Re: [patch] aio: remove aio-max-nr and instead use the memlock rlimit to limit the number of pages pinned for the aio completion ring

[Eric Dumazet <dada1@cosmosbay.com>]

Benjamin LaHaise a écrit :
> On Thu, Mar 12, 2009 at 03:39:51AM +0100, Eric Dumazet wrote:
>> Assuming that __fput() cannot be called from interrupt context.
>> -> fput() should not be called from interrupt context as well.
> 
> It doesn't get called in irq context.  The atomic dec is done in irq 
> context, and if that is the last user of the file descriptor, it's 
> placed on a work queue and then released in process context.
> 

really_put_req() itself can be called from interrupt context, unless I am wrong.
(its 4:24 am here ;) )

        if (unlikely(atomic_long_dec_and_test(&req->ki_filp->f_count))) {
                get_ioctx(ctx);
                spin_lock(&fput_lock);
                list_add(&req->ki_list, &fput_head);
                spin_unlock(&fput_lock);
                queue_work(aio_wq, &fput_work);
        } else
                really_put_req(ctx, req); /* from interrupt context */



static inline void really_put_req(struct kioctx *ctx, struct kiocb *req)
{
        assert_spin_locked(&ctx->ctx_lock);

        if (!IS_ERR(req->ki_eventfd))
                fput(req->ki_eventfd);   /* BANG : can be called from interrupt context */
        if (req->ki_dtor)
                req->ki_dtor(req);
        if (req->ki_iovec != &req->ki_inline_vec)
                kfree(req->ki_iovec);
        kmem_cache_free(kiocb_cachep, req);
        ctx->reqs_active--;

        if (unlikely(!ctx->reqs_active && ctx->dead))
                wake_up(&ctx->wait);
}

Thank you