svn commit: r227394 - in head/sys: amd64/amd64 i386/i386

[Kostik Belousov]

On Thu, Nov 10, 2011 at 07:23:23AM +0100, Pawel Jakub Dawidek wrote:
> On Wed, Nov 09, 2011 at 07:31:28PM +0200, Kostik Belousov wrote:
> > On Wed, Nov 09, 2011 at 05:25:43PM +0000, Konstantin Belousov wrote:
> > > Author: kib
> > > Date: Wed Nov  9 17:25:43 2011
> > > New Revision: 227394
> > > URL: http://svn.freebsd.org/changeset/base/227394
> > > 
> > > Log:
> > >   Stopped process may legitimately have some threads sleeping and not
> > >   suspended, if the sleep is uninterruptible.
> > Even more, stopped process might have some threads still running in the
> > kernel mode, or inhibited due to wait on blockable locks. I was unable
> > to design an expression that would assert that such thread will be stopped
> > at the kernel->user boundary.
> > 
> > The assertion itself is useful and catched several bugs, but theoretically
> > can cause false positives. If any report of the fired assert for kernel-mode
> > thread is provided, I will remove the assertions.
> 
> If in five years some change will make it to fire and you won't be
> around, nobody will remember this e-mail of yours and someone will spend
> time on looking for a bug that doesn't exist.
> 
> If we know it can cause false positives, I'd be in favour of removing it
> or changing it into a warning or at the very least moving this
> information into commit log.
Yes, I do agree that removing the assertion is the most straightforward
solution, but on the assertion already catched more real bugs then
caused headaches (my headache, BTW). Either I change the third or condition
to plain P_SHOULDSTOP(), or create a way to assert exactly what I need.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/svn-src-head/attachments/20111110/6bd8fb80/attachment-0001.pgp