Re: [PATCH] relay: Add global mode support for buffer-only channels

[Goel, Akash]

On 7/4/2016 1:30 PM, Chris Wilson wrote:
> On Sun, Jul 03, 2016 at 09:45:27PM +0530, akash.goel@xxxxxxxxx wrote:
> > From: Akash Goel <akash.goel@xxxxxxxxx>
> >
> > The following patch added support to use channels with no associated files.
> > 	relay: add buffer-only channels; useful for early logging
> > This is useful when the exact location of relay file is not known or the
> > the parent directory of relay file is not available, while creating the
> > channel and the logging has to start right from the boot.
> >
> > But there was no provision to use global mode with buffer-only channels,
> > which is added by this patch, without modifying the interface where initially
> > there will be a dummy invocation of create_buf_file callback through which
> > kernel client can convey the need of a global buffer.
> >
> > For the use case where drivers/kernel clients want a simple interface for the
> > userspace, which enables them to capture data/logs from relay file in order &
> > without any post processing, support of Global buffer mode is warranted.
> >
> > Cc: Eduard - Gabriel Munteanu <eduard.munteanu@xxxxxxxxxxx>
> > Cc: Tom Zanussi <tzanussi@xxxxxxxxx>
> > Cc: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
> > Signed-off-by: Akash Goel <akash.goel@xxxxxxxxx>
> > ---
> >  kernel/relay.c | 32 ++++++++++++++++++++++++++++++++
> >  1 file changed, 32 insertions(+)
> >
> > diff --git a/kernel/relay.c b/kernel/relay.c
> > index 04d7cf3..b267384 100644
> > --- a/kernel/relay.c
> > +++ b/kernel/relay.c
> > @@ -451,6 +451,16 @@ static struct rchan_buf *relay_open_buf(struct rchan *chan, unsigned int cpu)
> >  		if (!dentry)
> >  			goto free_buf;
> >  		relay_set_buf_dentry(buf, dentry);
> > +	} else {
>
> The trailing block can be replaced with:
>
> 		/* Only retrieve global info, nothing more, nothing less */

Thanks, will do like this.
> > +		dentry = chan->cb->create_buf_file(NULL, NULL,
> > +					S_IRUSR, buf,
> > +					&chan->is_global);
>
> Watch alignment, align to the opening bracket.
>
> > +		if (WARN_ON(dentry))
> > +			goto free_buf;
> >  	}
> >
> >   	buf->cpu = cpu;
> > @@ -666,6 +676,27 @@ int relay_late_setup_files(struct rchan *chan,
> >  	}
> >  	chan->has_base_filename = 1;
> >  	chan->parent = parent;
> > +
> > +	if (chan->is_global) {
> > +		if (unlikely(!chan->buf[0])) {
>
> if (WARN_ON_ONCE(!chan->buf[0])) {
>
> > +			WARN_ONCE(1, KERN_ERR "CPU 0 has no buffer!\n");
>
> The WARN includes the unlikely and the message here isn't any more
> informative than the line itself, plus has the erroneous KERN_ERR
Thanks will combine the WARN and if check.

> > +			mutex_unlock(&relay_channels_mutex);
> > +			return -EINVAL;
> > +                }
> > +
> > +		dentry = relay_create_buf_file(chan, chan->buf[0], 0);
> > +
> > +		if (unlikely(!dentry))
> > +			err = -EINVAL;
> > +		else if (WARN_ON(!chan->is_global))
>
> ?

Though have checked 'is_global' value before, but can't rule out the
clients toggling its value (inadvertently or maliciously).

Actually create_buf_file() callback will be invoked again from the
relay_create_buf_file() and pointer of 'chan->is_global' is passed in
that.
So to be on safer side, thought to validate the 'chan->is_global' value 
after the call to relay_create_buf_file().

> > +			err = -EINVAL;
> > +		else
> > +			relay_set_buf_dentry(chan->buf[0], dentry);
> > +
> > +		mutex_unlock(&relay_channels_mutex);
> > +		return err;
>
> This could be written:
Thanks this is much better.


Best Regards
Akash

> if (chan->is_global) {
> 	err = -EINVAL;
> 	if (!WARN_ON_ONCE(!chan->buf[0])) {
> 		dentry = relay_create_buf_file(chan, chan->buf[0], 0))
> 		if (dentry) {
> 			relay_set_buf_dentry(chan->buf[0], dentry);
> 			err = 0;
> 		}
> 	}
> 	mutex_unlock();
> 	return err;
> }
>
> for a single exit/unlock path in this block.
> -Chris