by Uwe Kleine-König

[permalink] [raw]

Subject: Re: [PATCH 4/8] input: Remove OOM message after input_allocate_device

On Thu, Oct 24, 2013 at 11:48:48AM -0700, Joe Perches wrote:
> On Thu, 2013-10-24 at 20:46 +0200, Uwe Kleine-K?nig wrote:
> > On Thu, Oct 24, 2013 at 11:43:38AM -0700, Joe Perches wrote:
> []
> > > Any k.alloc without __GFP_NOWARN does a generic OOM message
> > > and a dump_stack() so there could already be 2 messages anyway.
> > Then mention that in the commit log if you still want this patch?!
>
> How do you think this commit message should be changed:
>
> Emitting an OOM message isn't necessary after input_allocate_device
> as there's a generic OOM and a dump_stack already done.
after input_allocate_device and/or k.alloc ...

Best regards
Uwe

--
Pengutronix e.K. | Uwe Kleine-K?nig |
Industrial Linux Solutions | http://www.pengutronix.de/ |

2013-10-24 19:20:33

by Joe Perches

[permalink] [raw]

Subject: Re: [PATCH 4/8] input: Remove OOM message after input_allocate_device

On Thu, 2013-10-24 at 12:10 -0700, Dmitry Torokhov wrote:
> On Thu, Oct 24, 2013 at 11:45:39AM -0700, Joe Perches wrote:
> > On Thu, 2013-10-24 at 11:37 -0700, Dmitry Torokhov wrote:
> > > Hi Joe,
> > >
> > > On Wed, Oct 23, 2013 at 12:14:50PM -0700, Joe Perches wrote:
> > > > Emitting an OOM message isn't necessary after input_allocate_device
> > > > as there's a generic OOM and a dump_stack already done.
> > >
> > > No, please don't. The kzalloc may get changed in the future to not dump
> > > stack (that was added originally because not everyone was handling OOM
> > > properly, right?), input core might get changed to use something else
> > > than kzalloc, etc, etc.
> > >
> > > The majority of errors use dev_err so we also get idea what device
> > > failed (if there are several), and more.
> >
> > I think that's not valuable as input_allocate_device already has
> > dozens of locations that don't emit a specific OOM and centralizing
> > the location for any generic message would work anyway.
>
> Not having diagnostic messages in some of the drivers is hardly a
> justification to remove them from everywhere else.

But standardization is.

If a diagnostic message is really desired
without the already existing dump_stack(),
it's a small matter of adding something like:

drivers/input/input.c | 26 +++++++++++++++-----------
1 file changed, 15 insertions(+), 11 deletions(-)

diff --git a/drivers/input/input.c b/drivers/input/input.c
index c044699..98570c2 100644
--- a/drivers/input/input.c
+++ b/drivers/input/input.c
@@ -1736,19 +1736,23 @@ struct input_dev *input_allocate_device(void)
{
struct input_dev *dev;

- dev = kzalloc(sizeof(struct input_dev), GFP_KERNEL);
- if (dev) {
- dev->dev.type = &input_dev_type;
- dev->dev.class = &input_class;
- device_initialize(&dev->dev);
- mutex_init(&dev->mutex);
- spin_lock_init(&dev->event_lock);
- INIT_LIST_HEAD(&dev->h_list);
- INIT_LIST_HEAD(&dev->node);
-
- __module_get(THIS_MODULE);
+ dev = kzalloc(sizeof(struct input_dev), GFP_KERNEL | __GFP_NOWARN);
+ if (!dev) {
+ pr_err("%pf: input_allocate_device failed\n",
+ __builtin_return_address(1));
+ return NULL;
}

+ dev->dev.type = &input_dev_type;
+ dev->dev.class = &input_class;
+ device_initialize(&dev->dev);
+ mutex_init(&dev->mutex);
+ spin_lock_init(&dev->event_lock);
+ INIT_LIST_HEAD(&dev->h_list);
+ INIT_LIST_HEAD(&dev->node);
+
+ __module_get(THIS_MODULE);
+
return dev;
}
EXPORT_SYMBOL(input_allocate_device);