Message-ID: <486DB511.2090905@bull.net>
Date:	Fri, 04 Jul 2008 07:28:49 +0200
From:	Nadia Derbey <Nadia.Derbey@...l.net>
To:	Matt Helsley <matthltc@...ibm.com>
Cc:	akpm@...ux-foundation.org, linux-kernel@...r.kernel.org,
	Solofo.Ramangalahy@...l.net
Subject: Re: [PATCH 1/1] IPC - Do not use a negative value to re-enable	msgmni
 automatic recomputing

Matt Helsley wrote:
> On Thu, 2008-07-03 at 14:15 +0200, Nadia.Derbey@...l.net wrote:
> 
>>plain text document attachment (auto_msgmni_proc_file.patch)
>>[PATCH 01/01]
>>
>>This patch proposes an alternative to the "magical positive-versus-negative
>>number trick" Andrew complained about last week in
>>http://lkml.org/lkml/2008/6/24/418
>>
>>This had been introduced with the patches that scale msgmni to the amount of
>>lowmem. With these patches, msgmni has a registered notification routine
>>that recomputes msgmni value upon memory add/remove or ipc namespace creation/
>>removal.
>>
>>When msgmni is changed from user space (i.e. value written to the proc file),
>>that notification routine is unregistered, and the way to make it registered
>>back is to write a negative value into the proc file. This is the "magical
>>positive-versus-negative number trick".
>>
>>To fix this, a new proc file is introduced: /proc/sys/kernel/auto_msgmni.
>>This file acts as ON/OFF for msgmni automatic recomputing.
>>
>>With this patch, the process is the following:
>>1) kernel boots in "automatic recomputing mode"
>>   /proc/sys/kernel/msgmni contains the value that has been computed (depends
>>                           on lowmem)
>>   /proc/sys/kernel/automatic_msgmni contains "1"
>>
>>2) echo <val> > /proc/sys/kernel/msgmni
>>   . sets msg_ctlmni to <val>
>>   . de-activates automatic recomputing (i.e. if, say, some memory is added
>>     msgmni won't be recomputed anymore)
>>   . /proc/sys/kernel/automatic_msgmni now contains "0"
>>
>>3) echo "0" > /proc/sys/kernel/automatic_msgmni
>>   . de-activates msgmni automatic recomputing
>>     this has the same effect as 2) except that msg_ctlmni's value stays
>>     blocked at its current value)
>>
>>3) echo "1" > /proc/sys/kernel/automatic_msgmni
>>   . recomputes msgmni's value based on the current available memory size
>>     and number of ipc namespaces
>>   . re-activates automatic recomputing for msgmni.
>>
>>This patch applies to 2.6.26-rc5-mm3.
> 
> 
> This makes sense to me.
> 
> 
>>Signed-off-by: Nadia Derbey <Nadia.Derbey@...l.net>
>>
>>---
>> include/linux/ipc_namespace.h |    1 
>> ipc/ipc_sysctl.c              |   75 ++++++++++++++++++++++++++++++++++--------
>> ipc/ipcns_notifier.c          |   19 ++++++++--
>> 3 files changed, 78 insertions(+), 17 deletions(-)
>>
>>Index: linux-2.6.26-rc5-mm3/ipc/ipc_sysctl.c
>>===================================================================
>>--- linux-2.6.26-rc5-mm3.orig/ipc/ipc_sysctl.c	2008-06-16 09:12:57.000000000 +0200
>>+++ linux-2.6.26-rc5-mm3/ipc/ipc_sysctl.c	2008-07-03 13:29:50.000000000 +0200
>>@@ -27,15 +27,17 @@ static void *get_ipc(ctl_table *table)
>> }
>>
>> /*
>>- * Routine that is called when a tunable has successfully been changed by
>>- * hand and it has a callback routine registered on the ipc namespace notifier
>>- * chain: we don't want such tunables to be recomputed anymore upon memory
>>- * add/remove or ipc namespace creation/removal.
>>- * They can come back to a recomputable state by being set to a <0 value.
>>+ * Routine that is called when the file "auto_msgmni" has successfully been
>>+ * written.
>>+ * Two values are allowed:
>>+ * 0: unregister msgmni's callback routine from the ipc namespace notifier
>>+ *    chain. This means that msgmni won't be recomputed anymore upon memory
>>+ *    add/remove or ipc namespace creation/removal.
>>+ * 1: register back the callback routine.
>>  */
>>-static void tunable_set_callback(int val)
>>+static void ipc_auto_callback(int val)
>> {
>>-	if (val >= 0)
>>+	if (!val)
>> 		unregister_ipcns_notifier(current->nsproxy->ipc_ns);
>> 	else {
>> 		/*
>>@@ -71,7 +73,15 @@ static int proc_ipc_callback_dointvec(ct
>> 	rc = proc_dointvec(&ipc_table, write, filp, buffer, lenp, ppos);
>>
>> 	if (write && !rc && lenp_bef == *lenp)
>>-		tunable_set_callback(*((int *)(ipc_table.data)));
>>+		/*
>>+		 * Tunable has successfully been changed by hand and it has a
>>+		 * callback routine registered on the ipc namespace notifier
>>+		 * chain: we don't want this tunable to be recomputed anymore
>>+		 * upon memory add/remove or ipc namespace creation/removal.
>>+		 * It can come back to a recomputable state if the
>>+		 * corresponding auto_ file is set to 1.
>>+		 */
> 
> 
> The register_ipcns_notifier() code tells us what will trigger the
> recalculation. If that code gets changed you'd need to update this
> comment. Also your comment at the top of the function describes what 0/1
> mean when written to this file. So I think this comment could be greatly
> simplified:
> 
> /*
>  * Disabling automatic adjustment of msgmni simply requires
>  * unregistering the notifiers that trigger recalculation.
>  */
> 
> 
>>+		unregister_ipcns_notifier(current->nsproxy->ipc_ns);
>>
>> 	return rc;
>> }
>>@@ -87,10 +97,39 @@ static int proc_ipc_doulongvec_minmax(ct
>> 					lenp, ppos);
>> }
>>
>>+static int proc_ipcauto_dointvec_minmax(ctl_table *table, int write,
>>+	struct file *filp, void __user *buffer, size_t *lenp, loff_t *ppos)
>>+{
>>+	struct ctl_table ipc_table;
>>+	size_t lenp_bef = *lenp;
>>+	int oldval;
>>+	int rc;
>>+
>>+	memcpy(&ipc_table, table, sizeof(ipc_table));
>>+	ipc_table.data = get_ipc(table);
>>+	oldval = *((int *)(ipc_table.data));
>>+
>>+	rc = proc_dointvec_minmax(&ipc_table, write, filp, buffer, lenp, ppos);
>>+
>>+	if (write && !rc && lenp_bef == *lenp) {
>>+		int newval = *((int *)(ipc_table.data));
>>+		/*
>>+		 * The file "auto_msgmni" has correctly been set.
>>+		 * React by (un)registering the corresponding tunable, if the
>>+		 * value has changed.
>>+		 */
>>+		if (newval != oldval)
>>+			ipc_auto_callback(newval);
>>+	}
>>+
>>+	return rc;
>>+}
>>+
>> #else
>> #define proc_ipc_doulongvec_minmax NULL
>> #define proc_ipc_dointvec	   NULL
>> #define proc_ipc_callback_dointvec NULL
>>+#define proc_ipcauto_dointvec_minmax NULL
>> #endif
>>
>> #ifdef CONFIG_SYSCTL_SYSCALL
>>@@ -142,14 +181,11 @@ static int sysctl_ipc_registered_data(ct
>> 	rc = sysctl_ipc_data(table, name, nlen, oldval, oldlenp, newval,
>> 		newlen);
>>
>>-	if (newval && newlen && rc > 0) {
>>+	if (newval && newlen && rc > 0)
>> 		/*
>> 		 * Tunable has successfully been changed from userland
>> 		 */
>>-		int *data = get_ipc(table);
>>-
>>-		tunable_set_callback(*data);
>>-	}
>>+		unregister_ipcns_notifier(current->nsproxy->ipc_ns);
>>
>> 	return rc;
>> }
>>@@ -158,6 +194,9 @@ static int sysctl_ipc_registered_data(ct
>> #define sysctl_ipc_registered_data NULL
>> #endif
>>
>>+static int zero;
>>+static int one = 1;
>>+
>> static struct ctl_table ipc_kern_table[] = {
>> 	{
>> 		.ctl_name	= KERN_SHMMAX,
>>@@ -222,6 +261,16 @@ static struct ctl_table ipc_kern_table[]
>> 		.proc_handler	= proc_ipc_dointvec,
>> 		.strategy	= sysctl_ipc_data,
>> 	},
>>+	{
>>+		.ctl_name	= CTL_UNNUMBERED,
>>+		.procname	= "auto_msgmni",
>>+		.data		= &init_ipc_ns.auto_msgmni,
>>+		.maxlen		= sizeof(int),
>>+		.mode		= 0644,
>>+		.proc_handler	= proc_ipcauto_dointvec_minmax,
>>+		.extra1		= &zero,
>>+		.extra2		= &one,
>>+	},
>> 	{}
>> };
>>
>>Index: linux-2.6.26-rc5-mm3/include/linux/ipc_namespace.h
>>===================================================================
>>--- linux-2.6.26-rc5-mm3.orig/include/linux/ipc_namespace.h	2008-06-16 09:12:03.000000000 +0200
>>+++ linux-2.6.26-rc5-mm3/include/linux/ipc_namespace.h	2008-07-03 08:33:56.000000000 +0200
>>@@ -36,6 +36,7 @@ struct ipc_namespace {
>> 	int		msg_ctlmni;
>> 	atomic_t	msg_bytes;
>> 	atomic_t	msg_hdrs;
>>+	int		auto_msgmni;
>>
>> 	size_t		shm_ctlmax;
>> 	size_t		shm_ctlall;
>>Index: linux-2.6.26-rc5-mm3/ipc/ipcns_notifier.c
>>===================================================================
>>--- linux-2.6.26-rc5-mm3.orig/ipc/ipcns_notifier.c	2008-06-16 09:12:57.000000000 +0200
>>+++ linux-2.6.26-rc5-mm3/ipc/ipcns_notifier.c	2008-07-03 11:38:07.000000000 +0200
>>@@ -55,25 +55,36 @@ static int ipcns_callback(struct notifie
>>
>> int register_ipcns_notifier(struct ipc_namespace *ns)
>> {
>>+	int rc;
>>+
>> 	memset(&ns->ipcns_nb, 0, sizeof(ns->ipcns_nb));
>> 	ns->ipcns_nb.notifier_call = ipcns_callback;
>> 	ns->ipcns_nb.priority = IPCNS_CALLBACK_PRI;
>>-	return blocking_notifier_chain_register(&ipcns_chain, &ns->ipcns_nb);
>>+	rc = blocking_notifier_chain_register(&ipcns_chain, &ns->ipcns_nb);
>>+	if (!rc)
>>+		ns->auto_msgmni = 1;
>>+	return rc;
>> }
>>
>> int cond_register_ipcns_notifier(struct ipc_namespace *ns)
>> {
>>+	int rc;
>>+
>> 	memset(&ns->ipcns_nb, 0, sizeof(ns->ipcns_nb));
>> 	ns->ipcns_nb.notifier_call = ipcns_callback;
>> 	ns->ipcns_nb.priority = IPCNS_CALLBACK_PRI;
>>-	return blocking_notifier_chain_cond_register(&ipcns_chain,
>>+	rc = blocking_notifier_chain_cond_register(&ipcns_chain,
>> 							&ns->ipcns_nb);
>>+	if (!rc)
>>+		ns->auto_msgmni = 1;
>>+	return rc;
>> }
>>
>> int unregister_ipcns_notifier(struct ipc_namespace *ns)
>> {
>>-	return blocking_notifier_chain_unregister(&ipcns_chain,
>>-						&ns->ipcns_nb);
>>+	blocking_notifier_chain_unregister(&ipcns_chain, &ns->ipcns_nb);
>>+	ns->auto_msgmni = 0;
>>+	return 0;
>> }
> 
> 
> This looks odd -- we're no longer returning the return code from
> blocking_notifier_chain_unregister().
> From what I can see in the patch,
> the return value is unused. Perhaps it ought to be removed? Otherwise it
> might make sense to do the same as you did with "register":
> 
>         rc = blocking_notifier_chain_unregister(&ipcns_chain, &ns->ipcns_nb);
>         if (!rc)
>         	ns->auto_msgmni = 0;
>         return rc;
> 
> Cheers,
> 	-Matt Helsley
> 
> 
> 

Well, I thought that the registration routines might evolve in the 
future, even though they are now unconditionally returning 0. That's why 
I'm setting the flag to 1 only if 
blocking_notifier_chain_cond_register() succeeded.
Now, talking about unregister: it acutally returns a ENOENT, but there's 
no need to test the return code, since even in that case the flag should 
be set to 0.
I'll change unregister_ipcns_notifier() into a void.

Thanks for the review, Matt!

Regards,
Nadia


--
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/

Powered by blists - more mailing lists