Re: Thread flags modified without set_thread_flag() (non atomically)

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 28 Feb 2007 22:23:41 -0800 (PST) David Miller <davem@davemloft.net> wrote:

> > That all looks rather deliberate.
> > 
> > > powerpc/kernel/process.c:               t->flags ^= (_TIF_ABI_PENDING | 
> > > _TIF_32BIT);
> > >
> > > ia64/kernel/mca.c:      ti->flags = _TIF_MCA_INIT;
> > >
> > > avr32/kernel/ptrace.c:          ti->flags |= _TIF_BREAKPOINT;
> > 
> > No, I don't immediately see anything in the flush_old_exec() code path
> > which tells us that nobody else can look up this thread_info (or be holding
> > a ref to it) in this context.
> 
> Provide the counter example, what other threads of control can modify
> relevant flags while a thread is exec()'ing?  It's essentially frozen
> outside of these code paths, otherwise we wouldn't be able to make all
> of these modifications to the task state.

As Mathieu points out, resched_task(), perhaps.  It can be called from interrupt.