From: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
To: shli@kernel.org
Cc: linux-raid@vger.kernel.org,
Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Subject: [PATCH 2/4] raid5-ppl: move no_mem_stripes to struct ppl_conf
Date: Tue, 4 Apr 2017 13:13:56 +0200 [thread overview]
Message-ID: <20170404111358.14829-3-artur.paszkiewicz@intel.com> (raw)
In-Reply-To: <20170404111358.14829-1-artur.paszkiewicz@intel.com>
Use a single no_mem_stripes list instead of per member device lists for
handling stripes that need retrying in case of failed io_unit
allocation. Because io_units are allocated from a memory pool shared
between all member disks, the no_mem_stripes list should be checked when
an io_unit for any member is freed. This fixes a deadlock that could
happen if there are stripes in more than one no_mem_stripes list.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
---
drivers/md/raid5-ppl.c | 36 +++++++++++++++++++++++-------------
1 file changed, 23 insertions(+), 13 deletions(-)
diff --git a/drivers/md/raid5-ppl.c b/drivers/md/raid5-ppl.c
index 42e43467d1e8..0a39a6bbcbde 100644
--- a/drivers/md/raid5-ppl.c
+++ b/drivers/md/raid5-ppl.c
@@ -106,6 +106,10 @@ struct ppl_conf {
/* used only for recovery */
int recovered_entries;
int mismatch_count;
+
+ /* stripes to retry if failed to allocate io_unit */
+ struct list_head no_mem_stripes;
+ spinlock_t no_mem_stripes_lock;
};
struct ppl_log {
@@ -118,8 +122,6 @@ struct ppl_log {
* always at the end of io_list */
spinlock_t io_list_lock;
struct list_head io_list; /* all io_units of this log */
- struct list_head no_mem_stripes;/* stripes to retry if failed to
- * allocate io_unit */
};
#define PPL_IO_INLINE_BVECS 32
@@ -374,9 +376,9 @@ int ppl_write_stripe(struct r5conf *conf, struct stripe_head *sh)
atomic_inc(&sh->count);
if (ppl_log_stripe(log, sh)) {
- spin_lock_irq(&log->io_list_lock);
- list_add_tail(&sh->log_list, &log->no_mem_stripes);
- spin_unlock_irq(&log->io_list_lock);
+ spin_lock_irq(&ppl_conf->no_mem_stripes_lock);
+ list_add_tail(&sh->log_list, &ppl_conf->no_mem_stripes);
+ spin_unlock_irq(&ppl_conf->no_mem_stripes_lock);
}
mutex_unlock(&log->io_mutex);
@@ -517,25 +519,32 @@ void ppl_write_stripe_run(struct r5conf *conf)
static void ppl_io_unit_finished(struct ppl_io_unit *io)
{
struct ppl_log *log = io->log;
+ struct ppl_conf *ppl_conf = log->ppl_conf;
unsigned long flags;
pr_debug("%s: seq: %llu\n", __func__, io->seq);
- spin_lock_irqsave(&log->io_list_lock, flags);
+ local_irq_save(flags);
+ spin_lock(&log->io_list_lock);
list_del(&io->log_sibling);
- mempool_free(io, log->ppl_conf->io_pool);
+ spin_unlock(&log->io_list_lock);
+
+ mempool_free(io, ppl_conf->io_pool);
+
+ spin_lock(&ppl_conf->no_mem_stripes_lock);
+ if (!list_empty(&ppl_conf->no_mem_stripes)) {
+ struct stripe_head *sh;
- if (!list_empty(&log->no_mem_stripes)) {
- struct stripe_head *sh = list_first_entry(&log->no_mem_stripes,
- struct stripe_head,
- log_list);
+ sh = list_first_entry(&ppl_conf->no_mem_stripes,
+ struct stripe_head, log_list);
list_del_init(&sh->log_list);
set_bit(STRIPE_HANDLE, &sh->state);
raid5_release_stripe(sh);
}
+ spin_unlock(&ppl_conf->no_mem_stripes_lock);
- spin_unlock_irqrestore(&log->io_list_lock, flags);
+ local_irq_restore(flags);
}
void ppl_stripe_write_finished(struct stripe_head *sh)
@@ -1154,6 +1163,8 @@ int ppl_init_log(struct r5conf *conf)
}
atomic64_set(&ppl_conf->seq, 0);
+ INIT_LIST_HEAD(&ppl_conf->no_mem_stripes);
+ spin_lock_init(&ppl_conf->no_mem_stripes_lock);
if (!mddev->external) {
ppl_conf->signature = ~crc32c_le(~0, mddev->uuid, sizeof(mddev->uuid));
@@ -1169,7 +1180,6 @@ int ppl_init_log(struct r5conf *conf)
mutex_init(&log->io_mutex);
spin_lock_init(&log->io_list_lock);
INIT_LIST_HEAD(&log->io_list);
- INIT_LIST_HEAD(&log->no_mem_stripes);
log->ppl_conf = ppl_conf;
log->rdev = rdev;
--
2.11.0
next prev parent reply other threads:[~2017-04-04 11:13 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-04-04 11:13 [PATCH 0/4] PPL fixes and improvements Artur Paszkiewicz
2017-04-04 11:13 ` [PATCH 1/4] raid5-ppl: use a single mempool for ppl_io_unit and header_page Artur Paszkiewicz
2017-04-10 19:09 ` Shaohua Li
2017-04-11 8:28 ` Artur Paszkiewicz
2017-04-11 9:20 ` Artur Paszkiewicz
2017-04-11 15:41 ` Shaohua Li
2017-04-11 18:50 ` [PATCH v2] " Artur Paszkiewicz
2017-04-11 21:58 ` Shaohua Li
2017-04-04 11:13 ` Artur Paszkiewicz [this message]
2017-04-04 11:13 ` [PATCH 3/4] raid5-ppl: use resize_stripes() when enabling or disabling ppl Artur Paszkiewicz
2017-04-04 11:13 ` [PATCH 4/4] raid5-ppl: partial parity calculation optimization Artur Paszkiewicz
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=20170404111358.14829-3-artur.paszkiewicz@intel.com \
--to=artur.paszkiewicz@intel.com \
--cc=linux-raid@vger.kernel.org \
--cc=shli@kernel.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).