Kristian Høgsberg wrote:
> On Mon, Aug 17, 2009 at 10:28 AM, Thomas Hellstrom<thellst...@vmware.com>
> wrote:
>
>> The device directory will be the base directory of the
>> sysfs representation of other ttm subsystems.
>>
>> Signed-off-by: Thomas Hellstrom <thellst...@vmware.com>
>> ---
>> drivers/gpu/drm/ttm/ttm_module.c | 56
>> +++++++++++++++++++++++++++++++++++++-
>> 1 files changed, 55 insertions(+), 1 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/ttm/ttm_module.c
>> b/drivers/gpu/drm/ttm/ttm_module.c
>> index 59ce819..4197467 100644
>> --- a/drivers/gpu/drm/ttm/ttm_module.c
>> +++ b/drivers/gpu/drm/ttm/ttm_module.c
>> @@ -29,16 +29,70 @@
>> * Jerome Glisse
>> */
>> #include <linux/module.h>
>> -#include <ttm/ttm_module.h>
>> +#include <linux/device.h>
>> +#include <linux/sched.h>
>> +#include "ttm/ttm_module.h"
>> +#include "drm_sysfs.h"
>> +
>> +atomic_t device_released;
>> +
>> +static struct device_type ttm_drm_class_type = {
>> + .name = "ttm",
>> + /**
>> + * Add pm ops here.
>> + */
>> +};
>> +
>> +static void ttm_drm_class_device_release(struct device *dev)
>> +{
>> + atomic_set(&device_released, 1);
>> +}
>> +
>> +static struct device ttm_drm_class_device = {
>> + .type = &ttm_drm_class_type,
>> + .release = &ttm_drm_class_device_release
>> +};
>> +
>> +struct kobject *ttm_get_kobj(void)
>> +{
>> + struct kobject *kobj = &ttm_drm_class_device.kobj;
>> + BUG_ON(kobj == NULL);
>> + return kobj;
>> +}
>>
>> static int __init ttm_init(void)
>> {
>> + int ret;
>> +
>> + ret = dev_set_name(&ttm_drm_class_device, "ttm");
>> + if (unlikely(ret != 0))
>> + return ret;
>> +
>> ttm_global_init();
>> +
>> + atomic_set(&device_released, 0);
>> + ret = drm_class_device_register(&ttm_drm_class_device);
>> + if (unlikely(ret != 0))
>> + goto out_no_dev_reg;
>> +
>> return 0;
>> +out_no_dev_reg:
>> + atomic_set(&device_released, 1);
>> + ttm_global_release();
>> + return ret;
>> }
>>
>> static void __exit ttm_exit(void)
>> {
>> + DECLARE_WAIT_QUEUE_HEAD_ONSTACK(exit_q);
>> + drm_class_device_unregister(&ttm_drm_class_device);
>> +
>> + /**
>> + * Refuse to unload until the TTM device is released.
>> + * Not sure this is 100% needed.
>> + */
>> +
>> + wait_event(exit_q, atomic_read(&device_released) == 1);
>>
>
> How is this supposed to work? You need to make the wait queue public
> so that the drm_class_device_release function can wake up waiters on
> the queue. The atomic is just a plain integer variable, decreasing it
> doesn't wake up anything.
>
> cheers,
> Kristian
>
>
Ouch.
You're right. Will fix.
/Thomas
>> ttm_global_release();
>> }
>>
>
> ------------------------------------------------------------------------------
> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
> trial. Simplify your report design, integration and deployment - and focus on
> what you do best, core application coding. Discover what's new with
> Crystal Reports now. http://p.sf.net/sfu/bobj-july
> --
> _______________________________________________
> Dri-devel mailing list
> Dri-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/dri-devel
>
------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july
--
_______________________________________________
Dri-devel mailing list
Dri-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dri-devel
