| Message ID | 1351613409-21186-13-git-send-email-tomi.valkeinen@ti.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tuesday 30 October 2012 09:40 PM, Tomi Valkeinen wrote: > We currently get the decision whether to use PRCM or DSI PLL clock for > DPI from the board file. This is not a good way to handle it, and it > won't work with device tree. > > This patch changes DPI to always use DSI PLL if it's available. > > Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com> > --- > drivers/video/omap2/dss/dpi.c | 64 ++++++++++++++++++++++++----------------- > 1 file changed, 37 insertions(+), 27 deletions(-) > > diff --git a/drivers/video/omap2/dss/dpi.c b/drivers/video/omap2/dss/dpi.c > index 267caf0..32e7dd5 100644 > --- a/drivers/video/omap2/dss/dpi.c > +++ b/drivers/video/omap2/dss/dpi.c > @@ -49,28 +49,30 @@ static struct { > struct omap_dss_output output; > } dpi; > > -static struct platform_device *dpi_get_dsidev(enum omap_dss_clk_source clk) > +static struct platform_device *dpi_get_dsidev(enum omap_channel channel) > { > - int dsi_module; > - > - dsi_module = clk == OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DISPC ? 0 : 1; > - > - return dsi_get_dsidev_from_id(dsi_module); > + switch (channel) { > + case OMAP_DSS_CHANNEL_LCD: > + return dsi_get_dsidev_from_id(0); > + case OMAP_DSS_CHANNEL_LCD2: > + return dsi_get_dsidev_from_id(1); > + default: > + return NULL; > + } > } > > -static bool dpi_use_dsi_pll(struct omap_dss_device *dssdev) > +static enum omap_dss_clk_source dpi_get_alt_clk_src(enum omap_channel channel) > { > - if (dssdev->clocks.dispc.dispc_fclk_src == > - OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DISPC || > - dssdev->clocks.dispc.dispc_fclk_src == > - OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DISPC || > - dssdev->clocks.dispc.channel.lcd_clk_src == > - OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DISPC || > - dssdev->clocks.dispc.channel.lcd_clk_src == > - OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DISPC) > - return true; > - else > - return false; > + switch (channel) { > + case OMAP_DSS_CHANNEL_LCD: > + return OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DISPC; > + case OMAP_DSS_CHANNEL_LCD2: > + return OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DISPC; > + default: > + /* this shouldn't happen */ > + WARN_ON(1); > + return OMAP_DSS_CLK_SRC_FCK; > + } > } > > static int dpi_set_dsi_clk(struct omap_dss_device *dssdev, > @@ -92,7 +94,7 @@ static int dpi_set_dsi_clk(struct omap_dss_device *dssdev, > return r; > > dss_select_lcd_clk_source(mgr->id, > - dssdev->clocks.dispc.channel.lcd_clk_src); > + dpi_get_alt_clk_src(mgr->id)); > > dpi.mgr_config.clock_info = dispc_cinfo; > > @@ -385,6 +387,8 @@ static int __init dpi_verify_dsi_pll(struct platform_device *dsidev) > > static int __init dpi_init_display(struct omap_dss_device *dssdev) > { > + struct platform_device *dsidev; > + > DSSDBG("init_display\n"); > > if (dss_has_feature(FEAT_DPI_USES_VDDS_DSI) && > @@ -401,17 +405,23 @@ static int __init dpi_init_display(struct omap_dss_device *dssdev) > dpi.vdds_dsi_reg = vdds_dsi; > } > > - if (dpi_use_dsi_pll(dssdev)) { > - enum omap_dss_clk_source dispc_fclk_src = > - dssdev->clocks.dispc.dispc_fclk_src; > - dpi.dsidev = dpi_get_dsidev(dispc_fclk_src); > + /* > + * XXX We shouldn't need dssdev->channel for this. The dsi pll clock > + * source for DPI is SoC integration detail, not something that should > + * be configured in the dssdev > + */ It is SoC integration detail, but it's flexible, it depends on which manager is connected to DPI output. If it's connected to LCD1, the source can be DSI1 PLL, if it's LCD2, it's source can be DSI2 PLL, if it's connected to TV manager, it's source "has to be" HDMI PLL. And these connections vary based on which OMAP revision we are on. We can only be certain on OMAP3 that the source for DPI pixel clock can be either PRCM or DSI PLL. At the point of probe, we really don't know which manager is the DPI output connected to. Hence, we sort of use dssdev->channel to make a guess what manager we connect to in the future. The right approach would be to figure this out at the time of enable, where we know which manager the DPI output is connected to. We could probably move the verification there too. Archit > + dsidev = dpi_get_dsidev(dssdev->channel); > > - if (dpi_verify_dsi_pll(dpi.dsidev)) { > - dpi.dsidev = NULL; > - DSSWARN("DSI PLL not operational\n"); > - } > + if (dpi_verify_dsi_pll(dsidev)) { > + dsidev = NULL; > + DSSWARN("DSI PLL not operational\n"); > } > > + if (dsidev) > + DSSDBG("using DSI PLL for DPI clock\n"); > + > + dpi.dsidev = dsidev; > + > return 0; > } > > -- To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2012-10-31 09:26, Archit Taneja wrote: > On Tuesday 30 October 2012 09:40 PM, Tomi Valkeinen wrote: >> - if (dpi_use_dsi_pll(dssdev)) { >> - enum omap_dss_clk_source dispc_fclk_src = >> - dssdev->clocks.dispc.dispc_fclk_src; >> - dpi.dsidev = dpi_get_dsidev(dispc_fclk_src); >> + /* >> + * XXX We shouldn't need dssdev->channel for this. The dsi pll clock >> + * source for DPI is SoC integration detail, not something that >> should >> + * be configured in the dssdev >> + */ > > It is SoC integration detail, but it's flexible, it depends on which > manager is connected to DPI output. If it's connected to LCD1, the Hmm, yes, the comment is a bit misleading. The DSI PLL is not used for DPI, but for DISPC's LCD output. And DPI uses one of those LCD outputs. > source can be DSI1 PLL, if it's LCD2, it's source can be DSI2 PLL, if > it's connected to TV manager, it's source "has to be" HDMI PLL. And > these connections vary based on which OMAP revision we are on. We can > only be certain on OMAP3 that the source for DPI pixel clock can be > either PRCM or DSI PLL. On OMAP2 we can be certain the clock is PRCM =). > At the point of probe, we really don't know which manager is the DPI > output connected to. Hence, we sort of use dssdev->channel to make a > guess what manager we connect to in the future. Yep. My point was mainly that dssdev needs to go away, and we should somehow decide the used channel internally. > The right approach would be to figure this out at the time of enable, > where we know which manager the DPI output is connected to. We could > probably move the verification there too. Who chooses which manager to use for DPI? I'm not sure... I would really like to manage the basic setup, acquiring the resources, etc. at probe time, and enable would only enable the display. That means that we should somehow get a manager for DPI at probe time, which may be a bit difficult, as we don't know what other displays there will be. So if, say, DPI can use LCD1 or LCD2, but DSI can only use LCD2, and at DPI's probe (presuming it's before DSI) we pick LCD2, DSI won't work. Anyway, while I'm not sure how to solve the above problem, I think we should improve our init a bit. For DPI there are the following steps done, in order: - DPI device added - DPI driver probe - DPI panel device added - DPI panel driver probe We currently add the panel device in DPI driver's probe, and figure out the DSI PLL at the same time. I think that should happen only when the panel driver probe happens. The panel driver should call something like dpi_get_output() or whatever, which acquires the DPI bus for the panel driver, and this would probably also choose the manager. Tomi
On Friday 02 November 2012 03:38 PM, Tomi Valkeinen wrote: > On 2012-10-31 09:26, Archit Taneja wrote: >> On Tuesday 30 October 2012 09:40 PM, Tomi Valkeinen wrote: > >>> - if (dpi_use_dsi_pll(dssdev)) { >>> - enum omap_dss_clk_source dispc_fclk_src = >>> - dssdev->clocks.dispc.dispc_fclk_src; >>> - dpi.dsidev = dpi_get_dsidev(dispc_fclk_src); >>> + /* >>> + * XXX We shouldn't need dssdev->channel for this. The dsi pll clock >>> + * source for DPI is SoC integration detail, not something that >>> should >>> + * be configured in the dssdev >>> + */ >> >> It is SoC integration detail, but it's flexible, it depends on which >> manager is connected to DPI output. If it's connected to LCD1, the > > Hmm, yes, the comment is a bit misleading. The DSI PLL is not used for > DPI, but for DISPC's LCD output. And DPI uses one of those LCD outputs. > >> source can be DSI1 PLL, if it's LCD2, it's source can be DSI2 PLL, if >> it's connected to TV manager, it's source "has to be" HDMI PLL. And >> these connections vary based on which OMAP revision we are on. We can >> only be certain on OMAP3 that the source for DPI pixel clock can be >> either PRCM or DSI PLL. > > On OMAP2 we can be certain the clock is PRCM =). > >> At the point of probe, we really don't know which manager is the DPI >> output connected to. Hence, we sort of use dssdev->channel to make a >> guess what manager we connect to in the future. > > Yep. My point was mainly that dssdev needs to go away, and we should > somehow decide the used channel internally. > >> The right approach would be to figure this out at the time of enable, >> where we know which manager the DPI output is connected to. We could >> probably move the verification there too. > > Who chooses which manager to use for DPI? If you are asking in terms of HW. The value in the DSS_CTRL bitfield decides which manager to use. If you meant who/how should we choose this in software, then I don't know either. > > I'm not sure... I would really like to manage the basic setup, acquiring > the resources, etc. at probe time, and enable would only enable the display. One thing we could do is to grab all the possible resources that DPI can use for its pixel clock, and when it's enable time, see what all options it has. So, for example, for DPI on omap4, we could try to grab and verify all DSI PLL, HDMI PLL and PRCM. And then later choose the most appropriate one while enabling. > > That means that we should somehow get a manager for DPI at probe time, > which may be a bit difficult, as we don't know what other displays there > will be. So if, say, DPI can use LCD1 or LCD2, but DSI can only use > LCD2, and at DPI's probe (presuming it's before DSI) we pick LCD2, DSI > won't work. Yes, it's not easy to know this at probe time. We could try to allocate resources at the time mgr->set_output() is called. We could have an output specific op. dss_mgr_set_output() could look like: dss_mgr_set_output(mgr, output) { /* Do the older stuff */ ... ... output->get_and_config_resources(output); } In dpi.c: dss_dpi_get_and_config_resources(output) { switch (output->manager->id) { case LCD: Get DSI 1 PLL; case LCD2: Get DSI2 PLL; case TV: Get HDMI PLL; } /* * Also set the DSS_CTRL bits here to tell which manager * we need to connect to */ dss_select_dpi_manager(output->manager->id); } omapdss_output_ops dpi_ops = { .get_and_config_resources = dss_dpi_get_and_config_resources, ... }; However, even though this approach might be correct in the sense that we confiugre dpi when we know what manager we are connected to, I have to say that its not nice at all. Especially because setting manager output links is very omapdss-apply-ish. But I feel we would need to do something similar in omapdrm too. > > Anyway, while I'm not sure how to solve the above problem, I think we > should improve our init a bit. For DPI there are the following steps > done, in order: > > - DPI device added > - DPI driver probe > - DPI panel device added > - DPI panel driver probe > > We currently add the panel device in DPI driver's probe, and figure out > the DSI PLL at the same time. I think that should happen only when the > panel driver probe happens. The panel driver should call something like > dpi_get_output() or whatever, which acquires the DPI bus for the panel > driver, and this would probably also choose the manager. Hmm, that makes sense. Anyway, I don't think it's really bad if we refer to dssdev->channel for now. I think we could have a clearer picture of this when we understand how omapdrm sets the links between its entities and how CPF would link the output-panel side of things. Archit -- To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2012-11-02 12:44, Archit Taneja wrote: > Hmm, that makes sense. Anyway, I don't think it's really bad if we refer > to dssdev->channel for now. It is, because dssdev->channel doesn't exist with DT. With DT we either need to figure out the channel in omapdss at runtime, or add a property to the DT data telling the channel. And adding such a property is not correct, as DT should be about describing the HW. Tomi
On Friday 02 November 2012 04:19 PM, Tomi Valkeinen wrote: > On 2012-11-02 12:44, Archit Taneja wrote: > >> Hmm, that makes sense. Anyway, I don't think it's really bad if we refer >> to dssdev->channel for now. > > It is, because dssdev->channel doesn't exist with DT. > > With DT we either need to figure out the channel in omapdss at runtime, > or add a property to the DT data telling the channel. And adding such a > property is not correct, as DT should be about describing the HW. Ok. I don't totally agree with your idea of figuring out the manager in panel the panel's probe. If it's done in the panel driver's probe itself, then by this point of time we have already set mgr->output->device links. If omapdss only does this stuff, then omapfb/omapdrm have just the job of connecting the overlays to the manager. Do you think that's okay? Archit -- To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2012-11-02 13:09, Archit Taneja wrote: > On Friday 02 November 2012 04:19 PM, Tomi Valkeinen wrote: >> On 2012-11-02 12:44, Archit Taneja wrote: >> >>> Hmm, that makes sense. Anyway, I don't think it's really bad if we refer >>> to dssdev->channel for now. >> >> It is, because dssdev->channel doesn't exist with DT. >> >> With DT we either need to figure out the channel in omapdss at runtime, >> or add a property to the DT data telling the channel. And adding such a >> property is not correct, as DT should be about describing the HW. > > Ok. > > I don't totally agree with your idea of figuring out the manager in > panel the panel's probe. If it's done in the panel driver's probe > itself, then by this point of time we have already set > mgr->output->device links. If omapdss only does this stuff, then Hmm, I'm not sure I understand what's your point above? If figuring out the mgr is done in panel's probe, the mgr->output link is not yet made before that time. > omapfb/omapdrm have just the job of connecting the overlays to the > manager. Do you think that's okay? Yes, that's how I think it should be. I don't see why omapfb/omapdrm should care about which manager is being used for the output, it doesn't really matter as long there is one and it works. Then again, I don't have anything against omapfb/omapdrm choosing the manager, but I don't see how they would have any better idea of which manager to use than omapdss. But as doing the connections at probe time is a bit problematic, perhaps we should have a new step in this whole sequence. Something like "connect" or whatever, which would lock the required blocks in the whole pipeline, and acquire the required resources that couldn't be gotten at probe time. But even then, choosing the manager is not easy, as whoever chooses the manager needs to observe all the possible displays used at the same time... Tomi
On Friday 02 November 2012 04:58 PM, Tomi Valkeinen wrote: > On 2012-11-02 13:09, Archit Taneja wrote: >> On Friday 02 November 2012 04:19 PM, Tomi Valkeinen wrote: >>> On 2012-11-02 12:44, Archit Taneja wrote: >>> >>>> Hmm, that makes sense. Anyway, I don't think it's really bad if we refer >>>> to dssdev->channel for now. >>> >>> It is, because dssdev->channel doesn't exist with DT. >>> >>> With DT we either need to figure out the channel in omapdss at runtime, >>> or add a property to the DT data telling the channel. And adding such a >>> property is not correct, as DT should be about describing the HW. >> >> Ok. >> >> I don't totally agree with your idea of figuring out the manager in >> panel the panel's probe. If it's done in the panel driver's probe >> itself, then by this point of time we have already set >> mgr->output->device links. If omapdss only does this stuff, then > > Hmm, I'm not sure I understand what's your point above? If figuring out > the mgr is done in panel's probe, the mgr->output link is not yet made > before that time. My point is that we are trying to find a manager at panel's probe itself. It think that's what we do now. But one of your recent patch moves that to omapfb. > >> omapfb/omapdrm have just the job of connecting the overlays to the >> manager. Do you think that's okay? > > Yes, that's how I think it should be. I don't see why omapfb/omapdrm > should care about which manager is being used for the output, it doesn't > really matter as long there is one and it works. > > Then again, I don't have anything against omapfb/omapdrm choosing the > manager, but I don't see how they would have any better idea of which > manager to use than omapdss. > > But as doing the connections at probe time is a bit problematic, perhaps > we should have a new step in this whole sequence. Something like > "connect" or whatever, which would lock the required blocks in the whole > pipeline, and acquire the required resources that couldn't be gotten at > probe time. > > But even then, choosing the manager is not easy, as whoever chooses the > manager needs to observe all the possible displays used at the same time... Right. I was wondering if omapfb/omapdrm could understand the 'all possible displays information' better compared to a panel's probe. Even omapdrm/omafb can't be perfect because we could insert a panel driver module at any time, and omapfb/omapdrm may miss that out. Archit -- To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2012-11-02 13:56, Archit Taneja wrote: > On Friday 02 November 2012 04:58 PM, Tomi Valkeinen wrote: >> On 2012-11-02 13:09, Archit Taneja wrote: >>> On Friday 02 November 2012 04:19 PM, Tomi Valkeinen wrote: >>>> On 2012-11-02 12:44, Archit Taneja wrote: >>>> >>>>> Hmm, that makes sense. Anyway, I don't think it's really bad if we >>>>> refer >>>>> to dssdev->channel for now. >>>> >>>> It is, because dssdev->channel doesn't exist with DT. >>>> >>>> With DT we either need to figure out the channel in omapdss at runtime, >>>> or add a property to the DT data telling the channel. And adding such a >>>> property is not correct, as DT should be about describing the HW. >>> >>> Ok. >>> >>> I don't totally agree with your idea of figuring out the manager in >>> panel the panel's probe. If it's done in the panel driver's probe >>> itself, then by this point of time we have already set >>> mgr->output->device links. If omapdss only does this stuff, then >> >> Hmm, I'm not sure I understand what's your point above? If figuring out >> the mgr is done in panel's probe, the mgr->output link is not yet made >> before that time. > > My point is that we are trying to find a manager at panel's probe > itself. It think that's what we do now. But one of your recent patch > moves that to omapfb. Ah. Yes, that's true. >>> omapfb/omapdrm have just the job of connecting the overlays to the >>> manager. Do you think that's okay? >> >> Yes, that's how I think it should be. I don't see why omapfb/omapdrm >> should care about which manager is being used for the output, it doesn't >> really matter as long there is one and it works. >> >> Then again, I don't have anything against omapfb/omapdrm choosing the >> manager, but I don't see how they would have any better idea of which >> manager to use than omapdss. >> >> But as doing the connections at probe time is a bit problematic, perhaps >> we should have a new step in this whole sequence. Something like >> "connect" or whatever, which would lock the required blocks in the whole >> pipeline, and acquire the required resources that couldn't be gotten at >> probe time. >> >> But even then, choosing the manager is not easy, as whoever chooses the >> manager needs to observe all the possible displays used at the same >> time... > > Right. I was wondering if omapfb/omapdrm could understand the 'all > possible displays information' better compared to a panel's probe. > > Even omapdrm/omafb can't be perfect because we could insert a panel > driver module at any time, and omapfb/omapdrm may miss that out. True, omapdrm/fb may have a better idea. It's still unclear though. Currently we have quite strict order in the sequence the modules need to be loaded, which is quite bad and causes issues. We should make things more dynamic, so that the initialization of the drivers could happen more freely. But that creates more problems: when booting up, omapfb starts. But omapfb can't know if all the panel drivers have already been loaded. omapfb may see that DVI is the default display, but what should it do if DVI doesn't have a driver yet? It could wait, but perhaps the driver for DVI will never even be loaded. Tomi
On 11/05/2012 02:55 AM, Tomi Valkeinen wrote: >>> But even then, choosing the manager is not easy, as whoever chooses the >>> >>manager needs to observe all the possible displays used at the same >>> >>time... >> > >> >Right. I was wondering if omapfb/omapdrm could understand the 'all >> >possible displays information' better compared to a panel's probe. >> > >> >Even omapdrm/omafb can't be perfect because we could insert a panel >> >driver module at any time, and omapfb/omapdrm may miss that out. > True, omapdrm/fb may have a better idea. It's still unclear though. > Currently we have quite strict order in the sequence the modules need to > be loaded, which is quite bad and causes issues. We should make things > more dynamic, so that the initialization of the drivers could happen > more freely. > > But that creates more problems: when booting up, omapfb starts. But > omapfb can't know if all the panel drivers have already been loaded. > omapfb may see that DVI is the default display, but what should it do if > DVI doesn't have a driver yet? It could wait, but perhaps the driver for > DVI will never even be loaded. The encoder which is connected to the crtc (manager) is picked by combination of encoder->possible_crtcs bitmask and connector->best_encoder(). We could keep things limited so that the association of crtc to encoder (manager to output, roughly) never changes, but this isn't really the right thing to do. It is better that the dssdev not rely on knowing the manager it is attached to at probe time, but instead grab resources more dynamically. Also, at the moment we don't really have any notification to userspace about new encoders/connectors showing up (or conversely, being removed). Only about existing connectors being plugged/unplugged. The closest analogy is perhaps the USB display devices, but even there it is only the entire drm device that is plugged/unplugged. And TBH I don't really see the point in supporting panel drivers being dynamically loaded. It isn't like someone is dynamically soldering on a new display connector to some board that is running. I think omapfb or omapdrm probe should trigger registering the compiled-in panel drivers, so that it can be sure that the dssdev's pop up before it goes and creates drm connector objects. Currently we have to hack around this in omapdrm with late_initcall() to ensure the panel drivers are probed first, but that is an ugly hack that I'd like to get rid of. BR, -R -- To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2012-11-05 16:21, Rob Clark wrote: > On 11/05/2012 02:55 AM, Tomi Valkeinen wrote: >>>> But even then, choosing the manager is not easy, as whoever chooses the >>>> >>manager needs to observe all the possible displays used at the same >>>> >>time... >>> > >>> >Right. I was wondering if omapfb/omapdrm could understand the 'all >>> >possible displays information' better compared to a panel's probe. >>> > >>> >Even omapdrm/omafb can't be perfect because we could insert a panel >>> >driver module at any time, and omapfb/omapdrm may miss that out. >> True, omapdrm/fb may have a better idea. It's still unclear though. >> Currently we have quite strict order in the sequence the modules need to >> be loaded, which is quite bad and causes issues. We should make things >> more dynamic, so that the initialization of the drivers could happen >> more freely. >> >> But that creates more problems: when booting up, omapfb starts. But >> omapfb can't know if all the panel drivers have already been loaded. >> omapfb may see that DVI is the default display, but what should it do if >> DVI doesn't have a driver yet? It could wait, but perhaps the driver for >> DVI will never even be loaded. > > The encoder which is connected to the crtc (manager) is picked by > combination of encoder->possible_crtcs bitmask and > connector->best_encoder(). We could keep things limited so that the > association of crtc to encoder (manager to output, roughly) never > changes, but this isn't really the right thing to do. It is better that > the dssdev not rely on knowing the manager it is attached to at probe > time, but instead grab resources more dynamically. > > Also, at the moment we don't really have any notification to userspace > about new encoders/connectors showing up (or conversely, being > removed). Only about existing connectors being plugged/unplugged. The > closest analogy is perhaps the USB display devices, but even there it is > only the entire drm device that is plugged/unplugged. And TBH I don't > really see the point in supporting panel drivers being dynamically > loaded. It isn't like someone is dynamically soldering on a new display > connector to some board that is running. I think omapfb or omapdrm > probe should trigger registering the compiled-in panel drivers, so that > it can be sure that the dssdev's pop up before it goes and creates drm > connector objects. Currently we have to hack around this in omapdrm > with late_initcall() to ensure the panel drivers are probed first, but > that is an ugly hack that I'd like to get rid of. We have panel devices and panel drivers, each of which can appear at any time. Both are needed for the panel probe to happen. If we don't support device hotplugging (dynamic creation of devices), we need to use late_initcall for omapfb/drm. At least I don't see any other option. You say that omapdrm should trigger registering of the drivers. How would that work? Do you mean that the panel drivers would register themselves to some common list, and omapdrm would go through this list when drm is loaded, calling probe for the items in the list? I guess that's doable, but... It's not how kernel drivers are supposed to work, and so doesn't sound very clean approach to me. I think we should support proper hotplugging of the panels. This would fix the problem about init order, but it would also give us device hotplug support. Obviously nobody is going to solder panel to a running board, but I don't see any reason why panels, or, more likely, panels on an add-on boards (like the capes being discussed in omap ml) would not be hotpluggable using whatever connector is used on the particular use case. And even if we don't support removing of the devices, things like the add-on capes could cause the panel on the cape to be identified at some late time (the panel is not described in the board file or DT data, but found at runtime depending on the ID of the cape). This would add another step to the init sequence that should be just right, if we don't support hotplug. Yes, I know it's not simple =). And I'm fine with simpler approach for the time being, but I'd like full hotplug to be the future goal. At least the common panel framework should not create restrictions about this, even if drm wouldn't allow device hotplug. Tomi -- To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Nov 6, 2012 at 7:41 AM, Tomi Valkeinen <tomba@iki.fi> wrote: > On 2012-11-05 16:21, Rob Clark wrote: >> On 11/05/2012 02:55 AM, Tomi Valkeinen wrote: >>>>> But even then, choosing the manager is not easy, as whoever chooses the >>>>> >>manager needs to observe all the possible displays used at the same >>>>> >>time... >>>> > >>>> >Right. I was wondering if omapfb/omapdrm could understand the 'all >>>> >possible displays information' better compared to a panel's probe. >>>> > >>>> >Even omapdrm/omafb can't be perfect because we could insert a panel >>>> >driver module at any time, and omapfb/omapdrm may miss that out. >>> True, omapdrm/fb may have a better idea. It's still unclear though. >>> Currently we have quite strict order in the sequence the modules need to >>> be loaded, which is quite bad and causes issues. We should make things >>> more dynamic, so that the initialization of the drivers could happen >>> more freely. >>> >>> But that creates more problems: when booting up, omapfb starts. But >>> omapfb can't know if all the panel drivers have already been loaded. >>> omapfb may see that DVI is the default display, but what should it do if >>> DVI doesn't have a driver yet? It could wait, but perhaps the driver for >>> DVI will never even be loaded. >> >> The encoder which is connected to the crtc (manager) is picked by >> combination of encoder->possible_crtcs bitmask and >> connector->best_encoder(). We could keep things limited so that the >> association of crtc to encoder (manager to output, roughly) never >> changes, but this isn't really the right thing to do. It is better that >> the dssdev not rely on knowing the manager it is attached to at probe >> time, but instead grab resources more dynamically. >> >> Also, at the moment we don't really have any notification to userspace >> about new encoders/connectors showing up (or conversely, being >> removed). Only about existing connectors being plugged/unplugged. The >> closest analogy is perhaps the USB display devices, but even there it is >> only the entire drm device that is plugged/unplugged. And TBH I don't >> really see the point in supporting panel drivers being dynamically >> loaded. It isn't like someone is dynamically soldering on a new display >> connector to some board that is running. I think omapfb or omapdrm >> probe should trigger registering the compiled-in panel drivers, so that >> it can be sure that the dssdev's pop up before it goes and creates drm >> connector objects. Currently we have to hack around this in omapdrm >> with late_initcall() to ensure the panel drivers are probed first, but >> that is an ugly hack that I'd like to get rid of. > > We have panel devices and panel drivers, each of which can appear at any > time. Both are needed for the panel probe to happen. If we don't support > device hotplugging (dynamic creation of devices), we need to use > late_initcall for omapfb/drm. At least I don't see any other option. > > You say that omapdrm should trigger registering of the drivers. How > would that work? Do you mean that the panel drivers would register > themselves to some common list, and omapdrm would go through this list > when drm is loaded, calling probe for the items in the list? I guess > that's doable, but... It's not how kernel drivers are supposed to work, > and so doesn't sound very clean approach to me. I mean, similar to how we handle the subdev for dmm.. the omap_drm_init() does the platform_driver_register() for the dmm device before the platform_driver_register() for omapdrm itself, so we know if there is a dmm device, the driver gets probed first before omapdrm. It could be a matter of iterating through a list, or something like this.. that is basically an implementation detail. But the end result is that the order the drivers are registered is controlled so the probe sequence works out properly (not to mention suspend/resume sequence). > I think we should support proper hotplugging of the panels. This would > fix the problem about init order, but it would also give us device > hotplug support. Obviously nobody is going to solder panel to a running > board, but I don't see any reason why panels, or, more likely, panels on > an add-on boards (like the capes being discussed in omap ml) would not > be hotpluggable using whatever connector is used on the particular use case. > > And even if we don't support removing of the devices, things like the > add-on capes could cause the panel on the cape to be identified at some > late time (the panel is not described in the board file or DT data, but > found at runtime depending on the ID of the cape). This would add > another step to the init sequence that should be just right, if we don't > support hotplug. If capes are really hot-pluggable, then maybe it is worth thinking about how to make this more dynamic. Although it is a bigger problem, which involves userspace being aware that connectors can dynamically appear/disappear. And the dynamic disappearing is something I worry about more.. it adds the possibility of all sorts of interesting race conditions, such as connectors disappearing in the middle of modeset. I prefer not making things more complicated and error prone than they need to be. If there is not a legitimate use case for connector hw dynamically appearing/disappearing then I don't think we should go there. It sounds nice and simple and clean, but in reality I think it just introduces a whole lot of ways for things to go wrong. A wise man once said: https://github.com/robclark/kernel-omap4/blob/master/Documentation/SubmittingPatches#L700 BR, -R > Yes, I know it's not simple =). And I'm fine with simpler approach for > the time being, but I'd like full hotplug to be the future goal. At > least the common panel framework should not create restrictions about > this, even if drm wouldn't allow device hotplug. > > Tomi > > -- > To unsubscribe from this list: send the line "unsubscribe linux-omap" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2012-11-06 16:40, Rob Clark wrote: > I mean, similar to how we handle the subdev for dmm.. the > omap_drm_init() does the platform_driver_register() for the dmm device > before the platform_driver_register() for omapdrm itself, so we know > if there is a dmm device, the driver gets probed first before omapdrm. Well, I consider that a bit hacky too. That's not how linux device framework is supposed to be used. I know it makes life easier to do the registering like that, though. > It could be a matter of iterating through a list, or something like > this.. that is basically an implementation detail. But the end result > is that the order the drivers are registered is controlled so the > probe sequence works out properly (not to mention suspend/resume > sequence). I feel that this kind of solution just tries to solve the generic problem of init/suspend ordering in a single driver, instead of fixing the device framework itself. Or, of course it's possible that our drive architecture just sucks, and the device framework is fine. In that case the workaround is even worse, and we should fix our drivers. >> I think we should support proper hotplugging of the panels. This would >> fix the problem about init order, but it would also give us device >> hotplug support. Obviously nobody is going to solder panel to a running >> board, but I don't see any reason why panels, or, more likely, panels on >> an add-on boards (like the capes being discussed in omap ml) would not >> be hotpluggable using whatever connector is used on the particular use case. >> >> And even if we don't support removing of the devices, things like the >> add-on capes could cause the panel on the cape to be identified at some >> late time (the panel is not described in the board file or DT data, but >> found at runtime depending on the ID of the cape). This would add >> another step to the init sequence that should be just right, if we don't >> support hotplug. > > If capes are really hot-pluggable, then maybe it is worth thinking > about how to make this more dynamic. Although it is a bigger problem, > which involves userspace being aware that connectors can dynamically > appear/disappear. And the dynamic disappearing is something I worry > about more.. it adds the possibility of all sorts of interesting race > conditions, such as connectors disappearing in the middle of modeset. > I prefer not making things more complicated and error prone than they > need to be. If there is not a legitimate use case for connector hw > dynamically appearing/disappearing then I don't think we should go > there. It sounds nice and simple and clean, but in reality I think it > just introduces a whole lot of ways for things to go wrong. A wise Yes, I agree that it complicates things. > man once said: > > https://github.com/robclark/kernel-omap4/blob/master/Documentation/SubmittingPatches#L700 I've done things simple lots of times in the omapdss driver, only to have to rewrite the thing in more complex way later to accommodate new scenarios. I think it's good to write the code in a bit more generic way than the use case at the moment of writing requires, because more often than not, it'll save time in the future. Hotplugging is not some abstract future scenario, we already have hardware that could use it. For example, omap3 SDP board has a switchable output to DVI or LCD panel. In this case we know what the two options are, but the disabled component is still effectually removed from the system, and plugged back in when it's enabled. Hotplug is not a high priority item, but I do wish we get it supported in common panel framework. Then it's at least possible to extend drm in the future to support it. Anyway, this makes me wonder... omapdrm currently maps the elements of the whole video pipeline to drm elements (encoder, connector, etc). Would it make more sense to just map the DISPC to these drm elements? Connector would then be the output from DISPC. This would map the drm elements to the static hardware blocks, and the meaning of those blocks would be quite similar to what they are in the desktop world (I guess). The panel driver, the external chips, and the DSS internal output blocks (dsi, dpi, ...) would be handled separately from those drm elements. The DSS internal blocks are static, of course, but they can be effectively considered the same way as external chips. The omapdrm driver needs of course to access those separate elements also, but that shouldn't be a problem. If omapdrm needs to call a function in the panel driver, all it needs to do is go through the chain to find the panel. Well, except if one output connected two two panels via a bridge chip... And if drm is at some point extended to support panel drivers, or chains of external display entities, it would be easier to add that support. What would it require the manage the elements like that? Would it help? It sounds to me that this would simplify the model. Tomi
On Wed, Nov 7, 2012 at 4:01 AM, Tomi Valkeinen <tomi.valkeinen@ti.com> wrote: > On 2012-11-06 16:40, Rob Clark wrote: > >> I mean, similar to how we handle the subdev for dmm.. the >> omap_drm_init() does the platform_driver_register() for the dmm device >> before the platform_driver_register() for omapdrm itself, so we know >> if there is a dmm device, the driver gets probed first before omapdrm. > > Well, I consider that a bit hacky too. That's not how linux device > framework is supposed to be used. I know it makes life easier to do the > registering like that, though. > >> It could be a matter of iterating through a list, or something like >> this.. that is basically an implementation detail. But the end result >> is that the order the drivers are registered is controlled so the >> probe sequence works out properly (not to mention suspend/resume >> sequence). > > I feel that this kind of solution just tries to solve the generic > problem of init/suspend ordering in a single driver, instead of fixing > the device framework itself. > > Or, of course it's possible that our drive architecture just sucks, and > the device framework is fine. In that case the workaround is even worse, > and we should fix our drivers. well, I guess by splitting things up into many pieces, we push the device framework in ways that it is not pushed on desktop. But I've enough on my plate so I wasn't going to volunteer to try and fix that ;-) controlling the order that the drivers are registered is currently the most straightforward way to not have problems w/ probe/init/suspend.. >>> I think we should support proper hotplugging of the panels. This would >>> fix the problem about init order, but it would also give us device >>> hotplug support. Obviously nobody is going to solder panel to a running >>> board, but I don't see any reason why panels, or, more likely, panels on >>> an add-on boards (like the capes being discussed in omap ml) would not >>> be hotpluggable using whatever connector is used on the particular use case. >>> >>> And even if we don't support removing of the devices, things like the >>> add-on capes could cause the panel on the cape to be identified at some >>> late time (the panel is not described in the board file or DT data, but >>> found at runtime depending on the ID of the cape). This would add >>> another step to the init sequence that should be just right, if we don't >>> support hotplug. >> >> If capes are really hot-pluggable, then maybe it is worth thinking >> about how to make this more dynamic. Although it is a bigger problem, >> which involves userspace being aware that connectors can dynamically >> appear/disappear. And the dynamic disappearing is something I worry >> about more.. it adds the possibility of all sorts of interesting race >> conditions, such as connectors disappearing in the middle of modeset. >> I prefer not making things more complicated and error prone than they >> need to be. If there is not a legitimate use case for connector hw >> dynamically appearing/disappearing then I don't think we should go >> there. It sounds nice and simple and clean, but in reality I think it >> just introduces a whole lot of ways for things to go wrong. A wise > > Yes, I agree that it complicates things. > >> man once said: >> >> https://github.com/robclark/kernel-omap4/blob/master/Documentation/SubmittingPatches#L700 > > I've done things simple lots of times in the omapdss driver, only to > have to rewrite the thing in more complex way later to accommodate new > scenarios. I think it's good to write the code in a bit more generic way > than the use case at the moment of writing requires, because more often > than not, it'll save time in the future. well, if it is not effecting userspace ABI, then I think, when in doubt I prefer to start simple. It can always be changed later if needed. > Hotplugging is not some abstract future scenario, we already have > hardware that could use it. For example, omap3 SDP board has a > switchable output to DVI or LCD panel. In this case we know what the two > options are, but the disabled component is still effectually removed > from the system, and plugged back in when it's enabled. I would look at this as two different connectors which can not be used at the same time. You have this scenario with desktop graphics cards. > Hotplug is not a high priority item, but I do wish we get it supported > in common panel framework. Then it's at least possible to extend drm in > the future to support it. > > > > Anyway, this makes me wonder... omapdrm currently maps the elements of > the whole video pipeline to drm elements (encoder, connector, etc). > Would it make more sense to just map the DISPC to these drm elements? > Connector would then be the output from DISPC. I think: plane->overlay crtc->manager is pretty clear. And really encoder->output should be the way it is.. on the branch w/ omapdss/omapdrm kms re-write, this is how it is for plane/crtc, except for now: encoder+connector->dssdev Basically the encoder is doing the "control" stuff (power on/off, set timings, etc), and the connector is only doing non control stuff (detect, reading edid, etc). But I think this will probably change a bit as CFP comes into the picture. Currently the drm connector is somewhat a "passive" element, but I think this will have to change a bit w/ CFP. > This would map the drm elements to the static hardware blocks, and the > meaning of those blocks would be quite similar to what they are in the > desktop world (I guess). > > The panel driver, the external chips, and the DSS internal output blocks > (dsi, dpi, ...) would be handled separately from those drm elements. The > DSS internal blocks are static, of course, but they can be effectively > considered the same way as external chips. I think dsi/dpi/etc map to encoder. The big question is where the panel's fit. But to userspace somehow this should look like connectors. I think: encoder->output connector->panel could work.. although connector is less passive than KMS currently assumes. And "panel" could really be a whole chain in the case of bridge chips, etc. I don't know, maybe there are better ways. But I think userspace really just wants to know "which monitor" which is basically connector. > The omapdrm driver needs of course to access those separate elements > also, but that shouldn't be a problem. If omapdrm needs to call a > function in the panel driver, all it needs to do is go through the chain > to find the panel. Well, except if one output connected two two panels > via a bridge chip... yeah, that is a really ugly case in our hw since it is quite non-transparent (ie. implications about use of planes, etc). > And if drm is at some point extended to support panel drivers, or chains > of external display entities, it would be easier to add that support. > > What would it require the manage the elements like that? Would it help? > It sounds to me that this would simplify the model. I'm not really entirely sure.. other than at least other drivers supporting CFP will have the same requirements ;-) I guess the two best options are either bury some sort of chain of panel drivers in the connector, or introduce some internal elements in DRM which are not necessarily visible to userspace. (Or at least userspace should have the option to ignore it for backwards compatibility. For atomic pageflip/modeset, the converting of everything to properties makes it easier to think about exposing new KMS mode object types to userspace.) Either way, I think we need to break this into stages. And first stage is to ignore chains of panel drivers and just support single panel drivers (ignoring bridges, etc). I'm not entirely sure what the second stage should be, but am open to suggestions and seeing how it evolves. BR, -R > Tomi > > -- To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2012-11-07 16:32, Rob Clark wrote: > On Wed, Nov 7, 2012 at 4:01 AM, Tomi Valkeinen <tomi.valkeinen@ti.com> wrote: >> Hotplugging is not some abstract future scenario, we already have >> hardware that could use it. For example, omap3 SDP board has a >> switchable output to DVI or LCD panel. In this case we know what the two >> options are, but the disabled component is still effectually removed >> from the system, and plugged back in when it's enabled. > > I would look at this as two different connectors which can not be used > at the same time. You have this scenario with desktop graphics cards. Yes, that's an option with fixed amount of display devices. But doesn't work for capes. >> Hotplug is not a high priority item, but I do wish we get it supported >> in common panel framework. Then it's at least possible to extend drm in >> the future to support it. >> >> >> >> Anyway, this makes me wonder... omapdrm currently maps the elements of >> the whole video pipeline to drm elements (encoder, connector, etc). >> Would it make more sense to just map the DISPC to these drm elements? >> Connector would then be the output from DISPC. > > I think: > > plane->overlay > crtc->manager > > is pretty clear. And really > > encoder->output > > should be the way it is.. on the branch w/ omapdss/omapdrm kms I'm not so sure. The output (dpi/dsi/hdmi...) is the second step in our chain. The primary "output" is in the DISPC module, the overlay manager. That's where the timings, pixel clock, etc. are programmed. The second step, our output, is really a converter IP. It receives the primary output, converts it and outputs something else. Just like an external converter chip would do. And the output can be quite a bit anything. For example, with DBI or DSI command mode outputs we don't have any of the conventional video timings. It doesn't make sense to program, say, video blanking periods to those outputs. But even with DBI and DSI we do have video blanking periods in the DISPC's output, the ovl mgr. Of course, at the end of the chain we have a panel that uses normal video timings (well, most likely but not necessarily), and so we could program those timings at the end of the chain, in the block before the panel. But even then the encoder doesn't really map to the DSS's output block, as the DSS's output block may not have the conventional timings (like DBI), or they may be something totally different than what we get in the end of the chain to the panel. So I think mapping encoder to output will not work with multiple display blocks in a chain. Thus I'd see the encoder would better match the DISPC's output, or alternatively perhaps the block which is just before the panel (whatever that is, sometimes it can be OMAP's DSI/HDMI/etc). However, the latter may be a bit strange as the block could be an external component, possibly hotpluggable. > re-write, this is how it is for plane/crtc, except for now: > > encoder+connector->dssdev > > Basically the encoder is doing the "control" stuff (power on/off, set > timings, etc), and the connector is only doing non control stuff > (detect, reading edid, etc). > > But I think this will probably change a bit as CFP comes into the > picture. Currently the drm connector is somewhat a "passive" element, > but I think this will have to change a bit w/ CFP. > >> This would map the drm elements to the static hardware blocks, and the >> meaning of those blocks would be quite similar to what they are in the >> desktop world (I guess). >> >> The panel driver, the external chips, and the DSS internal output blocks >> (dsi, dpi, ...) would be handled separately from those drm elements. The >> DSS internal blocks are static, of course, but they can be effectively >> considered the same way as external chips. > > I think dsi/dpi/etc map to encoder. The big question is where the > panel's fit. But to userspace somehow this should look like > connectors. I think: > > encoder->output > connector->panel > > could work.. although connector is less passive than KMS currently > assumes. And "panel" could really be a whole chain in the case of > bridge chips, etc. I don't know, maybe there are better ways. But I > think userspace really just wants to know "which monitor" which is > basically connector. Hmm yes. Well, even if we map encoder and connector to the ovl manager, the userspace could see which monitor there is. Obviously we need to make changes for that to work, but as a model it feels a lot more natural to me than using output and panel for encoder and connector. Perhaps it's wrong to say "map connector to ovl mgr". It would be more like "this connector observes the chain connected to this ovl mgr", even though the connector wouldn't observe any block in the chain directly. Just observing the plug in/out status, etc. But I think encoder really maps quite well directly to the output side of overlay manager. >> The omapdrm driver needs of course to access those separate elements >> also, but that shouldn't be a problem. If omapdrm needs to call a >> function in the panel driver, all it needs to do is go through the chain >> to find the panel. Well, except if one output connected two two panels >> via a bridge chip... > > yeah, that is a really ugly case in our hw since it is quite > non-transparent (ie. implications about use of planes, etc). Not really in this case. You're perhaps thinking about connecting two outputs to a single panel. Which is problematic also. We don't have in sights a board that splits one output to two panels, so I think we should just ignore that for now. But two outputs for one panel is on the table. >> And if drm is at some point extended to support panel drivers, or chains >> of external display entities, it would be easier to add that support. >> >> What would it require the manage the elements like that? Would it help? >> It sounds to me that this would simplify the model. > > I'm not really entirely sure.. other than at least other drivers > supporting CFP will have the same requirements ;-) > > I guess the two best options are either bury some sort of chain of > panel drivers in the connector, or introduce some internal elements in > DRM which are not necessarily visible to userspace. (Or at least > userspace should have the option to ignore it for backwards > compatibility. For atomic pageflip/modeset, the converting of > everything to properties makes it easier to think about exposing new > KMS mode object types to userspace.) Yes, I don't think we should or need to expose these new elements to userspace, at least in the first place. Tomi
On Wed, Nov 7, 2012 at 9:13 AM, Tomi Valkeinen <tomi.valkeinen@ti.com> wrote: > On 2012-11-07 16:32, Rob Clark wrote: >> On Wed, Nov 7, 2012 at 4:01 AM, Tomi Valkeinen <tomi.valkeinen@ti.com> wrote: > >>> Hotplugging is not some abstract future scenario, we already have >>> hardware that could use it. For example, omap3 SDP board has a >>> switchable output to DVI or LCD panel. In this case we know what the two >>> options are, but the disabled component is still effectually removed >>> from the system, and plugged back in when it's enabled. >> >> I would look at this as two different connectors which can not be used >> at the same time. You have this scenario with desktop graphics cards. > > Yes, that's an option with fixed amount of display devices. But doesn't > work for capes. Only if capes are hotpluggable.. otherwise probe what cape(s?) are present at boot time (is this possible to detect a cape from sw?), and create the associated connector(s). Anyways, I think we are stretching a bit hard for use cases for hot-pluggable panels.. I just prefer to ignore hotplug for now and come back to it when there is a more legitimate use-case. >>> Hotplug is not a high priority item, but I do wish we get it supported >>> in common panel framework. Then it's at least possible to extend drm in >>> the future to support it. >>> >>> >>> >>> Anyway, this makes me wonder... omapdrm currently maps the elements of >>> the whole video pipeline to drm elements (encoder, connector, etc). >>> Would it make more sense to just map the DISPC to these drm elements? >>> Connector would then be the output from DISPC. >> >> I think: >> >> plane->overlay >> crtc->manager >> >> is pretty clear. And really >> >> encoder->output >> >> should be the way it is.. on the branch w/ omapdss/omapdrm kms > > I'm not so sure. The output (dpi/dsi/hdmi...) is the second step in our > chain. The primary "output" is in the DISPC module, the overlay manager. > That's where the timings, pixel clock, etc. are programmed. The second > step, our output, is really a converter IP. It receives the primary > output, converts it and outputs something else. Just like an external > converter chip would do. > the timings, pixel clock, vblank/framedone irqs, that all maps to crtc. encoder == converter, so I think this fits. "An encoder takes pixel data from a CRTC and converts it to a format suitable for any attached connectors" (from drm docbook) > And the output can be quite a bit anything. For example, with DBI or DSI > command mode outputs we don't have any of the conventional video > timings. It doesn't make sense to program, say, video blanking periods > to those outputs. But even with DBI and DSI we do have video blanking > periods in the DISPC's output, the ovl mgr. the encoder has mode_fixup() which can alter the timings that end up getting set, which might be a way to account for this. I guess really it should be the panel driver that is telling the encoder what adjusted timings to give to the crtc.. so the panel driver doesn't quite map to connector. > Of course, at the end of the chain we have a panel that uses normal > video timings (well, most likely but not necessarily), and so we could > program those timings at the end of the chain, in the block before the > panel. But even then the encoder doesn't really map to the DSS's output > block, as the DSS's output block may not have the conventional timings > (like DBI), or they may be something totally different than what we get > in the end of the chain to the panel. > > So I think mapping encoder to output will not work with multiple display > blocks in a chain. Thus I'd see the encoder would better match the > DISPC's output, or alternatively perhaps the block which is just before > the panel (whatever that is, sometimes it can be OMAP's DSI/HDMI/etc). > However, the latter may be a bit strange as the block could be an > external component, possibly hotpluggable. > >> re-write, this is how it is for plane/crtc, except for now: >> >> encoder+connector->dssdev >> >> Basically the encoder is doing the "control" stuff (power on/off, set >> timings, etc), and the connector is only doing non control stuff >> (detect, reading edid, etc). >> >> But I think this will probably change a bit as CFP comes into the >> picture. Currently the drm connector is somewhat a "passive" element, >> but I think this will have to change a bit w/ CFP. >> >>> This would map the drm elements to the static hardware blocks, and the >>> meaning of those blocks would be quite similar to what they are in the >>> desktop world (I guess). >>> >>> The panel driver, the external chips, and the DSS internal output blocks >>> (dsi, dpi, ...) would be handled separately from those drm elements. The >>> DSS internal blocks are static, of course, but they can be effectively >>> considered the same way as external chips. >> >> I think dsi/dpi/etc map to encoder. The big question is where the >> panel's fit. But to userspace somehow this should look like >> connectors. I think: >> >> encoder->output >> connector->panel >> >> could work.. although connector is less passive than KMS currently >> assumes. And "panel" could really be a whole chain in the case of >> bridge chips, etc. I don't know, maybe there are better ways. But I >> think userspace really just wants to know "which monitor" which is >> basically connector. > > Hmm yes. Well, even if we map encoder and connector to the ovl manager, > the userspace could see which monitor there is. Obviously we need to > make changes for that to work, but as a model it feels a lot more > natural to me than using output and panel for encoder and connector. well, what we call 'ovl mgr' is really where the (vblank) interrupt generation is. This really should be the crtc. I tried it differently initially, and it wasn't really working out. > Perhaps it's wrong to say "map connector to ovl mgr". It would be more > like "this connector observes the chain connected to this ovl mgr", even > though the connector wouldn't observe any block in the chain directly. > Just observing the plug in/out status, etc. > > But I think encoder really maps quite well directly to the output side > of overlay manager. > >>> The omapdrm driver needs of course to access those separate elements >>> also, but that shouldn't be a problem. If omapdrm needs to call a >>> function in the panel driver, all it needs to do is go through the chain >>> to find the panel. Well, except if one output connected two two panels >>> via a bridge chip... >> >> yeah, that is a really ugly case in our hw since it is quite >> non-transparent (ie. implications about use of planes, etc). > > Not really in this case. You're perhaps thinking about connecting two > outputs to a single panel. Which is problematic also. yes, sorry, I read that backwards.. > We don't have in sights a board that splits one output to two panels, so > I think we should just ignore that for now. But two outputs for one > panel is on the table. > >>> And if drm is at some point extended to support panel drivers, or chains >>> of external display entities, it would be easier to add that support. >>> >>> What would it require the manage the elements like that? Would it help? >>> It sounds to me that this would simplify the model. >> >> I'm not really entirely sure.. other than at least other drivers >> supporting CFP will have the same requirements ;-) >> >> I guess the two best options are either bury some sort of chain of >> panel drivers in the connector, or introduce some internal elements in >> DRM which are not necessarily visible to userspace. (Or at least >> userspace should have the option to ignore it for backwards >> compatibility. For atomic pageflip/modeset, the converting of >> everything to properties makes it easier to think about exposing new >> KMS mode object types to userspace.) > > Yes, I don't think we should or need to expose these new elements to > userspace, at least in the first place. It *might* be useful someday, if there are ever any settings for a bridge chip, etc, which we'd want to expose to userspace as properties.. but I think we can come back to that later. BR, -R > Tomi > > -- To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2012-11-07 21:18, Rob Clark wrote: > On Wed, Nov 7, 2012 at 9:13 AM, Tomi Valkeinen <tomi.valkeinen@ti.com> wrote: >> On 2012-11-07 16:32, Rob Clark wrote: >>> On Wed, Nov 7, 2012 at 4:01 AM, Tomi Valkeinen <tomi.valkeinen@ti.com> wrote: >> >>>> Hotplugging is not some abstract future scenario, we already have >>>> hardware that could use it. For example, omap3 SDP board has a >>>> switchable output to DVI or LCD panel. In this case we know what the two >>>> options are, but the disabled component is still effectually removed >>>> from the system, and plugged back in when it's enabled. >>> >>> I would look at this as two different connectors which can not be used >>> at the same time. You have this scenario with desktop graphics cards. >> >> Yes, that's an option with fixed amount of display devices. But doesn't >> work for capes. > > Only if capes are hotpluggable.. otherwise probe what cape(s?) are > present at boot time (is this possible to detect a cape from sw?), and > create the associated connector(s). Well, a cape can be anything. For beaglebone they have capes with eeprom, and you can detect it. The reason I'd like to have hotplug is that it would simplify panel drivers in the case where we have multiple possible panels for the same output, like the DVI/LCD case above for omap3 SDP. If we don't have hotplug, then both DVI and LCD panel devices are present at the same time, and they will share resources. In the minimum they are sharing the video output, but more often than not they share gpios/powers/etc. It's normal that a driver will acquire resources for its device in its probe, and thus we would have two drivers acquiring the same resources at boot time, leading to the other driver failing. We currently manage this by acquiring the resources late, only when the panel is being enabled. But I think that's rather ugly. It would be much cleaner if the panel device does not exist at all if the panel is disconnected, and is created only when it is connected. This of course creates the problem of who is responsible for creating the panel device, and what triggers it. I think that's case specific, and for capes, it'd be the cape driver. But then again, I guess it's acceptable that we don't allow changing the plugged-in panels at runtime. The user would have to select them with kernel parameters or such. I guess this would be ok for capes and development boards. I'm not aware of a production board that would switch panels at runtime, although I know these were on the table in Nokia. > Anyways, I think we are stretching a bit hard for use cases for > hot-pluggable panels.. I just prefer to ignore hotplug for now and > come back to it when there is a more legitimate use-case. Ok, fair enough. But let's keep hotplug in mind, and if we're going to create code that would make hotplug impossible to implement, let's stop for a moment and think if we can do that in some other way. >> I'm not so sure. The output (dpi/dsi/hdmi...) is the second step in our >> chain. The primary "output" is in the DISPC module, the overlay manager. >> That's where the timings, pixel clock, etc. are programmed. The second >> step, our output, is really a converter IP. It receives the primary >> output, converts it and outputs something else. Just like an external >> converter chip would do. >> > > the timings, pixel clock, vblank/framedone irqs, that all maps to crtc. > > encoder == converter, so I think this fits. "An encoder takes pixel > data from a CRTC and converts it to a format suitable for any attached > connectors" (from drm docbook) Oh, ok. Then what you say makes sense. I thought encoder contains the timings, as you said previously: "Basically the encoder is doing the "control" stuff (power on/off, set timings, etc)". In the case we have a long chain of display blocks, encoder could cover all of them, not just the output block of OMAP? I don't know where the panel goes, though. >> And the output can be quite a bit anything. For example, with DBI or DSI >> command mode outputs we don't have any of the conventional video >> timings. It doesn't make sense to program, say, video blanking periods >> to those outputs. But even with DBI and DSI we do have video blanking >> periods in the DISPC's output, the ovl mgr. > > the encoder has mode_fixup() which can alter the timings that end up > getting set, which might be a way to account for this. I guess really > it should be the panel driver that is telling the encoder what > adjusted timings to give to the crtc.. so the panel driver doesn't > quite map to connector. The panel can't say what timings to give to crtc, it depends on what's between the crtc and the panel. In case of OMAP DSI, the dsi driver needs to specify the timings for crtc, based on the panel. Tomi
diff --git a/drivers/video/omap2/dss/dpi.c b/drivers/video/omap2/dss/dpi.c index 267caf0..32e7dd5 100644 --- a/drivers/video/omap2/dss/dpi.c +++ b/drivers/video/omap2/dss/dpi.c @@ -49,28 +49,30 @@ static struct { struct omap_dss_output output; } dpi; -static struct platform_device *dpi_get_dsidev(enum omap_dss_clk_source clk) +static struct platform_device *dpi_get_dsidev(enum omap_channel channel) { - int dsi_module; - - dsi_module = clk == OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DISPC ? 0 : 1; - - return dsi_get_dsidev_from_id(dsi_module); + switch (channel) { + case OMAP_DSS_CHANNEL_LCD: + return dsi_get_dsidev_from_id(0); + case OMAP_DSS_CHANNEL_LCD2: + return dsi_get_dsidev_from_id(1); + default: + return NULL; + } } -static bool dpi_use_dsi_pll(struct omap_dss_device *dssdev) +static enum omap_dss_clk_source dpi_get_alt_clk_src(enum omap_channel channel) { - if (dssdev->clocks.dispc.dispc_fclk_src == - OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DISPC || - dssdev->clocks.dispc.dispc_fclk_src == - OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DISPC || - dssdev->clocks.dispc.channel.lcd_clk_src == - OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DISPC || - dssdev->clocks.dispc.channel.lcd_clk_src == - OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DISPC) - return true; - else - return false; + switch (channel) { + case OMAP_DSS_CHANNEL_LCD: + return OMAP_DSS_CLK_SRC_DSI_PLL_HSDIV_DISPC; + case OMAP_DSS_CHANNEL_LCD2: + return OMAP_DSS_CLK_SRC_DSI2_PLL_HSDIV_DISPC; + default: + /* this shouldn't happen */ + WARN_ON(1); + return OMAP_DSS_CLK_SRC_FCK; + } } static int dpi_set_dsi_clk(struct omap_dss_device *dssdev, @@ -92,7 +94,7 @@ static int dpi_set_dsi_clk(struct omap_dss_device *dssdev, return r; dss_select_lcd_clk_source(mgr->id, - dssdev->clocks.dispc.channel.lcd_clk_src); + dpi_get_alt_clk_src(mgr->id)); dpi.mgr_config.clock_info = dispc_cinfo; @@ -385,6 +387,8 @@ static int __init dpi_verify_dsi_pll(struct platform_device *dsidev) static int __init dpi_init_display(struct omap_dss_device *dssdev) { + struct platform_device *dsidev; + DSSDBG("init_display\n"); if (dss_has_feature(FEAT_DPI_USES_VDDS_DSI) && @@ -401,17 +405,23 @@ static int __init dpi_init_display(struct omap_dss_device *dssdev) dpi.vdds_dsi_reg = vdds_dsi; } - if (dpi_use_dsi_pll(dssdev)) { - enum omap_dss_clk_source dispc_fclk_src = - dssdev->clocks.dispc.dispc_fclk_src; - dpi.dsidev = dpi_get_dsidev(dispc_fclk_src); + /* + * XXX We shouldn't need dssdev->channel for this. The dsi pll clock + * source for DPI is SoC integration detail, not something that should + * be configured in the dssdev + */ + dsidev = dpi_get_dsidev(dssdev->channel); - if (dpi_verify_dsi_pll(dpi.dsidev)) { - dpi.dsidev = NULL; - DSSWARN("DSI PLL not operational\n"); - } + if (dpi_verify_dsi_pll(dsidev)) { + dsidev = NULL; + DSSWARN("DSI PLL not operational\n"); } + if (dsidev) + DSSDBG("using DSI PLL for DPI clock\n"); + + dpi.dsidev = dsidev; + return 0; }
We currently get the decision whether to use PRCM or DSI PLL clock for DPI from the board file. This is not a good way to handle it, and it won't work with device tree. This patch changes DPI to always use DSI PLL if it's available. Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com> --- drivers/video/omap2/dss/dpi.c | 64 ++++++++++++++++++++++++----------------- 1 file changed, 37 insertions(+), 27 deletions(-)