Re: [PATCH 1/2] xen-pciback: Use dev_printk() when possible

From: Boris Ostrovsky
Date: Wed May 27 2020 - 19:19:07 EST

On 5/27/20 6:50 PM, Bjorn Helgaas wrote:
> On Wed, May 27, 2020 at 03:34:26PM -0700, Boris Ostrovsky wrote:
>> On 5/27/20 1:43 PM, Bjorn Helgaas wrote:
>>> @@ -155,8 +157,8 @@ int xen_pcibk_config_read(struct pci_dev *dev, int offset, int size,
>>> u32 value = 0, tmp_val;
>>> if (unlikely(verbose_request))
>>> - printk(KERN_DEBUG DRV_NAME ": %s: read %d bytes at 0x%x\n",
>>> - pci_name(dev), size, offset);
>>> + dev_printk(KERN_DEBUG, &dev->dev, "read %d bytes at 0x%x\n",
>>> + size, offset);
>> Maybe then dev_dbg() ?
> printk(KERN_DEBUG) always produces output, so I used
> dev_printk(KERN_DEBUG) to retain that behavior.
> dev_dbg() does not always produce output, since it depends on DEBUG or
> CONFIG_DYNAMIC_DEBUG and the dynamic debug settings.

Oh, I didn't realize it needs either of these.

> If dev_dbg() seems like the right thing, I would probably add a
> separate patch on top to convert dev_printk(KERN_DEBUG) to dev_dbg().

I think anyone who wants to see those messages ought to have at least
CONFIG_DYNAMIC_DEBUG, especially since they are under verbose_request
(which also should go away IMO). In fact, I wonder whether this code
predates dynamic debugging, it's been there for almost 10 years.

I'll leave it to you whether you want to add another patch.

Reviewed-by: Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>