On 12/15/2014 10:32 AM, Dave Airlie wrote:
> On 15 December 2014 at 17:59, Daniel Vetter <daniel at ffwll.ch> wrote:
>> On Sun, Dec 14, 2014 at 03:10:17PM +0100, Christian König wrote:
>>> Am 14.12.2014 um 14:35 schrieb Oded Gabbay:
>>>> When an ioctl function returns -EAGAIN, don't print error in kfd_ioctl()
>>>
>>> You most likely want to handle -ERESTARTSYS the same way.
>>
>> Please just reuse drmIoctl or at least copy it perfectly. We've had too
>> many tears about ioctl restarting going badly wrong. Also make sure you
>> never do a raw ioctl call anywhere for amdkfd. Adding Dave.
>
> Also please don't make a user triggerable printk.
>
> If the user can throw crap at the ioctl and get msgs in dmesg,
> then its annoying as hell.
>
> Copy the drm.debug stuff and code as well, and for userspace,
> yes do what Daniel says and use drmIoctl wrapper or something like
> that, though Daniel I believe one of the main consumers on i915
> insists on opencoding his ioctls.
>
> Dave.
>
Hi Dave, Daniel
I just sent a patch-set that copies the drm_ioctl() handling to kfd_ioctl(), as
you requested.
All error prints have been converted to debug prints.
This is the first part and I'm now going to change the userspace as well.
Oded
