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

[John Baldwin]

On Monday, March 30, 2015 06:26:25 PM Randall Stewart wrote:
> 
> On Mar 30, 2015, at 9:16 AM, John Baldwin <jhb at freebsd.org> wrote:
> 
> > On Saturday, March 28, 2015 12:50:24 PM Randall Stewart wrote:
> >> Author: rrs
> >> Date: Sat Mar 28 12:50:24 2015
> >> New Revision: 280785
> >> URL: https://svnweb.freebsd.org/changeset/base/280785
> >> 
> >> Log:
> >>  Change the callout to supply -1 to indicate we are not changing
> >>  CPU, also add protection against invalid CPU's as well as
> >>  split c_flags and c_iflags so that if a user plays with the active
> >>  flag (the one expected to be played with by callers in MPSAFE) without
> >>  a lock, it won't adversely affect the callout system by causing a corrupt
> >>  list. This also means that all callers need to use the macros and *not*
> >>  play with the falgs directly (like netgraph used to).
> >> 
> >>  Differential Revision: htts://reviews.freebsd.org/D1894
> >>  Reviewed by: .. timed out but looked at by jhb, imp, adrian hselasky
> >>               tested by hiren and netflix.
> >>  Sponsored by:	Netflix Inc.
> > 
> > Please use NOCPU rather than -1 directly for the CPU field when not
> > moving a callout.
> > 
> 
> John:
> 
> I have made *all* of your suggested changes, adopting the comments and
> moving migration to kern_timeout.c.. thanks..
> 
> Now as to the 
> 
> -1 -> NOCPU
> 
> This is like pulling on a string on your sweater.. the only sensible solution that
> I could come up with after chatting with Lawrence is to add 
> #include <sys/proc.h>
> to everyone that uses the callout.h and does not have it already… (putting it into callout.h does not work) .. sigh..
> 
> Now for this cosmetic change I end up with the following changes (and as yet I have
> not built LINT or universe so there may be more).. I have spent about 2 hours on this
> so far and I can at least build a kernel with the change for amd64 :-0
> 
> Here is what has to change, do you really think that this is worth it?
> 
> Note I did not look into moving NOCPU in proc.h it says it means no CPU is present
> which is sort of the meaning we want.. I am not sure if the define could be moved .. but
> that too may be yet another string...
> 
> Is this worth it, or do you have another idea on how best to do this???

Ugh. :(  I guess leave it at -1 for now.  Another alternative would be to
move NOCPU to <sys/param.h> (at least for the kernel) if bde@ would allow it.
I don't think changing all those files is appropriate.

(Also, <sys/proc.h> seems like an odd place for NOCPU now, it should really be
in <sys/smp.h> if not param.h.  I understand why it was first added in proc.h,
but it is now used in many more places than just td_oncpu.)

-- 
John Baldwin