Re: [PATCH] Platform lockdown information in SYSFS

From: Greg Kroah-Hartman
Date: Tue Aug 04 2020 - 11:51:16 EST


On Tue, Aug 04, 2020 at 12:05:25PM -0300, Daniel Gutson wrote:
> On Tue, Aug 4, 2020 at 11:43 AM Greg Kroah-Hartman
> <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> >
> > On Tue, Aug 04, 2020 at 11:37:02AM -0300, Daniel Gutson wrote:
> > > On Tue, Aug 4, 2020 at 11:23 AM Greg Kroah-Hartman <
> > > gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > > >
> > > > On Tue, Aug 04, 2020 at 10:50:13AM -0300, Daniel Gutson wrote:
> > > > > On Tue, Aug 4, 2020 at 3:41 AM Greg Kroah-Hartman
> > > > > <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > > > > >
> > > > > > On Mon, Aug 03, 2020 at 07:04:56PM -0300, Daniel Gutson wrote:
> > > > > > > > > > Think of this as an input device. You don't put the random
> > > input
> > > > > > > > > > attributes all in one place, you create a new device that
> > > represents the
> > > > > > > > > > input interface and register that.
> > > > > > >
> > > > > > > I'm having trouble with this. What's the dev_t for the child
> > > devices?
> > > > > > > I'm doing
> > > > > > > child_device = device_create(&my_class, &pdev->dev, MKDEV(0, 0),
> > > > > > > NULL, "child");
> > > > > > > pdev is the pci_device (intel-spi-pci)
> > > > > > > dmesg shows
> > > > > > >
> > > > > > > sysfs: cannot create duplicate filename '/class/my-class'
> > > > > > > (call trace)
> > > > > > > kobject_add_internal failed for my-class with -EEXIST, don't try
> > > > > > > to register things with the same name in the same directory.
> > > > > >
> > > > > > Without seeing all of your code, I can't tell you what you are doing
> > > > > > wrong, but the kernel should be giving you a huge hint here...
> > > > > >
> > > > > > Don't create duplicate names in the same subdirectory.
> > > > >
> > > > > I'm not doing that. One of my questions is if MKDEV(0, 0) is valid for
> > > > > create_device, which I inferred so from the documentation.
> > > >
> > > > Yes it is, but that's not the error given to you :)
> > > >
> > > > Many in-kernel users call device_create() with MKDEV(0, 0)
> > > >
> > > > > Here is the listing
> > > >
> > > > <snip>
> > > >
> > > > It's not in any format to read, please never strip leading whitespace,
> > > > it hurts my brain...
> > >
> > > (trying again)
> > > Also, this is in pastebin: https://pastebin.com/8Ye9eUm5
> > >
> > > #include <linux/kobject.h>
> > > #include <linux/sysfs.h>
> > > #include <linux/module.h>
> > > #include <linux/init.h>
> > > #include <linux/list.h>
> > > #include <linux/slab.h>
> > > #include <linux/device.h>
> > > #include <linux/pci.h>
> > >
> > > static ssize_t howareyou_show(struct class *class, struct class_attribute
> > > *attr,
> > > char *buf)
> > > {
> > > return sprintf(buf, "%s\n", "How are you?");
> > > }
> > > static CLASS_ATTR_RO(howareyou);
> >
> > These are rare, as they are "global" for a class, are you sure you want
> > that?
>
> I'm no longer using class attributes, this is from my previous
> experiment. Sorry for the confusion.
> In the "real" code I'll use DEVICE_ATTR_RO.
>
> >
> > >
> > > static struct class my_class = {
> > > .name = "my-class",
> > > .owner = THIS_MODULE,
> > > };
> > >
> > > struct device* child_device;
> > >
> > > static int mypci_probe(struct pci_dev *pdev,
> > > const struct pci_device_id *id)
> > > {
> > > int ret;
> > >
> > > ret = pcim_enable_device(pdev);
> > > if (ret)
> > > return ret;
> > >
> > > ret = class_register(&my_class);
> > > if (ret < 0)
> > > return ret;
> > >
> > >
> > > pr_info("DFG: Recognized. DID: %lx\n", (unsigned long
> > > int)id->driver_data);
> > > pr_info("DFG: device DID: %lx\n", (unsigned long int)pdev->device);
> > >
> > > ret = class_create_file(&my_class, &class_attr_howareyou);
> > > if (ret != 0) {
> > > pr_err("DFG class create file error: %d\n", ret);
> > > class_unregister(&my_class);
> > > return ret;
> > > }
> > >
> > > child_device = device_create(&my_class, &pdev->dev, MKDEV(0, 0),
> > > NULL, "child");
> > > if (child_device == NULL) {
> > > pr_err("DFG error child device NULL");
> > > }
> > >
> > > return ret;
> > > }
> > >
> >
> >
> > Looks sane, what does your kernel log say when you load this?
> first insmod, OK.
> rmmod, OK.
> Second insmod:
>
> [ 4149.389133] sysfs: cannot create duplicate filename
> '/devices/pci0000:00/0000:00:1f.0/my-class'

Ok, that means you did not clean up properly.

After rmmod see if you really did clean up all of the directories/files
you created.

The kernel is trying to be nice and give you a hint here :)

thanks,

greg k-h