From: James Simmons <jsimmons@infradead.org>
To: lustre-devel@lists.lustre.org
Subject: [lustre-devel] [PATCH 09/37] lustre: obdclass: fix module load locking.
Date: Sun, 24 Feb 2019 17:04:24 +0000 (GMT) [thread overview]
Message-ID: <alpine.LFD.2.21.1902241704150.22752@casper.infradead.org> (raw)
In-Reply-To: <155053494536.24125.16645429190038913975.stgit@noble.brown>
> Safe module loading requires that we try_module_get() in a context
> where the module cannot be unloaded, typically protected by
> a spinlock that module-unload has to take.
> This doesn't currently happen in class_get_type().
>
> As free_module() calls synchronize_rcu() between calling the
> exit function and freeing the module, we can use rcu_read_lock()
> to check if the exit function has been called, and try_module_get()
> if it hasn't.
>
> We must also check the return status of try_module_get().
Reviewed-by: James Simmons <jsimmons@infradead.org>
> Signed-off-by: NeilBrown <neilb@suse.com>
> ---
> drivers/staging/lustre/lustre/obdclass/genops.c | 24 ++++++++++++++++++++---
> 1 file changed, 21 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/staging/lustre/lustre/obdclass/genops.c b/drivers/staging/lustre/lustre/obdclass/genops.c
> index d013e9a358f1..1d1a6b2d436e 100644
> --- a/drivers/staging/lustre/lustre/obdclass/genops.c
> +++ b/drivers/staging/lustre/lustre/obdclass/genops.c
> @@ -98,21 +98,31 @@ static struct obd_type *class_get_type(const char *name)
> {
> struct obd_type *type;
>
> + rcu_read_lock();
> type = class_search_type(name);
>
> if (!type) {
> const char *modname = name;
>
> + rcu_read_unlock();
> if (!request_module("%s", modname)) {
> CDEBUG(D_INFO, "Loaded module '%s'\n", modname);
> - type = class_search_type(name);
> } else {
> LCONSOLE_ERROR_MSG(0x158, "Can't load module '%s'\n",
> modname);
> }
> + rcu_read_lock();
> + type = class_search_type(name);
> }
> if (type) {
> - if (try_module_get(type->typ_dt_ops->owner)) {
> + /*
> + * Holding rcu_read_lock() matches the synchronize_rcu() call
> + * in free_module() and ensures that if type->typ_dt_ops is
> + * not yet NULL, then the module won't be freed until after
> + * we rcu_read_unlock().
> + */
> + const struct obd_ops *dt_ops = READ_ONCE(type->typ_dt_ops);
> + if (dt_ops && try_module_get(dt_ops->owner)) {
> refcount_inc(&type->typ_refcnt);
> /* class_search_type() returned a counted reference,
> * but we don't need that count any more as
> @@ -124,6 +134,7 @@ static struct obd_type *class_get_type(const char *name)
> type = NULL;
> }
> }
> + rcu_read_unlock();
> return type;
> }
>
> @@ -213,11 +224,18 @@ int class_unregister_type(const char *name)
> return -EINVAL;
> }
>
> + /*
> + * Ensure that class_get_type doesn't try to get the module
> + * as it could be freed before the obd_type is released.
> + * synchronize_rcu() will be called before the module
> + * is freed.
> + */
> + type->typ_dt_ops = NULL;
> +
> if (refcount_read(&type->typ_refcnt)) {
> CERROR("type %s has refcount (%d)\n", name, refcount_read(&type->typ_refcnt));
> /* This is a bad situation, let's make the best of it */
> /* Remove ops, but leave the name for debugging */
> - type->typ_dt_ops = NULL;
> type->typ_md_ops = NULL;
> kobject_put(&type->typ_kobj);
> return -EBUSY;
>
>
>
next prev parent reply other threads:[~2019-02-24 17:04 UTC|newest]
Thread overview: 105+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-02-19 0:09 [lustre-devel] [PATCH 00/37] More lustre patches from obdclass NeilBrown
2019-02-19 0:09 ` [lustre-devel] [PATCH 01/37] lustre: obdclass: char obd_ioctl_getdata type NeilBrown
2019-02-24 18:35 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 02/37] lustre: llite: don't use class_setup_tunables() NeilBrown
2019-02-24 16:35 ` James Simmons
2019-02-25 22:27 ` NeilBrown
2019-02-26 22:18 ` James Simmons
2019-02-24 16:52 ` [lustre-devel] [PATCH 03/37] lustre: embed typ_kobj if obd_type James Simmons
2019-02-25 22:38 ` NeilBrown
2019-02-26 20:41 ` Simmons, James A.
2019-02-19 0:09 ` [lustre-devel] [PATCH 14/37] lustre: llog: change lgh_refcount to struct kref NeilBrown
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 13/37] lustre: llog: remove lgh_hdr_lock NeilBrown
2019-02-24 20:29 ` James Simmons
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 07/37] lustre: obd_type: discard obd_type_lock NeilBrown
2019-02-24 17:02 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 08/37] lustre: obdclass: don't copy ops structures in to new type NeilBrown
2019-02-24 17:03 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 16/37] lustre: obdclass: typo: Banlance -> Balance NeilBrown
2019-02-24 17:39 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 05/37] lustre: obd_type: use typ_kobj.name as typ_name NeilBrown
2019-02-24 16:56 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 03/37] lustre: embed typ_kobj if obd_type NeilBrown
2019-02-19 0:09 ` [lustre-devel] [PATCH 09/37] lustre: obdclass: fix module load locking NeilBrown
2019-02-24 17:04 ` James Simmons [this message]
2019-02-19 0:09 ` [lustre-devel] [PATCH 17/37] lustre: simplify lprocfs_read_frac_helper NeilBrown
2019-02-24 17:52 ` James Simmons
2019-02-26 23:59 ` NeilBrown
2019-02-27 1:06 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 18/37] lustre: obdclass: discard lprocfs_single/seq_release NeilBrown
2019-02-24 17:53 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 12/37] lustre: remove unused function in linkea NeilBrown
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 10/37] lustre: kernelcomm: pass correct gfp_t to kmalloc NeilBrown
2019-02-24 17:05 ` James Simmons
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 21/37] lustre: remove several MAX_STRING_SIZE defines NeilBrown
2019-02-24 19:07 ` James Simmons
2019-02-27 0:41 ` NeilBrown
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 19/37] lustre: discard lprocfs_strnstr() NeilBrown
2019-02-24 17:53 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 15/37] lustre: llog_obd: Convert loc_refcount to refcount_t NeilBrown
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 11/37] lustre: kernelcomm: make libcfs_kkuc_msg_put static NeilBrown
2019-02-24 17:15 ` James Simmons
2019-02-26 23:45 ` NeilBrown
2019-02-27 22:36 ` James Simmons
2019-02-27 22:37 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 04/37] lustre: collect all resource releasing for obj_type NeilBrown
2019-02-24 16:54 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 06/37] lustre: obd_type: discard obd_types linked list NeilBrown
2019-02-24 17:00 ` James Simmons
2019-02-19 0:09 ` [lustre-devel] [PATCH 20/37] lustre: convert rsi_sem to a spinlock NeilBrown
2019-02-25 18:16 ` Andreas Dilger
2019-02-27 0:22 ` NeilBrown
2019-02-27 1:00 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 32/37] lustre: portals_handle: rename ops to owner NeilBrown
2019-02-19 0:09 ` [lustre-devel] [PATCH 22/37] lustre: lprocfs: use log2.h macros instead of shift loop NeilBrown
2019-02-24 18:09 ` James Simmons
2019-02-26 20:55 ` James Simmons
2019-02-27 0:51 ` NeilBrown
2019-02-27 0:54 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 25/37] lustre: deprecate libcfs_debug_vmsg2 NeilBrown
2019-02-24 20:02 ` James Simmons
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 30/37] lustre: handle: move refcount into the lustre_handle NeilBrown
2019-02-27 6:32 ` Andreas Dilger
2019-02-27 21:48 ` NeilBrown
2019-02-27 22:14 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 28/37] lustre: remove scope and source from class_incref and class_decref NeilBrown
2019-02-27 6:52 ` Andreas Dilger
2019-02-28 0:39 ` NeilBrown
2019-02-19 0:09 ` [lustre-devel] [PATCH 26/37] lustre: remove libcfs_debug_vmsg2 NeilBrown
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 23/37] lustre: prefer to use tabs for alignment NeilBrown
2019-02-24 18:51 ` James Simmons
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 27/37] lustre: discard lu_ref NeilBrown
2019-02-24 20:28 ` James Simmons
2019-02-27 1:17 ` NeilBrown
2019-02-27 5:35 ` Andreas Dilger
2019-03-01 6:45 ` Mike Pershin
2019-02-19 0:09 ` [lustre-devel] [PATCH 29/37] lustre: handles: discard h_owner in favour of h_ops NeilBrown
2019-02-27 6:37 ` Andreas Dilger
2019-02-27 21:41 ` NeilBrown
2019-02-28 6:41 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 33/37] lustre: portals_handle: remove locking from class_handle2object() NeilBrown
2019-02-19 0:09 ` [lustre-devel] [PATCH 34/37] lustre: portals_handle: use hlist for hash lists NeilBrown
2019-02-19 0:09 ` [lustre-devel] [PATCH 31/37] lustre: discard OBD_FREE_RCU NeilBrown
2019-02-19 0:09 ` [lustre-devel] [PATCH 24/37] lustre: lu_object: remove extra newline from debug printing NeilBrown
2019-02-24 19:08 ` James Simmons
2019-02-25 18:16 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 35/37] lustre: portals_handle: discard h_lock NeilBrown
2019-02-19 0:09 ` [lustre-devel] [PATCH 37/37] lustre: obd_sysfs: error-check value stored in jobid_var NeilBrown
2019-02-27 6:17 ` Andreas Dilger
2019-03-01 2:35 ` NeilBrown
2019-03-01 8:32 ` Andreas Dilger
2019-03-01 14:30 ` Patrick Farrell
2019-03-14 0:34 ` NeilBrown
2019-03-14 14:12 ` Patrick Farrell
2019-03-14 22:56 ` NeilBrown
2019-03-14 23:05 ` Andreas Dilger
2019-02-19 0:09 ` [lustre-devel] [PATCH 36/37] lustre: remove unused fields from struct obd_device NeilBrown
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=alpine.LFD.2.21.1902241704150.22752@casper.infradead.org \
--to=jsimmons@infradead.org \
--cc=lustre-devel@lists.lustre.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).