On 07/30/2014 07:03 PM, Boaz Harrosh wrote:
<>
>> +long bdev_direct_access(struct block_device *bdev, sector_t sector,
>> + void **addr, unsigned long *pfn, long size)
>> +{
>> + const struct block_device_operations *ops = bdev->bd_disk->fops;
>> + if (!ops->direct_access)
>> + return -EOPNOTSUPP;
>
> You need to check alignment on PAGE_SIZE since this API requires it, do
> to pfn defined to a page_number.
>
> (Unless you want to define another output-param like page_offset.
> but this exercise can be left to the caller)
>
> You also need to check against the partition boundary. so something like:
>
> + if (sector & (PAGE_SECTORS-1))
> + return -EINVAL;
> + if (unlikely(sector + size > part_nr_sects_read(bdev->bd_part)))
Off course I was wrong here size is in bytes not in sectors. Which points
out that maybe this API needs to be in sectors.
[Actually it needs to be in pages both size and offset, because of return of
pfn, but its your call.]
Anyway my code above needs to be fixed with
(size + 512 -1) / 512
Thanks
Boaz
> + return -ERANGE;
>
> Then perhaps you can remove that check from drivers
>
>> + return ops->direct_access(bdev, sector + get_start_sect(bdev), addr,
>> + pfn, size);
>> +}
>> +EXPORT_SYMBOL_GPL(bdev_direct_access);
<>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>