[PATCH v3] driver-staging: vsoc.c: Add sysfs support for examining the permissions of regions.

[PATCH v3] driver-staging: vsoc.c: Add sysfs support for examining the permissions of regions.

[Jerry Lin]

Add a attribute called permissions under vsoc device node for examining
current granted permissions in vsoc_device.

This file will display permissions in following format:
  begin_offset  end_offset  owner_offset  owned_value
            %x          %x            %x           %x

Signed-off-by: Jerry Lin <wahahab11@gmail.com>
---
 drivers/staging/android/vsoc.c | 48 +++++++++++++++++++++++++++++++++++++++---
 1 file changed, 45 insertions(+), 3 deletions(-)

diff --git a/drivers/staging/android/vsoc.c b/drivers/staging/android/vsoc.c
index 22571ab..8ce3604 100644
--- a/drivers/staging/android/vsoc.c
+++ b/drivers/staging/android/vsoc.c
@@ -128,9 +128,10 @@ struct vsoc_device {
 
 static struct vsoc_device vsoc_dev;
 
-/*
- * TODO(ghartman): Add a /sys filesystem entry that summarizes the permissions.
- */
+static ssize_t permissions_show(struct device *dev,
+				struct device_attribute *attr,
+				char *buf);
+static DEVICE_ATTR_RO(permissions);
 
 struct fd_scoped_permission_node {
 	struct fd_scoped_permission permission;
@@ -718,6 +719,37 @@ static ssize_t vsoc_write(struct file *filp, const char __user *buffer,
 	return len;
 }
 
+static ssize_t permissions_show(struct device *dev,
+				struct device_attribute *attr,
+				char *buffer)
+{
+	struct fd_scoped_permission_node *node;
+	char *row;
+	int ret;
+	ssize_t written = 0;
+
+	row = kzalloc(sizeof(char) * 128, GFP_KERNEL);
+	if (!row)
+		return 0;
+	mutex_lock(&vsoc_dev.mtx);
+	list_for_each_entry(node, &vsoc_dev.permissions, list) {
+		ret = snprintf(row, 128, "%x\t%x\t%x\t%x\n",
+			       node->permission.begin_offset,
+			       node->permission.end_offset,
+			       node->permission.owner_offset,
+			       node->permission.owned_value);
+		if (ret < 0)
+			goto done;
+		memcpy(buffer + written, row, ret);
+		written += ret;
+	}
+
+done:
+	mutex_unlock(&vsoc_dev.mtx);
+	kfree(row);
+	return written;
+}
+
 static irqreturn_t vsoc_interrupt(int irq, void *region_data_v)
 {
 	struct vsoc_region_data *region_data =
@@ -942,6 +974,15 @@ static int vsoc_probe_device(struct pci_dev *pdev,
 		}
 		vsoc_dev.regions_data[i].device_created = true;
 	}
+	/*
+	 * Create permission attribute on device node.
+	 */
+	result = device_create_file(&pdev->dev, &dev_attr_permissions);
+	if (result) {
+		dev_err(&vsoc_dev.dev->dev, "device_create_file failed\n");
+		vsoc_remove_device(pdev);
+		return -EFAULT;
+	}
 	return 0;
 }
 
@@ -967,6 +1008,7 @@ static void vsoc_remove_device(struct pci_dev *pdev)
 	if (!pdev || !vsoc_dev.dev)
 		return;
 	dev_info(&pdev->dev, "remove_device\n");
+	device_remove_file(&pdev->dev, &dev_attr_permissions);
 	if (vsoc_dev.regions_data) {
 		for (i = 0; i < vsoc_dev.layout->region_count; ++i) {
 			if (vsoc_dev.regions_data[i].device_created) {
-- 
2.7.4

Re: [PATCH v3] driver-staging: vsoc.c: Add sysfs support for examining the permissions of regions.

[Greg Kroah-Hartman]

On Wed, Nov 07, 2018 at 10:30:43AM +0800, Jerry Lin wrote:
> Add a attribute called permissions under vsoc device node for examining
> current granted permissions in vsoc_device.
> 
> This file will display permissions in following format:
>   begin_offset  end_offset  owner_offset  owned_value
>             %x          %x            %x           %x
> 
> Signed-off-by: Jerry Lin <wahahab11@gmail.com>
> ---
>  drivers/staging/android/vsoc.c | 48 +++++++++++++++++++++++++++++++++++++++---
>  1 file changed, 45 insertions(+), 3 deletions(-)

What changed from v2?  And where was v2?  What about v1?

You need a change log here of what you did different from the previous
patches.

And why ignore my response saying that this type of sysfs file is not ok
at all?

thanks,

greg k-h

Re: [PATCH v3] driver-staging: vsoc.c: Add sysfs support for examining the permissions of regions.

[wahahab]

> On 7 Nov 2018, at 5:15 PM, Greg Kroah-Hartman <greg@kroah.com> wrote:
> 
> On Wed, Nov 07, 2018 at 10:30:43AM +0800, Jerry Lin wrote:
>> Add a attribute called permissions under vsoc device node for examining
>> current granted permissions in vsoc_device.
>> 
>> This file will display permissions in following format:
>> begin_offset  end_offset  owner_offset  owned_value
>>           %x          %x            %x           %x
>> 
>> Signed-off-by: Jerry Lin <wahahab11@gmail.com>
>> ---
>> drivers/staging/android/vsoc.c | 48 +++++++++++++++++++++++++++++++++++++++---
>> 1 file changed, 45 insertions(+), 3 deletions(-)
> 
> What changed from v2?  And where was v2?  What about v1?
> 
> You need a change log here of what you did different from the previous
> patches.

Sorry for the mistakes I made, I shall read the document about patches more carefully.
Here is the change logs:

Changes in v2:
	- Display permissions information in sysfs insureds of debufs.
Changes in v3:
	- Remove unnecessary null terminator after snprintf.

> 
> And why ignore my response saying that this type of sysfs file is not ok
> at all?
> 

I didn’t mean to ignore it but I haven’t receive the response you described,
May you send the response to me again so I can do further revision as well as
change logs and resubmit the patch again?

Thanks for your patience

Jerry

Re: [PATCH v3] driver-staging: vsoc.c: Add sysfs support for examining the permissions of regions.

[Greg Kroah-Hartman]

On Thu, Nov 08, 2018 at 08:49:41AM +0800, wahahab wrote:
> 
> > On 7 Nov 2018, at 5:15 PM, Greg Kroah-Hartman <greg@kroah.com> wrote:
> > 
> > On Wed, Nov 07, 2018 at 10:30:43AM +0800, Jerry Lin wrote:
> >> Add a attribute called permissions under vsoc device node for examining
> >> current granted permissions in vsoc_device.
> >> 
> >> This file will display permissions in following format:
> >> begin_offset  end_offset  owner_offset  owned_value
> >>           %x          %x            %x           %x
> >> 
> >> Signed-off-by: Jerry Lin <wahahab11@gmail.com>
> >> ---
> >> drivers/staging/android/vsoc.c | 48 +++++++++++++++++++++++++++++++++++++++---
> >> 1 file changed, 45 insertions(+), 3 deletions(-)
> > 
> > What changed from v2?  And where was v2?  What about v1?
> > 
> > You need a change log here of what you did different from the previous
> > patches.
> 
> Sorry for the mistakes I made, I shall read the document about patches more carefully.
> Here is the change logs:
> 
> Changes in v2:
> 	- Display permissions information in sysfs insureds of debufs.
> Changes in v3:
> 	- Remove unnecessary null terminator after snprintf.
> 
> > 
> > And why ignore my response saying that this type of sysfs file is not ok
> > at all?
> > 
> 
> I didn’t mean to ignore it but I haven’t receive the response you described,
> May you send the response to me again so I can do further revision as well as
> change logs and resubmit the patch again?

You can not have multiple values in a single sysfs file.  sysfs is "one
value per file".  This needs to be individual files, if you really need
this.  And never a "header" for a sysfs file, that's never something
that should ever be in a sysfs file.

And finally, you need a Documentation/ABI/ update for any sysfs file
changes.

thanks,

greg k-h

Re: [PATCH v3] driver-staging: vsoc.c: Add sysfs support for examining the permissions of regions.

[Dan Carpenter]

On Wed, Nov 07, 2018 at 10:30:43AM +0800, Jerry Lin wrote:
> Add a attribute called permissions under vsoc device node for examining
> current granted permissions in vsoc_device.
> 
> This file will display permissions in following format:
>   begin_offset  end_offset  owner_offset  owned_value
>             %x          %x            %x           %x
> 

(I'm not totally an expert on sysfs rules).

Sysfs are supposed to be one value per file, so instead of doing this
you would create a directory with four files like
vsoc_device/begin_offset vsoc_device/end_offset etc.  And each would
just hae a %x output.

> +static ssize_t permissions_show(struct device *dev,
> +				struct device_attribute *attr,
> +				char *buffer)
> +{
> +	struct fd_scoped_permission_node *node;
> +	char *row;
> +	int ret;
> +	ssize_t written = 0;
> +
> +	row = kzalloc(sizeof(char) * 128, GFP_KERNEL);
> +	if (!row)
> +		return 0;

Don't allocate this, just snprintf() directly into the buffer.

regards,
dan carpenter

Re: [PATCH v3] driver-staging: vsoc.c: Add sysfs support for examining the permissions of regions.

[Greg Kroah-Hartman]

On Mon, Nov 12, 2018 at 04:49:30PM +0800, wahahab wrote:
> 
> > On 10 Nov 2018, at 1:15 AM, Dan Carpenter <dan.carpenter@oracle.com <mailto:dan.carpenter@oracle.com>> wrote:
> > 
> > On Wed, Nov 07, 2018 at 10:30:43AM +0800, Jerry Lin wrote:
> >> Add a attribute called permissions under vsoc device node for examining
> >> current granted permissions in vsoc_device.
> >> 
> >> This file will display permissions in following format:
> >>  begin_offset  end_offset  owner_offset  owned_value
> >>            %x          %x            %x           %x
> >> 
> > 
> > (I'm not totally an expert on sysfs rules).
> > 
> > Sysfs are supposed to be one value per file, so instead of doing this
> > you would create a directory with four files like
> > vsoc_device/begin_offset vsoc_device/end_offset etc.  And each would
> > just hae a %x output.
> 
> Thanks for your advice. I have started with this approach.
> 
> But when I trying to create this kind of sysfs hierarchy. I encountered the difficulties of file organizing.
> 
> My current thought is to create a folder under the device node called permissions and user can examine
> permission though file path like vsoc_device/permissions/permission1/begin_offset. But there comes a
> problem that it seems hard to determine the correct index of permission to make folder name unique.
> 
> The solution I come up with is to use memory address of permission node to be the index of permission.
> So the path will be something like vsoc_device/permissions/0x4d23f/begin_offset.
> Is this OK for sysfs?

Ick, that is messy.  What exactly are you trying to export and why use
sysfs?  Is this just debugging information?  Who is going to use this
data?

If it is just for debugging, use debugfs, that is what it is there for.
If you use sysfs, you then have to support your new api for a very long
time, as you can not break userspace reliance on it.

thanks,

greg k-h

Re: [PATCH v3] driver-staging: vsoc.c: Add sysfs support for examining the permissions of regions.

[wahahab]

> On 12 Nov 2018, at 8:40 PM, Greg Kroah-Hartman <greg@kroah.com> wrote:
> 
> On Mon, Nov 12, 2018 at 04:49:30PM +0800, wahahab wrote:
>> 
>>> On 10 Nov 2018, at 1:15 AM, Dan Carpenter <dan.carpenter@oracle.com <mailto:dan.carpenter@oracle.com>> wrote:
>>> 
>>> On Wed, Nov 07, 2018 at 10:30:43AM +0800, Jerry Lin wrote:
>>>> Add a attribute called permissions under vsoc device node for examining
>>>> current granted permissions in vsoc_device.
>>>> 
>>>> This file will display permissions in following format:
>>>> begin_offset  end_offset  owner_offset  owned_value
>>>>          %x          %x            %x           %x
>>>> 
>>> 
>>> (I'm not totally an expert on sysfs rules).
>>> 
>>> Sysfs are supposed to be one value per file, so instead of doing this
>>> you would create a directory with four files like
>>> vsoc_device/begin_offset vsoc_device/end_offset etc.  And each would
>>> just hae a %x output.
>> 
>> Thanks for your advice. I have started with this approach.
>> 
>> But when I trying to create this kind of sysfs hierarchy. I encountered the difficulties of file organizing.
>> 
>> My current thought is to create a folder under the device node called permissions and user can examine
>> permission though file path like vsoc_device/permissions/permission1/begin_offset. But there comes a
>> problem that it seems hard to determine the correct index of permission to make folder name unique.
>> 
>> The solution I come up with is to use memory address of permission node to be the index of permission.
>> So the path will be something like vsoc_device/permissions/0x4d23f/begin_offset.
>> Is this OK for sysfs?
> 
> Ick, that is messy.  What exactly are you trying to export and why use
> sysfs?  Is this just debugging information?  Who is going to use this
> data?

I felt that exporting these information in sysfs will add lots of complexities in this driver.
And I’m not sure these informations are for user space or just for debugging.

It seems there is a conflict of TODO messages between TODO file and the
comment in vsoc.c.

Should I use debugfs first for this patch?

Thanks

Jerry Lin

> 
> If it is just for debugging, use debugfs, that is what it is there for.
> If you use sysfs, you then have to support your new api for a very long
> time, as you can not break userspace reliance on it.
> 
> thanks,
> 
> greg k-h

Re: [PATCH v3] driver-staging: vsoc.c: Add sysfs support for examining the permissions of regions.

[Greg Kroah-Hartman]

On Tue, Nov 13, 2018 at 11:32:51AM +0800, wahahab wrote:
> 
> > On 12 Nov 2018, at 8:40 PM, Greg Kroah-Hartman <greg@kroah.com> wrote:
> > 
> > On Mon, Nov 12, 2018 at 04:49:30PM +0800, wahahab wrote:
> >> 
> >>> On 10 Nov 2018, at 1:15 AM, Dan Carpenter <dan.carpenter@oracle.com <mailto:dan.carpenter@oracle.com>> wrote:
> >>> 
> >>> On Wed, Nov 07, 2018 at 10:30:43AM +0800, Jerry Lin wrote:
> >>>> Add a attribute called permissions under vsoc device node for examining
> >>>> current granted permissions in vsoc_device.
> >>>> 
> >>>> This file will display permissions in following format:
> >>>> begin_offset  end_offset  owner_offset  owned_value
> >>>>          %x          %x            %x           %x
> >>>> 
> >>> 
> >>> (I'm not totally an expert on sysfs rules).
> >>> 
> >>> Sysfs are supposed to be one value per file, so instead of doing this
> >>> you would create a directory with four files like
> >>> vsoc_device/begin_offset vsoc_device/end_offset etc.  And each would
> >>> just hae a %x output.
> >> 
> >> Thanks for your advice. I have started with this approach.
> >> 
> >> But when I trying to create this kind of sysfs hierarchy. I encountered the difficulties of file organizing.
> >> 
> >> My current thought is to create a folder under the device node called permissions and user can examine
> >> permission though file path like vsoc_device/permissions/permission1/begin_offset. But there comes a
> >> problem that it seems hard to determine the correct index of permission to make folder name unique.
> >> 
> >> The solution I come up with is to use memory address of permission node to be the index of permission.
> >> So the path will be something like vsoc_device/permissions/0x4d23f/begin_offset.
> >> Is this OK for sysfs?
> > 
> > Ick, that is messy.  What exactly are you trying to export and why use
> > sysfs?  Is this just debugging information?  Who is going to use this
> > data?
> 
> I felt that exporting these information in sysfs will add lots of complexities in this driver.
> And I’m not sure these informations are for user space or just for debugging.
> 
> It seems there is a conflict of TODO messages between TODO file and the
> comment in vsoc.c.
> 
> Should I use debugfs first for this patch?

If it is for debugging, yes.  As I have no idea what this code is doing,
or what wants that information, it is hard to determine, sorry.

good luck!

greg k-h