linux-kernel - Re: [PATCH 1/5] rcusync: introduce struct rcu_sync

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: <20131008091339.GN3081@twins.programming.kicks-ass.net>
Date:	Tue, 8 Oct 2013 11:13:39 +0200
From:	Peter Zijlstra <peterz@...radead.org>
To:	Oleg Nesterov <oleg@...hat.com>
Cc:	Linus Torvalds <torvalds@...ux-foundation.org>,
	Steven Rostedt <rostedt@...dmis.org>,
	Paul McKenney <paulmck@...ux.vnet.ibm.com>,
	Mel Gorman <mgorman@...e.de>, Rik van Riel <riel@...hat.com>,
	Srikar Dronamraju <srikar@...ux.vnet.ibm.com>,
	Ingo Molnar <mingo@...nel.org>,
	Andrea Arcangeli <aarcange@...hat.com>,
	Johannes Weiner <hannes@...xchg.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/5] rcusync: introduce struct rcu_sync_ops

On Sat, Oct 05, 2013 at 07:17:46PM +0200, Oleg Nesterov wrote:
> To me the annoying part is that this patch exports rcu_sync_ops*.
> We can add "enum rcu_sync_type rcu_sync_struct->gp_type" instead,
> see the patch below. But then the next "add the CONFIG_PROVE_RCU
> checks" needs to uninline rcu_sync_is_idle() for CONFIG_PROVE_RCU.
> 
> Or do you still think we should do something else?

> @@ -44,7 +45,7 @@ void rcu_sync_enter(struct rcu_sync_struct *rss)
>  	BUG_ON(need_wait && need_sync);
>  
>  	if (need_sync) {
> -		rss->sync();
> +		gp_ops[rss->gp_type].sync();
>  		rss->gp_state = GP_PASSED;
>  		wake_up_all(&rss->gp_wait);
>  	} else if (need_wait) {
> @@ -81,7 +82,7 @@ static void rcu_sync_func(struct rcu_head *rcu)
>  		 * to catch a later GP.
>  		 */
>  		rss->cb_state = CB_PENDING;
> -		rss->call(&rss->cb_head, rcu_sync_func);
> +		gp_ops[rss->gp_type].call(&rss->cb_head, rcu_sync_func);
>  	} else {
>  		/*
>  		 * We're at least a GP after rcu_sync_exit(); eveybody will now
> @@ -99,7 +100,7 @@ void rcu_sync_exit(struct rcu_sync_struct *rss)
>  	if (!--rss->gp_count) {
>  		if (rss->cb_state == CB_IDLE) {
>  			rss->cb_state = CB_PENDING;
> -			rss->call(&rss->cb_head, rcu_sync_func);
> +			gp_ops[rss->gp_type].call(&rss->cb_head, rcu_sync_func);
>  		} else if (rss->cb_state == CB_PENDING) {
>  			rss->cb_state = CB_REPLAY;
>  		}
> 

I think Linus meant to have rcu_sync_{enter,exit} as inlines with a
const enum argument for the gp_type.

That said; yes that will generate better code, but also more code, and
like Steven already argued performance isn't really an issue here since
we're going to potentially sleep for a rather long time.


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/