On Mon, 08 Aug 2016 10:37:21 +0200, Takashi Sakamoto wrote:
On Aug 8 2016 16:04, Takashi Iwai wrote:
On Sun, 07 Aug 2016 11:48:47 +0200, Takashi Sakamoto wrote:
In former commit, actual operations of each ioctl command get argument in kernel space. Copying from/to user space is performed outside of the function.
This commit optimizes to the new design.
Signed-off-by: Takashi Sakamoto o-takashi@sakamocchi.jp
While it's OK to split to small patches if you prefer, you don't have to do so. Basically all the rest are doing the same thing (strip copy_*_user() and replace to the pointer accesses), and it's rather boring to read repeated mails.
The mail server of alsa-project.org has a limitation of the size of one message. It shrinks developers to split commit to a small parts. Not from my taste.
Yeah, that's sometimes annoying, indeed. Jaroslav, could you raise the bar to allow larger patches, or drop this restriction?
Meanwhile, the current limitation is reasonable in most cases. Patches with thousands of lines make review more difficult. If your patch exceeds the size, it already indicates that it may be potentially a bad patch. The logic is something similar like 80 chars limitation we have in general.
Sometimes a big patch is unavoidable (e.g. writing a new driver from scratch). Or sometimes a bigger patch is acceptable (e.g. when changes are machinery, done purely by a script). But if you wrote some patch by hands and it exceeds the size, you should doubt it at first.
Takashi