- 25 Feb, 2014 8 commits
-
-
Mark Hounschell authored
There is a lot of cleanup work to do on these digi drivers and merging as much as is possible will make it easier. I also notice that many merged drivers are single source and header. Signed-off-by:
Mark Hounschell <markh@compro.net> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Mark Hounschell authored
There is a lot of cleanup work to do on these digi drivers and merging as much as is possible will make it easier. I also notice that many merged drivers are single source and header. Signed-off-by:
-
Mark Hounschell authored
There is a lot of cleanup work to do on these digi drivers and merging as much as is possible will make it easier. I also notice that many merged drivers are single source and header. Signed-off-by:
-
Mark Hounschell authored
This patch removes userland source code files downld.c and dgap_downld.h Signed-off-by:
-
Mark Hounschell authored
This patch removes all the original CVS tags because they are in my way Signed-off-by:
-
John Stultz authored
Add a more clear explanation of the option in the prompt, and make the config depend on ANDROID_BINDER_IPC being selected. Also sets the default to y, which matches AOSP. Cc: Colin Cross <ccross@android.com> Cc: Arve Hjønnevåg <arve@android.com> Cc: Serban Constantinescu <serban.constantinescu@arm.com> Cc: Android Kernel Team <kernel-team@android.com> Signed-off-by:
John Stultz <john.stultz@linaro.org> Signed-off-by:
-
Arve Hjønnevåg authored
For 64bit systems we want to use the same binder interface for 32bit and 64bit processes. Thus the size and the layout of the structures passed between the kernel and the userspace has to be the same for both 32 and 64bit processes. This change replaces all the uses of void* and size_t with binder_uintptr_t and binder_size_t. These are then typedefed to specific sizes depending on the use of the interface, as follows: * __u32 - on legacy 32bit only userspace * __u64 - on mixed 32/64bit userspace where all processes use the same interface. This change also increments the BINDER_CURRENT_PROTOCOL_VERSION to 8 and hooks the compat_ioctl entry for the mixed 32/64bit Android userspace. This patch also provides a CONFIG_ANDROID_BINDER_IPC_32BIT option for compatability, which if set which enables the old protocol, setting BINDER_CURRENT_PROTOCOL_VERSION to 7, on 32 bit systems. Please note that all 64bit kernels will use the 64bit Binder ABI. Cc: Colin Cross <ccross@android.com> Cc: Arve Hjønnevåg <arve@android.com> Cc: Serban Constantinescu <serban.constantinescu@arm.com> Cc: Android Kernel Team <kernel-team@android.com> Signed-off-by:Arve Hjønnevåg <arve@android.com> [jstultz: Merged with upstream type changes. Various whitespace fixes and longer Kconfig description for checkpatch. Included improved commit message from Serban (with a few tweaks).] Signed-off-by:
-
Serban Constantinescu authored
BC_REQUEST_DEATH_NOTIFICATION and BC_CLEAR_DEATH_NOTIFICATION were defined with the wrong structure that did not match the code. Since a binder pointer and handle are the same size on 32 bit systems, this change does not affect them. The two commands claimed they were using struct binder_ptr_cookie but they are using a 32bit handle and a pointer. The main purpose of this patch is to add the binder_handle_cookie struct so the service manager does not have to define its own version (libbinder writes one field at a time so it does not use the struct). On 32bit systems the payload size is the same as the size of struct binder_ptr_cookie. On 64bit systems, the size does differ, and the ioctl number does change. However, there are no known 64bit users of this interface, and any 64bit systems will need the following patch to run 32 bit processes anyway, so it is not expected that anyone will ship a 64bit system without this change, so this change should not affect any existing systems. Cc: Colin Cross <ccross@android.com> Cc: Arve Hjønnevåg <arve@android.com> Cc: Serban Constantinescu <serban.constantinescu@arm.com> Cc: Android Kernel Team <kernel-team@android.com> Signed-off-by:
Serban Constantinescu <serban.constantinescu@arm.com> [jstultz: Few 80+ col fixes for checkpatch, improved commit message with help from Serban, and included rational from Arve's email] Signed-off-by:
-
- 24 Feb, 2014 32 commits
-
-
git://ftp.arm.linux.org.uk/~rmk/linux-armGreg Kroah-Hartman authored
Russell writes: These changes, which convert imx-drm to use the recently merged component infrastructure, have been reviewed and acked by Philipp Zabel, Shawn Guo and Fabio Estevam, and are now deemed to be ready.
-
Greg Kroah-Hartman authored
We want those fixes here as well.
-
Russell King authored
Add hotplug support. We have to make the interrupt handler threaded so we can call drm_helper_hpd_irq_event(). Keeping in mind that we will want to share the interrupt with other HDMI interface drivers (eg, audio and CEC) put the groundwork in now for that, rather than just using IRQF_ONESHOT. Also, we must not call drm_helper_hpd_irq_event() until we have fully setup the connector; keep the interrupt(s) muted until after that point. Acked-by:
Philipp Zabel <p.zabel@pengutronix.de> Acked-by:
Shawn Guo <shawn.guo@linaro.org> Reviewed-by:
Fabio Estevam <fabio.estevam@freescale.com> Signed-off-by:
Russell King <rmk+kernel@arm.linux.org.uk>
-
Russell King authored
Add core imx-drm support for hotplug connector support. We need to setup the poll helper after we've setup the connectors; the helper scans the connectors to determine their capabilities. Acked-by:
-
Russell King authored
Various cleanups are possible after the previous round of changes; these have no real functional bearing other than tidying up the code. Acked-by:
-
Russell King authored
It is now no longer necessary to keep this structure around; we can allocate it upon DRM driver load and destroy it thereafter without affecting the other components now. Acked-by:
-
Russell King authored
This mutex doesn't protect anything anymore; get rid of it. Acked-by:
-
Russell King authored
Since we're now operating like a conventional DRM driver, doing all the initialisation within the driver's ->load callback, we don't need to mess around with the mode groups - we can rely on the one in the DRM platform code. Acked-by:
-
Russell King authored
The core imx_drm_connector and imx_drm_encoder code is no longer required - the connectors and encoders are all using the component support, so we can remove this. Acked-by:
-
Russell King authored
Now that our bind function is only ever called during the main DRM driver ->load callback, we don't need to have the imx_drm_connector or imx_drm_encoder abstractions anymore. So let's get rid of it, and move the DRM connector and encoder setup into the connector support files. Acked-by:
-
Russell King authored
Provide two helper functions to assist with cleaning up imx-drm connectors and encoders. Acked-by:
-
Russell King authored
Provide a helper function to parse possible crtcs before the encoder is registered. The crtc mask is derived from the position of the CRTCs registered in the drm_device. Acked-by:
-
Russell King authored
The component helper provides us the drm_device which is being registered. Rather than having to reference a global in imx-drm-core, use this to get the imxdrm device, and also use it to register the CRTC against. This means we never have CRTCs/encoders/connectors without the drivers private data being accessible. Remove the module owner field as well; this provides no protection against the device being unbound. Acked-by:
-
Russell King authored
imx-fb.c doesn't need to be separate from imx-drm-core.c - all it is doing is setting up the minimum and maximum sizes of the scanout buffers, and setting up the mode_config function pointers. Move the contents into imx-drm-core.c and kill this file. Acked-by:
-
Russell King authored
Now that we know when the components of the imx-drm subsystem will be initialised, we can move the fbdev helper initialisation and teardown into imx-drm-core. This gives us the required ordering that DRM wants in both driver load and unload methods. We can also stop exporting the imx_drm_device_get() and imx_drm_device_put() methods; nothing but the fbdev helper was making use of these. Acked-by:
-
Russell King authored
Delay publishing sysfs connector entries until all components have initialised. This reduces the probability of generating false hotplug events when we're uncertain whether the driver can fully initialise. This also pulls that code out of the individual imx-drm connector drivers. Acked-by:
-
Russell King authored
Use the componentised device support for imx-drm. This requires all the sub-components and the master device to register with the component device support. Acked-by:
-
Russell King authored
The encoder format passed into imx_drm_crtc_panel_format*() is the encoder format used for DRM in most cases; the HDMI encoder sets this to none, but this is incorrect, it should be TMDS. Since this is the case, we can pass the drm_encoder structure directly into this function and use the supplied fields there to configure the CRTC. Acked-by:
-
Russell King authored
Provide a common connector mode validation function, which can be used to limit the available modes according to other components in the system. Acked-by:
-
Russell King authored
The DRM core indexes vblank by number, so there's little point maintaining a list, and have to scan the list to find the appropriate structure. Instead, use an array of pointers to the CRTCs. Acked-by:
-
Russell King authored
Address the following issues: - imx_drm_encoder_get_mux_id() searches the CRTC list for the matching CRTC, and returns the position within this list as the MUX programming value for encoders. This is sub-optimal for two reasons: 1. It relies upon the CRTC list not changing during the lifetime of the driver. 2. It is dependent on the initialisation order of the CRTCs. We address (1) in this patch, leaving (2) until a better solution can be found, as (2) requires larger changes. - imx_drm_encoder is unused. Instead, pass the drm_encoder which is slightly more useful; all callers pass encoder->crtc as the required crtc, so move this inside the function. Acked-by: -
Russell King authored
Acked-by:
-
Russell King authored
Extracted from another patch by Fabio Estevam, this adds the DT configuration for HDMI output on the IMX6 SoCs Acked-by:
-
Russell King authored
There's no point in using the clk API for this; we end up having to violate the layering this provides. Acked-by:
-
Russell King authored
This is nonsense; clk_round_rate() is just clk_set_rate() without the side effect of changing the hardware. Acked-by:
-
Russell King authored
The DI clock selection was very rudimentary: it would statically use either the IPU internal clock or the DI external clock depending on which "encoder" was being used. In the case of HDMI, it would always use the IPU clock. Moreover, using the IPU clock resulted in fractional divisors, which are achieved by skipping clock pulses. This can result in the HDMI PHY PLL being frequency modulated, and the attached device is then unable to properly lock on to the TMDS clock. We need at least 1% accurate and stable clocks for HDMI. Arrange for the DI clock to be sourced from the IPU internal clock if it can satisfy our requirements, otherwise switch to the DI external clock and try and set the external clock to our desired pixel clock rate. Acked-by:
-
Russell King authored
Many of the variables for the audio clock regenerator (CTS/N) were not initialised in any way. The pixel rate which was being used also wasn't being adjusted at all when the display mode is modified. Get rid of the seaprate 'pixel_clk_rate', and use the stored pixel clock rate instead. Pass this desired pixel clock rate into hdmi_set_clk_regenerator(). Collapse down hdmi_init_clk_regenerator() since it is a copy of hdmi_set_clk_regenerator(), and pass a default pixel clock rate. Acked-by:
-
Russell King authored
Acked-by:
-
Russell King authored
There are a load of read-modify-write patterns to change bitfields in various registers in this driver; provide a helper to perform this manipulation. Acked-by:
-
Russell King authored
Rather than manually writing each register sequentially, we can use a loop to reduce the amount of code. Acked-by:
-
Russell King authored
Rather than having large if() and switch() statements, provide a table to look up the register settings for various clock rates. Acked-by:
-
Linus Torvalds authored
-
