Needed to bring blk-mq uptodate, since changes have been going in
since for-3.14/core was established.
Fixup merge issues related to the immutable biovec changes.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Conflicts:
block/blk-flush.c
fs/btrfs/check-integrity.c
fs/btrfs/extent_io.c
fs/btrfs/scrub.c
fs/logfs/dev_bdev.c
In the blkif_release function the bdget_disk() call might returns
a NULL ptr which might be dereferenced on bdev->bd_openers checking
Signed-off-by: Felipe Pena <felipensp@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
[v2: Added WARN per Roger's suggestion]
pfn cannot actually be used unless (!info->feature_persistent), nor is
pfn accessed in get_grant() unless (!info->feature_persistent), but silence
this warning anyway. gcc-4.8
drivers/block/xen-blkfront.c: In function 'do_blkif_request':
drivers/block/xen-blkfront.c:508:20: warning: 'pfn' may be used uninitialized in this function [-Wmaybe-uninitialized]
gnt_list_entry = get_grant(&gref_head, pfn, info);
^
drivers/block/xen-blkfront.c:492:19: note: 'pfn' was declared here
unsigned long pfn;
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
When persistent grants were added they were always used, even if the
backend doesn't have this feature (there's no harm in always using the
same set of pages). This restores the old data path when the backend
doesn't have persistent grants, removing the burden of doing a memcpy
when it is not actually needed.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reported-by: Felipe Franciosi <felipe.franciosi@citrix.com>
Cc: Felipe Franciosi <felipe.franciosi@citrix.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
[v2: Fix up whitespace issues]
Improve the calculation of required grants to process a request by
using nr_phys_segments instead of always assuming a request is going
to use all posible segments.
nr_phys_segments contains the number of scatter-gather DMA addr+len
pairs, which is basically what we put at every granted page.
for_each_sg iterates over the DMA addr+len pairs and uses a grant
page for each of them.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: David Vrabel <david.vrabel@citrix.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
There's no need to keep the foreign access in a grant if it is not
persistently mapped by the backend. This allows us to free grants that
are not mapped by the backend, thus preventing blkfront from hoarding
all grants.
The main effect of this is that blkfront will only persistently map
the same grants as the backend, and it will always try to use grants
that are already mapped by the backend. Also the number of persistent
grants in blkfront is the same as in blkback (and is controlled by the
value in blkback).
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: David Vrabel <david.vrabel@citrix.com>
Acked-by: Matt Wilson <msw@amazon.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Konrad writes:
It has the 'feature-max-indirect-segments' implemented in both backend
and frontend. The current problem with the backend and frontend is that the
segment size is limited to 11 pages. It means we can at most squeeze in 44kB per
request. The ring can hold 32 (next power of two below 36) requests, meaning we
can do 1.4M of outstanding requests. Nowadays that is not enough.
The problem in the past was addressed in two ways - but neither one went upstream.
The first solution to this proposed by Justin from Spectralogic was to negotiate
the segment size. This means that the ‘struct blkif_sring_entry’ is now a variable size.
It can expand from 112 bytes (cover 11 pages of data - 44kB) to 1580 bytes
(256 pages of data - so 1MB). It is a simple extension by just making the array in the
request expand from 11 to a variable size negotiated. But it had limits: this extension
still limits the number of segments per request to 255 (as the total number must be
specified in the request, which only has an 8-bit field for that purpose).
The other solution (from Intel - Ronghui) was to create one extra ring that only has the
‘struct blkif_request_segment’ in them. The ‘struct blkif_request’ would be changed to have
an index in said ‘segment ring’. There is only one segment ring. This means that the size of
the initial ring is still the same. The requests would point to the segment and enumerate out
how many of the indexes it wants to use. The limit is of course the size of the segment.
If one assumes a one-page segment this means we can in one request cover ~4MB.
Those patches were posted as RFC and the author never followed up on the ideas on changing
it to be a bit more flexible.
There is yet another mechanism that could be employed (which these patches implement) - and it
borrows from VirtIO protocol. And that is the ‘indirect descriptors’. This very similar to
what Intel suggests, but with a twist. The twist is to negotiate how many of these
'segment' pages (aka indirect descriptor pages) we want to support (in reality we negotiate
how many entries in the segment we want to cover, and we module the number if it is
bigger than the segment size).
This means that with the existing 36 slots in the ring (single page) we can cover:
32 slots * each blkif_request_indirect covers: 512 * 4096 ~= 64M. Since we ample space
in the blkif_request_indirect to span more than one indirect page, that number (64M)
can be also multiplied by eight = 512MB.
Roger Pau Monne took the idea and implemented them in these patches. They work
great and the corner cases (migration between backends with and without this extension)
work nicely. The backend has a limit right now off how many indirect entries
it can handle: one indirect page, and at maximum 256 entries (out of 512 - so 50% of the page
is used). That comes out to 32 slots * 256 entries in a indirect page * 1 indirect page
per request * 4096 = 32MB.
This is a conservative number that can change in the future. Right now it strikes
a good balance between giving excellent performance, memory usage in the backend, and
balancing the needs of many guests.
In the patchset there is also the split of the blkback structure to be per-VBD.
This means that the spinlock contention we had with many guests trying to do I/O and
all the blkback threads hitting the same lock has been eliminated.
Also there are bug-fixes to deal with oddly sized sectors, insane amounts on
th ring, and also a security fix (posted earlier).
Now that indirect segments are enabled blk_queue_max_hw_sectors must
be set to match the maximum number of sectors we can handle in a
request.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reported-by: Felipe Franciosi <felipe.franciosi@citrix.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Currently xen-blkback passes the logical sector size over xenbus and
xen-blkfront sets up the paravirt disk with that logical block size.
But newer drives usually have the logical sector size set to 512 for
compatibility reasons and would show the actual sector size only in
physical sector size.
This results in the device being partitioned and accessed in dom0 with
the correct sector size, but the guest thinks 512 bytes is the correct
block size. And that results in poor performance.
To fix this, blkback gets modified to pass also physical-sector-size
over xenbus and blkfront to use both values to set up the paravirt
disk. I did not just change the passed in sector-size because I am
not sure having a bigger logical sector size than the physical one
is valid (and that would happen if a newer dom0 kernel hits an older
domU kernel). Also this way a domU set up before should still be
accessible (just some tools might detect the unaligned setup).
[v2: Make xenbus write failure non-fatal]
[v3: Use xenbus_scanf instead of xenbus_gather]
[v4: Rebased against segment changes]
Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
The max module parameter (by default 32) is the maximum number of
segments that the frontend will negotiate with the backend for indirect
descriptors. Higher value means more potential throughput but more
memory usage. The backend picks the minimum of the frontend and its
default backend value.
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
In blkif_queue_request blkfront iterates over the scatterlist in order
to set the segments of the request, and in blkif_completion blkfront
iterates over the raw request, which makes it hard to know the exact
position of the source and destination memory positions.
This can be solved by allocating a scatterlist for each request, that
will be keep until the request is finished, allowing us to copy the
data back to the original memory without having to iterate over the
raw request.
Oracle-Bug: 16660413 - LARGE ASYNCHRONOUS READS APPEAR BROKEN ON 2.6.39-400
CC: stable@vger.kernel.org
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reported-and-Tested-by: Anne Milicia <anne.milicia@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
The value passed is 0 in all but "it can never happen" cases (and those
only in a couple of drivers) *and* it would've been lost on the way
out anyway, even if something tried to pass something meaningful.
Just don't bother.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Indirect descriptors introduce a new block operation
(BLKIF_OP_INDIRECT) that passes grant references instead of segments
in the request. This grant references are filled with arrays of
blkif_request_segment_aligned, this way we can send more segments in a
request.
The proposed implementation sets the maximum number of indirect grefs
(frames filled with blkif_request_segment_aligned) to 256 in the
backend and 32 in the frontend. The value in the frontend has been
chosen experimentally, and the backend value has been set to a sane
value that allows expanding the maximum number of indirect descriptors
in the frontend if needed.
The migration code has changed from the previous implementation, in
which we simply remapped the segments on the shared ring. Now the
maximum number of segments allowed in a request can change depending
on the backend, so we have to requeue all the requests in the ring and
in the queue and split the bios in them if they are bigger than the
new maximum number of segments.
[v2: Fixed minor comments by Konrad.
[v1: Added padding to make the indirect request 64bit aligned.
Added some BUGs, comments; fixed number of indirect pages in
blkif_get_x86_{32/64}_req. Added description about the indirect operation
in blkif.h]
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
[v3: Fixed spaces and tabs mix ups]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
We already have the frame (pfn of the grant page) stored inside struct
grant, so there's no need to keep an aditional list of mapped frames
for a specific request. This reduces memory usage in blkfront.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: xen-devel@lists.xen.org
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
This prevents us from having to call alloc_page while we are preparing
the request. Since blkfront was calling alloc_page with a spinlock
held we used GFP_ATOMIC, which can fail if we are requesting a lot of
pages since it is using the emergency memory pools.
Allocating all the pages at init prevents us from having to call
alloc_page, thus preventing possible failures.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: xen-devel@lists.xen.org
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
The git commit f84adf4921
(xen-blkfront: drop the use of llist_for_each_entry_safe)
was a stop-gate to fix a GCC4.1 bug. The appropiate way
is to actually use an list instead of using an llist.
As such this patch replaces the usage of llist with an
list.
Since we always manipulate the list while holding the io_lock, there's
no need for additional locking (llist used previously is safe to use
concurrently without additional locking).
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
CC: stable@vger.kernel.org
[v1: Redid the git commit description]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
The benefits are:
* code is cleaner
* kmemdup adds additional debugging info useful for tracking the real
place where memory was allocated (CONFIG_DEBUG_SLAB).
Signed-off-by: Mihnea Dobrescu-Balaur <mihneadb@gmail.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Replace llist_for_each_entry_safe with a while loop.
llist_for_each_entry_safe can trigger a bug in GCC 4.1, so it's best
to remove it and use a while loop and do the deletion manually.
Specifically this bug can be triggered by hot-unplugging a disk, either
by doing xm block-detach or by save/restore cycle.
BUG: unable to handle kernel paging request at fffffffffffffff0
IP: [<ffffffffa0047223>] blkif_free+0x63/0x130 [xen_blkfront]
The crash call trace is:
...
bad_area_nosemaphore+0x13/0x20
do_page_fault+0x25e/0x4b0
page_fault+0x25/0x30
? blkif_free+0x63/0x130 [xen_blkfront]
blkfront_resume+0x46/0xa0 [xen_blkfront]
xenbus_dev_resume+0x6c/0x140
pm_op+0x192/0x1b0
device_resume+0x82/0x1e0
dpm_resume+0xc9/0x1a0
dpm_resume_end+0x15/0x30
do_suspend+0x117/0x1e0
When drilling down to the assembler code, on newer GCC it does
.L29:
cmpq $-16, %r12 #, persistent_gnt check
je .L30 #, out of the loop
.L25:
... code in the loop
testq %r13, %r13 # n
je .L29 #, back to the top of the loop
cmpq $-16, %r12 #, persistent_gnt check
movq 16(%r12), %r13 # <variable>.node.next, n
jne .L25 #, back to the top of the loop
.L30:
While on GCC 4.1, it is:
L78:
... code in the loop
testq %r13, %r13 # n
je .L78 #, back to the top of the loop
movq 16(%rbx), %r13 # <variable>.node.next, n
jmp .L78 #, back to the top of the loop
Which basically means that the exit loop condition instead of
being:
&(pos)->member != NULL;
is:
;
which makes the loop unbound.
Since xen-blkfront is the only user of the llist_for_each_entry_safe
macro remove it from llist.h.
Orabug: 16263164
CC: stable@vger.kernel.org
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Currently blkfront fails to handle cases in blkif_completion like the
following:
1st loop in rq_for_each_segment
* bv_offset: 3584
* bv_len: 512
* offset += bv_len
* i: 0
2nd loop:
* bv_offset: 0
* bv_len: 512
* i: 0
In the second loop i should be 1, since we assume we only wanted to
read a part of the previous page. This patches fixes this cases where
only a part of the shared page is read, and blkif_completion assumes
that if the bv_offset of a bvec is less than the previous bv_offset
plus the bv_size we have to switch to the next shared page.
Reported-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Cc: linux-kernel@vger.kernel.org
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Implement a safe version of llist_for_each_entry, and use it in
blkif_free. Previously grants where freed while iterating the list,
which lead to dereferences when trying to fetch the next item.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
[v2: Move the llist_for_each_entry_safe in llist.h]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
This patch contains fixes for persistent grants implementation v2:
* handle == 0 is a valid handle, so initialize grants in blkback
setting the handle to BLKBACK_INVALID_HANDLE instead of 0. Reported
by Konrad Rzeszutek Wilk.
* new_map is a boolean, use "true" or "false" instead of 1 and 0.
Reported by Konrad Rzeszutek Wilk.
* blkfront announces the persistent-grants feature as
feature-persistent-grants, use feature-persistent instead which is
consistent with blkback and the public Xen headers.
* Add a consistency check in blkfront to make sure we don't try to
access segments that have not been set.
Reported-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Roger Pau Monne <roger.pau@citrix.com>
[v1: The new_map int->bool had already been changed]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
This patch implements persistent grants for the xen-blk{front,back}
mechanism. The effect of this change is to reduce the number of unmap
operations performed, since they cause a (costly) TLB shootdown. This
allows the I/O performance to scale better when a large number of VMs
are performing I/O.
Previously, the blkfront driver was supplied a bvec[] from the request
queue. This was granted to dom0; dom0 performed the I/O and wrote
directly into the grant-mapped memory and unmapped it; blkfront then
removed foreign access for that grant. The cost of unmapping scales
badly with the number of CPUs in Dom0. An experiment showed that when
Dom0 has 24 VCPUs, and guests are performing parallel I/O to a
ramdisk, the IPIs from performing unmap's is a bottleneck at 5 guests
(at which point 650,000 IOPS are being performed in total). If more
than 5 guests are used, the performance declines. By 10 guests, only
400,000 IOPS are being performed.
This patch improves performance by only unmapping when the connection
between blkfront and back is broken.
On startup blkfront notifies blkback that it is using persistent
grants, and blkback will do the same. If blkback is not capable of
persistent mapping, blkfront will still use the same grants, since it
is compatible with the previous protocol, and simplifies the code
complexity in blkfront.
To perform a read, in persistent mode, blkfront uses a separate pool
of pages that it maps to dom0. When a request comes in, blkfront
transmutes the request so that blkback will write into one of these
free pages. Blkback keeps note of which grefs it has already
mapped. When a new ring request comes to blkback, it looks to see if
it has already mapped that page. If so, it will not map it again. If
the page hasn't been previously mapped, it is mapped now, and a record
is kept of this mapping. Blkback proceeds as usual. When blkfront is
notified that blkback has completed a request, it memcpy's from the
shared memory, into the bvec supplied. A record that the {gref, page}
tuple is mapped, and not inflight is kept.
Writes are similar, except that the memcpy is peformed from the
supplied bvecs, into the shared pages, before the request is put onto
the ring.
Blkback stores a mapping of grefs=>{page mapped to by gref} in
a red-black tree. As the grefs are not known apriori, and provide no
guarantees on their ordering, we have to perform a search
through this tree to find the page, for every gref we receive. This
operation takes O(log n) time in the worst case. In blkfront grants
are stored using a single linked list.
The maximum number of grants that blkback will persistenly map is
currently set to RING_SIZE * BLKIF_MAX_SEGMENTS_PER_REQUEST, to
prevent a malicios guest from attempting a DoS, by supplying fresh
grefs, causing the Dom0 kernel to map excessively. If a guest
is using persistent grants and exceeds the maximum number of grants to
map persistenly the newly passed grefs will be mapped and unmaped.
Using this approach, we can have requests that mix persistent and
non-persistent grants, and we need to handle them correctly.
This allows us to set the maximum number of persistent grants to a
lower value than RING_SIZE * BLKIF_MAX_SEGMENTS_PER_REQUEST, although
setting it will lead to unpredictable performance.
In writing this patch, the question arrises as to if the additional
cost of performing memcpys in the guest (to/from the pool of granted
pages) outweigh the gains of not performing TLB shootdowns. The answer
to that question is `no'. There appears to be very little, if any
additional cost to the guest of using persistent grants. There is
perhaps a small saving, from the reduced number of hypercalls
performed in granting, and ending foreign access.
Signed-off-by: Oliver Chick <oliver.chick@citrix.com>
Signed-off-by: Roger Pau Monne <roger.pau@citrix.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
[v1: Fixed up the misuse of bool as int]
Jens Axboe
Felipe Pena
Tim Gardner
Kent Overstreet
Roger Pau Monne
Stefan Bader
Konrad Rzeszutek Wilk
Al Viro
Mihnea Dobrescu-Balaur