| Message ID | 1522307533-31629-1-git-send-email-yang.a.shi@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Quoting Yang (2018-03-29 08:12:13) > From: Yang Shi <yang.a.shi@intel.com> > > snd_hdac driver would use the component interface from i915 driver. > if i915 driver do the audio component intialization too late, snd_hdac > driver will meet ipanic. > > Signed-off-by: Bo He <bo.he@intel.com> > Signed-off-by: Yang Shi <yang.a.shi@intel.com> > --- > drivers/gpu/drm/i915/i915_drv.c | 2 -- > drivers/gpu/drm/i915/intel_display.c | 2 ++ > 2 files changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c > index 2f5209d..9d25d7e 100644 > --- a/drivers/gpu/drm/i915/i915_drv.c > +++ b/drivers/gpu/drm/i915/i915_drv.c > @@ -1243,8 +1243,6 @@ static void i915_driver_register(struct drm_i915_private *dev_priv) > if (IS_GEN5(dev_priv)) > intel_gpu_ips_init(dev_priv); > > - intel_audio_init(dev_priv); > - > /* > * Some ports require correctly set-up hpd registers for detection to > * work properly (leading to ghost connected connector status), e.g. VGA > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c > index f288bcc..a471c88 100644 > --- a/drivers/gpu/drm/i915/intel_display.c > +++ b/drivers/gpu/drm/i915/intel_display.c > @@ -14468,6 +14468,8 @@ int intel_modeset_init(struct drm_device *dev) > > dev->mode_config.funcs = &intel_mode_funcs; > > + intel_audio_init(dev_priv); Has info->num_pipes been finalized yet? Does the component framework expose the device to the external clients (if so, it can not be done before we are ready). -Chris
On Thu, 29 Mar 2018, Chris Wilson <chris@chris-wilson.co.uk> wrote: > Quoting Yang (2018-03-29 08:12:13) >> From: Yang Shi <yang.a.shi@intel.com> >> >> snd_hdac driver would use the component interface from i915 driver. >> if i915 driver do the audio component intialization too late, snd_hdac >> driver will meet ipanic. >> >> Signed-off-by: Bo He <bo.he@intel.com> >> Signed-off-by: Yang Shi <yang.a.shi@intel.com> >> --- >> drivers/gpu/drm/i915/i915_drv.c | 2 -- >> drivers/gpu/drm/i915/intel_display.c | 2 ++ >> 2 files changed, 2 insertions(+), 2 deletions(-) >> >> diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c >> index 2f5209d..9d25d7e 100644 >> --- a/drivers/gpu/drm/i915/i915_drv.c >> +++ b/drivers/gpu/drm/i915/i915_drv.c >> @@ -1243,8 +1243,6 @@ static void i915_driver_register(struct drm_i915_private *dev_priv) >> if (IS_GEN5(dev_priv)) >> intel_gpu_ips_init(dev_priv); >> >> - intel_audio_init(dev_priv); >> - >> /* >> * Some ports require correctly set-up hpd registers for detection to >> * work properly (leading to ghost connected connector status), e.g. VGA >> diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c >> index f288bcc..a471c88 100644 >> --- a/drivers/gpu/drm/i915/intel_display.c >> +++ b/drivers/gpu/drm/i915/intel_display.c >> @@ -14468,6 +14468,8 @@ int intel_modeset_init(struct drm_device *dev) >> >> dev->mode_config.funcs = &intel_mode_funcs; >> >> + intel_audio_init(dev_priv); > > Has info->num_pipes been finalized yet? Does the component framework > expose the device to the external clients (if so, it can not be done > before we are ready). Hmmh, the same patch seems to have been sent twice. I replied at [1]. BR, Jani. [1] http://mid.mail-archive.com/87k1tvciec.fsf@intel.com > -Chris > _______________________________________________ > Intel-gfx mailing list > Intel-gfx@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
> -----Original Message----- > From: Chris Wilson [mailto:chris@chris-wilson.co.uk] > Sent: Thursday, March 29, 2018 7:56 PM > To: Shi, Yang A <yang.a.shi@intel.com>; intel-gfx@lists.freedesktop.org > Cc: Shi, Yang A <yang.a.shi@intel.com>; He, Bo <bo.he@intel.com> > Subject: Re: [Intel-gfx] [PATCH 1/1] drm/i915: move audio component intialization > before audio driver use it > > Quoting Yang (2018-03-29 08:12:13) > > From: Yang Shi <yang.a.shi@intel.com> > > > > snd_hdac driver would use the component interface from i915 driver. > > if i915 driver do the audio component intialization too late, snd_hdac > > driver will meet ipanic. > > > > Signed-off-by: Bo He <bo.he@intel.com> > > Signed-off-by: Yang Shi <yang.a.shi@intel.com> > > --- > > drivers/gpu/drm/i915/i915_drv.c | 2 -- > > drivers/gpu/drm/i915/intel_display.c | 2 ++ > > 2 files changed, 2 insertions(+), 2 deletions(-) > > > > diff --git a/drivers/gpu/drm/i915/i915_drv.c > > b/drivers/gpu/drm/i915/i915_drv.c index 2f5209d..9d25d7e 100644 > > --- a/drivers/gpu/drm/i915/i915_drv.c > > +++ b/drivers/gpu/drm/i915/i915_drv.c > > @@ -1243,8 +1243,6 @@ static void i915_driver_register(struct drm_i915_private > *dev_priv) > > if (IS_GEN5(dev_priv)) > > intel_gpu_ips_init(dev_priv); > > > > - intel_audio_init(dev_priv); > > - > > /* > > * Some ports require correctly set-up hpd registers for detection to > > * work properly (leading to ghost connected connector > > status), e.g. VGA diff --git a/drivers/gpu/drm/i915/intel_display.c > > b/drivers/gpu/drm/i915/intel_display.c > > index f288bcc..a471c88 100644 > > --- a/drivers/gpu/drm/i915/intel_display.c > > +++ b/drivers/gpu/drm/i915/intel_display.c > > @@ -14468,6 +14468,8 @@ int intel_modeset_init(struct drm_device *dev) > > > > dev->mode_config.funcs = &intel_mode_funcs; > > > > + intel_audio_init(dev_priv); > > Has info->num_pipes been finalized yet? Does the component framework expose the > device to the external clients (if so, it can not be done before we are ready). > -Chris Yes, info->num_pipes is finalized yet. This audio component will be used in audio driver snd_hdac_i915_init.
>-----Original Message----- >From: Jani Nikula [mailto:jani.nikula@linux.intel.com] >Sent: Thursday, March 29, 2018 9:16 PM >To: Chris Wilson <chris@chris-wilson.co.uk>; Shi, Yang A <yang.a.shi@intel.com>; >intel-gfx@lists.freedesktop.org >Cc: Shi, Yang A <yang.a.shi@intel.com>; He, Bo <bo.he@intel.com> >Subject: Re: [Intel-gfx] [PATCH 1/1] drm/i915: move audio component intialization >before audio driver use it > >On Thu, 29 Mar 2018, Chris Wilson <chris@chris-wilson.co.uk> wrote: >> Quoting Yang (2018-03-29 08:12:13) >>> From: Yang Shi <yang.a.shi@intel.com> >>> >>> snd_hdac driver would use the component interface from i915 driver. >>> if i915 driver do the audio component intialization too late, >>> snd_hdac driver will meet ipanic. >>> >>> Signed-off-by: Bo He <bo.he@intel.com> >>> Signed-off-by: Yang Shi <yang.a.shi@intel.com> >>> --- >>> drivers/gpu/drm/i915/i915_drv.c | 2 -- >>> drivers/gpu/drm/i915/intel_display.c | 2 ++ >>> 2 files changed, 2 insertions(+), 2 deletions(-) >>> >>> diff --git a/drivers/gpu/drm/i915/i915_drv.c >>> b/drivers/gpu/drm/i915/i915_drv.c index 2f5209d..9d25d7e 100644 >>> --- a/drivers/gpu/drm/i915/i915_drv.c >>> +++ b/drivers/gpu/drm/i915/i915_drv.c >>> @@ -1243,8 +1243,6 @@ static void i915_driver_register(struct drm_i915_private >*dev_priv) >>> if (IS_GEN5(dev_priv)) >>> intel_gpu_ips_init(dev_priv); >>> >>> - intel_audio_init(dev_priv); >>> - >>> /* >>> * Some ports require correctly set-up hpd registers for detection to >>> * work properly (leading to ghost connected connector >>> status), e.g. VGA diff --git a/drivers/gpu/drm/i915/intel_display.c >>> b/drivers/gpu/drm/i915/intel_display.c >>> index f288bcc..a471c88 100644 >>> --- a/drivers/gpu/drm/i915/intel_display.c >>> +++ b/drivers/gpu/drm/i915/intel_display.c >>> @@ -14468,6 +14468,8 @@ int intel_modeset_init(struct drm_device >>> *dev) >>> >>> dev->mode_config.funcs = &intel_mode_funcs; >>> >>> + intel_audio_init(dev_priv); >> >> Has info->num_pipes been finalized yet? Does the component framework >> expose the device to the external clients (if so, it can not be done >> before we are ready). > >Hmmh, the same patch seems to have been sent twice. I replied at [1]. > >BR, >Jani. issue: snd_soc_skl meet "failed to add i915 component master (-19)" when platform don't connect any display output. i915 do initialization before than skl_probe, but if there is no display output connect, in function drm_dp_dpcd_access, there is a 32 retry for aux i2c transactions. It will meet timeout and do usleep. Then skl_probe function will be scheduled. It will call snd_hdac_i915_init, and it will meet "failed to add i915 component master" error. And whole snd_soc_skl initialization will be failed, audio can't work normally anymore. So i915 driver need to move intel_audio_init at the beginning of intel_modeset_init. This will make sure i915_audio_component_init process before snd_hdac_i915_init call it. BR. Yang. > > >[1] http://mid.mail-archive.com/87k1tvciec.fsf@intel.com > > >> -Chris >> _______________________________________________ >> Intel-gfx mailing list >> Intel-gfx@lists.freedesktop.org >> https://lists.freedesktop.org/mailman/listinfo/intel-gfx > >-- >Jani Nikula, Intel Open Source Technology Center
On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: > issue: snd_soc_skl meet "failed to add i915 component master (-19)" > when platform don't connect any display output. > > i915 do initialization before than skl_probe, but if there is no > display output connect, in function drm_dp_dpcd_access, there is a 32 > retry for aux i2c transactions. It will meet timeout and do > usleep. Then skl_probe function will be scheduled. It will call > snd_hdac_i915_init, and it will meet "failed to add i915 component > master" error. And whole snd_soc_skl initialization will be failed, > audio can't work normally anymore. > > So i915 driver need to move intel_audio_init at the beginning of > intel_modeset_init. This will make sure i915_audio_component_init > process before snd_hdac_i915_init call it. We do intel_audio_init() and register the audio component when we are ready to handle the audio component calls. We are ready at i915_driver_register(). We are not ready at intel_modeset_init(). BR, Jani.
>-----Original Message----- >From: Jani Nikula [mailto:jani.nikula@linux.intel.com] >Sent: Tuesday, April 10, 2018 3:57 PM >To: Shi, Yang A <yang.a.shi@intel.com>; Chris Wilson <chris@chris-wilson.co.uk>; >intel-gfx@lists.freedesktop.org >Cc: He, Bo <bo.he@intel.com> >Subject: RE: [Intel-gfx] [PATCH 1/1] drm/i915: move audio component intialization >before audio driver use it > >On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >> issue: snd_soc_skl meet "failed to add i915 component master (-19)" >> when platform don't connect any display output. >> >> i915 do initialization before than skl_probe, but if there is no >> display output connect, in function drm_dp_dpcd_access, there is a 32 >> retry for aux i2c transactions. It will meet timeout and do usleep. >> Then skl_probe function will be scheduled. It will call >> snd_hdac_i915_init, and it will meet "failed to add i915 component >> master" error. And whole snd_soc_skl initialization will be failed, >> audio can't work normally anymore. >> >> So i915 driver need to move intel_audio_init at the beginning of >> intel_modeset_init. This will make sure i915_audio_component_init >> process before snd_hdac_i915_init call it. > >We do intel_audio_init() and register the audio component when we are ready to handle >the audio component calls. We are ready at i915_driver_register(). We are not ready at >intel_modeset_init(). > >BR, >Jani. Thanks to comments my patch. After I check the whole driver code, I think all ops in i915_audio_component_ops should be ready at the beginning of function intel_modeset_init. So can we move intel_audio_init as early as we can. Would you like to suggest a better place to do intel_audio_init? Thanks. BR. Yang. > > >-- >Jani Nikula, Intel Open Source Technology Center
On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >>On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >>> issue: snd_soc_skl meet "failed to add i915 component master (-19)" >>> when platform don't connect any display output. >>> >>> i915 do initialization before than skl_probe, but if there is no >>> display output connect, in function drm_dp_dpcd_access, there is a 32 >>> retry for aux i2c transactions. It will meet timeout and do usleep. >>> Then skl_probe function will be scheduled. It will call >>> snd_hdac_i915_init, and it will meet "failed to add i915 component >>> master" error. And whole snd_soc_skl initialization will be failed, >>> audio can't work normally anymore. >>> >>> So i915 driver need to move intel_audio_init at the beginning of >>> intel_modeset_init. This will make sure i915_audio_component_init >>> process before snd_hdac_i915_init call it. >> >>We do intel_audio_init() and register the audio component when we are >>ready to handle the audio component calls. We are ready at >>i915_driver_register(). We are not ready at intel_modeset_init(). >> >>BR, >>Jani. > > Thanks to comments my patch. > After I check the whole driver code, I think all ops in > i915_audio_component_ops should be ready at the beginning of function > intel_modeset_init. So can we move intel_audio_init as early as we > can. No, that's not true. Just as an example, dev_priv->cdclk.hw.cdclk hasn't been initialized. > Would you like to suggest a better place to do intel_audio_init? I think the call is already where it is supposed to be. We expose ourselves to the rest of the system when we are ready. If it takes long, it takes long. I think you have a race in your driver, and you need to deal with it properly in your driver. In snd_hdac_i915_init(), I don't think there are any guarantees that the request_module() call is the one actually probing i915. We might already be mid-probe. You don't even check or log request_module() return value. I'm also not 100% sure at what point of driver loading request_module() returns. I think it's when the module init hook returns, which should be all right, but again, I don't think you can count on that if it isn't your request_module() that actually probes i915. I think the patch at hand is a hack that reduces the window for the race, and not a real fix. Moreover, it makes the i915 audio component code fragile by introducing tricky probe order dependencies that we've been systematically trying to reduce by placing the call where it is now. Cc: Lucas for any further input on module probing. BR, Jani.
On Tue, 10 Apr 2018, Jani Nikula <jani.nikula@linux.intel.com> wrote: > On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >>>On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >>>> issue: snd_soc_skl meet "failed to add i915 component master (-19)" >>>> when platform don't connect any display output. >>>> >>>> i915 do initialization before than skl_probe, but if there is no >>>> display output connect, in function drm_dp_dpcd_access, there is a 32 >>>> retry for aux i2c transactions. It will meet timeout and do usleep. >>>> Then skl_probe function will be scheduled. It will call >>>> snd_hdac_i915_init, and it will meet "failed to add i915 component >>>> master" error. And whole snd_soc_skl initialization will be failed, >>>> audio can't work normally anymore. >>>> >>>> So i915 driver need to move intel_audio_init at the beginning of >>>> intel_modeset_init. This will make sure i915_audio_component_init >>>> process before snd_hdac_i915_init call it. >>> >>>We do intel_audio_init() and register the audio component when we are >>>ready to handle the audio component calls. We are ready at >>>i915_driver_register(). We are not ready at intel_modeset_init(). >>> >>>BR, >>>Jani. >> >> Thanks to comments my patch. >> After I check the whole driver code, I think all ops in >> i915_audio_component_ops should be ready at the beginning of function >> intel_modeset_init. So can we move intel_audio_init as early as we >> can. > > No, that's not true. Just as an example, dev_priv->cdclk.hw.cdclk hasn't > been initialized. > >> Would you like to suggest a better place to do intel_audio_init? > > I think the call is already where it is supposed to be. We expose > ourselves to the rest of the system when we are ready. If it takes long, > it takes long. I think you have a race in your driver, and you need to > deal with it properly in your driver. > > In snd_hdac_i915_init(), I don't think there are any guarantees that the > request_module() call is the one actually probing i915. We might already > be mid-probe. You don't even check or log request_module() return value. > > I'm also not 100% sure at what point of driver loading request_module() > returns. I think it's when the module init hook returns, which should be > all right, but again, I don't think you can count on that if it isn't > your request_module() that actually probes i915. > > I think the patch at hand is a hack that reduces the window for the > race, and not a real fix. Moreover, it makes the i915 audio component > code fragile by introducing tricky probe order dependencies that we've > been systematically trying to reduce by placing the call where it is > now. > > Cc: Lucas for any further input on module probing. Apparently there was also a bug in some version of kmod/modprobe which could have lead to what you're experiencing. Are you running the fixed version? See [1]. BR, Jani. [1] https://github.com/lucasdemarchi/kmod/commit/fd44a98ae2eb5eb32161088954ab21e58e19dfc4
On Tue, Apr 10, 2018 at 10:58:28AM -0300, Jani Nikula wrote: > On Tue, 10 Apr 2018, Jani Nikula <jani.nikula@linux.intel.com> wrote: > > On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: > >>>On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: > >>>> issue: snd_soc_skl meet "failed to add i915 component master (-19)" > >>>> when platform don't connect any display output. > >>>> > >>>> i915 do initialization before than skl_probe, but if there is no > >>>> display output connect, in function drm_dp_dpcd_access, there is a 32 > >>>> retry for aux i2c transactions. It will meet timeout and do usleep. You should not really rely on "module is loaded, I can use it". It may not be bound to the device yet in case of async probe for example. You should rather wait for the service you want to use from the driver to be ready. See documentation of __request_module(). > >>>> Then skl_probe function will be scheduled. It will call > >>>> snd_hdac_i915_init, and it will meet "failed to add i915 component > >>>> master" error. And whole snd_soc_skl initialization will be failed, > >>>> audio can't work normally anymore. Humn... I fail to see how the usleep() would cause any of this. If the probe is synchronous (see below) we an eventual call to request_module() will/should wait for the first probe to finish nonetheless. To me it seems the error is elsewhere. > >>>> > >>>> So i915 driver need to move intel_audio_init at the beginning of > >>>> intel_modeset_init. This will make sure i915_audio_component_init > >>>> process before snd_hdac_i915_init call it. > >>> > >>>We do intel_audio_init() and register the audio component when we are > >>>ready to handle the audio component calls. We are ready at > >>>i915_driver_register(). We are not ready at intel_modeset_init(). > >>> > >>>BR, > >>>Jani. > >> > >> Thanks to comments my patch. > >> After I check the whole driver code, I think all ops in > >> i915_audio_component_ops should be ready at the beginning of function > >> intel_modeset_init. So can we move intel_audio_init as early as we > >> can. > > > > No, that's not true. Just as an example, dev_priv->cdclk.hw.cdclk hasn't > > been initialized. > > > >> Would you like to suggest a better place to do intel_audio_init? > > > > I think the call is already where it is supposed to be. We expose > > ourselves to the rest of the system when we are ready. If it takes long, > > it takes long. I think you have a race in your driver, and you need to > > deal with it properly in your driver. > > > > In snd_hdac_i915_init(), I don't think there are any guarantees that the > > request_module() call is the one actually probing i915. We might already This should not be relevant in this case. Even if the module is already mid-probe, request_module() will only return after modprobe returns (since request_module forces UMH_WAIT_PROC). If a second call races with the first, it will try to find the module in the module list and end up waiting for it to complete. See add_unformed_module(). > > be mid-probe. You don't even check or log request_module() return value. Checking the return value of request_module() is absolutely the first thing to do. It's plain broken otherwise... there are many reasons why a call to request_module() could fail. If your driver is being loaded from a work queue, if the modprobe binary is not where it should be, if there's a bug in modprobe, etc etc etc. > > > > I'm also not 100% sure at what point of driver loading request_module() > > returns. I think it's when the module init hook returns, which should be > > all right, but again, I don't think you can count on that if it isn't > > your request_module() that actually probes i915. You can, except if the pci probe is async... In this case modprobe would return before the rest of the initialization. I'm not totally sure about this without giving it a deeper look. But then, moving the call anywhere in that function should produce the same results with enough tries. > > > > I think the patch at hand is a hack that reduces the window for the > > race, and not a real fix. Moreover, it makes the i915 audio component > > code fragile by introducing tricky probe order dependencies that we've > > been systematically trying to reduce by placing the call where it is > > now. > > > > Cc: Lucas for any further input on module probing. > > Apparently there was also a bug in some version of kmod/modprobe which > could have lead to what you're experiencing. Are you running the fixed > version? See [1]. Yep, this might explain as well. But only if i915 is builtin rather than a module. Is this the case here? Lucas De Marchi > > BR, > Jani. > > > [1] https://github.com/lucasdemarchi/kmod/commit/fd44a98ae2eb5eb32161088954ab21e58e19dfc4 > > > -- > Jani Nikula, Intel Open Source Technology Center
>-----Original Message----- >From: Jani Nikula [mailto:jani.nikula@linux.intel.com] >Sent: Tuesday, April 10, 2018 6:57 PM >To: Shi, Yang A <yang.a.shi@intel.com>; Chris Wilson <chris@chris-wilson.co.uk>; >intel-gfx@lists.freedesktop.org >Cc: He, Bo <bo.he@intel.com>; De Marchi, Lucas <lucas.demarchi@intel.com> >Subject: RE: [Intel-gfx] [PATCH 1/1] drm/i915: move audio component intialization >before audio driver use it > >On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >>>On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >>>> issue: snd_soc_skl meet "failed to add i915 component master (-19)" >>>> when platform don't connect any display output. >>>> >>>> i915 do initialization before than skl_probe, but if there is no >>>> display output connect, in function drm_dp_dpcd_access, there is a >>>> 32 retry for aux i2c transactions. It will meet timeout and do usleep. >>>> Then skl_probe function will be scheduled. It will call >>>> snd_hdac_i915_init, and it will meet "failed to add i915 component >>>> master" error. And whole snd_soc_skl initialization will be failed, >>>> audio can't work normally anymore. >>>> >>>> So i915 driver need to move intel_audio_init at the beginning of >>>> intel_modeset_init. This will make sure i915_audio_component_init >>>> process before snd_hdac_i915_init call it. >>> >>>We do intel_audio_init() and register the audio component when we are >>>ready to handle the audio component calls. We are ready at >>>i915_driver_register(). We are not ready at intel_modeset_init(). >>> >>>BR, >>>Jani. >> >> Thanks to comments my patch. >> After I check the whole driver code, I think all ops in >> i915_audio_component_ops should be ready at the beginning of function >> intel_modeset_init. So can we move intel_audio_init as early as we >> can. > >No, that's not true. Just as an example, dev_priv->cdclk.hw.cdclk hasn't been initialized. i915_load_modeset_init->intel_power_domains_init_hw->bxt_display_core_init->bxt_init_cdclk dev_priv->cdclk.hw.cdclk will be initialized here. It happened before driver enter into intel_modeset_init. So it should be safe to be used from audio driver. > >> Would you like to suggest a better place to do intel_audio_init? > >I think the call is already where it is supposed to be. We expose ourselves to the rest of >the system when we are ready. If it takes long, it takes long. I think you have a race in >your driver, and you need to deal with it properly in your driver. > No, display driver consume too much time to do initialization when it don't connect any display Output device.It influence whole audio driver and cause whole audio driver can't works well. This don't make sense. >In snd_hdac_i915_init(), I don't think there are any guarantees that the >request_module() call is the one actually probing i915. We might already be mid-probe. >You don't even check or log request_module() return value. This is the issue in audio driver itself. Need be checked by audio driver owner. BR. Yang. Yang Shi, Intel Open Source Technology Center > >I'm also not 100% sure at what point of driver loading request_module() returns. I think >it's when the module init hook returns, which should be all right, but again, I don't think >you can count on that if it isn't your request_module() that actually probes i915. > >I think the patch at hand is a hack that reduces the window for the race, and not a real >fix. Moreover, it makes the i915 audio component code fragile by introducing tricky >probe order dependencies that we've been systematically trying to reduce by placing the >call where it is now. > >Cc: Lucas for any further input on module probing. > >BR, >Jani. > >-- >Jani Nikula, Intel Open Source Technology Center
>-----Original Message----- >From: De Marchi, Lucas >Sent: Wednesday, April 11, 2018 9:37 AM >To: Jani Nikula <jani.nikula@linux.intel.com> >Cc: Shi, Yang A <yang.a.shi@intel.com>; Chris Wilson <chris@chris-wilson.co.uk>; >intel-gfx@lists.freedesktop.org; He, Bo <bo.he@intel.com>; Deak, Imre ><imre.deak@intel.com> >Subject: Re: [Intel-gfx] [PATCH 1/1] drm/i915: move audio component intialization >before audio driver use it > >On Tue, Apr 10, 2018 at 10:58:28AM -0300, Jani Nikula wrote: >> On Tue, 10 Apr 2018, Jani Nikula <jani.nikula@linux.intel.com> wrote: >> > On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >> >>>On Tue, 10 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >> >>>> issue: snd_soc_skl meet "failed to add i915 component master (-19)" >> >>>> when platform don't connect any display output. >> >>>> >> >>>> i915 do initialization before than skl_probe, but if there is no >> >>>> display output connect, in function drm_dp_dpcd_access, there is >> >>>> a 32 retry for aux i2c transactions. It will meet timeout and do usleep. > >You should not really rely on "module is loaded, I can use it". It may not be bound to >the device yet in case of async probe for example. >You should rather wait for the service you want to use from the driver to be ready. See >documentation of __request_module(). > >> >>>> Then skl_probe function will be scheduled. It will call >> >>>> snd_hdac_i915_init, and it will meet "failed to add i915 >> >>>> component master" error. And whole snd_soc_skl initialization >> >>>> will be failed, audio can't work normally anymore. > >Humn... I fail to see how the usleep() would cause any of this. If the probe is >synchronous (see below) we an eventual call to request_module() will/should wait for >the first probe to finish nonetheless. >To me it seems the error is elsewhere. > >> >>>> >> >>>> So i915 driver need to move intel_audio_init at the beginning of >> >>>> intel_modeset_init. This will make sure i915_audio_component_init >> >>>> process before snd_hdac_i915_init call it. >> >>> >> >>>We do intel_audio_init() and register the audio component when we >> >>>are ready to handle the audio component calls. We are ready at >> >>>i915_driver_register(). We are not ready at intel_modeset_init(). >> >>> >> >>>BR, >> >>>Jani. >> >> >> >> Thanks to comments my patch. >> >> After I check the whole driver code, I think all ops in >> >> i915_audio_component_ops should be ready at the beginning of >> >> function intel_modeset_init. So can we move intel_audio_init as >> >> early as we can. >> > >> > No, that's not true. Just as an example, dev_priv->cdclk.hw.cdclk >> > hasn't been initialized. >> > >> >> Would you like to suggest a better place to do intel_audio_init? >> > >> > I think the call is already where it is supposed to be. We expose >> > ourselves to the rest of the system when we are ready. If it takes >> > long, it takes long. I think you have a race in your driver, and you >> > need to deal with it properly in your driver. >> > >> > In snd_hdac_i915_init(), I don't think there are any guarantees that >> > the >> > request_module() call is the one actually probing i915. We might >> > already > >This should not be relevant in this case. Even if the module is already mid-probe, >request_module() will only return after modprobe returns (since request_module forces >UMH_WAIT_PROC). > This issue is not related to request_module. When issue happened, request_module get i915 Correctly and return 0 successfully. It just be caused by acomp->ops is null in function snd_hdac_i915_init after it called Component_master_add_with_match. If intel_audio_init can be called before it check acomp->ops. Everything will goes well. But intel_audio_init is called too late because i915 driver go to usleep. BR. Yang. >If a second call races with the first, it will try to find the module in the module list and >end up waiting for it to complete. See add_unformed_module(). > >> > be mid-probe. You don't even check or log request_module() return value. > >Checking the return value of request_module() is absolutely the first thing to do. It's >plain broken otherwise... there are many reasons why a call to request_module() could >fail. If your driver is being loaded from a work queue, if the modprobe binary is not >where it should be, if there's a bug in modprobe, etc etc etc. > >> > >> > I'm also not 100% sure at what point of driver loading >> > request_module() returns. I think it's when the module init hook >> > returns, which should be all right, but again, I don't think you can >> > count on that if it isn't your request_module() that actually probes i915. > >You can, except if the pci probe is async... In this case modprobe would return before >the rest of the initialization. I'm not totally sure about this without giving it a deeper >look. But then, moving the call anywhere in that function should produce the same >results with enough tries. > >> > >> > I think the patch at hand is a hack that reduces the window for the >> > race, and not a real fix. Moreover, it makes the i915 audio >> > component code fragile by introducing tricky probe order >> > dependencies that we've been systematically trying to reduce by >> > placing the call where it is now. >> > >> > Cc: Lucas for any further input on module probing. >> >> Apparently there was also a bug in some version of kmod/modprobe which >> could have lead to what you're experiencing. Are you running the fixed >> version? See [1]. > >Yep, this might explain as well. But only if i915 is builtin rather than a module. Is this >the case here? > >Lucas De Marchi > >> >> BR, >> Jani. >> >> >> [1] >> https://github.com/lucasdemarchi/kmod/commit/fd44a98ae2eb5eb3216108895 >> 4ab21e58e19dfc4 >> >> >> -- >> Jani Nikula, Intel Open Source Technology Center
On Wed, 11 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >>No, that's not true. Just as an example, dev_priv->cdclk.hw.cdclk >>hasn't been initialized. > i915_load_modeset_init->intel_power_domains_init_hw->bxt_display_core_init->bxt_init_cdclk > dev_priv->cdclk.hw.cdclk will be initialized here. > It happened before driver enter into intel_modeset_init. > So it should be safe to be used from audio driver. Was it easy to figure that out? Imagine double checking everything whenever a) intel_audio.c gets changed, or b) anything in our probe that might impact intel_audio.c gets changed. It's untenable. Our probe ordering is what it is to manage the complexity of the driver. When everything is done, we expose the driver to the rest of the system. Same with sysfs, debugfs, etc. >>> Would you like to suggest a better place to do intel_audio_init? >> >>I think the call is already where it is supposed to be. We expose >>ourselves to the rest of the system when we are ready. If it takes >>long, it takes long. I think you have a race in your driver, and you >>need to deal with it properly in your driver. >> > No, display driver consume too much time to do initialization when it > don't connect any display Output device.It influence whole audio > driver and cause whole audio driver can't works well. This don't make > sense. Our probe takes as long as it takes. It's a complicated driver with complicated probing to do. Why does your design require i915 to be probed as part of your probe? The component interface supports the master and client binding in any order, asynchronously. If you have issues because of the timing, it's probably a race condition in your driver. Doing the audio init earlier on our side just papers over your race condition without actually fixing it. If we were to optimize our probe, it would probably mean more parallelism and asynchronous probing, and that in turn means even stricter requirements for finding the root cause instead of just fiddling with the call site. BR, Jani.
On Wed, 11 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: > This issue is not related to request_module. When issue happened, > request_module get i915 Correctly and return 0 successfully. It just > be caused by acomp->ops is null in function snd_hdac_i915_init after > it called Component_master_add_with_match. > > If intel_audio_init can be called before it check > acomp->ops. Everything will goes well. But intel_audio_init is called > too late because i915 driver go to usleep. Please add drm.debug=14 module parameter, enable dev_dbg for your driver, and send us the dmesg. BR, Jani.
>-----Original Message----- >From: Jani Nikula [mailto:jani.nikula@linux.intel.com] >Sent: Wednesday, April 11, 2018 8:25 PM >To: Shi, Yang A <yang.a.shi@intel.com>; De Marchi, Lucas ><lucas.demarchi@intel.com> >Cc: Chris Wilson <chris@chris-wilson.co.uk>; intel-gfx@lists.freedesktop.org; He, Bo ><bo.he@intel.com>; Deak, Imre <imre.deak@intel.com> >Subject: RE: [Intel-gfx] [PATCH 1/1] drm/i915: move audio component intialization >before audio driver use it > >On Wed, 11 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >> This issue is not related to request_module. When issue happened, >> request_module get i915 Correctly and return 0 successfully. It just >> be caused by acomp->ops is null in function snd_hdac_i915_init after >> it called Component_master_add_with_match. >> >> If intel_audio_init can be called before it check >> acomp->ops. Everything will goes well. But intel_audio_init is called >> too late because i915 driver go to usleep. > >Please add drm.debug=14 module parameter, enable dev_dbg for your driver, and send >us the dmesg. > >BR, >Jani. Dmesg is as following: i915 driver load at about 0.214362. snd_soc_skl driver is loaded at about 0.853961. it meet error here. Then i915 continue to do driver load. [ 0.214362] [drm:i915_driver_load] No PCH found. [ 0.214377] [drm:intel_power_domains_init] Allowed DC state mask 09 ... [ 0.217374] [drm:i915_driver_load] ppgtt mode: 3 [ 0.217377] [drm:i915_driver_load] use GPU semaphores? no [ 0.217381] [drm:i915_driver_load] preemption enabled? no [ 0.217409] [drm] Memory usable by graphics device = 4078M [ 0.217414] [drm:i915_ggtt_probe_hw] GMADR size = 256M [ 0.217418] [drm:i915_ggtt_probe_hw] GTT stolen size = 64M [ 0.217453] [drm:i915_gem_init_stolen] Memory reserved for graphics device: 65536K, usable: 64512K [ 0.217467] [drm:sanitize_rc6_option] BIOS enabled RC states: HW_CTRL off HW_RC6 off SW_TARGET_STATE 4 [ 0.217496] [drm:intel_opregion_setup] graphic opregion physical addr: 0x7aedd000 [ 0.217533] [drm:intel_opregion_setup] ASLE supported [ 0.217541] [drm:intel_opregion_setup] Found valid VBT in ACPI OpRegion (RVDA) [ 0.217630] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013). [ 0.217633] [drm] Driver supports precise vblank timestamp query. ... [ 0.853961] snd_soc_skl 0000:00:0e.0: enabling device (0000 -> 0002) [ 0.859009] snd_soc_skl 0000:00:0e.0: failed to add i915 component master (-19) [ 0.888176] init: Starting service 'earlyaudioalsa'... [ 0.888471] init: write_file: Unable to open '/sys/kernel/debug/snd_soc_skl/nhlt/control': No such file or directory [ 0.902024] trusty trusty: trusty version: 14:57:13 Feb 7 2018 [ 0.902043] trusty trusty: selected api version: 2 (requested 3) [ 0.904308] trusty-irq trusty-irq: irq from lk = 49 [ 0.910886] trusty-virtio trusty-virtio: initializing [ 0.911177] trusty_ipc virtio0: vring0: va(pa) ffff88007487a000(0) qsz 32 notifyid 1 [ 0.911185] trusty_ipc virtio0: vring1: va(pa) ffff880074894000(0) qsz 32 notifyid 2 [ 0.911415] trusty-virtio trusty-virtio: initializing done [ 0.911695] trusty_ipc virtio0: is online [ 0.916608] trusty-wall trusty-wall: smwall: initialized 4096 bytes [ 0.919025] trusty-timer trusty-timer: initialized [ 0.929655] rpmb_sim rpmb_sim: registered RPMB capacity = 131072 of 512 blocks [ 0.930057] init: Service 'earlyaudioalsa' (pid 1925) exited with status 1 [ 0.930405] init: Starting service 'logd'... [ 0.941825] [drm:intel_dp_aux_ch] dp_aux_ch timeout status 0x7d4003ff [ 0.950299] [drm:intel_dp_aux_ch] dp_aux_ch timeout status 0x7d4003ff [ 0.952397] logd.auditd: start [ 0.952416] logd.klogd: 910077775 [ 0.958767] [drm:intel_dp_aux_ch] dp_aux_ch timeout status 0x7d4003ff [ 0.967223] [drm:intel_dp_aux_ch] dp_aux_ch timeout status 0x7d4003ff BR. Yang. Yang Shi, Intel Open Source Technology Center > >-- >Jani Nikula, Intel Open Source Technology Center
On Thu, 12 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote:
> Dmesg is as following:
Full dmesg from boot please.
BR,
Jani.
>-----Original Message----- >From: Jani Nikula [mailto:jani.nikula@linux.intel.com] >Sent: Thursday, April 12, 2018 4:06 PM >To: Shi, Yang A <yang.a.shi@intel.com>; De Marchi, Lucas ><lucas.demarchi@intel.com> >Cc: Chris Wilson <chris@chris-wilson.co.uk>; intel-gfx@lists.freedesktop.org; He, Bo ><bo.he@intel.com>; Deak, Imre <imre.deak@intel.com> >Subject: RE: [Intel-gfx] [PATCH 1/1] drm/i915: move audio component intialization >before audio driver use it > >On Thu, 12 Apr 2018, "Shi, Yang A" <yang.a.shi@intel.com> wrote: >> Dmesg is as following: > >Full dmesg from boot please. Done, I already send it directly to you. Thanks. BR. Yang. > >BR, >Jani. > > >-- >Jani Nikula, Intel Open Source Technology Center
On Thu, 29 Mar 2018, Yang <yang.a.shi@intel.com> wrote: > From: Yang Shi <yang.a.shi@intel.com> > > snd_hdac driver would use the component interface from i915 driver. > if i915 driver do the audio component intialization too late, snd_hdac > driver will meet ipanic. To follow-up, we figured out off-list that the root cause here were local async probe changes, and the issue that this patch addresses is not present upstream. However, it seems likely that the patch [1] recently proposed by Chris would cause similar issues. Let's continue the discussion on that thread. BR, Jani. [1] http://patchwork.freedesktop.org/patch/msgid/20180323083048.13327-1-chris@chris-wilson.co.uk > > Signed-off-by: Bo He <bo.he@intel.com> > Signed-off-by: Yang Shi <yang.a.shi@intel.com> > --- > drivers/gpu/drm/i915/i915_drv.c | 2 -- > drivers/gpu/drm/i915/intel_display.c | 2 ++ > 2 files changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c > index 2f5209d..9d25d7e 100644 > --- a/drivers/gpu/drm/i915/i915_drv.c > +++ b/drivers/gpu/drm/i915/i915_drv.c > @@ -1243,8 +1243,6 @@ static void i915_driver_register(struct drm_i915_private *dev_priv) > if (IS_GEN5(dev_priv)) > intel_gpu_ips_init(dev_priv); > > - intel_audio_init(dev_priv); > - > /* > * Some ports require correctly set-up hpd registers for detection to > * work properly (leading to ghost connected connector status), e.g. VGA > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c > index f288bcc..a471c88 100644 > --- a/drivers/gpu/drm/i915/intel_display.c > +++ b/drivers/gpu/drm/i915/intel_display.c > @@ -14468,6 +14468,8 @@ int intel_modeset_init(struct drm_device *dev) > > dev->mode_config.funcs = &intel_mode_funcs; > > + intel_audio_init(dev_priv); > + > init_llist_head(&dev_priv->atomic_helper.free_list); > INIT_WORK(&dev_priv->atomic_helper.free_work, > intel_atomic_helper_free_state_worker);
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index 2f5209d..9d25d7e 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -1243,8 +1243,6 @@ static void i915_driver_register(struct drm_i915_private *dev_priv) if (IS_GEN5(dev_priv)) intel_gpu_ips_init(dev_priv); - intel_audio_init(dev_priv); - /* * Some ports require correctly set-up hpd registers for detection to * work properly (leading to ghost connected connector status), e.g. VGA diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index f288bcc..a471c88 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -14468,6 +14468,8 @@ int intel_modeset_init(struct drm_device *dev) dev->mode_config.funcs = &intel_mode_funcs; + intel_audio_init(dev_priv); + init_llist_head(&dev_priv->atomic_helper.free_list); INIT_WORK(&dev_priv->atomic_helper.free_work, intel_atomic_helper_free_state_worker);