| Message ID | 1445235370-6717-1-git-send-email-tiwai@suse.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Takashi, Removing FW_LOADER_USER_HELPER_FALLBACK leads to a failure in our HQVDP firmware execution. Indeed, our firmware is not built-in. It is a proprietary firmware uploaded into the file system that's why we need the USER_HELPER_FALLBACK to be able to load it once file system is available. BR Vincent On 10/19/2015 08:16 AM, Takashi Iwai wrote: > The commit [4fdbc678fe4d: drm: sti: add HQVDP plane] added the select > of CONFIG_FW_LOADER_USER_HELPER_FALLBACK by some unwritten reason. > But this config is known to be harmful, and is present only for > compatibility reason for an old exotic system that mandates udev > interaction which isn't supposed to be selected by a driver. > Let's remove it. > > Fixes: 4fdbc678fe4d ('drm: sti: add HQVDP plane') > Cc: <stable@vger.kernel.org> > Signed-off-by: Takashi Iwai <tiwai@suse.de> > --- > drivers/gpu/drm/sti/Kconfig | 1 - > 1 file changed, 1 deletion(-) > > diff --git a/drivers/gpu/drm/sti/Kconfig b/drivers/gpu/drm/sti/Kconfig > index fbccc105819b..a18159074b76 100644 > --- a/drivers/gpu/drm/sti/Kconfig > +++ b/drivers/gpu/drm/sti/Kconfig > @@ -6,7 +6,6 @@ config DRM_STI > select DRM_GEM_CMA_HELPER > select DRM_KMS_CMA_HELPER > select DRM_PANEL > - select FW_LOADER_USER_HELPER_FALLBACK > help > Choose this option to enable DRM on STM stiH41x chipset > >
On Thu, 29 Oct 2015 15:21:35 +0100, Vincent ABRIOU wrote: > > Hi Takashi, > > Removing FW_LOADER_USER_HELPER_FALLBACK leads to a failure in our HQVDP > firmware execution. > Indeed, our firmware is not built-in. It is a proprietary firmware > uploaded into the file system that's why we need the > USER_HELPER_FALLBACK to be able to load it once file system is available. It's not the option a driver can set. It's a global control option that is applied *all* drivers. If the fallback behavior is mandatory by a weird reason, you'd need to adjust in a different way; e.g. modify the firmware loader code to provide a new API for the driver to give a fallback (i.e. the another flip of request_firmware_direct()) and use that API explicitly. But I don't believe it's the case. Judging from your comment, it's your system setup that requires the user-space fallback. So don't make the fallback thing mandatory. thanks, Takashi > > BR > Vincent > > On 10/19/2015 08:16 AM, Takashi Iwai wrote: > > The commit [4fdbc678fe4d: drm: sti: add HQVDP plane] added the select > > of CONFIG_FW_LOADER_USER_HELPER_FALLBACK by some unwritten reason. > > But this config is known to be harmful, and is present only for > > compatibility reason for an old exotic system that mandates udev > > interaction which isn't supposed to be selected by a driver. > > Let's remove it. > > > > Fixes: 4fdbc678fe4d ('drm: sti: add HQVDP plane') > > Cc: <stable@vger.kernel.org> > > Signed-off-by: Takashi Iwai <tiwai@suse.de> > > --- > > drivers/gpu/drm/sti/Kconfig | 1 - > > 1 file changed, 1 deletion(-) > > > > diff --git a/drivers/gpu/drm/sti/Kconfig b/drivers/gpu/drm/sti/Kconfig > > index fbccc105819b..a18159074b76 100644 > > --- a/drivers/gpu/drm/sti/Kconfig > > +++ b/drivers/gpu/drm/sti/Kconfig > > @@ -6,7 +6,6 @@ config DRM_STI > > select DRM_GEM_CMA_HELPER > > select DRM_KMS_CMA_HELPER > > select DRM_PANEL > > - select FW_LOADER_USER_HELPER_FALLBACK > > help > > Choose this option to enable DRM on STM stiH41x chipset > > > >
On 29 October 2015 at 14:21, Vincent ABRIOU <vincent.abriou@st.com> wrote: > Hi Takashi, > > Removing FW_LOADER_USER_HELPER_FALLBACK leads to a failure in our HQVDP > firmware execution. > Indeed, our firmware is not built-in. It is a proprietary firmware > uploaded into the file system that's why we need the > USER_HELPER_FALLBACK to be able to load it once file system is available. > Hmm most other DRM drivers also require firmware. Whist some allow the firmware to be picked in initrd it's not a strict requirement. So I'm wondering how come there hasn't been (m)any reports, considering that neither one sets USER_HELPER_FALLBACK. Perhaps they also need it, or something in the sti module is done differently ? Just some food for thought. Regards, Emil
On Thu, 29 Oct 2015 15:37:51 +0100, Emil Velikov wrote: > > On 29 October 2015 at 14:21, Vincent ABRIOU <vincent.abriou@st.com> wrote: > > Hi Takashi, > > > > Removing FW_LOADER_USER_HELPER_FALLBACK leads to a failure in our HQVDP > > firmware execution. > > Indeed, our firmware is not built-in. It is a proprietary firmware > > uploaded into the file system that's why we need the > > USER_HELPER_FALLBACK to be able to load it once file system is available. > > > Hmm most other DRM drivers also require firmware. Whist some allow the > firmware to be picked in initrd it's not a strict requirement. > So I'm wondering how come there hasn't been (m)any reports, > considering that neither one sets USER_HELPER_FALLBACK. > > Perhaps they also need it, or something in the sti module is done > differently ? Just some food for thought. It's the option each user decides to set or not, depending on the deployed system. Most of PCs don't need them, and actually enabling this option causes troubles for them. On other embedded systems, this might be still needed. So, it's the system setup issue, and not the thing a driver needs to care. Imagine that your driver has "select EXT3_FS" because your system requires it; without that option, it won't boot, OMG! Is it the right thing? Obviously no. The same logic is applied to this case, too. Takashi
Hi Takashi, I made a rework to load the firmware at a different time, when file system is available. I am now able to remove FW_LOADER_USER_HELPER_FALLBACK. I will submit a new patch for this. Vincent On 10/29/2015 03:53 PM, Takashi Iwai wrote: > On Thu, 29 Oct 2015 15:37:51 +0100, > Emil Velikov wrote: >> >> On 29 October 2015 at 14:21, Vincent ABRIOU <vincent.abriou@st.com> wrote: >>> Hi Takashi, >>> >>> Removing FW_LOADER_USER_HELPER_FALLBACK leads to a failure in our HQVDP >>> firmware execution. >>> Indeed, our firmware is not built-in. It is a proprietary firmware >>> uploaded into the file system that's why we need the >>> USER_HELPER_FALLBACK to be able to load it once file system is available. >>> >> Hmm most other DRM drivers also require firmware. Whist some allow the >> firmware to be picked in initrd it's not a strict requirement. >> So I'm wondering how come there hasn't been (m)any reports, >> considering that neither one sets USER_HELPER_FALLBACK. >> >> Perhaps they also need it, or something in the sti module is done >> differently ? Just some food for thought. > > It's the option each user decides to set or not, depending on the > deployed system. Most of PCs don't need them, and actually enabling > this option causes troubles for them. On other embedded systems, this > might be still needed. So, it's the system setup issue, and not the > thing a driver needs to care. > > Imagine that your driver has "select EXT3_FS" because your system > requires it; without that option, it won't boot, OMG! > Is it the right thing? Obviously no. The same logic is applied to > this case, too. > > > Takashi > _______________________________________________ > dri-devel mailing list > dri-devel@lists.freedesktop.org > http://lists.freedesktop.org/mailman/listinfo/dri-devel >
diff --git a/drivers/gpu/drm/sti/Kconfig b/drivers/gpu/drm/sti/Kconfig index fbccc105819b..a18159074b76 100644 --- a/drivers/gpu/drm/sti/Kconfig +++ b/drivers/gpu/drm/sti/Kconfig @@ -6,7 +6,6 @@ config DRM_STI select DRM_GEM_CMA_HELPER select DRM_KMS_CMA_HELPER select DRM_PANEL - select FW_LOADER_USER_HELPER_FALLBACK help Choose this option to enable DRM on STM stiH41x chipset
The commit [4fdbc678fe4d: drm: sti: add HQVDP plane] added the select of CONFIG_FW_LOADER_USER_HELPER_FALLBACK by some unwritten reason. But this config is known to be harmful, and is present only for compatibility reason for an old exotic system that mandates udev interaction which isn't supposed to be selected by a driver. Let's remove it. Fixes: 4fdbc678fe4d ('drm: sti: add HQVDP plane') Cc: <stable@vger.kernel.org> Signed-off-by: Takashi Iwai <tiwai@suse.de> --- drivers/gpu/drm/sti/Kconfig | 1 - 1 file changed, 1 deletion(-)