| Message ID | 20170502000455.13240-5-computersforpeace@gmail.com |
|---|---|
| State | Rejected |
| Delegated to: | Brian Norris |
| Headers | show |
On Mon, 1 May 2017 17:04:54 -0700 Brian Norris <computersforpeace@gmail.com> wrote: > Recent patches to this driver changed the error handling for missing > clocks. Now we'll print an error if the clock causes us to defer > probing. Let's not do that. > > Cc: Simon Baatz <gmbnomis@gmail.com> > Signed-off-by: Brian Norris <computersforpeace@gmail.com> Acked-by: Boris Brezillon <boris.brezillon@free-electrons.com> > --- > Compile tested only > > drivers/mtd/nand/orion_nand.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/drivers/mtd/nand/orion_nand.c b/drivers/mtd/nand/orion_nand.c > index f8e463a97b9e..e2bfb37f11df 100644 > --- a/drivers/mtd/nand/orion_nand.c > +++ b/drivers/mtd/nand/orion_nand.c > @@ -161,7 +161,8 @@ static int __init orion_nand_probe(struct platform_device *pdev) > if (ret == -ENOENT) { > info->clk = NULL; > } else { > - dev_err(&pdev->dev, "failed to get clock!\n"); > + if (ret != -EPROBE_DEFER) > + dev_err(&pdev->dev, "failed to get clock!\n"); > return ret; > } > }
Hi Brian, On Mon, May 01, 2017 at 05:04:54PM -0700, Brian Norris wrote: > Recent patches to this driver changed the error handling for missing > clocks. Now we'll print an error if the clock causes us to defer > probing. Let's not do that. > This is by intention. Probe deferral is not supported by drivers using module_platform_driver_probe(). EPROBE_DEFER results in not being able to get the clock and, thus, is handled like any other error. - Simon
On Tue, May 02, 2017 at 10:07:31AM +0200, Simon Baatz wrote: > Hi Brian, > > On Mon, May 01, 2017 at 05:04:54PM -0700, Brian Norris wrote: > > Recent patches to this driver changed the error handling for missing > > clocks. Now we'll print an error if the clock causes us to defer > > probing. Let's not do that. > > > > This is by intention. Probe deferral is not supported by drivers > using module_platform_driver_probe(). EPROBE_DEFER results in not > being able to get the clock and, thus, is handled like any other > error. Ah, good point. I overlooked that part. Seems kind of inconsistent driver writing (what happens if someone wants to use this driver on a more complex system, where some of the dependent resources are not available at init time?), but I see in the git logs that apparently this is intentional (to save memory?). So, I'll NAK this patch. Brian
diff --git a/drivers/mtd/nand/orion_nand.c b/drivers/mtd/nand/orion_nand.c index f8e463a97b9e..e2bfb37f11df 100644 --- a/drivers/mtd/nand/orion_nand.c +++ b/drivers/mtd/nand/orion_nand.c @@ -161,7 +161,8 @@ static int __init orion_nand_probe(struct platform_device *pdev) if (ret == -ENOENT) { info->clk = NULL; } else { - dev_err(&pdev->dev, "failed to get clock!\n"); + if (ret != -EPROBE_DEFER) + dev_err(&pdev->dev, "failed to get clock!\n"); return ret; } }
Recent patches to this driver changed the error handling for missing clocks. Now we'll print an error if the clock causes us to defer probing. Let's not do that. Cc: Simon Baatz <gmbnomis@gmail.com> Signed-off-by: Brian Norris <computersforpeace@gmail.com> --- Compile tested only drivers/mtd/nand/orion_nand.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)