svn commit: r280785 - in head/sys: kern netgraph/atm/sscop netgraph/atm/uni sys

[Randall Stewart]

John:

As I just said, Warner and I feel we can get by with making the int -> short/short so
we preserver the KPI and at the same time achieve the objective ..

My big concern was no intel platforms but Warner gave me a green light there ;-)

R
On Mar 30, 2015, at 10:10 AM, John Baldwin <jhb at freebsd.org> wrote:

> On Saturday, March 28, 2015 01:48:49 PM Davide Italiano wrote:
>> On Sat, Mar 28, 2015 at 12:04 PM, Randall Stewart <rrs at netflix.com> wrote:
>>> Hmm does the KPI include the size of the callout structure (which is
>>> private)?
>>> 
>> 
>> It's KBI, not KPI.
>> 
>>> If so I suppose we could change the c_flags/c_iflags to
>>> ushort …
>>> 
>> 
>> I'm not familiar with your changes enough to suggest/propose a fix.
>> As I previously mentioned, this looks suspicious KBI-wise to me, and I
>> reported to you.
> 
> The KBI issue makes it much harder to MFC.  In particular, even though
> you could move 'c_iflags' to the end to fix most things (though not
> callout_pending()), many drivers have 'struct callout foo' inside their
> softc, etc. and this would break existing modules.
> 
> OTOH, the uglier hack of adding locking in callout_deactivate() would
> preserve the KBI and be safe to merge.
> 
> -- 
> John Baldwin

--------
Randall Stewart
rrs at netflix.com
803-317-4952