Thread (25 messages) 25 messages, 11 authors, 2014-10-24

Re: [PATCH] staging: android: binder: move to the "real" part of the kernel

From: Pavel Machek <hidden>
Date: 2014-10-21 20:05:29
Also in: lkml

On Tue 2014-10-21 16:12:24, Arnd Bergmann wrote:
On Tuesday 21 October 2014 12:36:22 Pavel Machek wrote:
quoted
On Fri 2014-10-17 01:12:21, Greg Kroah-Hartman wrote:
quoted
On Thu, Oct 16, 2014 at 10:09:04AM -0700, John Stultz wrote:
quoted
On Thu, Oct 16, 2014 at 5:47 AM, Greg Kroah-Hartman
[off-list ref] wrote:
quoted
From: Greg Kroah-Hartman <redacted>
quoted
quoted
Are the Android guys comfortable with the ABI stability rules they'll
now face?
Just because something is in staging, doesn't mean you don't have to
follow the same ABI stability rules as the rest of the kernel.  If a
change had happened to this code that broke userspace in the past, I
would have reverted it.  So this should not be anything different from
what has been happening inthe past.
Actually, there's big difference.

If Al Viro changes core filesystem in a way that breaks
staging/binder, binder is broken, and if it can't be fixed... well it
can't be fixed.

If Al Viro changes core filesystem in a way that breaks
drivers/binder, Al's change is going to be reverted.
One might have argued that we'd have to do that already, but the reasons
for doing that with binder in the main kernel are certainly stronger.
quoted
It is really hard to review without API documentation. Normally, API
documentation is required for stuff like this.

For example: does it add new files in /proc?

Given that it is stable, can we get rid of binder_debug() and
especially BINDER_DEBUG_ENTRY stuff?
Good point. We require documentation for every single sysfs attribute
that gets added to a driver (some escape the review, but that doesn't
change the rule), so we should not make an exception for a new procfs
file here.
Actually, it looked like it is debugfs file. Code was messy enough
that I was not sure.
quoted
Could binder_transcation() be split to smaller functions according to
CodingStyle? 17 goto targets at the end of function are not exactly
easy to read.

ginder_thread_read/write also needs splitting.
Yes, in principle, but this is still a detail that would mainly serve
to simplify review. The problem is more the lack of review and
documentation of the API.
Yes, the problem is that code is impossible to review without API
documentation.

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help