Bcache fixes for 3.10

Hey Jens, this is everything I've got ready for 3.10 - there's _still_ one more bug I'm trying to track down.

Andrew - I've got patches that rip out the pkey() and pbtree() macros, but they're somewhat tied up with some other nontrivial refactorings so I think I'm going to wait a bit on those.

The following changes since commit cef5279735d3f6f0243e626963e6d5c84efade0a:

bcache: Disable broken btree fuzz tester (2013-04-08 13:33:49 -0700)

are available in the git repository at:

http://evilpiepirate.org/git/linux-bcache.git bcache-for-upstream

for you to fetch changes up to cd66f1f69ed40270bb34e20b25a8111bc72a1b2a:

bcache: Use bd_link_disk_holder() (2013-04-24 13:15:17 -0700)

----------------------------------------------------------------
Kent Overstreet (8):
bcache: Take data offset from the bdev superblock.
bcache: Set ra_pages based on backing device's ra_pages
bcache: Hack around stuff that clones up to bi_max_vecs
bcache: Correctly check against BIO_MAX_PAGES
bcache: Fix merge_bvec_fn usage for when it modifies the bvm
bcache: Make sure blocksize isn't smaller than device blocksize
bcache: Allocator cleanup/fixes
bcache: Use bd_link_disk_holder()

 drivers/md/bcache/alloc.c   |  74 +++++++++++++-------
 drivers/md/bcache/bcache.h  |  48 ++++++++++---
 drivers/md/bcache/btree.c   |   8 +--
 drivers/md/bcache/io.c      |  35 ++++++----
 drivers/md/bcache/request.c |   2 +-
 drivers/md/bcache/super.c   | 166 +++++++++++++++++++++++++++-----------------
 6 files changed, 215 insertions(+), 118 deletions(-)

Quoted 30 lines · Kent Overstreet

Hey Jens, this is everything I've got ready for 3.10 - there's _still_ one more bug I'm trying to track down.

Andrew - I've got patches that rip out the pkey() and pbtree() macros, but they're somewhat tied up with some other nontrivial refactorings so I think I'm going to wait a bit on those.

The following changes since commit cef5279735d3f6f0243e626963e6d5c84efade0a:

bcache: Disable broken btree fuzz tester (2013-04-08 13:33:49 -0700)

are available in the git repository at:

http://evilpiepirate.org/git/linux-bcache.git bcache-for-upstream

for you to fetch changes up to cd66f1f69ed40270bb34e20b25a8111bc72a1b2a:

bcache: Use bd_link_disk_holder() (2013-04-24 13:15:17 -0700)

----------------------------------------------------------------
Kent Overstreet (8):
bcache: Take data offset from the bdev superblock.
bcache: Set ra_pages based on backing device's ra_pages
bcache: Hack around stuff that clones up to bi_max_vecs
bcache: Correctly check against BIO_MAX_PAGES
bcache: Fix merge_bvec_fn usage for when it modifies the bvm
bcache: Make sure blocksize isn't smaller than device blocksize
bcache: Allocator cleanup/fixes
bcache: Use bd_link_disk_holder()

Quoted 7 lines

 drivers/md/bcache/alloc.c   |  74 +++++++++++++-------
 drivers/md/bcache/bcache.h  |  48 ++++++++++---
 drivers/md/bcache/btree.c   |   8 +--
 drivers/md/bcache/io.c      |  35 ++++++----
 drivers/md/bcache/request.c |   2 +-
 drivers/md/bcache/super.c   | 166 +++++++++++++++++++++++++++-----------------
 6 files changed, 215 insertions(+), 118 deletions(-)

This is what I got, when I pulled it in:

Merge made by the 'recursive' strategy.

 drivers/md/bcache/alloc.c   |  72 +++++++++++++------
 drivers/md/bcache/bcache.h  |  47 ++++++++++---
 drivers/md/bcache/btree.c   |   3 +-
 drivers/md/bcache/io.c      |  35 ++++++----
 drivers/md/bcache/request.c |   2 +-
 drivers/md/bcache/super.c   | 166 +++++++++++++++++++++++++++-----------------
 6 files changed, 213 insertions(+), 112 deletions(-)

I only see the same 8 commits there. What happened? Please check for-3.10/drivers and verify it looks as you expect.

Quoted 50 lines · Jens Axboe

Quoted 30 lines · Kent Overstreet

Hey Jens, this is everything I've got ready for 3.10 - there's _still_ one more bug I'm trying to track down.

Andrew - I've got patches that rip out the pkey() and pbtree() macros, but they're somewhat tied up with some other nontrivial refactorings so I think I'm going to wait a bit on those.

The following changes since commit cef5279735d3f6f0243e626963e6d5c84efade0a:

bcache: Disable broken btree fuzz tester (2013-04-08 13:33:49 -0700)

are available in the git repository at:

http://evilpiepirate.org/git/linux-bcache.git bcache-for-upstream

for you to fetch changes up to cd66f1f69ed40270bb34e20b25a8111bc72a1b2a:

bcache: Use bd_link_disk_holder() (2013-04-24 13:15:17 -0700)

----------------------------------------------------------------
Kent Overstreet (8):
bcache: Take data offset from the bdev superblock.
bcache: Set ra_pages based on backing device's ra_pages
bcache: Hack around stuff that clones up to bi_max_vecs
bcache: Correctly check against BIO_MAX_PAGES
bcache: Fix merge_bvec_fn usage for when it modifies the bvm
bcache: Make sure blocksize isn't smaller than device blocksize
bcache: Allocator cleanup/fixes
bcache: Use bd_link_disk_holder()

Quoted 7 lines

 drivers/md/bcache/alloc.c   |  74 +++++++++++++-------
 drivers/md/bcache/bcache.h  |  48 ++++++++++---
 drivers/md/bcache/btree.c   |   8 +--
 drivers/md/bcache/io.c      |  35 ++++++----
 drivers/md/bcache/request.c |   2 +-
 drivers/md/bcache/super.c   | 166 +++++++++++++++++++++++++++-----------------
 6 files changed, 215 insertions(+), 118 deletions(-)

This is what I got, when I pulled it in:

Merge made by the 'recursive' strategy.

 drivers/md/bcache/alloc.c   |  72 +++++++++++++------
 drivers/md/bcache/bcache.h  |  47 ++++++++++---
 drivers/md/bcache/btree.c   |   3 +-
 drivers/md/bcache/io.c      |  35 ++++++----
 drivers/md/bcache/request.c |   2 +-
 drivers/md/bcache/super.c   | 166 +++++++++++++++++++++++++++-----------------
 6 files changed, 213 insertions(+), 112 deletions(-)

Quoted 2 lines

I only see the same 8 commits there. What happened? Please check for-3.10/drivers and verify it looks as you expect.

Oh, were you expecting me to rebase? Result looks good, though.

Nope, in fact your series was on top of the last commit I got from you, cef52797. Only non-bcache in between. Just wondering how that 215 insertions and 118 deletions becames 213/112 for that case.

Quoted 3 lines · Jens Axboe

Nope, in fact your series was on top of the last commit I got from you, cef52797. Only non-bcache in between. Just wondering how that 215 insertions and 118 deletions becames 213/112 for that case.

Oh, that. That was a last minute bugfix, just backed out part of a change. Sorry :P