lkml.org 
[lkml]   [2012]   [Jul]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: [PATCH v4 RESEND] mm: cma: add a simple kernel module as the helper to test CMA
Hi Arnd,

2012/7/5 Arnd Bergmann <arnd@arndb.de>:
>
> I think it should be in mm/cma-test.c, along with kmemleak-test.c. It would be nice
> if you could add some code that just runs at boot time (or when the module is
> loaded) and allocates and frees memory using CMA.
>
>> +static struct device *cma_dev;
>
> Do you actually need this device? It's not connected to hardware so
> it doesn't actually do DMA and we might as well pass a NULL pointer
> into dma_alloc_*().
>
>> +static const struct file_operations cma_test_fops = {
>> + .owner = THIS_MODULE,
>> + .read = cma_test_read,
>> + .write = cma_test_write,
>> +};
>> +
>> +static struct miscdevice cma_test_misc = {
>> + .name = "cma_test",
>> + .fops = &cma_test_fops,
>> +};
>> +
>> +static int __init cma_test_init(void)
>> +{
>> + int ret = misc_register(&cma_test_misc);
>
> A better place for this is really debugfs. The driver is not meant as a stable
> kernel interface that applications can rely on, it's purely a debugging help.
>
> Just make this
>
> ret = debugfs_create_file("cma-test", 0600, NULL, NULL, &cma_test_fops);
>

we did have some discussions about this at:
http://www.spinics.net/lists/arm-kernel/msg163548.html

i'd say both are acceptable to me.

> Arnd
>

-barry


\
 
 \ /
  Last update: 2012-07-05 11:41    [W:0.053 / U:0.980 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site