If we have two processes with different ioprio_class, but the same
ioprio_data, their async requests will fall into the same queue. I guess
such behavior is not expected, because it's not right to put real-time
requests and best-effort requests in the same queue.
The attached patch fixes the problem by introducing additional *cfqq
fields on cfqd, pointing to per-(class,priority) async queues.
Signed-off-by: Jens Axboe <[email protected]>
OpenVZ Linux kernel team has discovered the problem with 32bit quota tools
working on 64bit architectures. In 2.6.10 kernel sys32_quotactl() function
was replaced by sys_quotactl() with the comment "sys_quotactl seems to be
32/64bit clean, enable it for 32bit" However this isn't right. Look at
if_dqblk structure:
struct if_dqblk {
__u64 dqb_bhardlimit;
__u64 dqb_bsoftlimit;
__u64 dqb_curspace;
__u64 dqb_ihardlimit;
__u64 dqb_isoftlimit;
__u64 dqb_curinodes;
__u64 dqb_btime;
__u64 dqb_itime;
__u32 dqb_valid;
};
For 32 bit quota tools sizeof(if_dqblk) == 0x44.
But for 64 bit kernel its size is 0x48, 'cause of alignment!
Thus we got a problem. Attached patch reintroduce sys32_quotactl() function,
that handles this and related situations.
[[email protected]: build fix]
[[email protected]: Make it link with CONFIG_QUOTA=n]
Signed-off-by: Vasily Tarasov <[email protected]>
Cc: Andi Kleen <[email protected]>
Cc: "Luck, Tony" <[email protected]>
Cc: Jan Kara <[email protected]>
Cc: <[email protected]>
Signed-off-by: Michal Piotrowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
cfq hash is no more necessary. We always can get cfqq from io context.
cfq_get_io_context_noalloc() function is introduced, because we don't
want to allocate cic on merging and checking may_queue. In order to
identify sync queue we've used hash key = CFQ_KEY_ASYNC. Since hash is
eliminated we need to use other criterion: sync flag for queue is added.
In all places where we dig in rb_tree we're in current context, so no
additional locking is required.
Advantages of this patch: no additional memory for hash, no seeking in
hash, code is cleaner. But it is necessary now to seek cic in per-ioc
rbtree, but it is faster:
- most processes work only with few devices
- most systems have only few block devices
- it is a rb-tree
Signed-off-by: Vasily Tarasov <[email protected]>
Changes by me:
- Merge into CFQ devel branch
- Get rid of cfq_get_io_context_noalloc()
- Fix various bugs with dereferencing cic->cfqq[] with offset other
than 0 or 1.
- Fix bug in cfqq setup, is_sync condition was reversed.
- Fix bug where only bio_sync() is used, we need to check for a READ too
Signed-off-by: Jens Axboe <[email protected]>
There is a small problem in handling page bounce.
At the moment blk_max_pfn equals max_pfn, which is in fact not maximum
possible _number_ of a page frame, but the _amount_ of page frames. For
example for the 32bit x86 node with 4Gb RAM, max_pfn = 0x100000, but not
0xFFFF.
request_queue structure has a member q->bounce_pfn and queue needs bounce
pages for the pages _above_ this limit. This routine is handled by
blk_queue_bounce(), where the following check is produced:
if (q->bounce_pfn >= blk_max_pfn)
return;
Assume, that a driver has set q->bounce_pfn to 0xFFFF, but blk_max_pfn
equals 0x10000. In such situation the check above fails and for each bio
we always fall down for iterating over pages tied to the bio.
I want to notice, that for quite a big range of device drivers (ide, md,
...) such problem doesn't happen because they use BLK_BOUNCE_ANY for
bounce_pfn. BLK_BOUNCE_ANY is defined as blk_max_pfn << PAGE_SHIFT, and
then the check above doesn't fail. But for other drivers, which obtain
reuired value from drivers, it fails. For example sata_nv uses
ATA_DMA_MASK or dev->dma_mask.
I propose to use (max_pfn - 1) for blk_max_pfn. And the same for
blk_max_low_pfn. The patch also cleanses some checks related with
bounce_pfn.
Signed-off-by: Vasily Tarasov <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
Currently ioprio_best function first checks wethere aioprio or bioprio equals
IOPRIO_CLASS_NONE (ioprio_valid() macros does that) and if it is so it returns
bioprio/aioprio appropriately. Thus the next four lines, that set aclass/bclass
to IOPRIO_CLASS_BE, if aclass/bclass == IOPRIO_CLASS_NONE, are never executed.
The second problem: if aioprio from class IOPRIO_CLASS_NONE and bioprio from
class IOPRIO_CLASS_IDLE are passed to ioprio_best function, it will return
IOPRIO_CLASS_IDLE. It means that during __make_request we can merge two
requests and set the priority of merged request to IDLE, while one of
the initial requests originates from a process with NONE (default) priority.
So we can get a situation when a process with default ioprio will experience
IO starvation, while there is no process from real-time class in the system.
Just removing ioprio_valid check should correct situation.
Signed-off-by: Vasily Tarasov <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
elv_iosched_show function iterates other elv_list, hence
elv_list_lock should be got.
Signed-off-by: Vasily Tarasov <[email protected]>
Signed-off-by: Vasily Tarasov <[email protected]>
We can easily produce search through the elevator list
without introducing additional elevator_type variable.
Signed-off-by: Vasily Tarasov <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
Vasily Tarasov