Thread (14 messages) 14 messages, 2 authors, 2012-03-21

Re: [PATCH 00/11] Various cleanups and minor fixes

flat view

From: NeilBrown <hidden>
Date: 2012-03-20 21:11:30

On Tue, 20 Mar 2012 17:54:05 +0100 Jes.Sorensen@redhat.com wrote:
From: Jes Sorensen <redacted>

Hi,

This patch set implements a number of smaller cleanups, as well as
fixing a couple of cases where we don't zero out the full buffer we
just obtained via malloc.
All applied and push out - thanks.  There were a few conflict which some
stuff I was working on, but nothing serious.
The most controversial is probably patch 06 which caches the device
block size for aligned read/writes in super1.c to avoid calling the
ioctl() on each write.
Yeah, it's a bit ugly, but seems reasonable.
I also introduced some generalized ROUND_UP macros to avoid the manual
pointer and size manipulation in various places.

The ddf code I am unable to test as I don't have any raids with ddf
layout, but it does compile.

Cheers,
Jes
I think it might be nearly time for 3.2.4.  Do you have any thoughts about
that?

Thanks,
NeilBrown


Jes Sorensen (11):
  super1.c don't keep recalculating bitmap pointer
  Define and use SUPER1_SIZE for allocations
  init_super1() memset full buffer allocated for superblock
  match_metadata_desc1(): Use calloc instead of malloc+memset
  Use 4K buffer alignment for superblock allocations
  Use struct align_fd to cache fd's block size for aligned reads/writes
  match_metadata_desc0(): Use calloc instead of malloc+memset
  Generalize ROUND_UP() macro and introduce matching ROUND_UP_PTR()
  super1.c: use ROUND_UP/ROUND_UP_PTR
  super-intel.c: Use ROUND_UP() instead of manually coding it
  __write_init_super_ddf(): Use posix_memalign() instead of static
    aligned buffer

 mdadm.h       |    8 ++--
 super-ddf.c   |   19 ++++++----
 super-intel.c |    2 +-
 super0.c      |    6 ++--
 super1.c      |  111 +++++++++++++++++++++++++++++++++++----------------------
 5 files changed, 87 insertions(+), 59 deletions(-)
  

Attachments

Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help