| 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
| ||
|
Message-ID: <20081030045547.GA27381@Krystal>
Date: Thu, 30 Oct 2008 00:55:47 -0400
From: Mathieu Desnoyers <mathieu.desnoyers@...ymtl.ca>
To: Lai Jiangshan <laijs@...fujitsu.com>
Cc: Ingo Molnar <mingo@...e.hu>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/2] tracepoint: simplify for tracepoint using RCU
* Lai Jiangshan (laijs@...fujitsu.com) wrote:
>
> Now, unused memory is handled by struct tp_probes.
>
> old code use these three field to handle unused memory.
> struct tracepoint_entry {
> ...
> struct rcu_head rcu;
> void *oldptr;
> unsigned char rcu_pending:1;
> ...
> };
> in this way, unused memory is handled by struct tracepoint_entry.
> it bring reenter bug(it was fixed) and tracepoint.c is filled
> full of ".*rcu.*" code statements. this patch remove all these.
>
> and:
> rcu_barrier_sched() is removed.
> Do not need regain tracepoints_mutex after tracepoint_update_probes()
> several little cleanup.
>
Ok, I had to bring my head into this code again. So yes, it looks good.
I originally used the RCU the way I did because this code evolved from
markers, where I had special-cases for 0/1 probes connected which
required to way a quiescent state between two consecutive
register/unregister on the same probe.
In the tracepoints, because the probe array is replaced by a newly
allocated memory entry each time, and the old array can wait for any
given time to be freed by RCU, we don't have to wait.
So yes, this patch is good.
Thanks Lay!
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@...ymtl.ca>
> Signed-off-by: Lai Jiangshan <laijs@...fujitsu.com>
> ---
> tracepoint.c | 108 ++++++++++++++++++++---------------------------------------
> 1 file changed, 38 insertions(+), 70 deletions(-)
> diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c
> index f2b7c28..76d4571 100644
> --- a/kernel/tracepoint.c
> +++ b/kernel/tracepoint.c
> @@ -43,6 +43,7 @@ static DEFINE_MUTEX(tracepoints_mutex);
> */
> #define TRACEPOINT_HASH_BITS 6
> #define TRACEPOINT_TABLE_SIZE (1 << TRACEPOINT_HASH_BITS)
> +static struct hlist_head tracepoint_table[TRACEPOINT_TABLE_SIZE];
>
> /*
> * Note about RCU :
> @@ -54,40 +55,40 @@ struct tracepoint_entry {
> struct hlist_node hlist;
> void **funcs;
> int refcount; /* Number of times armed. 0 if disarmed. */
> - struct rcu_head rcu;
> - void *oldptr;
> - unsigned char rcu_pending:1;
> char name[0];
> };
>
> -static struct hlist_head tracepoint_table[TRACEPOINT_TABLE_SIZE];
> +struct tp_probes {
> + struct rcu_head rcu;
> + void *probes[0];
> +};
>
> -static void free_old_closure(struct rcu_head *head)
> +static inline void *allocate_probes(int count)
> {
> - struct tracepoint_entry *entry = container_of(head,
> - struct tracepoint_entry, rcu);
> - kfree(entry->oldptr);
> - /* Make sure we free the data before setting the pending flag to 0 */
> - smp_wmb();
> - entry->rcu_pending = 0;
> + struct tp_probes *p = kmalloc(count * sizeof(void *)
> + + sizeof(struct tp_probes), GFP_KERNEL);
> + return p == NULL ? NULL : p->probes;
> }
>
> -static void tracepoint_entry_free_old(struct tracepoint_entry *entry, void *old)
> +static void rcu_free_old_probes(struct rcu_head *head)
> {
> - if (!old)
> - return;
> - entry->oldptr = old;
> - entry->rcu_pending = 1;
> - /* write rcu_pending before calling the RCU callback */
> - smp_wmb();
> - call_rcu_sched(&entry->rcu, free_old_closure);
> + kfree(container_of(head, struct tp_probes, rcu));
> +}
> +
> +static inline void release_probes(void *old)
> +{
> + if (old) {
> + struct tp_probes *tp_probes = container_of(old,
> + struct tp_probes, probes[0]);
> + call_rcu(&tp_probes->rcu, rcu_free_old_probes);
> + }
> }
>
> static void debug_print_probes(struct tracepoint_entry *entry)
> {
> int i;
>
> - if (!tracepoint_debug)
> + if (!tracepoint_debug || !entry->funcs)
> return;
>
> for (i = 0; entry->funcs[i]; i++)
> @@ -111,12 +112,13 @@ tracepoint_entry_add_probe(struct tracepoint_entry *entry, void *probe)
> return ERR_PTR(-EEXIST);
> }
> /* + 2 : one for new probe, one for NULL func */
> - new = kzalloc((nr_probes + 2) * sizeof(void *), GFP_KERNEL);
> + new = allocate_probes(nr_probes + 2);
> if (new == NULL)
> return ERR_PTR(-ENOMEM);
> if (old)
> memcpy(new, old, nr_probes * sizeof(void *));
> new[nr_probes] = probe;
> + new[nr_probes + 1] = NULL;
> entry->refcount = nr_probes + 1;
> entry->funcs = new;
> debug_print_probes(entry);
> @@ -148,13 +150,13 @@ tracepoint_entry_remove_probe(struct tracepoint_entry *entry, void *probe)
> int j = 0;
> /* N -> M, (N > 1, M > 0) */
> /* + 1 for NULL */
> - new = kzalloc((nr_probes - nr_del + 1)
> - * sizeof(void *), GFP_KERNEL);
> + new = allocate_probes(nr_probes - nr_del + 1);
> if (new == NULL)
> return ERR_PTR(-ENOMEM);
> for (i = 0; old[i]; i++)
> if ((probe && old[i] != probe))
> new[j++] = old[i];
> + new[nr_probes - nr_del] = NULL;
> entry->refcount = nr_probes - nr_del;
> entry->funcs = new;
> }
> @@ -212,7 +214,6 @@ static struct tracepoint_entry *add_tracepoint(const char *name)
> memcpy(&e->name[0], name, name_len);
> e->funcs = NULL;
> e->refcount = 0;
> - e->rcu_pending = 0;
> hlist_add_head(&e->hlist, head);
> return e;
> }
> @@ -221,32 +222,10 @@ static struct tracepoint_entry *add_tracepoint(const char *name)
> * Remove the tracepoint from the tracepoint hash table. Must be called with
> * mutex_lock held.
> */
> -static int remove_tracepoint(const char *name)
> +static inline void remove_tracepoint(struct tracepoint_entry *e)
> {
> - struct hlist_head *head;
> - struct hlist_node *node;
> - struct tracepoint_entry *e;
> - int found = 0;
> - size_t len = strlen(name) + 1;
> - u32 hash = jhash(name, len-1, 0);
> -
> - head = &tracepoint_table[hash & (TRACEPOINT_TABLE_SIZE - 1)];
> - hlist_for_each_entry(e, node, head, hlist) {
> - if (!strcmp(name, e->name)) {
> - found = 1;
> - break;
> - }
> - }
> - if (!found)
> - return -ENOENT;
> - if (e->refcount)
> - return -EBUSY;
> hlist_del(&e->hlist);
> - /* Make sure the call_rcu_sched has been executed */
> - if (e->rcu_pending)
> - rcu_barrier_sched();
> kfree(e);
> - return 0;
> }
>
> /*
> @@ -340,25 +319,17 @@ int tracepoint_probe_register(const char *name, void *probe)
> goto end;
> }
> }
> - /*
> - * If we detect that a call_rcu_sched is pending for this tracepoint,
> - * make sure it's executed now.
> - */
> - if (entry->rcu_pending)
> - rcu_barrier_sched();
> old = tracepoint_entry_add_probe(entry, probe);
> if (IS_ERR(old)) {
> + if (!entry->refcount)
> + remove_tracepoint(entry);
> ret = PTR_ERR(old);
> goto end;
> }
> mutex_unlock(&tracepoints_mutex);
> tracepoint_update_probes(); /* may update entry */
> - mutex_lock(&tracepoints_mutex);
> - entry = get_tracepoint(name);
> - WARN_ON(!entry);
> - if (entry->rcu_pending)
> - rcu_barrier_sched();
> - tracepoint_entry_free_old(entry, old);
> + release_probes(old);
> + return 0;
> end:
> mutex_unlock(&tracepoints_mutex);
> return ret;
> @@ -385,20 +356,17 @@ int tracepoint_probe_unregister(const char *name, void *probe)
> entry = get_tracepoint(name);
> if (!entry)
> goto end;
> - if (entry->rcu_pending)
> - rcu_barrier_sched();
> old = tracepoint_entry_remove_probe(entry, probe);
> + if (IS_ERR(old)) {
> + ret = PTR_ERR(old);
> + goto end;
> + }
> + if (!entry->refcount)
> + remove_tracepoint(entry);
> mutex_unlock(&tracepoints_mutex);
> tracepoint_update_probes(); /* may update entry */
> - mutex_lock(&tracepoints_mutex);
> - entry = get_tracepoint(name);
> - if (!entry)
> - goto end;
> - if (entry->rcu_pending)
> - rcu_barrier_sched();
> - tracepoint_entry_free_old(entry, old);
> - remove_tracepoint(name); /* Ignore busy error message */
> - ret = 0;
> + release_probes(old);
> + return 0;
> end:
> mutex_unlock(&tracepoints_mutex);
> return ret;
>
--
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68
--
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