This one was the working commit for 3.10:
>From 023d8db36affa2489a83052dbb7c969d0941f06f Mon Sep 17 00:00:00 2001
From: Kent Overstreet <koverstreet@xxxxxxxxxx>
Date: Wed, 26 Jun 2013 17:25:38 -0700
Subject: [PATCH 17/17] bcache: FUA fixes
Journal writes need to be marked FUA, not just REQ_FLUSH. And btree node
writes have... weird ordering requirements.
Signed-off-by: Kent Overstreet <koverstreet@xxxxxxxxxx>
---
drivers/md/bcache/btree.c | 19 +++++++++++++++++--
drivers/md/bcache/journal.c | 2 +-
2 files changed, 18 insertions(+), 3 deletions(-)
diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c
index 09fb8a2..803af59 100644
--- a/drivers/md/bcache/btree.c
+++ b/drivers/md/bcache/btree.c
@@ -328,10 +328,25 @@ static void do_btree_node_write(struct btree *b)
b->bio->bi_end_io = btree_node_write_endio;
b->bio->bi_private = &b->io.cl;
- b->bio->bi_rw = REQ_META|WRITE_SYNC;
- b->bio->bi_size = set_blocks(i, b->c) * block_bytes(b->c);
+ b->bio->bi_rw = REQ_META|WRITE_SYNC|REQ_FUA;
+ b->bio->bi_size = set_blocks(i, b->c) * block_bytes(b->c);
bch_bio_map(b->bio, i);
+ /*
+ * If we're appending to a leaf node, we don't technically need
FUA -
+ * this write just needs to be persisted before the next journal
write,
+ * which will be marked FLUSH|FUA.
+ *
+ * Similarly if we're writing a new btree root - the pointer is
going to
+ * be in the next journal entry.
+ *
+ * But if we're writing a new btree node (that isn't a root) or
+ * appending to a non leaf btree node, we need either FUA or a flush
+ * when we write the parent with the new pointer. FUA is cheaper
than a
+ * flush, and writes appending to leaf nodes aren't blocking
anything so
+ * just make all btree node writes FUA to keep things sane.
+ */
+
bkey_copy(&k.key, &b->key);
SET_PTR_OFFSET(&k.key, 0, PTR_OFFSET(&k.key, 0) + bset_offset(b,
i));
diff --git a/drivers/md/bcache/journal.c b/drivers/md/bcache/journal.c
index 5ca2214..4b25066 100644
--- a/drivers/md/bcache/journal.c
+++ b/drivers/md/bcache/journal.c
@@ -620,7 +620,7 @@ static void journal_write_unlocked(struct closure *cl)
bio_reset(bio);
bio->bi_sector = PTR_OFFSET(k, i);
bio->bi_bdev = ca->bdev;
- bio->bi_rw = REQ_WRITE|REQ_SYNC|REQ_META|REQ_FLUSH;
+ bio->bi_rw =
REQ_WRITE|REQ_SYNC|REQ_META|REQ_FLUSH|REQ_FUA;
bio->bi_size = sectors << 9;
bio->bi_end_io = journal_write_endio;
--
1.7.10.4
Stefan
Am 12.07.2013 10:42, schrieb Stefan Priebe - Profihost AG:
> it's this one:
> - bch_writeback_add(dc, bio_sectors(bio));
> + bch_writeback_add(dc);
>
> from cache: FUA fixes
>
> Journal writes need to be marked FUA, not just REQ_FLUSH. And btree node
> writes have... weird ordering requirements.
>
> in 3.10 bch_writeback_add still needs to arguments.
>
> Stefan
>
> Am 12.07.2013 10:38, schrieb Stefan Priebe - Profihost AG:
>> Does not seem to build:
>> CC [M] drivers/md/bcache/request.o
>> drivers/md/bcache/request.c: In function ‘request_write’:
>> drivers/md/bcache/request.c:1057: error: too few arguments to function
>> ‘bch_writeback_add’
>> make[3]: *** [drivers/md/bcache/request.o] Error 1
>> make[2]: *** [drivers/md/bcache] Error 2
>>
>> Am 12.07.2013 10:31, schrieb Stefan Priebe - Profihost AG:
>>> Am 12.07.2013 10:19, schrieb Kent Overstreet:
>>>> On Fri, Jul 12, 2013 at 08:11:45AM +0200, Stefan Priebe - Profihost AG wrote:
>>>>> Am 12.07.2013 03:40, schrieb Kent Overstreet:
>>>>>> On Thu, Jul 04, 2013 at 01:45:42PM +0200, Stefan Priebe - Profihost AG wrote:
>>>>>>> fix is this one: http://pastebin.com/raw.php?i=LBFJRvEt
>>>>>>
>>>>>> Oh I see, the patch that introduced this is an old version of the patch
>>>>>> - the bcache-for-3.11 had this fixed, and I forgot to update that
>>>>>> branch. Thanks, I'll apply it there.
>>>>>
>>>>> thanks - which commit id is the fix in? i can't find it.
>>>>
>>>> The patch that introduced it wasn't ever destined for 3.10 anyways, and
>>>> I just deleted the bcache/bcache-3.9 branches because they'd gotten out
>>>> of sync and weren't getting updated with bcache-dev/bcache-for-3.11.
>>>
>>> Thanks!
>>>
>>>> So just stick with bcache-for-3.10, I just backported all the important
>>>> bugfixes to it - I'll be sending that stuff to Greg K-H for 3.10.1 in
>>>> the next couple days too.
>>> Thanks!
>>>
>>> Stefan
>>>
--
To unsubscribe from this list: send the line "unsubscribe linux-bcache" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
