netdev - Re: [net V2 05/15] net/mlx5: Add retry mechanism to the command entry index allocation

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <cda8be83538b9cac786b12ea8324248e88f62be1.camel@kernel.org>
Date:   Fri, 02 Oct 2020 10:03:14 -0700
From:   Saeed Mahameed <saeed@...nel.org>
To:     Jakub Kicinski <kuba@...nel.org>
Cc:     "David S. Miller" <davem@...emloft.net>, netdev@...r.kernel.org,
        Eran Ben Elisha <eranbe@...dia.com>,
        Moshe Shemesh <moshe@...dia.com>
Subject: Re: [net V2 05/15] net/mlx5: Add retry mechanism to the command
 entry index allocation

On Thu, 2020-10-01 at 16:23 -0700, Jakub Kicinski wrote:
> On Thu,  1 Oct 2020 12:52:37 -0700 saeed@...nel.org wrote:
> > +static int cmd_alloc_index_retry(struct mlx5_cmd *cmd)
> > +{
> > +	unsigned long alloc_end = jiffies + msecs_to_jiffies(1000);
> > +	int idx;
> > +
> > +retry:
> > +	idx = cmd_alloc_index(cmd);
> > +	if (idx < 0 && time_before(jiffies, alloc_end)) {
> > +		/* Index allocation can fail on heavy load of commands.
> > This is a temporary
> > +		 * situation as the current command already holds the
> > semaphore, meaning that
> > +		 * another command completion is being handled and it
> > is expected to release
> > +		 * the entry index soon.
> > +		 */
> > +		cond_resched();
> > +		goto retry;
> > +	}
> > +	return idx;
> > +}
> 
> This looks excessive. At least add some cpu_relax(), or udelay()?

cpu_relax() should also work fine, it is just that we have 100%
certainty that the allocation will success real soon.