[Devel] [PATCH RHEL9 COMMIT] dm-ploop: Use READ_ONCE/WRITE_ONCE to access md page data
Konstantin Khorenko
khorenko at virtuozzo.com
Mon Jan 27 16:12:32 MSK 2025
The commit is pushed to "branch-rh9-5.14.0-427.44.1.vz9.80.x-ovz" and will appear at git at bitbucket.org:openvz/vzkernel.git
after rh9-5.14.0-427.44.1.vz9.80.6
------>
commit 329b26c848f1c0a6dce91678dc87b13b36e5ff8c
Author: Alexander Atanasov <alexander.atanasov at virtuozzo.com>
Date: Fri Jan 24 17:35:36 2025 +0200
dm-ploop: Use READ_ONCE/WRITE_ONCE to access md page data
Prepare to reduce locking by using atomic 32 bit access to the fields.
To ensure this we need to use the _ONCE macros.
https://virtuozzo.atlassian.net/browse/VSTOR-91659
Signed-off-by: Alexander Atanasov <alexander.atanasov at virtuozzo.com>
======
Patchset description:
ploop: optimistations and scalling
Ploop processes requsts in a different threads in parallel
where possible which results in significant improvement in
performance and makes further optimistations possible.
Known bugs:
- delayed metadata writeback is not working and is missing error handling
- patch to disable it until fixed
- fast path is not working - causes rcu lockups - patch to disable it
Further improvements:
- optimize md pages lookups
Alexander Atanasov (50):
dm-ploop: md_pages map all pages at creation time
dm-ploop: Use READ_ONCE/WRITE_ONCE to access md page data
dm-ploop: fsync after all pios are sent
dm-ploop: move md status to use proper bitops
dm-ploop: convert wait_list and wb_batch_llist to use lockless lists
dm-ploop: convert enospc handling to use lockless lists
dm-ploop: convert suspended_pios list to use lockless list
dm-ploop: convert the rest of the lists to use llist variant
dm-ploop: combine processing of pios thru prepare list and remove
fsync worker
dm-ploop: move from wq to kthread
dm-ploop: move preparations of pios into the caller from worker
dm-ploop: fast path execution for reads
dm-ploop: do not use a wrapper for set_bit to make a page writeback
dm-ploop: BAT use only one list for writeback
dm-ploop: make md writeback timeout to be per page
dm-ploop: add interface to disable bat writeback delay
dm-ploop: convert wb_batch_list to lockless variant
dm-ploop: convert high_prio to status
dm-ploop: split cow processing into two functions
dm-ploop: convert md page rw lock to spin lock
dm-ploop: convert bat_rwlock to bat_lock spinlock
dm-ploop: prepare bat updates under bat_lock
dm-ploop: make ploop_bat_write_complete ready for parallel pio
completion
dm-ploop: make ploop_submit_metadata_writeback return number of
requests sent
dm-ploop: introduce pio runner threads
dm-ploop: add pio list ids to be used when passing pios to runners
dm-ploop: process pios via runners
dm-ploop: disable metadata writeback delay
dm-ploop: disable fast path
dm-ploop: use lockless lists for chained cow updates list
dm-ploop: use lockless lists for data ready pios
dm-ploop: give runner threads better name
dm-ploop: resize operation - add holes bitmap locking
dm-ploop: remove unnecessary operations
dm-ploop: use filp per thread
dm-ploop: catch if we try to advance pio past bio end
dm-ploop: support REQ_FUA for data pios
dm-ploop: proplerly access nr_bat_entries
dm-ploop: fix locking and improve error handling when submitting pios
dm-ploop: fix how ENOTBLK is handled
dm-ploop: sync when suspended or stopping
dm-ploop: rework bat completion logic
dm-ploop: rework logic in pio processing
dm-ploop: end fsync pios in parallel
dm-ploop: make filespace preallocations async
dm-ploop: resubmit enospc pios from dispatcher thread
dm-ploop: dm-ploop: simplify discard completion
dm-ploop: use GFP_ATOMIC instead of GFP_NOIO
dm-ploop: fix locks used in mixed context
dm-ploop: fix how current flags are managed inside threads
Andrey Zhadchenko (13):
dm-ploop: do not flush after metadata writes
dm-ploop: set IOCB_DSYNC on all FUA requests
dm-ploop: remove extra ploop_cluster_is_in_top_delta()
dm-ploop: introduce per-md page locking
dm-ploop: reduce BAT accesses on discard completion
dm-ploop: simplify llseek
dm-ploop: speed up ploop_prepare_bat_update()
dm-ploop: make new allocations immediately visible in BAT
dm-ploop: drop ploop_cluster_is_in_top_delta()
dm-ploop: do not wait for BAT update for non-FUA requests
dm-ploop: add delay for metadata writeback
dm-ploop: submit all postponed metadata on REQ_OP_FLUSH
dm-ploop: handle REQ_PREFLUSH
Feature: dm-ploop: ploop target driver
---
drivers/md/dm-ploop-bat.c | 20 ++++++++++----------
drivers/md/dm-ploop-cmd.c | 28 ++++++++++++++--------------
drivers/md/dm-ploop-map.c | 24 +++++++++++++-----------
drivers/md/dm-ploop.h | 2 +-
4 files changed, 38 insertions(+), 36 deletions(-)
diff --git a/drivers/md/dm-ploop-bat.c b/drivers/md/dm-ploop-bat.c
index f28bd5d7b322..886a05cdd23a 100644
--- a/drivers/md/dm-ploop-bat.c
+++ b/drivers/md/dm-ploop-bat.c
@@ -143,9 +143,9 @@ bool ploop_try_update_bat_entry(struct ploop *ploop, u32 clu, u8 level, u32 dst_
clu = ploop_bat_clu_idx_in_page(clu); /* relative offset */
- if (md->bat_levels[clu] == level) {
+ if (READ_ONCE(md->bat_levels[clu]) == level) {
bat_entries = md->kmpage;
- bat_entries[clu] = dst_clu;
+ WRITE_ONCE(bat_entries[clu], dst_clu);
return true;
}
return false;
@@ -287,11 +287,11 @@ static int ploop_convert_bat_entries(struct ploop *ploop, struct rb_root *md_roo
page_id++;
for (; i <= end; i++) {
- if (bat_entries[i] > max_file_clu)
+ if (READ_ONCE(bat_entries[i]) > max_file_clu)
ret = -EPROTO;
- if (!bat_entries[i])
- bat_entries[i] = BAT_ENTRY_NONE;
- if (bat_entries[i] < bat_clusters)
+ if (!READ_ONCE(bat_entries[i]))
+ WRITE_ONCE(bat_entries[i], BAT_ENTRY_NONE);
+ if (READ_ONCE(bat_entries[i]) < bat_clusters)
ret = -EXDEV;
}
@@ -431,21 +431,21 @@ static void ploop_apply_delta_mappings(struct ploop *ploop,
if (clu == nr_be - 1)
stop = true;
- if (bat_entries[i] != BAT_ENTRY_NONE) {
+ if (READ_ONCE(bat_entries[i]) != BAT_ENTRY_NONE) {
/* md0 is already populated */
WARN_ON_ONCE(md->id && is_top_level);
goto set_not_hole;
}
if (!is_raw)
- dst_clu = d_bat_entries[i];
+ dst_clu = READ_ONCE(d_bat_entries[i]);
else
dst_clu = clu;
if (dst_clu == BAT_ENTRY_NONE)
continue;
- md->bat_levels[i] = level;
- bat_entries[i] = dst_clu;
+ WRITE_ONCE(md->bat_levels[i], level);
+ WRITE_ONCE(bat_entries[i], dst_clu);
set_not_hole:
if (is_top_level)
ploop_set_not_hole(ploop, bat_entries[i]);
diff --git a/drivers/md/dm-ploop-cmd.c b/drivers/md/dm-ploop-cmd.c
index d9d41d79d937..25ca9db4eeb8 100644
--- a/drivers/md/dm-ploop-cmd.c
+++ b/drivers/md/dm-ploop-cmd.c
@@ -47,7 +47,7 @@ static void ploop_advance_holes_bitmap(struct ploop *ploop,
for (; i <= end; i++) {
if (!ploop_md_page_cluster_is_in_top_delta(ploop, md, i))
continue;
- dst_clu = bat_entries[i];
+ dst_clu = READ_ONCE(bat_entries[i]);
/* This may happen after grow->shrink->(now) grow */
if (dst_clu < ploop->hb_nr &&
test_bit(dst_clu, ploop->holes_bitmap)) {
@@ -162,7 +162,7 @@ static u32 ploop_find_bat_entry(struct ploop *ploop, u32 dst_clu, bool *is_locke
ploop_init_be_iter(ploop, md->id, &i, &end);
bat_entries = md->kmpage;
for (; i <= end; i++) {
- if (bat_entries[i] != dst_clu)
+ if (READ_ONCE(bat_entries[i]) != dst_clu)
continue;
if (ploop_md_page_cluster_is_in_top_delta(ploop, md, i)) {
clu = ploop_page_clu_idx_to_bat_clu(md->id, i);
@@ -735,17 +735,17 @@ static void notify_delta_merged(struct ploop *ploop, u8 level,
stop = true;
/* deltas above @level become renumbered */
- if (bat_entries[i] != BAT_ENTRY_NONE &&
- md->bat_levels[i] > level) {
+ if (READ_ONCE(bat_entries[i]) != BAT_ENTRY_NONE &&
+ READ_ONCE(md->bat_levels[i]) > level) {
md->bat_levels[i]--;
continue;
}
- if (bat_entries[i] != BAT_ENTRY_NONE &&
- md->bat_levels[i] < level)
+ if (READ_ONCE(bat_entries[i]) != BAT_ENTRY_NONE &&
+ READ_ONCE(md->bat_levels[i]) < level)
continue;
- if (d_bat_entries[i] == BAT_ENTRY_NONE) {
+ if (READ_ONCE(d_bat_entries[i]) == BAT_ENTRY_NONE) {
WARN_ON_ONCE(bat_entries[i] != BAT_ENTRY_NONE);
continue;
}
@@ -757,9 +757,9 @@ static void notify_delta_merged(struct ploop *ploop, u8 level,
*/
WRITE_ONCE(bat_entries[i], READ_ONCE(d_bat_entries[i]));
if (!forward)
- md->bat_levels[i] = level - 1;
+ WRITE_ONCE(md->bat_levels[i], level - 1);
else
- md->bat_levels[i] = level;
+ WRITE_ONCE(md->bat_levels[i], level);
}
if (stop)
break;
@@ -985,13 +985,13 @@ static int process_flip_upper_deltas(struct ploop *ploop)
ploop_init_be_iter(ploop, md->id, &i, &end);
bat_entries = md->kmpage;
for (; i <= end; i++) {
- if (bat_entries[i] == BAT_ENTRY_NONE)
+ if (READ_ONCE(bat_entries[i]) == BAT_ENTRY_NONE)
continue;
- if (md->bat_levels[i] == level) {
- md->bat_levels[i] = ploop_top_level(ploop);
+ if (READ_ONCE(md->bat_levels[i]) == level) {
+ WRITE_ONCE(md->bat_levels[i], ploop_top_level(ploop));
clear_bit(bat_entries[i], holes_bitmap);
- } else if (md->bat_levels[i] == ploop_top_level(ploop)) {
- md->bat_levels[i] = level;
+ } else if (READ_ONCE(md->bat_levels[i]) == ploop_top_level(ploop)) {
+ WRITE_ONCE(md->bat_levels[i], level);
}
}
}
diff --git a/drivers/md/dm-ploop-map.c b/drivers/md/dm-ploop-map.c
index 634f0679be3f..221e97ba84a9 100644
--- a/drivers/md/dm-ploop-map.c
+++ b/drivers/md/dm-ploop-map.c
@@ -723,9 +723,9 @@ static void ploop_release_cluster(struct ploop *ploop, u32 clu)
clu = ploop_bat_clu_idx_in_page(clu); /* relative to page */
bat_entries = md->kmpage;
- dst_clu = bat_entries[clu];
- bat_entries[clu] = BAT_ENTRY_NONE;
- md->bat_levels[clu] = 0;
+ dst_clu = READ_ONCE(bat_entries[clu]);
+ WRITE_ONCE(bat_entries[clu], BAT_ENTRY_NONE);
+ WRITE_ONCE(md->bat_levels[clu], 0);
ploop_hole_set_bit(dst_clu, ploop);
}
@@ -793,8 +793,8 @@ static void ploop_advance_local_after_bat_wb(struct ploop *ploop,
}
if (success) {
- bat_entries[i] = dst_clu[i];
- md->bat_levels[i] = ploop_top_level(ploop);
+ WRITE_ONCE(bat_entries[i], READ_ONCE(dst_clu[i]));
+ WRITE_ONCE(md->bat_levels[i], ploop_top_level(ploop));
} else {
ploop_hole_set_bit(i + off, ploop);
}
@@ -977,7 +977,7 @@ static void ploop_bat_page_zero_cluster(struct ploop *ploop,
clu = ploop_bat_clu_idx_in_page(clu);
to = kmap_local_page(piwb->bat_page);
- to[clu] = 0;
+ WRITE_ONCE(to[clu], 0);
kunmap_local(to);
}
@@ -1085,14 +1085,16 @@ static int ploop_alloc_cluster(struct ploop *ploop, struct ploop_index_wb *piwb,
bool already_alloced = false;
map_index_t *to;
int ret = 0;
+ u32 tmp_clu;
/* Cluster index related to the page[page_id] start */
clu -= piwb->page_id * PAGE_SIZE / sizeof(map_index_t) - PLOOP_MAP_OFFSET;
to = kmap_local_page(piwb->bat_page);
- if (to[clu]) {
+ tmp_clu = READ_ONCE(to[clu]);
+ if (tmp_clu) {
/* Already mapped by one of previous bios */
- *dst_clu = to[clu];
+ *dst_clu = tmp_clu;
already_alloced = true;
}
kunmap_local(to);
@@ -1105,7 +1107,7 @@ static int ploop_alloc_cluster(struct ploop *ploop, struct ploop_index_wb *piwb,
goto out;
to = kmap_local_page(piwb->bat_page);
- to[clu] = *dst_clu;
+ WRITE_ONCE(to[clu], *dst_clu);
kunmap_local(to);
out:
return ret;
@@ -1388,7 +1390,7 @@ static void ploop_submit_cow_index_wb(struct ploop_cow *cow)
to = kmap_local_page(piwb->bat_page);
WARN_ON(to[clu]);
- to[clu] = cow->dst_clu;
+ WRITE_ONCE(to[clu], cow->dst_clu);
kunmap_local(to);
/* Prevent double clearing of holes_bitmap bit on complete_cow() */
@@ -1725,7 +1727,7 @@ static void ploop_process_one_discard_pio(struct ploop *ploop, struct pio *pio)
pio->bi_status = BLK_STS_IOERR;
goto err;
} else {
- to[clu] = 0;
+ WRITE_ONCE(to[clu], 0);
list_add_tail(&pio->list, &piwb->ready_data_pios);
}
kunmap_local(to);
diff --git a/drivers/md/dm-ploop.h b/drivers/md/dm-ploop.h
index 3783d42a6162..b42baccf9a00 100644
--- a/drivers/md/dm-ploop.h
+++ b/drivers/md/dm-ploop.h
@@ -427,7 +427,7 @@ static inline u32 ploop_bat_entries(struct ploop *ploop, u32 clu,
clu = ploop_bat_clu_idx_in_page(clu);
if (bat_level)
- *bat_level = md->bat_levels[clu];
+ *bat_level = READ_ONCE(md->bat_levels[clu]);
if (md_ret)
*md_ret = md;
More information about the Devel
mailing list