[Intel-gfx] [PATCH v2 08/13] drm/i915: Make shared dpll functions take crtc_state

[Maarten Lankhorst]

Op 04-10-18 om 17:24 schreef Ville Syrjälä:
> On Thu, Oct 04, 2018 at 03:25:42PM +0200, Maarten Lankhorst wrote:
>> Op 04-10-18 om 14:57 schreef Ville Syrjälä:
>>> On Thu, Oct 04, 2018 at 11:45:59AM +0200, Maarten Lankhorst wrote:
>>>> Do not rely on crtc->config any more. Remove the assertion from
>>>> ibx_pch_dpll_disable, because we the dpll state tracking should
>>>> already handle this case correctly.
>>>>
>>>> Signed-off-by: Maarten Lankhorst <maarten.lankhorst at linux.intel.com>
>>>> Reviewed-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
>>>> ---
>>>>  drivers/gpu/drm/i915/intel_display.c  | 14 ++++++------
>>>>  drivers/gpu/drm/i915/intel_dpll_mgr.c | 31 +++++++++++----------------
>>>>  drivers/gpu/drm/i915/intel_dpll_mgr.h |  6 +++---
>>>>  3 files changed, 22 insertions(+), 29 deletions(-)
>>>>
>>>> diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
>>>> index fe8ccbdd4ea1..4df4293a7917 100644
>>>> --- a/drivers/gpu/drm/i915/intel_display.c
>>>> +++ b/drivers/gpu/drm/i915/intel_display.c
>>>> @@ -4749,7 +4749,7 @@ static void ironlake_pch_enable(const struct intel_atomic_state *state,
>>>>  	 * Note that enable_shared_dpll tries to do the right thing, but
>>>>  	 * get_shared_dpll unconditionally resets the pll - we need that to have
>>>>  	 * the right LVDS enable sequence. */
>>>> -	intel_enable_shared_dpll(crtc);
>>>> +	intel_enable_shared_dpll(crtc_state);
>>>>  
>>>>  	/* set transcoder timing, panel must allow it */
>>>>  	assert_panel_unlocked(dev_priv, pipe);
>>>> @@ -5591,8 +5591,8 @@ static void ironlake_crtc_enable(struct intel_crtc_state *pipe_config,
>>>>  	intel_set_cpu_fifo_underrun_reporting(dev_priv, pipe, false);
>>>>  	intel_set_pch_fifo_underrun_reporting(dev_priv, pipe, false);
>>>>  
>>>> -	if (intel_crtc->config->has_pch_encoder)
>>>> -		intel_prepare_shared_dpll(intel_crtc);
>>>> +	if (pipe_config->has_pch_encoder)
>>>> +		intel_prepare_shared_dpll(pipe_config);
>>>>  
>>>>  	if (intel_crtc_has_dp_encoder(intel_crtc->config))
>>>>  		intel_dp_set_m_n(intel_crtc, M1_N1);
>>>> @@ -5710,8 +5710,8 @@ static void haswell_crtc_enable(struct intel_crtc_state *pipe_config,
>>>>  
>>>>  	intel_encoders_pre_pll_enable(crtc, pipe_config, old_state);
>>>>  
>>>> -	if (intel_crtc->config->shared_dpll)
>>>> -		intel_enable_shared_dpll(intel_crtc);
>>>> +	if (pipe_config->shared_dpll)
>>>> +		intel_enable_shared_dpll(pipe_config);
>>>>  
>>>>  	if (INTEL_GEN(dev_priv) >= 11)
>>>>  		icl_map_plls_to_ports(crtc, pipe_config, old_state);
>>>> @@ -6286,7 +6286,7 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc,
>>>>  
>>>>  	intel_fbc_disable(intel_crtc);
>>>>  	intel_update_watermarks(intel_crtc);
>>>> -	intel_disable_shared_dpll(intel_crtc);
>>>> +	intel_disable_shared_dpll(crtc_state);
>>> I guess this is the explosion on i915g/i965g. We've already freed this
>>> crtc state at this point. So I guess s/crtc_state/crtc->state/ and it
>>> should be fine?
>>>
>> No, the explosion is at
>>
>> -	if (WARN_ON(pll == NULL))
>> +	if (!WARN_ON(!pll))
>>  		return;
>>
>> Oops! Should be fixed in v2 I'm sending now.
> <4>[   11.181254] R10: 0000000000000000 R11: 0000000000000000 R12: 6b6b6b6b6b6b6b6b
> <4>[   11.181141] RIP: 0010:intel_disable_shared_dpll+0x12/0x130 [i915]
> <4>[   11.181157] Code: c7 e9 67 25 a0 e8 9e da ec e0 0f 0b e9 18 ff ff ff 0f 1f 80 00 00 00 00 41 56 41 55 41 54 55 53 4c 8b 27 48 8b 9f 98 02 00 00 <4d> 8b 34 24 41 80 be 5c 0d 00 00 04 76 05 48 85 db 75 09 5b 5d 41
>
> 72a:   4d 8b 34 24             mov    (%r12),%r14
>
> Looks like it's still trying to drink poison.
>
Yeah, probably intel_crtc_disable_noatomic

Seems I need to change crtc_state (which is freed) to to_intel_crtc_state(crtc->state)