| Message ID | 20151116122947.GA9125@amd (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Mon, Nov 16, 2015 at 01:29:47PM +0100, Pavel Machek wrote: > Hi! > > > > > Every single sound driver gets this right, none of them assume the name > > > > is global. What makes you say that they assume names are global? > > > > > Ok, so you are saying that if I fix mfd initialization, sound will > > > automagically switch from global regulators to device-specific > > > regulators and things will start working? > > > > Yes. > > Ok, so something like this should be applied? > > (I'm not sure how to test it, as audio works before and after the > patch.) > Apologies this all going down over the weekend I have only just caught up. I will probably send a few other inline replies to the thread. > Thanks, > Pavel > > Signed-off-by: Pavel Machek <pavel@denx.de> > > commit d6005263acb94343645e719ee90b8940cb2545df > Author: Pavel <pavel@ucw.cz> > Date: Mon Nov 16 13:19:21 2015 +0100 > > regulator_bulk_register() needs device to be already > registered. Reorganize the code to make it so. > > diff --git a/drivers/mfd/mfd-core.c b/drivers/mfd/mfd-core.c > index 14fd5cb..e891f10 100644 > --- a/drivers/mfd/mfd-core.c > +++ b/drivers/mfd/mfd-core.c > @@ -147,12 +147,6 @@ static int mfd_add_device(struct device *parent, int id, > pdev->dev.dma_parms = parent->dma_parms; > pdev->dev.coherent_dma_mask = parent->coherent_dma_mask; > > - ret = regulator_bulk_register_supply_alias( > - &pdev->dev, cell->parent_supplies, > - parent, cell->parent_supplies, > - cell->num_parent_supplies); > - if (ret < 0) > - goto fail_res; This does look like it is too early, but really all we are doing it adding the device pointer and the supply name into a lookup table, neither of those things change across the stuff below. The printout for the mapping does indeed appear to be broken (because you can't dev_name on the device yet, until after it has been added), but that will be all that should get fixed by this move. > > if (parent->of_node && cell->of_compatible) { > for_each_child_of_node(parent->of_node, np) { > @@ -169,12 +176,12 @@ static int mfd_add_device(struct device *parent, int id, > ret = platform_device_add_data(pdev, > cell->platform_data, cell->pdata_size); > if (ret) > - goto fail_alias; > + goto fail_res; > } > > ret = mfd_platform_add_cell(pdev, cell, usage_count); > if (ret) > - goto fail_alias; > + goto fail_res; > > for (r = 0; r < cell->num_resources; r++) { > res[r].name = cell->resources[r].name; > @@ -210,22 +217,29 @@ static int mfd_add_device(struct device *parent, int id, > if (has_acpi_companion(&pdev->dev)) { > ret = acpi_check_resource_conflict(&res[r]); > if (ret) > - goto fail_alias; > + goto fail_res; > } > } > } > > ret = platform_device_add_resources(pdev, res, cell->num_resources); > if (ret) > - goto fail_alias; > + goto fail_res; > > ret = platform_device_add(pdev); > if (ret) > - goto fail_alias; > + goto fail_res; > > if (cell->pm_runtime_no_callbacks) > pm_runtime_no_callbacks(&pdev->dev); > > + ret = regulator_bulk_register_supply_alias( > + &pdev->dev, cell->parent_supplies, > + parent, cell->parent_supplies, > + cell->num_parent_supplies); > + if (ret < 0) > + goto fail_alias; > + Furthermore, the trouble is that you can't move this to here. The platform_device_add will usually cause the device to probe and most devices request their supplies in their probe. Thus if you only register the alias here then it will not be available when the device needs it. Thanks, Charles
On Mon, Nov 16, 2015 at 01:29:47PM +0100, Pavel Machek wrote: > > > Ok, so you are saying that if I fix mfd initialization, sound will > > > automagically switch from global regulators to device-specific > > > regulators and things will start working? > > Yes. > Ok, so something like this should be applied? No, moving the mapping of the aliases after the device has been instantiated won't help here since it means that we will try to probe the device without the mappings set up at all which guarantees that things will go wrong.
diff --git a/drivers/mfd/mfd-core.c b/drivers/mfd/mfd-core.c index 14fd5cb..e891f10 100644 --- a/drivers/mfd/mfd-core.c +++ b/drivers/mfd/mfd-core.c @@ -147,12 +147,6 @@ static int mfd_add_device(struct device *parent, int id, pdev->dev.dma_parms = parent->dma_parms; pdev->dev.coherent_dma_mask = parent->coherent_dma_mask; - ret = regulator_bulk_register_supply_alias( - &pdev->dev, cell->parent_supplies, - parent, cell->parent_supplies, - cell->num_parent_supplies); - if (ret < 0) - goto fail_res; if (parent->of_node && cell->of_compatible) { for_each_child_of_node(parent->of_node, np) { @@ -169,12 +176,12 @@ static int mfd_add_device(struct device *parent, int id, ret = platform_device_add_data(pdev, cell->platform_data, cell->pdata_size); if (ret) - goto fail_alias; + goto fail_res; } ret = mfd_platform_add_cell(pdev, cell, usage_count); if (ret) - goto fail_alias; + goto fail_res; for (r = 0; r < cell->num_resources; r++) { res[r].name = cell->resources[r].name; @@ -210,22 +217,29 @@ static int mfd_add_device(struct device *parent, int id, if (has_acpi_companion(&pdev->dev)) { ret = acpi_check_resource_conflict(&res[r]); if (ret) - goto fail_alias; + goto fail_res; } } } ret = platform_device_add_resources(pdev, res, cell->num_resources); if (ret) - goto fail_alias; + goto fail_res; ret = platform_device_add(pdev); if (ret) - goto fail_alias; + goto fail_res; if (cell->pm_runtime_no_callbacks) pm_runtime_no_callbacks(&pdev->dev); + ret = regulator_bulk_register_supply_alias( + &pdev->dev, cell->parent_supplies, + parent, cell->parent_supplies, + cell->num_parent_supplies); + if (ret < 0) + goto fail_alias; + kfree(res); return 0;