 ); + +/* Cancel any pending hot plug work */ +cancel_delayed_work_sync (&hdmi->hpd_work); +drm_bridge_remove (&hdmi->bridge);if (hdmi->audio && !IS_ERR (hdmi->audio))@@ -3539,9 +3612,6 @@ void dw_hdmi_remove(struct dw_hdmi *hdmi)if (!IS_ERR (hdmi->cec))platform_device_unregister (hdmi->cec); -/* Disable all interrupts */ -hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); -if (hdmi->i2c)i2c_del_adapter (&hdmi->i2c->adap);else
Jonas Karlman
reviewerPATCH v7 20/23 drm: bridge: dw_hdmi: Rework HDP and RXSENSE interrupt handling
In reply to Jonas KarlmanThe commit aeac23bda87f ("drm: bridge/dw_hdmi: improve HDMI enable/disable handling") added use of PHY RXSENSE indications to avoid triggering a full enable/disable of the HDMI block when a sink use a HPD low voltage level pulse to indicate changes of the EDID.
HDMI Specification Version 1.4b chapter 8.5 mentions:
An HDMI Sink shall indicate any change to the contents of the E-EDID by driving a low voltage level pulse on the Hot Plug Detect pin. This pulse shall be at least 100 msec.
A delayed work is now used to debounce reacting on a HPD low voltage level pulse when a sink changes the EDID. The delayed work triggers a hotplug uevent every time the connection status or EDID has changed.
Remove RXSENSE handling to simplify the HPD interrupt handling and instead depend on the delayed work to detect any connection status or EDID changes.
This also ensures the initial HPD interrupt polarity is based on current HPD status to avoid an unnecessary interrupt from being triggered immediately at probe or resume when a sink is connected.
---
v7: Remove clear of STAT0_RX_SENSE in dw_hdmi_remove() added in prior
patch
v6: Update commit message,
Collect t-b tag
v5: Add comment about interrupt generation
v4: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 147 ++++------------------ 1 file changed, 22 insertions(+), 125 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +22 -125
@@ -161,11 +161,7 @@ struct dw_hdmi {struct pinctrl_state *unwedge_state;struct mutex mutex;/* for state below */ -enum drm_connector_force force;/* mutex-protected force state */ struct drm_connector *curr_conn;/* current connector (only valid when !disabled) */ -bool disabled;/* DRM has disabled our bridge */ -bool rxsense;/* rxsense state */ -u8 phy_mask;/* desired phy int mask settings */ u8 mc_clkdis;/* clock disable register */ spinlock_t audio_lock;
@@ -196,14 +192,6 @@ const struct dw_hdmi_plat_data *dw_hdmi_to_plat_data(struct dw_hdmi *hdmi) }EXPORT_SYMBOL_GPL (dw_hdmi_to_plat_data); -#define HDMI_IH_PHY_STAT0_RX_SENSE \ - (HDMI_IH_PHY_STAT0_RX_SENSE0 | HDMI_IH_PHY_STAT0_RX_SENSE1 | \ - HDMI_IH_PHY_STAT0_RX_SENSE2 | HDMI_IH_PHY_STAT0_RX_SENSE3) - -#define HDMI_PHY_RX_SENSE \ - (HDMI_PHY_RX_SENSE0 | HDMI_PHY_RX_SENSE1 | \ - HDMI_PHY_RX_SENSE2 | HDMI_PHY_RX_SENSE3) -static inline void hdmi_writeb (struct dw_hdmi *hdmi,u8 val,int offset) {regmap_write (hdmi->regm, offset << hdmi->reg_shift, val);
@@ -1702,36 +1690,25 @@ EXPORT_SYMBOL_GPL(dw_hdmi_phy_read_hpd);void dw_hdmi_phy_update_hpd (struct dw_hdmi *hdmi,void *data,bool force,bool disabled,bool rxsense) { -u8 old_mask = hdmi->phy_mask; - -if (force || disabled || !rxsense) - hdmi->phy_mask |=HDMI_PHY_RX_SENSE ; -else - hdmi->phy_mask &= ~HDMI_PHY_RX_SENSE ; - -if (old_mask != hdmi->phy_mask) -hdmi_writeb (hdmi, hdmi->phy_mask,HDMI_PHY_MASK0 ); }EXPORT_SYMBOL_GPL (dw_hdmi_phy_update_hpd);void dw_hdmi_phy_setup_hpd (struct dw_hdmi *hdmi,void *data) {/* -* Configure the PHY RX SENSE and HPD interrupts polarities and clear -* any pending interrupt. +* Configure the PHY HPD interrupt polarity based on current HPD status +* and clear any pending interrupt. */ -hdmi_writeb (hdmi,HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ,HDMI_PHY_POL0 ); -hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , -HDMI_IH_PHY_STAT0 ); +hdmi_modb (hdmi,hdmi_readb (hdmi,HDMI_PHY_STAT0 ) &HDMI_PHY_HPD ? +0 :HDMI_PHY_HPD ,HDMI_PHY_HPD ,HDMI_PHY_POL0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_PHY_STAT0 );/* Enable cable hot plug irq. */ -hdmi_writeb (hdmi, hdmi->phy_mask,HDMI_PHY_MASK0 ); +hdmi_writeb (hdmi, ~HDMI_PHY_HPD ,HDMI_PHY_MASK0 );/* Clear and unmute interrupts. */ -hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , -HDMI_IH_PHY_STAT0 ); -hdmi_writeb (hdmi, ~(HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE ), -HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_PHY_STAT0 ); +hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 ); }EXPORT_SYMBOL_GPL (dw_hdmi_phy_setup_hpd);
@@ -2395,26 +2372,6 @@ static void dw_hdmi_poweroff(struct dw_hdmi *hdmi) } } -/* -* Adjust the detection of RXSENSE according to whether we have a forced -* connection mode enabled, or whether we have been disabled. There is -* no point processing RXSENSE interrupts if we have a forced connection -* state, or DRM has us disabled. -* -* We also disable rxsense interrupts when we think we're disconnected -* to avoid floating TDMS signals giving false rxsense interrupts. -* -* Note: we still need to listen for HPD interrupts even when DRM has us -* disabled so that we can detect a connect event. -*/ -static void dw_hdmi_update_phy_mask (struct dw_hdmi *hdmi) -{ -if (hdmi->phy.ops->update_hpd) - hdmi->phy.ops->update_hpd (hdmi, hdmi->phy.data, - hdmi->force, hdmi->disabled, - hdmi->rxsense); -} -static enum drm_connector_status dw_hdmi_detect (struct dw_hdmi *hdmi) {enum drm_connector_status result;
@@ -2512,9 +2469,7 @@ static void dw_hdmi_connector_force(struct drm_connector *connector) struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);mutex_lock (&hdmi->mutex); - hdmi->force = connector->force; hdmi->last_connector_result = connector->status; -dw_hdmi_update_phy_mask (hdmi);mutex_unlock (&hdmi->mutex);dw_hdmi_connector_status_update (hdmi, connector, connector->status);
@@ -2932,10 +2887,8 @@ static void dw_hdmi_bridge_atomic_disable(struct drm_bridge *bridge,struct dw_hdmi *hdmi = bridge->driver_private;mutex_lock (&hdmi->mutex); - hdmi->disabled = true; hdmi->curr_conn =NULL ;dw_hdmi_poweroff (hdmi); -dw_hdmi_update_phy_mask (hdmi);handle_plugged_change (hdmi, false);mutex_unlock (&hdmi->mutex); }
@@ -2954,10 +2907,8 @@ static void dw_hdmi_bridge_atomic_enable(struct drm_bridge *bridge, mode = &drm_atomic_get_new_crtc_state (state, crtc)->adjusted_mode;mutex_lock (&hdmi->mutex); - hdmi->disabled = false; hdmi->curr_conn = connector;dw_hdmi_poweron (hdmi, connector, mode); -dw_hdmi_update_phy_mask (hdmi);handle_plugged_change (hdmi, true);mutex_unlock (&hdmi->mutex); }
@@ -3060,78 +3011,29 @@ static irqreturn_t dw_hdmi_hardirq(int irq, void *dev_id)void dw_hdmi_setup_rx_sense (struct dw_hdmi *hdmi,bool hpd,bool rx_sense) { -mutex_lock (&hdmi->mutex); - -if (!hdmi->force) { -/* -* If the RX sense status indicates we're disconnected, -* clear the software rxsense status. -*/ -if (!rx_sense) - hdmi->rxsense = false; - -/* -* Only set the software rxsense status when both -* rxsense and hpd indicates we're connected. -* This avoids what seems to be bad behaviour in -* at least iMX6S versions of the phy. -*/ -if (hpd) - hdmi->rxsense = true; - -dw_hdmi_update_phy_mask (hdmi); - } -mutex_unlock (&hdmi->mutex); }EXPORT_SYMBOL_GPL (dw_hdmi_setup_rx_sense);static irqreturn_t dw_hdmi_irq (int irq,void *dev_id) {struct dw_hdmi *hdmi = dev_id; -u8 intr_stat, phy_int_pol, phy_pol_mask, phy_stat; -enum drm_connector_status status = connector_status_unknown; - - intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); - phy_int_pol =hdmi_readb (hdmi,HDMI_PHY_POL0 ); - phy_stat =hdmi_readb (hdmi,HDMI_PHY_STAT0 ); - - phy_pol_mask =0 ; -if (intr_stat &HDMI_IH_PHY_STAT0_HPD ) - phy_pol_mask |=HDMI_PHY_HPD ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE0 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE0 ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE1 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE1 ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE2 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE2 ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE3 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE3 ; - -if (phy_pol_mask) -hdmi_modb (hdmi, ~phy_int_pol, phy_pol_mask,HDMI_PHY_POL0 ); +u8 intr_stat;/* -* RX sense tells us whether the TDMS transmitters are detecting -* load - in other words, there's something listening on the -* other end of the link. Use this to decide whether we should -* power on the phy as HPD may be toggled by the sink to merely -* ask the source to re-read the EDID. +* Interrupt generation is accomplished in the following way: +* interrupt = (mask == 0) && (polarity == status) +* All interrupts are forwarded to the Interrupt Handler sticky bit +* register ih_phy_stat0 and muted using the register ih_mute_phy_stat0. */ -if (intr_stat & - (HDMI_IH_PHY_STAT0_RX_SENSE |HDMI_IH_PHY_STAT0_HPD )) { -dw_hdmi_setup_rx_sense (hdmi, - phy_stat &HDMI_PHY_HPD , - phy_stat &HDMI_PHY_RX_SENSE ); + intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); +if (intr_stat &HDMI_IH_PHY_STAT0_HPD ) { +enum drm_connector_status status; -if ((intr_stat &HDMI_IH_PHY_STAT0_HPD ) && - (phy_stat &HDMI_PHY_HPD )) - status = connector_status_connected; +/* Set HPD interrupt polarity based on current HPD status. */ + status =dw_hdmi_phy_read_hpd (hdmi, hdmi->phy.data); +hdmi_modb (hdmi, status == connector_status_connected ? +0 :HDMI_PHY_HPD ,HDMI_PHY_HPD ,HDMI_PHY_POL0 ); -if (!(phy_stat & (HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ))) - status = connector_status_disconnected; - } - -if (status != connector_status_unknown) {dev_dbg (hdmi->dev,"EVENT=%s\n" , status == connector_status_connected ?"plugin" :"plugout" );
@@ -3141,8 +3043,7 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) }hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 ); -hdmi_writeb (hdmi, ~(HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE ), -HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 );return IRQ_HANDLED ; }
@@ -3343,9 +3244,6 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev, hdmi->dev = dev; hdmi->sample_rate =48000 ; hdmi->channels =2 ; - hdmi->disabled = true; - hdmi->rxsense = true; - hdmi->phy_mask = (u8 )~(HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ); hdmi->mc_clkdis =0x7f ; hdmi->last_connector_result = connector_status_disconnected;
@@ -3599,8 +3497,7 @@ void dw_hdmi_remove(struct dw_hdmi *hdmi)/* Free, mute and clear phy interrupts */ devm_free_irq (hdmi->dev, irq, hdmi);hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); -hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , -HDMI_IH_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_PHY_STAT0 );/* Cancel any pending hot plug work */ cancel_delayed_work_sync (&hdmi->hpd_work);
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v7 20/23] drm: bridge: dw_hdmi: Rework HDP and RXSENSE interrupt handling
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] Race condition when clearing the HPD interrupt sticky bit can permanently break hotplug detection --
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Race condition when clearing the HPD interrupt sticky bit can permanently break hotplug detection --
commit cb7edad18b267f9862ac845a3672fd15ae20a0f1
Author: Jonas Karlman <[email protected]>
Subject: drm: bridge: dw_hdmi: Rework HDP and RXSENSE interrupt handling
This commit removes RXSENSE handling to simplify the HPD interrupt handling, instead relying on a delayed work mechanism to detect connection status or EDID changes. It also updates the HPD interrupt polarity initialization to match current HPD status.
Quoted 4 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-0
[ ... ]
Quoted 1 line
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-0@@ -3060,78 +3011,29 @@ static irqreturn_t dw_hdmi_hardirq(int irq, void *dev_id)
[ ... ]
Quoted 9 lines
static
irqreturn_tdw_hdmi_irq(int irq, void *dev_id){drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-6struct dw_hdmi *hdmi = dev_id; -u8 intr_stat, phy_int_pol, phy_pol_mask, phy_stat; -enum drm_connector_status status = connector_status_unknown; - - intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); - phy_int_pol =hdmi_readb (hdmi,HDMI_PHY_POL0 ); - phy_stat =hdmi_readb (hdmi,HDMI_PHY_STAT0 );
[ ... ]
Quoted 39 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +12
-18+u8 intr_stat;/* -* RX sense tells us whether the TDMS transmitters are detecting -* load - in other words, there's something listening on the -* other end of the link. Use this to decide whether we should -* power on the phy as HPD may be toggled by the sink to merely -* ask the source to re-read the EDID. +* Interrupt generation is accomplished in the following way: +* interrupt = (mask == 0) && (polarity == status) +* All interrupts are forwarded to the Interrupt Handler sticky bit +* register ih_phy_stat0 and muted using the register ih_mute_phy_stat0. */ -if (intr_stat & - (HDMI_IH_PHY_STAT0_RX_SENSE |HDMI_IH_PHY_STAT0_HPD )) { -dw_hdmi_setup_rx_sense (hdmi, - phy_stat &HDMI_PHY_HPD , - phy_stat &HDMI_PHY_RX_SENSE ); - -if ((intr_stat &HDMI_IH_PHY_STAT0_HPD ) && - (phy_stat &HDMI_PHY_HPD )) - status = connector_status_connected; + intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); +if (intr_stat &HDMI_IH_PHY_STAT0_HPD ) { +enum drm_connector_status status; -if (!(phy_stat & (HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ))) - status = connector_status_disconnected; - } +/* Set HPD interrupt polarity based on current HPD status. */ + status =dw_hdmi_phy_read_hpd (hdmi, hdmi->phy.data); +hdmi_modb (hdmi, status == connector_status_connected ? +0 :HDMI_PHY_HPD ,HDMI_PHY_HPD ,HDMI_PHY_POL0 ); -if (status != connector_status_unknown) {dev_dbg (hdmi->dev,"EVENT=%s\n" , status == connector_status_connected ?"plugin" :"plugout" );
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) · line 3045
Quoted 3 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-0@@ -3141,8 +3043,7 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) }hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 );
This problem wasn't introduced by this patch, but could this interrupt clearing sequence contain a race condition that breaks hotplug detection?
If the HPD line bounces after the polarity is updated but before the interrupt is cleared, the new state will match the newly configured polarity. This generates a new interrupt edge and sets the sticky bit.
The subsequent write to HDMI_IH_PHY_STAT0 using the originally read intr_stat will immediately clear this new sticky bit.
As a result, the hardware is left with a cleared interrupt while the line matches the polarity. The next physical edge will not trigger an interrupt because it transitions away from the matched polarity, leaving HPD permanently desynchronized.
Should HDMI_IH_PHY_STAT0 be cleared before reading the live status and updating the polarity to prevent losing interrupts?
Quoted 6 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +1
-2-hdmi_writeb (hdmi, ~(HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE ), -HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 );return IRQ_HANDLED ; }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=20
Neil Armstrong
maintainerRe: [PATCH v7 20/23] drm: bridge: dw_hdmi: Rework HDP and RXSENSE interrupt handling
In reply to Jonas KarlmanQuoted 15 lines · Jonas Karlman
The commit
aeac23bda87f("drm: bridge/dw_hdmi: improve HDMI enable/disable handling") added use of PHY RXSENSE indications to avoid triggering a full enable/disable of the HDMI block when a sink use a HPD low voltage level pulse to indicate changes of the EDID.HDMI Specification Version 1.4b chapter 8.5 mentions:
An HDMI Sink shall indicate any change to the contents of the E-EDID by driving a low voltage level pulse on the Hot Plug Detect pin. This pulse shall be at least 100 msec.
A delayed work is now used to debounce reacting on a HPD low voltage level pulse when a sink changes the EDID. The delayed work triggers a hotplug uevent every time the connection status or EDID has changed.
Quoted 7 lines
Remove RXSENSE handling to simplify the HPD interrupt handling and instead depend on the delayed work to detect any connection status or EDID changes.
This also ensures the initial HPD interrupt polarity is based on current HPD status to avoid an unnecessary interrupt from being triggered immediately at probe or resume when a sink is connected.
I'm still puzzled of the removal of RX_SENSE entirely as v1, and I since the rx_sense code is not easy to understand I don't have an opinion on that.
Can someone with more knowledge can comment on that ?
Neil
Quoted 270 lines
Signed-off-by Jonas Karlman <[email protected]>---
v7: Remove clear ofSTAT0_RX_SENSEindw_hdmi_remove()added in prior
patch
v6: Update commit message,
Collect t-b tag
v5: Add comment about interrupt generation
v4: New patch--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 147 ++++------------------ 1 file changed, 22 insertions(+), 125 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +22
-125@@ -161,11 +161,7 @@ struct dw_hdmi {struct pinctrl_state *unwedge_state;struct mutex mutex;/* for state below */ -enum drm_connector_force force;/* mutex-protected force state */ struct drm_connector *curr_conn;/* current connector (only valid when !disabled) */ -bool disabled;/* DRM has disabled our bridge */ -bool rxsense;/* rxsense state */ -u8 phy_mask;/* desired phy int mask settings */ u8 mc_clkdis;/* clock disable register */ spinlock_t audio_lock;@@ -196,14 +192,6 @@ const struct dw_hdmi_plat_data *dw_hdmi_to_plat_data(struct dw_hdmi *hdmi) }EXPORT_SYMBOL_GPL (dw_hdmi_to_plat_data); -#define HDMI_IH_PHY_STAT0_RX_SENSE \ - (HDMI_IH_PHY_STAT0_RX_SENSE0 | HDMI_IH_PHY_STAT0_RX_SENSE1 | \ - HDMI_IH_PHY_STAT0_RX_SENSE2 | HDMI_IH_PHY_STAT0_RX_SENSE3) - -#define HDMI_PHY_RX_SENSE \ - (HDMI_PHY_RX_SENSE0 | HDMI_PHY_RX_SENSE1 | \ - HDMI_PHY_RX_SENSE2 | HDMI_PHY_RX_SENSE3) -static inline void hdmi_writeb (struct dw_hdmi *hdmi,u8 val,int offset) {regmap_write (hdmi->regm, offset << hdmi->reg_shift, val);@@ -1702,36 +1690,25 @@ EXPORT_SYMBOL_GPL(dw_hdmi_phy_read_hpd);void dw_hdmi_phy_update_hpd (struct dw_hdmi *hdmi,void *data,bool force,bool disabled,bool rxsense) { -u8 old_mask = hdmi->phy_mask; - -if (force || disabled || !rxsense) - hdmi->phy_mask |=HDMI_PHY_RX_SENSE ; -else - hdmi->phy_mask &= ~HDMI_PHY_RX_SENSE ; - -if (old_mask != hdmi->phy_mask) -hdmi_writeb (hdmi, hdmi->phy_mask,HDMI_PHY_MASK0 ); }EXPORT_SYMBOL_GPL (dw_hdmi_phy_update_hpd);void dw_hdmi_phy_setup_hpd (struct dw_hdmi *hdmi,void *data) {/* -* Configure the PHY RX SENSE and HPD interrupts polarities and clear -* any pending interrupt. +* Configure the PHY HPD interrupt polarity based on current HPD status +* and clear any pending interrupt. */ -hdmi_writeb (hdmi,HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ,HDMI_PHY_POL0 ); -hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , -HDMI_IH_PHY_STAT0 ); +hdmi_modb (hdmi,hdmi_readb (hdmi,HDMI_PHY_STAT0 ) &HDMI_PHY_HPD ? +0 :HDMI_PHY_HPD ,HDMI_PHY_HPD ,HDMI_PHY_POL0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_PHY_STAT0 );/* Enable cable hot plug irq. */ -hdmi_writeb (hdmi, hdmi->phy_mask,HDMI_PHY_MASK0 ); +hdmi_writeb (hdmi, ~HDMI_PHY_HPD ,HDMI_PHY_MASK0 );/* Clear and unmute interrupts. */ -hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , -HDMI_IH_PHY_STAT0 ); -hdmi_writeb (hdmi, ~(HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE ), -HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_PHY_STAT0 ); +hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 ); }EXPORT_SYMBOL_GPL (dw_hdmi_phy_setup_hpd);@@ -2395,26 +2372,6 @@ static void dw_hdmi_poweroff(struct dw_hdmi *hdmi) } } -/* -* Adjust the detection of RXSENSE according to whether we have a forced -* connection mode enabled, or whether we have been disabled. There is -* no point processing RXSENSE interrupts if we have a forced connection -* state, or DRM has us disabled. -* -* We also disable rxsense interrupts when we think we're disconnected -* to avoid floating TDMS signals giving false rxsense interrupts. -* -* Note: we still need to listen for HPD interrupts even when DRM has us -* disabled so that we can detect a connect event. -*/ -static void dw_hdmi_update_phy_mask (struct dw_hdmi *hdmi) -{ -if (hdmi->phy.ops->update_hpd) - hdmi->phy.ops->update_hpd (hdmi, hdmi->phy.data, - hdmi->force, hdmi->disabled, - hdmi->rxsense); -} -static enum drm_connector_status dw_hdmi_detect (struct dw_hdmi *hdmi) {enum drm_connector_status result;@@ -2512,9 +2469,7 @@ static void dw_hdmi_connector_force(struct drm_connector *connector) struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);mutex_lock (&hdmi->mutex); - hdmi->force = connector->force; hdmi->last_connector_result = connector->status; -dw_hdmi_update_phy_mask (hdmi);mutex_unlock (&hdmi->mutex);dw_hdmi_connector_status_update (hdmi, connector, connector->status);@@ -2932,10 +2887,8 @@ static void dw_hdmi_bridge_atomic_disable(struct drm_bridge *bridge,struct dw_hdmi *hdmi = bridge->driver_private;mutex_lock (&hdmi->mutex); - hdmi->disabled = true; hdmi->curr_conn =NULL ;dw_hdmi_poweroff (hdmi); -dw_hdmi_update_phy_mask (hdmi);handle_plugged_change (hdmi, false);mutex_unlock (&hdmi->mutex); }@@ -2954,10 +2907,8 @@ static void dw_hdmi_bridge_atomic_enable(struct drm_bridge *bridge, mode = &drm_atomic_get_new_crtc_state (state, crtc)->adjusted_mode;mutex_lock (&hdmi->mutex); - hdmi->disabled = false; hdmi->curr_conn = connector;dw_hdmi_poweron (hdmi, connector, mode); -dw_hdmi_update_phy_mask (hdmi);handle_plugged_change (hdmi, true);mutex_unlock (&hdmi->mutex); }@@ -3060,78 +3011,29 @@ static irqreturn_t dw_hdmi_hardirq(int irq, void *dev_id)void dw_hdmi_setup_rx_sense (struct dw_hdmi *hdmi,bool hpd,bool rx_sense) { -mutex_lock (&hdmi->mutex); - -if (!hdmi->force) { -/* -* If the RX sense status indicates we're disconnected, -* clear the software rxsense status. -*/ -if (!rx_sense) - hdmi->rxsense = false; - -/* -* Only set the software rxsense status when both -* rxsense and hpd indicates we're connected. -* This avoids what seems to be bad behaviour in -* at least iMX6S versions of the phy. -*/ -if (hpd) - hdmi->rxsense = true; - -dw_hdmi_update_phy_mask (hdmi); - } -mutex_unlock (&hdmi->mutex); }EXPORT_SYMBOL_GPL (dw_hdmi_setup_rx_sense);static irqreturn_t dw_hdmi_irq (int irq,void *dev_id) {struct dw_hdmi *hdmi = dev_id; -u8 intr_stat, phy_int_pol, phy_pol_mask, phy_stat; -enum drm_connector_status status = connector_status_unknown; - - intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); - phy_int_pol =hdmi_readb (hdmi,HDMI_PHY_POL0 ); - phy_stat =hdmi_readb (hdmi,HDMI_PHY_STAT0 ); - - phy_pol_mask =0 ; -if (intr_stat &HDMI_IH_PHY_STAT0_HPD ) - phy_pol_mask |=HDMI_PHY_HPD ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE0 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE0 ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE1 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE1 ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE2 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE2 ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE3 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE3 ; - -if (phy_pol_mask) -hdmi_modb (hdmi, ~phy_int_pol, phy_pol_mask,HDMI_PHY_POL0 ); +u8 intr_stat;/* -* RX sense tells us whether the TDMS transmitters are detecting -* load - in other words, there's something listening on the -* other end of the link. Use this to decide whether we should -* power on the phy as HPD may be toggled by the sink to merely -* ask the source to re-read the EDID. +* Interrupt generation is accomplished in the following way: +* interrupt = (mask == 0) && (polarity == status) +* All interrupts are forwarded to the Interrupt Handler sticky bit +* register ih_phy_stat0 and muted using the register ih_mute_phy_stat0. */ -if (intr_stat & - (HDMI_IH_PHY_STAT0_RX_SENSE |HDMI_IH_PHY_STAT0_HPD )) { -dw_hdmi_setup_rx_sense (hdmi, - phy_stat &HDMI_PHY_HPD , - phy_stat &HDMI_PHY_RX_SENSE ); + intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); +if (intr_stat &HDMI_IH_PHY_STAT0_HPD ) { +enum drm_connector_status status; -if ((intr_stat &HDMI_IH_PHY_STAT0_HPD ) && - (phy_stat &HDMI_PHY_HPD )) - status = connector_status_connected; +/* Set HPD interrupt polarity based on current HPD status. */ + status =dw_hdmi_phy_read_hpd (hdmi, hdmi->phy.data); +hdmi_modb (hdmi, status == connector_status_connected ? +0 :HDMI_PHY_HPD ,HDMI_PHY_HPD ,HDMI_PHY_POL0 ); -if (!(phy_stat & (HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ))) - status = connector_status_disconnected; - } - -if (status != connector_status_unknown) {dev_dbg (hdmi->dev,"EVENT=%s\n" , status == connector_status_connected ?"plugin" :"plugout" );@@ -3141,8 +3043,7 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) }hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 ); -hdmi_writeb (hdmi, ~(HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE ), -HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 );return IRQ_HANDLED ; }@@ -3343,9 +3244,6 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev, hdmi->dev = dev; hdmi->sample_rate =48000 ; hdmi->channels =2 ; - hdmi->disabled = true; - hdmi->rxsense = true; - hdmi->phy_mask = (u8 )~(HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ); hdmi->mc_clkdis =0x7f ; hdmi->last_connector_result = connector_status_disconnected;@@ -3599,8 +3497,7 @@ void dw_hdmi_remove(struct dw_hdmi *hdmi)/* Free, mute and clear phy interrupts */ devm_free_irq (hdmi->dev, irq, hdmi);hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); -hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , -HDMI_IH_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_PHY_STAT0 );/* Cancel any pending hot plug work */ cancel_delayed_work_sync (&hdmi->hpd_work);
Re: [PATCH v7 20/23] drm: bridge: dw_hdmi: Rework HDP and RXSENSE interrupt handling
In reply to Jonas KarlmanQuoted 19 lines · Jonas Karlman
The commit
aeac23bda87f("drm: bridge/dw_hdmi: improve HDMI enable/disable handling") added use of PHY RXSENSE indications to avoid triggering a full enable/disable of the HDMI block when a sink use a HPD low voltage level pulse to indicate changes of the EDID.HDMI Specification Version 1.4b chapter 8.5 mentions:
An HDMI Sink shall indicate any change to the contents of the E-EDID by driving a low voltage level pulse on the Hot Plug Detect pin. This pulse shall be at least 100 msec.
A delayed work is now used to debounce reacting on a HPD low voltage level pulse when a sink changes the EDID. The delayed work triggers a hotplug uevent every time the connection status or EDID has changed.
Remove RXSENSE handling to simplify the HPD interrupt handling and instead depend on the delayed work to detect any connection status or EDID changes.
Quoted 6 lines
This also ensures the initial HPD interrupt polarity is based on current HPD status to avoid an unnecessary interrupt from being triggered immediately at probe or resume when a sink is connected.
Quoted 270 lines
---
v7: Remove clear ofSTAT0_RX_SENSEindw_hdmi_remove()added in prior
patch
v6: Update commit message,
Collect t-b tag
v5: Add comment about interrupt generation
v4: New patch--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 147 ++++------------------ 1 file changed, 22 insertions(+), 125 deletions(-)drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +22
-125@@ -161,11 +161,7 @@ struct dw_hdmi {struct pinctrl_state *unwedge_state;struct mutex mutex;/* for state below */ -enum drm_connector_force force;/* mutex-protected force state */ struct drm_connector *curr_conn;/* current connector (only valid when !disabled) */ -bool disabled;/* DRM has disabled our bridge */ -bool rxsense;/* rxsense state */ -u8 phy_mask;/* desired phy int mask settings */ u8 mc_clkdis;/* clock disable register */ spinlock_t audio_lock;@@ -196,14 +192,6 @@ const struct dw_hdmi_plat_data *dw_hdmi_to_plat_data(struct dw_hdmi *hdmi) }EXPORT_SYMBOL_GPL (dw_hdmi_to_plat_data); -#define HDMI_IH_PHY_STAT0_RX_SENSE \ - (HDMI_IH_PHY_STAT0_RX_SENSE0 | HDMI_IH_PHY_STAT0_RX_SENSE1 | \ - HDMI_IH_PHY_STAT0_RX_SENSE2 | HDMI_IH_PHY_STAT0_RX_SENSE3) - -#define HDMI_PHY_RX_SENSE \ - (HDMI_PHY_RX_SENSE0 | HDMI_PHY_RX_SENSE1 | \ - HDMI_PHY_RX_SENSE2 | HDMI_PHY_RX_SENSE3) -static inline void hdmi_writeb (struct dw_hdmi *hdmi,u8 val,int offset) {regmap_write (hdmi->regm, offset << hdmi->reg_shift, val);@@ -1702,36 +1690,25 @@ EXPORT_SYMBOL_GPL(dw_hdmi_phy_read_hpd);void dw_hdmi_phy_update_hpd (struct dw_hdmi *hdmi,void *data,bool force,bool disabled,bool rxsense) { -u8 old_mask = hdmi->phy_mask; - -if (force || disabled || !rxsense) - hdmi->phy_mask |=HDMI_PHY_RX_SENSE ; -else - hdmi->phy_mask &= ~HDMI_PHY_RX_SENSE ; - -if (old_mask != hdmi->phy_mask) -hdmi_writeb (hdmi, hdmi->phy_mask,HDMI_PHY_MASK0 ); }EXPORT_SYMBOL_GPL (dw_hdmi_phy_update_hpd);void dw_hdmi_phy_setup_hpd (struct dw_hdmi *hdmi,void *data) {/* -* Configure the PHY RX SENSE and HPD interrupts polarities and clear -* any pending interrupt. +* Configure the PHY HPD interrupt polarity based on current HPD status +* and clear any pending interrupt. */ -hdmi_writeb (hdmi,HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ,HDMI_PHY_POL0 ); -hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , -HDMI_IH_PHY_STAT0 ); +hdmi_modb (hdmi,hdmi_readb (hdmi,HDMI_PHY_STAT0 ) &HDMI_PHY_HPD ? +0 :HDMI_PHY_HPD ,HDMI_PHY_HPD ,HDMI_PHY_POL0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_PHY_STAT0 );/* Enable cable hot plug irq. */ -hdmi_writeb (hdmi, hdmi->phy_mask,HDMI_PHY_MASK0 ); +hdmi_writeb (hdmi, ~HDMI_PHY_HPD ,HDMI_PHY_MASK0 );/* Clear and unmute interrupts. */ -hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , -HDMI_IH_PHY_STAT0 ); -hdmi_writeb (hdmi, ~(HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE ), -HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_PHY_STAT0 ); +hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 ); }EXPORT_SYMBOL_GPL (dw_hdmi_phy_setup_hpd);@@ -2395,26 +2372,6 @@ static void dw_hdmi_poweroff(struct dw_hdmi *hdmi) } } -/* -* Adjust the detection of RXSENSE according to whether we have a forced -* connection mode enabled, or whether we have been disabled. There is -* no point processing RXSENSE interrupts if we have a forced connection -* state, or DRM has us disabled. -* -* We also disable rxsense interrupts when we think we're disconnected -* to avoid floating TDMS signals giving false rxsense interrupts. -* -* Note: we still need to listen for HPD interrupts even when DRM has us -* disabled so that we can detect a connect event. -*/ -static void dw_hdmi_update_phy_mask (struct dw_hdmi *hdmi) -{ -if (hdmi->phy.ops->update_hpd) - hdmi->phy.ops->update_hpd (hdmi, hdmi->phy.data, - hdmi->force, hdmi->disabled, - hdmi->rxsense); -} -static enum drm_connector_status dw_hdmi_detect (struct dw_hdmi *hdmi) {enum drm_connector_status result;@@ -2512,9 +2469,7 @@ static void dw_hdmi_connector_force(struct drm_connector *connector) struct dw_hdmi *hdmi = container_of(connector, struct dw_hdmi, connector);mutex_lock (&hdmi->mutex); - hdmi->force = connector->force; hdmi->last_connector_result = connector->status; -dw_hdmi_update_phy_mask (hdmi);mutex_unlock (&hdmi->mutex);dw_hdmi_connector_status_update (hdmi, connector, connector->status);@@ -2932,10 +2887,8 @@ static void dw_hdmi_bridge_atomic_disable(struct drm_bridge *bridge,struct dw_hdmi *hdmi = bridge->driver_private;mutex_lock (&hdmi->mutex); - hdmi->disabled = true; hdmi->curr_conn =NULL ;dw_hdmi_poweroff (hdmi); -dw_hdmi_update_phy_mask (hdmi);handle_plugged_change (hdmi, false);mutex_unlock (&hdmi->mutex); }@@ -2954,10 +2907,8 @@ static void dw_hdmi_bridge_atomic_enable(struct drm_bridge *bridge, mode = &drm_atomic_get_new_crtc_state (state, crtc)->adjusted_mode;mutex_lock (&hdmi->mutex); - hdmi->disabled = false; hdmi->curr_conn = connector;dw_hdmi_poweron (hdmi, connector, mode); -dw_hdmi_update_phy_mask (hdmi);handle_plugged_change (hdmi, true);mutex_unlock (&hdmi->mutex); }@@ -3060,78 +3011,29 @@ static irqreturn_t dw_hdmi_hardirq(int irq, void *dev_id)void dw_hdmi_setup_rx_sense (struct dw_hdmi *hdmi,bool hpd,bool rx_sense) { -mutex_lock (&hdmi->mutex); - -if (!hdmi->force) { -/* -* If the RX sense status indicates we're disconnected, -* clear the software rxsense status. -*/ -if (!rx_sense) - hdmi->rxsense = false; - -/* -* Only set the software rxsense status when both -* rxsense and hpd indicates we're connected. -* This avoids what seems to be bad behaviour in -* at least iMX6S versions of the phy. -*/ -if (hpd) - hdmi->rxsense = true; - -dw_hdmi_update_phy_mask (hdmi); - } -mutex_unlock (&hdmi->mutex); }EXPORT_SYMBOL_GPL (dw_hdmi_setup_rx_sense);static irqreturn_t dw_hdmi_irq (int irq,void *dev_id) {struct dw_hdmi *hdmi = dev_id; -u8 intr_stat, phy_int_pol, phy_pol_mask, phy_stat; -enum drm_connector_status status = connector_status_unknown; - - intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); - phy_int_pol =hdmi_readb (hdmi,HDMI_PHY_POL0 ); - phy_stat =hdmi_readb (hdmi,HDMI_PHY_STAT0 ); - - phy_pol_mask =0 ; -if (intr_stat &HDMI_IH_PHY_STAT0_HPD ) - phy_pol_mask |=HDMI_PHY_HPD ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE0 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE0 ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE1 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE1 ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE2 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE2 ; -if (intr_stat &HDMI_IH_PHY_STAT0_RX_SENSE3 ) - phy_pol_mask |=HDMI_PHY_RX_SENSE3 ; - -if (phy_pol_mask) -hdmi_modb (hdmi, ~phy_int_pol, phy_pol_mask,HDMI_PHY_POL0 ); +u8 intr_stat;/* -* RX sense tells us whether the TDMS transmitters are detecting -* load - in other words, there's something listening on the -* other end of the link. Use this to decide whether we should -* power on the phy as HPD may be toggled by the sink to merely -* ask the source to re-read the EDID. +* Interrupt generation is accomplished in the following way: +* interrupt = (mask == 0) && (polarity == status) +* All interrupts are forwarded to the Interrupt Handler sticky bit +* register ih_phy_stat0 and muted using the register ih_mute_phy_stat0. */ -if (intr_stat & - (HDMI_IH_PHY_STAT0_RX_SENSE |HDMI_IH_PHY_STAT0_HPD )) { -dw_hdmi_setup_rx_sense (hdmi, - phy_stat &HDMI_PHY_HPD , - phy_stat &HDMI_PHY_RX_SENSE ); + intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); +if (intr_stat &HDMI_IH_PHY_STAT0_HPD ) { +enum drm_connector_status status; -if ((intr_stat &HDMI_IH_PHY_STAT0_HPD ) && - (phy_stat &HDMI_PHY_HPD )) - status = connector_status_connected; +/* Set HPD interrupt polarity based on current HPD status. */ + status =dw_hdmi_phy_read_hpd (hdmi, hdmi->phy.data); +hdmi_modb (hdmi, status == connector_status_connected ? +0 :HDMI_PHY_HPD ,HDMI_PHY_HPD ,HDMI_PHY_POL0 ); -if (!(phy_stat & (HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ))) - status = connector_status_disconnected; - } - -if (status != connector_status_unknown) {dev_dbg (hdmi->dev,"EVENT=%s\n" , status == connector_status_connected ?"plugin" :"plugout" );@@ -3141,8 +3043,7 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) }hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 ); -hdmi_writeb (hdmi, ~(HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE ), -HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 );return IRQ_HANDLED ; }@@ -3343,9 +3244,6 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev, hdmi->dev = dev; hdmi->sample_rate =48000 ; hdmi->channels =2 ; - hdmi->disabled = true; - hdmi->rxsense = true; - hdmi->phy_mask = (u8 )~(HDMI_PHY_HPD |HDMI_PHY_RX_SENSE ); hdmi->mc_clkdis =0x7f ; hdmi->last_connector_result = connector_status_disconnected;@@ -3599,8 +3497,7 @@ void dw_hdmi_remove(struct dw_hdmi *hdmi)/* Free, mute and clear phy interrupts */ devm_free_irq (hdmi->dev, irq, hdmi);hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); -hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD |HDMI_IH_PHY_STAT0_RX_SENSE , -HDMI_IH_PHY_STAT0 ); +hdmi_writeb (hdmi,HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_PHY_STAT0 );/* Cancel any pending hot plug work */ cancel_delayed_work_sync (&hdmi->hpd_work);Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 21/23 drm: bridge: dw_hdmi: Remove the empty dw_hdmi_setup_rx_sense()
In reply to Jonas KarlmanThe dw_hdmi_setup_rx_sense() helper is empty and no longer needed after recent RXSENSE and HPD rework, remove it.
---
v7: No change
v6: Collect t-b tag
v5: No change
v4: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 5 ----- drivers/gpu/drm/meson/meson_dw_hdmi.c | 3 --- include/drm/bridge/dw_hdmi.h | 2 -- 3 files changed, 10 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0 -5
@@ -3009,11 +3009,6 @@ static irqreturn_t dw_hdmi_hardirq(int irq, void *dev_id)return ret; } -void dw_hdmi_setup_rx_sense (struct dw_hdmi *hdmi,bool hpd,bool rx_sense) -{ -} -EXPORT_SYMBOL_GPL (dw_hdmi_setup_rx_sense); -static irqreturn_t dw_hdmi_irq (int irq,void *dev_id) {struct dw_hdmi *hdmi = dev_id;
drivers/gpu/drm/meson/meson_dw_hdmi.c +0 -3
@@ -524,9 +524,6 @@ static irqreturn_t dw_hdmi_top_thread_irq(int irq, void *dev_id)if (stat &HDMITX_TOP_INTR_HPD_RISE ) hpd_connected = true; -dw_hdmi_setup_rx_sense (dw_hdmi->hdmi, hpd_connected, - hpd_connected); -drm_helper_hpd_irq_event (dw_hdmi->bridge->dev); drm_bridge_hpd_notify(dw_hdmi->bridge, hpd_connected ? connector_status_connected
include/drm/bridge/dw_hdmi.h +0 -2
@@ -186,8 +186,6 @@ struct dw_hdmi *dw_hdmi_bind(struct platform_device *pdev,void dw_hdmi_resume (struct dw_hdmi *hdmi); -void dw_hdmi_setup_rx_sense (struct dw_hdmi *hdmi,bool hpd,bool rx_sense); -int dw_hdmi_set_plugged_cb (struct dw_hdmi *hdmi,hdmi_codec_plugged_cb fn,struct device *codec_dev);void dw_hdmi_set_sample_non_pcm (struct dw_hdmi *hdmi,unsigned int non_pcm);
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 22/23 drm: bridge: dw_hdmi: Remove the empty dw_hdmi_phy_update_hpd()
In reply to Jonas KarlmanThe dw_hdmi_phy_update_hpd() helper is empty and no longer needed after recent RXSENSE and HPD rework, remove it.
---
v7: No change
v6: Collect t-b tag
v5: No change
v4: New patch
--- drivers/gpu/drm/bridge/imx/imx8mp-hdmi-tx.c | 1 - drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 7 ------- drivers/gpu/drm/rockchip/dw_hdmi-rockchip.c | 2 -- drivers/gpu/drm/sun4i/sun8i_hdmi_phy.c | 2 -- include/drm/bridge/dw_hdmi.h | 4 ---- 5 files changed, 16 deletions(-)
drivers/gpu/drm/bridge/imx/imx8mp-hdmi-tx.c +0 -1
@@ -78,7 +78,6 @@ static const struct dw_hdmi_phy_ops imx8mp_hdmi_phy_ops = { .disable = imx8mp_hdmi_phy_disable, .setup_hpd = im8mp_hdmi_phy_setup_hpd, .read_hpd = dw_hdmi_phy_read_hpd, - .update_hpd = dw_hdmi_phy_update_hpd, };static int imx8mp_dw_hdmi_bind (struct device *dev)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0 -7
@@ -1687,12 +1687,6 @@ enum drm_connector_status dw_hdmi_phy_read_hpd(struct dw_hdmi *hdmi, }EXPORT_SYMBOL_GPL (dw_hdmi_phy_read_hpd); -void dw_hdmi_phy_update_hpd (struct dw_hdmi *hdmi,void *data, -bool force,bool disabled,bool rxsense) -{ -} -EXPORT_SYMBOL_GPL (dw_hdmi_phy_update_hpd); -void dw_hdmi_phy_setup_hpd (struct dw_hdmi *hdmi,void *data) {/*
@@ -1716,7 +1710,6 @@ static const struct dw_hdmi_phy_ops dw_hdmi_synopsys_phy_ops = { .init = dw_hdmi_phy_init, .disable = dw_hdmi_phy_disable, .read_hpd = dw_hdmi_phy_read_hpd, - .update_hpd = dw_hdmi_phy_update_hpd, .setup_hpd = dw_hdmi_phy_setup_hpd, };
drivers/gpu/drm/rockchip/dw_hdmi-rockchip.c +0 -2
@@ -413,7 +413,6 @@ static const struct dw_hdmi_phy_ops rk3228_hdmi_phy_ops = { .init = dw_hdmi_rockchip_genphy_init, .disable = dw_hdmi_rockchip_genphy_disable, .read_hpd = dw_hdmi_phy_read_hpd, - .update_hpd = dw_hdmi_phy_update_hpd, .setup_hpd = dw_hdmi_rk3228_setup_hpd, };
@@ -449,7 +448,6 @@ static const struct dw_hdmi_phy_ops rk3328_hdmi_phy_ops = { .init = dw_hdmi_rockchip_genphy_init, .disable = dw_hdmi_rockchip_genphy_disable, .read_hpd = dw_hdmi_rk3328_read_hpd, - .update_hpd = dw_hdmi_phy_update_hpd, .setup_hpd = dw_hdmi_rk3328_setup_hpd, };
drivers/gpu/drm/sun4i/sun8i_hdmi_phy.c +0 -2
@@ -221,7 +221,6 @@ static const struct dw_hdmi_phy_ops sun8i_a83t_hdmi_phy_ops = { .init = sun8i_a83t_hdmi_phy_config, .disable = sun8i_a83t_hdmi_phy_disable, .read_hpd = dw_hdmi_phy_read_hpd, - .update_hpd = dw_hdmi_phy_update_hpd, .setup_hpd = dw_hdmi_phy_setup_hpd, };
@@ -395,7 +394,6 @@ static const struct dw_hdmi_phy_ops sun8i_h3_hdmi_phy_ops = { .init = sun8i_h3_hdmi_phy_config, .disable = sun8i_h3_hdmi_phy_disable, .read_hpd = dw_hdmi_phy_read_hpd, - .update_hpd = dw_hdmi_phy_update_hpd, .setup_hpd = dw_hdmi_phy_setup_hpd, };
include/drm/bridge/dw_hdmi.h +0 -4
@@ -118,8 +118,6 @@ struct dw_hdmi_phy_ops { const struct drm_display_mode *mode);void (*disable)(struct dw_hdmi *hdmi,void *data);enum drm_connector_status (*read_hpd)(struct dw_hdmi *hdmi,void *data); -void (*update_hpd)(struct dw_hdmi *hdmi,void *data, -bool force,bool disabled,bool rxsense);void (*setup_hpd)(struct dw_hdmi *hdmi,void *data); };
@@ -213,8 +211,6 @@ void dw_hdmi_phy_gen2_reset(struct dw_hdmi *hdmi);enum drm_connector_status dw_hdmi_phy_read_hpd (struct dw_hdmi *hdmi,void *data); -void dw_hdmi_phy_update_hpd (struct dw_hdmi *hdmi,void *data, -bool force,bool disabled,bool rxsense);void dw_hdmi_phy_setup_hpd (struct dw_hdmi *hdmi,void *data);bool dw_hdmi_bus_fmt_is_420 (struct dw_hdmi *hdmi);
Signature · 1 line
2.54.0
Jonas Karlman
reviewerPATCH v7 23/23 drm: bridge: dw_hdmi: Merge top and bottom half IRQ handlers
In reply to Jonas KarlmanThe bottom half IRQ handler only modify delay of or queue a delayed work used for HPD handling. The mod_delayed_work() called is documented as being safe to call from any context including IRQ handler.
Merge top and bottom half IRQ handlers to simplify IRQ handling now that HPD event is handled using a delayed work.
---
v7: No change
v6: Collect r-b tag
v5: No change
v4: New patch
--- drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 34 ++++++++--------------- 1 file changed, 11 insertions(+), 23 deletions(-)
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +11 -23
@@ -2993,30 +2993,18 @@ static irqreturn_t dw_hdmi_hardirq(int irq, void *dev_id)if (hdmi->i2c) ret =dw_hdmi_i2c_irq (hdmi); - intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); -if (intr_stat) { -hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); -return IRQ_WAKE_THREAD ; - } - -return ret; -} - -static irqreturn_t dw_hdmi_irq (int irq,void *dev_id) -{ -struct dw_hdmi *hdmi = dev_id; -u8 intr_stat; -/* * Interrupt generation is accomplished in the following way: * interrupt = (mask == 0) && (polarity == status) * All interrupts are forwarded to the Interrupt Handler sticky bit * register ih_phy_stat0 and muted using the register ih_mute_phy_stat0. */ - intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); -if (intr_stat &HDMI_IH_PHY_STAT0_HPD ) { + intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ) &HDMI_IH_PHY_STAT0_HPD ; +if (intr_stat) {enum drm_connector_status status; +hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); +/* Set HPD interrupt polarity based on current HPD status. */ status =dw_hdmi_phy_read_hpd (hdmi, hdmi->phy.data); hdmi_modb(hdmi, status == connector_status_connected ?
@@ -3028,12 +3016,13 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id)mod_delayed_work (system_percpu_wq, &hdmi->hpd_work,msecs_to_jiffies (HOTPLUG_DEBOUNCE_MS )); + +hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 ); +hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 ); + ret =IRQ_HANDLED ; } -hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 ); -hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 ); - -return IRQ_HANDLED ; +return ret; }static void dw_hdmi_hpd_work (struct work_struct *work)
@@ -3343,9 +3332,8 @@ struct dw_hdmi *dw_hdmi_probe(struct platform_device *pdev,INIT_DELAYED_WORK (&hdmi->hpd_work, dw_hdmi_hpd_work);disable_delayed_work (&hdmi->hpd_work); - ret =devm_request_threaded_irq (dev, irq, dw_hdmi_hardirq, - dw_hdmi_irq,IRQF_SHARED , -dev_name (dev), hdmi); + ret =devm_request_irq (dev, irq, dw_hdmi_hardirq,IRQF_SHARED , +dev_name (dev), hdmi);if (ret) goto err_res;
Signature · 1 line
2.54.0
[email protected]
AI reviewRe: [PATCH v7 23/23] drm: bridge: dw_hdmi: Merge top and bottom half IRQ handlers
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] Clearing the HPD interrupt status after modifying the interrupt polarity creates a race window…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Clearing the HPD interrupt status after modifying the interrupt polarity creates a race window where an edge transition is missed, breaking hotplug detection. --
commit b4c743935eb9892c0e2a6a962a52440a37e985de
Author: Jonas Karlman <[email protected]>
drm: bridge: dw_hdmi: Merge top and bottom half IRQ handlers
This commit simplifies IRQ handling by merging the top and bottom half IRQ handlers into a single handler. It transitions HPD event handling to use a delayed work queue instead of an IRQ thread.
Quoted 4 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +0
-0@@ -2993,30 +2993,18 @@ static irqreturn_t dw_hdmi_hardirq(int irq, void *dev_id)
[ ... ]
Quoted 12 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +4
-2- intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ); -if (intr_stat &HDMI_IH_PHY_STAT0_HPD ) { + intr_stat =hdmi_readb (hdmi,HDMI_IH_PHY_STAT0 ) &HDMI_IH_PHY_STAT0_HPD ; +if (intr_stat) {enum drm_connector_status status; +hdmi_writeb (hdmi, ~0 ,HDMI_IH_MUTE_PHY_STAT0 ); +/* Set HPD interrupt polarity based on current HPD status. */ status =dw_hdmi_phy_read_hpd (hdmi, hdmi->phy.data); hdmi_modb(hdmi, status == connector_status_connected ?
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c static irqreturn_t dw_hdmi_irq(int irq, void *dev_id) · line 3020
Quoted 8 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +1
-3@@ -3028,12 +3016,13 @@ static irqreturn_t dw_hdmi_irq(int irq, void *dev_id)mod_delayed_work (system_percpu_wq, &hdmi->hpd_work,msecs_to_jiffies (HOTPLUG_DEBOUNCE_MS )); - } -hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 ); -hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 ); +hdmi_writeb (hdmi, intr_stat,HDMI_IH_PHY_STAT0 );
Does clearing the HPD interrupt status after modifying the interrupt polarity create a race window where an edge transition is missed?
While this isn't a new regression introduced by this patch, it persists in the new merged handler:
1. The ISR reads HDMI_IH_PHY_STAT0 and finds the HPD bit set.
2. It reads the current PHY state (e.g. CONNECTED) and updates the polarity HDMI_PHY_POL0 to look for the opposite edge (e.g. PLUGOUT).
3. If a fast bounce occurs right here (the cable is unplugged), the hardware detects the edge and sets the HPD sticky bit to 1 again.
4. The ISR then writes the original intr_stat (with the HPD bit set) back to HDMI_IH_PHY_STAT0 (a write-1-to-clear register). This clears the newly latched unplug event.
This leaves the hardware waiting for a PLUGOUT event, but the cable is already disconnected. When the cable is plugged back in, the hardware ignores the PLUGIN edge because it is still looking for a PLUGOUT.
Could this lead to a regression where hotplug interrupts break until the next reboot? Should the W1C register be cleared before reading the PHY status and updating the polarity?
Quoted 7 lines
drivers/gpu/drm/bridge/synopsys/dw-hdmi.c +4
-1+hdmi_writeb (hdmi, ~HDMI_IH_PHY_STAT0_HPD ,HDMI_IH_MUTE_PHY_STAT0 ); + ret =IRQ_HANDLED ; + } -return IRQ_HANDLED ; +return ret; }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=23
Am Montag, 18. Mai 2026, 20:01:36 Mitteleuropäische Sommerzeit schrieb Jonas Karlman:
Quoted 15 lines
This is a revival of an old dw-hdmi series and is the first series part of a new effort to upstream old LibreELEC HDMI 2.0 patches for Rockchip RK33xx devices.
This series ensure poweron/poweroff and CEC phys addr invalidation is happening during normal DRM funcs, ensures EDID and CEC phys addr is updated in
detect()similar to how the bridge connector works with a HDMI bridge attached, and also changes to debounce hotplug processing to prevent a full disable/enable cycle during a HPD low voltage pulse.After this series HPD, EDID and CEC handling should work very similar regardless is the dw-hdmi connector or the bridge connector is used. It should also help ensure a smoother transition when dw-hdmi is fully converted into a HDMI bridge in a future series.
Quoted 4 lines
These changes have mainly been tested on Rockchip RK3328, RK3399 and RK3568 devices using both the dw-hdmi connector and also using a basic convert to use a bridge connector. The changes has also been tested on Amlogic S905X, S905Y2 and A311D devices that uses the bridge connector.
I now also gave this a spin on a rk3328 board and hdmi still workx :-)