Re: [PATCH 2/4] node_device_udev: Make udevGetStringProperty() return void
On Wed, Jun 02, 2021 at 09:48:46 +0200, Peter Krempa wrote:
On Wed, Jun 02, 2021 at 09:37:41 +0200, Michal Privoznik wrote:
> This function can't fail really as it's returning 0 no matter
> what. This is probably a residue from old days when we cared
> about propagating OOM errors. Now we just abort. Make its return
> type void then.
>
> Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
> ---
> src/node_device/node_device_udev.c | 46 ++++++++++--------------------
> 1 file changed, 15 insertions(+), 31 deletions(-)
>
> diff --git a/src/node_device/node_device_udev.c
b/src/node_device/node_device_udev.c
> index d5f3beb389..f48789d98f 100644
> --- a/src/node_device/node_device_udev.c
> +++ b/src/node_device/node_device_udev.c
[...]
> @@ -517,10 +515,9 @@ udevProcessUSBDevice(struct udev_device *device,
> if (udevGetUintProperty(device, "ID_VENDOR_ID",
&usb_dev->vendor, 16) < 0)
> return -1;
>
> - if (udevGetStringProperty(device,
> - "ID_VENDOR_FROM_DATABASE",
> - &usb_dev->vendor_name) < 0)
> - return -1;
> + udevGetStringProperty(device,
> + "ID_VENDOR_FROM_DATABASE",
> + &usb_dev->vendor_name);
>
> if (!usb_dev->vendor_name &&
Here you've kept the NULL check that should be now impossible to trigger
...
> udevGetStringSysfsAttr(device, "manufacturer",
> @@ -530,10 +527,9 @@ udevProcessUSBDevice(struct udev_device *device,
> if (udevGetUintProperty(device, "ID_MODEL_ID",
&usb_dev->product, 16) < 0)
> return -1;
>
> - if (udevGetStringProperty(device,
> - "ID_MODEL_FROM_DATABASE",
> - &usb_dev->product_name) < 0)
> - return -1;
> + udevGetStringProperty(device,
> + "ID_MODEL_FROM_DATABASE",
> + &usb_dev->product_name);
>
> if (!usb_dev->product_name &&
(same here)
> udevGetStringSysfsAttr(device, "product",
[...]
> @@ -965,8 +954,7 @@ udevProcessStorage(struct udev_device *device,
> * expected, so I don't see a problem with not having a property
> * for it. */
>
> - if (udevGetStringProperty(device, "ID_TYPE",
&storage->drive_type) < 0)
> - goto cleanup;
> + udevGetStringProperty(device, "ID_TYPE",
&storage->drive_type);
>
> if (!storage->drive_type ||
(and here, but with different logic)
> STREQ(def->caps->data.storage.drive_type, "generic")) {
> @@ -1010,9 +998,7 @@ static int
> udevProcessSCSIGeneric(struct udev_device *dev,
> virNodeDeviceDef *def)
> {
> - if (udevGetStringProperty(dev, "DEVNAME",
&def->caps->data.sg.path) < 0 ||
> - !def->caps->data.sg.path)
... but here you've removed it.
Okay, according to the manpage of 'udev_device_get_property_value' [1]
which is called from 'udevGetDeviceProperty' which is in turn called
from 'udevGetStringProperty' 'udev_device_get_property_value' can return
NULL on failure. g_strdup will faithfully duplicate the NULL,
thus this is changing the logic of the return value handling without
justification. Please keep the check in place or justify why it's okay.
[1]
On success, udev_device_get_property_value() and
udev_device_get_sysattr_value() return a pointer to a constant string of
the requested value. On error, NULL is returned. Attributes that may
contain NUL bytes should not be retrieved with
udev_device_get_sysattr_value(); instead, read them directly from the
files within the device's syspath.