Skip to contentSkip to conversation

block fixes for 3.9-rc

GIT PULL Jens Axboe 2 participants 6 messages Mar 29, 2013

lore
Share & tools
lore

Jens Axboe

author
#0 lore raw

Hi Linus,

Collection of fixes that have been queued up since the merge window opened, hence postponed until later in the cycle. The pull request contains:

- A bunch of fixes for the xen blk front/back driver.

- A round of fixes for the new IBM RamSan driver, fixing various nasty issues.

- Fixes for multiple drives from Wei Yongjun, bad handling of return values and wrong pointer math.

- A fix for loop properly killing partitions when being detached.

Please pull! As per usual, it's tagged and signed.

git://git.kernel.dk/linux-block.git for-linus

----------------------------------------------------------------
Alice Ferrazzi (1):
Block: blk-flush: Fixed indent code style

Chen Gang (1):
xen/xen-blkback: preq.dev is used without initialized

David Vrabel (1):
xen/blkback: correctly respond to unknown, non-native requests

Jan Beulich (1):
xen-blkback: fix dispatch_rw_block_io() error path

Jens Axboe (2):
rsxx: fix missing unlock on error return in rsxx_eeh_remap_dmas()
Merge branch 'stable/for-jens-3.9' of git://git.kernel.org/.../konrad/xen into for-linus

Mihnea Dobrescu-Balaur (1):
xen-blkfront: replace kmalloc and then memcpy with kmemdup

Philip J Kelleher (8):
block: IBM RamSan 70/80 trivial changes.
block: IBM RamSan 70/80 fixes inconsistent locking.
block: IBM RamSan 70/80 branding changes.
block: IBM RamSan 70/80 error message bug fix.
Adding in EEH support to the IBM FlashSystem 70/80 device driver
block: removes dynamic allocation on stack
rsxx: enable error return of rsxx_eeh_save_issued_dmas()
rsxx: remove unused variable

Phillip Susi (1):
loop: cleanup partitions when detaching loop device

Roger Pau Monne (5):
xen-blkback: fix foreach_grant_safe to handle empty lists
xen-blkfront: switch from llist to list
xen-blkback: don't store dev_bus_addr
xen-blkfront: pre-allocate pages for requests
xen-blkfront: remove frame list from blk_shadow

Wei Yongjun (4):
mtip32xx: fix error return code in mtip_pci_probe()
loop: fix error return code in loop_add()
cciss: fix invalid use of sizeof in cciss_find_cfgtables()
mg_disk: fix error return code in mg_probe()

Zoltan Kiss (1):
xen/blkback: Change statistics counter types to unsigned

 MAINTAINERS                         |   12 +-
 block/blk-flush.c                   |    2 +-
 block/partition-generic.c           |    1 +
 drivers/block/Kconfig               |    4 +-
 drivers/block/cciss.c               |    2 +-
 drivers/block/loop.c                |   22 +++-
 drivers/block/mg_disk.c             |    4 +-
 drivers/block/mtip32xx/mtip32xx.c   |    4 +-
 drivers/block/rsxx/Makefile         |    2 +-
 drivers/block/rsxx/config.c         |    8 +-
 drivers/block/rsxx/core.c           |  237 ++++++++++++++++++++++++++++++++--
 drivers/block/rsxx/cregs.c          |  112 ++++++++++------
 drivers/block/rsxx/dma.c            |  239 ++++++++++++++++++++++++-----------
 drivers/block/rsxx/rsxx.h           |    6 +-
 drivers/block/rsxx/rsxx_cfg.h       |    2 +-
 drivers/block/rsxx/rsxx_priv.h      |   34 ++++-
 drivers/block/xen-blkback/blkback.c |   68 +++++-----
 drivers/block/xen-blkback/common.h  |   40 ++++--
 drivers/block/xen-blkback/xenbus.c  |   14 +-
 drivers/block/xen-blkfront.c        |  154 +++++++++++++---------
 include/xen/interface/io/blkif.h    |   10 ++
 21 files changed, 716 insertions(+), 261 deletions(-)
Signature · 1 line
Jens Axboe
#1 lore raw
Quoted 2 lines · Jens Axboe

Please pull! As per usual, it's tagged and signed.

No it's not.

You have a *branch* called for-linus that contains the commits you mention.

And then you have a *tag* called for-linus, but that's just my old head that you've gone ahead and signed, which is entirely and utterly pointless.

Christ, the amount of confusion in that tree. Don't alias tags and branches. And if you do, make it clear which one you mean by *saying* so, ie "tags/for-linus", not just the ambiguous "for-linus". Now you've confused yourself, and because of the ambiguity you never even noticed, because the *branch* for-linus had the right commits.

Don't do this kind of thing. That branch is pointless, and just confused you.

Linus

author
#2 lore raw
Quoted 11 lines · Linus Torvalds
Quoted 2 lines · Jens Axboe

Please pull! As per usual, it's tagged and signed.

No it's not.

You have a *branch* called for-linus that contains the commits you mention.

And then you have a *tag* called for-linus, but that's just my old head that you've gone ahead and signed, which is entirely and utterly pointless.

Oops, that was a mistake. I must have been on the wrong branch at that point. In past requests, I've signed the for-linus (or topic) branch specifically.

Quoted 5 lines

Christ, the amount of confusion in that tree. Don't alias tags and branches. And if you do, make it clear which one you mean by *saying* so, ie "tags/for-linus", not just the ambiguous "for-linus". Now you've confused yourself, and because of the ambiguity you never even noticed, because the *branch* for-linus had the right commits.

Agree, the aliasing is nasty and git checkout usually doesn't like it a whole lot either...

Quoted 2 lines

Don't do this kind of thing. That branch is pointless, and just confused you.

Tag, you mean?

Signature · 1 line
Jens Axboe
#3 lore raw
Quoted 5 lines · Jens Axboe
Quoted 3 lines

Don't do this kind of thing. That branch is pointless, and just confused you.

Tag, you mean?

No. Branch. Why would you have a public branch called "for-linus", when you intend to tag the end result before sending it to me?

It would make much more sense if you just did your development in your "master" branch, perhaps with a separate branch for fixes during stabilization (so that you don't mix up fixes with your future development. Call it "fixes" or "stable" or whatever. No "for-linus" branches anywhere. After all, the branch isn't for me at all. It's for me only once you're ready. And at that point you tag it, so having a *tag* called for-linus makes sense and gives us the nice signing etc.

Hmm?

Linus

author
#4 lore raw
Quoted 9 lines · Linus Torvalds
Quoted 5 lines · Jens Axboe
Quoted 3 lines

Don't do this kind of thing. That branch is pointless, and just confused you.

Tag, you mean?

No. Branch. Why would you have a public branch called "for-linus", when you intend to tag the end result before sending it to me?

Outside of "that's what I have always done", the intent was for others that send patches through me to know what to base it off, if not previous release. All those folks know that for-3.x/drivers is for the next x release, and that for-linus is for the existing tree.

Quoted 7 lines

It would make much more sense if you just did your development in your "master" branch, perhaps with a separate branch for fixes during stabilization (so that you don't mix up fixes with your future development. Call it "fixes" or "stable" or whatever. No "for-linus" branches anywhere. After all, the branch isn't for me at all. It's for me only once you're ready. And at that point you tag it, so having a *tag* called for-linus makes sense and gives us the nice signing etc.

I see your point, it's not for you YET but it will be soon. I'd rather just keep that naming to avoid confusion with others, but tag it appropriately (for-linus-<date>?) when it's headed your way. Then use that proper signed tag name as the reference to you.

I prefer keeping 'master' pristine.

Signature · 1 line
Jens Axboe
#5 lore raw
Quoted 5 lines · Jens Axboe

Outside of "that's what I have always done", the intent was for others that send patches through me to know what to base it off, if not previous release. All those folks know that for-3.x/drivers is for the next x release, and that for-linus is for the existing tree.

Ok. Whatever. Feel free to continue doing it, but at least then make sure that the pull message refers to "tags/for-linus" and not that ambiguous "for-linus" thing.

Because you screwed up that pull request, and I argue that you screwed up exactly *because* it's ambiguous and confusing. The confusion may have happened before too, but because git will silently default to the branch for the ambiguous case, I just never noticed that I didn't actually get the proper signature. Now I noticed, because you said it was signed and yet my pull got the non-signed thing.

Linus