| Message ID | 1512786861-1014-2-git-send-email-mario.limonciello@dell.com (mailing list archive) |
|---|---|
| State | Changes Requested, archived |
| Delegated to: | Darren Hart |
| Headers | show |
--Andy > On Dec 8, 2017, at 6:34 PM, Mario Limonciello <mario.limonciello@dell.com> wrote: > > It's possible for the same GUID to show up on as system twice. > This means using solely the GUID for identify the file will not > be sufficient. Isn't the file already in a per-bus directory?
>> On Dec 8, 2017, at 6:34 PM, Mario Limonciello <mario.limonciello@dell.com> wrote: >> >> It's possible for the same GUID to show up on as system twice. >> This means using solely the GUID for identify the file will not >> be sufficient. > >Isn't the file already in a per-bus directory? Yep, but the symlink created in /sys/bus/wmi/devices isn't. That's where the kernel complains about duplicate sysfs attributes. It's not exactly a pretty path I submitted, but it does avoid those collisions. Example (with this in place from /sys/bus/wmi/devices): lrwxrwxrwx 1 root root 0 Dec 8 21:39 PNP0C14:04-70FE8229-D03B-4214-A1C6-1F884B1A892A -> ../../../devices/platform/PNP0C14:04/wmi_bus/wmi_bus-PNP0C14:04/PNP0C14:04-70FE8229-D03B-4214-A1C6-1F884B1A892A
On Fri, Dec 8, 2017 at 7:41 PM, <Mario.Limonciello@dell.com> wrote: >>> On Dec 8, 2017, at 6:34 PM, Mario Limonciello <mario.limonciello@dell.com> wrote: >>> >>> It's possible for the same GUID to show up on as system twice. >>> This means using solely the GUID for identify the file will not >>> be sufficient. >> >>Isn't the file already in a per-bus directory? > > Yep, but the symlink created in /sys/bus/wmi/devices isn't. > That's where the kernel complains about duplicate sysfs > attributes. > > It's not exactly a pretty path I submitted, but it does avoid > those collisions. > > Example (with this in place from /sys/bus/wmi/devices): > lrwxrwxrwx 1 root root 0 Dec 8 21:39 PNP0C14:04-70FE8229-D03B-4214-A1C6-1F884B1A892A -> ../../../devices/platform/PNP0C14:04/wmi_bus/wmi_bus-PNP0C14:04/PNP0C14:04-70FE8229-D03B-4214-A1C6-1F884B1A892A Right, I saw that in the cover letter right after sending this. Greg, is there a cleaner way to deal with this? There are two instances of the same bus type, each of which would like to have a device called "70FE8229-D03B-4214-A1C6-1F884B1A892A". Can we somehow rename the symlinks without renaming the device, or are we just supposed to prefix the device name like Mario is doing here?
On Sat, Dec 09, 2017 at 01:32:29PM -0800, Andy Lutomirski wrote: > On Fri, Dec 8, 2017 at 7:41 PM, <Mario.Limonciello@dell.com> wrote: > >>> On Dec 8, 2017, at 6:34 PM, Mario Limonciello <mario.limonciello@dell.com> wrote: > >>> > >>> It's possible for the same GUID to show up on as system twice. > >>> This means using solely the GUID for identify the file will not > >>> be sufficient. > >> > >>Isn't the file already in a per-bus directory? > > > > Yep, but the symlink created in /sys/bus/wmi/devices isn't. > > That's where the kernel complains about duplicate sysfs > > attributes. > > > > It's not exactly a pretty path I submitted, but it does avoid > > those collisions. > > > > Example (with this in place from /sys/bus/wmi/devices): > > lrwxrwxrwx 1 root root 0 Dec 8 21:39 PNP0C14:04-70FE8229-D03B-4214-A1C6-1F884B1A892A -> ../../../devices/platform/PNP0C14:04/wmi_bus/wmi_bus-PNP0C14:04/PNP0C14:04-70FE8229-D03B-4214-A1C6-1F884B1A892A > > Right, I saw that in the cover letter right after sending this. > > Greg, is there a cleaner way to deal with this? There are two > instances of the same bus type, each of which would like to have a > device called "70FE8229-D03B-4214-A1C6-1F884B1A892A". Can we somehow > rename the symlinks without renaming the device, or are we just > supposed to prefix the device name like Mario is doing here? > Mario, This one has been lingering at the bottom of my patch stack for a while now - sorry for not poking on this earlier. Is this still something you have a need for? If so, would you resend and add Greg to Cc - specifically directing Andy's question above to him?
On Sat, Dec 09, 2017 at 01:32:29PM -0800, Andy Lutomirski wrote: > On Fri, Dec 8, 2017 at 7:41 PM, <Mario.Limonciello@dell.com> wrote: > >>> On Dec 8, 2017, at 6:34 PM, Mario Limonciello <mario.limonciello@dell.com> wrote: > >>> > >>> It's possible for the same GUID to show up on as system twice. > >>> This means using solely the GUID for identify the file will not > >>> be sufficient. > >> > >>Isn't the file already in a per-bus directory? > > > > Yep, but the symlink created in /sys/bus/wmi/devices isn't. > > That's where the kernel complains about duplicate sysfs > > attributes. > > > > It's not exactly a pretty path I submitted, but it does avoid > > those collisions. > > > > Example (with this in place from /sys/bus/wmi/devices): > > lrwxrwxrwx 1 root root 0 Dec 8 21:39 PNP0C14:04-70FE8229-D03B-4214-A1C6-1F884B1A892A -> ../../../devices/platform/PNP0C14:04/wmi_bus/wmi_bus-PNP0C14:04/PNP0C14:04-70FE8229-D03B-4214-A1C6-1F884B1A892A > > Right, I saw that in the cover letter right after sending this. > > Greg, is there a cleaner way to deal with this? There are two > instances of the same bus type, each of which would like to have a > device called "70FE8229-D03B-4214-A1C6-1F884B1A892A". Can we somehow > rename the symlinks without renaming the device, or are we just > supposed to prefix the device name like Mario is doing here? > Mario, if this is still a concern (I presume it is) would you please resend the patch Cc gregkh directly?
diff --git a/drivers/platform/x86/wmi.c b/drivers/platform/x86/wmi.c index 791449a2370f..45d9010aafcf 100644 --- a/drivers/platform/x86/wmi.c +++ b/drivers/platform/x86/wmi.c @@ -1081,7 +1081,8 @@ static int wmi_create_device(struct device *wmi_bus_dev, wblock->dev.dev.bus = &wmi_bus_type; wblock->dev.dev.parent = wmi_bus_dev; - dev_set_name(&wblock->dev.dev, "%pUL", gblock->guid); + dev_set_name(&wblock->dev.dev, "%s-%pUL", + dev_name(&wblock->acpi_device->dev), gblock->guid); device_initialize(&wblock->dev.dev);
It's possible for the same GUID to show up on as system twice. This means using solely the GUID for identify the file will not be sufficient. Signed-off-by: Mario Limonciello <mario.limonciello@dell.com> --- drivers/platform/x86/wmi.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)