This patch allows Network Block Device to be mounted locally (nbd-client to
nbd-server over 127.0.0.1).
It creates a kthread to avoid the deadlock described in NBD tools
documentation. So, if nbd-client hangs waiting for pages, the kblockd thread
can continue its work and free pages.
I have tested the patch to verify that it avoids the hang that always occurs
when writing to a localhost nbd connection. I have also tested to verify that
no performance degradation results from the additional thread and queue.
Patch originally from Laurent Vivier.
Signed-off-by: Paul Clements <paul.clements@steeleye.com>
Signed-off-by: Laurent Vivier <Laurent.Vivier@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Use proc_create()/proc_create_data() to make sure that ->proc_fops and ->data
be setup before gluing PDE to main tree.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Peter Osterlund <petero2@telia.com>
Cc: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Cc: Dmitry Torokhov <dtor@mail.ru>
Cc: Neil Brown <neilb@suse.de>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Use creation by full path: "driver/foo".
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Introduced between 2.6.25-rc2 and -rc3
drivers/block/xen-blkfront.c:139:5: warning: symbol 'blkif_getgeo' was not declared. Should it be static?
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Replace init_module and cleanup_module with static functions and
module_init/module_exit.
Signed-off-by: Jon Schindler <jkschind@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Any path needs to call it to initialize the request.
This is a preparation for large command support, which needs to
initialize the request in a proper way (that is, just doing a memset()
will not work).
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
blk_get_request initializes rq->cmd (rq_init does) so the users don't
need to do that.
The purpose of this patch is to remove sizeof(rq->cmd) and &rq->cmd,
as a preparation for large command support, which changes rq->cmd from
the static array to a pointer. sizeof(rq->cmd) will not make sense and
&rq->cmd won't work.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Alasdair G Kergon <agk@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
The block layer initializes rq->cmd (queue_flush calls rq_init) so
prepare_flush_fn hooks don't need to do that.
The purpose of this patch is to remove sizeof(rq->cmd), as a
preparation for large command support, which changes rq->cmd from the
static array to a pointer. sizeof(rq->cmd) will not make sense.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
We can save some atomic ops in the IO path, if we clearly define
the rules of how to modify the queue flags.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
As ps3disk is a ppc64-only driver, sector_t equals to unsigned long, and the
cast is not needed.
Reuse in another (possibly 32-bit) driver is protected by the safety net called
`compiler warning' (with the cast, it may silently truncate to 32-bit).
If sector_t ever changes, we will get a compiler warning as well (with the
cast, we won't).
Signed-off-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Alter the block device ->direct_access() API to work with the new
get_xip_mem() API (that requires both kaddr and pfn are returned).
Some architectures will not do the right thing in their virt_to_page() for use
by XIP (to translate from the kernel virtual address returned by
direct_access(), to a user mappable pfn in XIP's page fault handler.
However, we can't switch it to just return the pfn and not the kaddr, because
we have no good way to get a kva from a pfn, and XIP requires the kva for its
read(2) and write(2) handlers. So we have to return both.
Signed-off-by: Jared Hulbert <jaredeh@gmail.com>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Carsten Otte <cotte@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: linux-mm@kvack.org
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Before getting merged, xen-blkfront was xenblk and
xen-netfront was xennet.
Temporarily adding compatibility module aliases
eases upgrades from older versions by e.g. allowing
mkinitrd to find the new version of the module.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Add module aliases to support autoprobing modules
for xen frontend devices.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
When the xen block frontend driver is built as a module the module load
is only synchronous up to the point where the frontend and the backend
become connected rather than when the disk is added.
This means that there can be a race on boot between loading the module and
loading the dm-* modules and doing the scan for LVM physical volumes (all
in the initrd). In the failure case the disk is not present until after the
scan for physical volumes is complete.
Taken from:
http://xenbits.xensource.com/linux-2.6.18-xen.hg?rev/11483a00c017
Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
info->dev is never initialized to anything, so bdget(info->dev) is
meaningless. Get rid of info->dev, and use bdget_disk on the gendisk.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Frontends are expected to write their protocol ABI to xenstore. Since
the protocol ABI defaults to the backend's native ABI, things work
fine without that as long as the frontend's native ABI is identical to
the backend's native ABI. This is not the case for xen-blkfront
running 32-on-64, because its ABI differs between 32 and 64 bit, and
thus needs this fix.
Based on http://xenbits.xensource.com/xen-unstable.hg?rev/c545932a18f3
and http://xenbits.xensource.com/xen-unstable.hg?rev/ffe52263b430 by
Gerd Hoffmann <kraxel@suse.de>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Jeremy Fitzhardinge <Jeremy.Fitzhardinge@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
While looking at the implementation of the Ram backed block device
driver, I stumbled across a write-only local variable, which makes
little sense, so I assume it should actually work like this:
Signed-off-by: Petr Tesarik <ptesarik@suse.cz>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* Fix oops on cciss rmmod due to calling pci_free_consistent with
irqs disabled.
Signed-off-by: Stephen M. Cameron <scameron@beardog.cca.cpqcorp.net>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Fix race condition between cciss_init_one(), cciss_update_drive_info(),
and cciss_check_queues().
Signed-off-by: Stephen M. Cameron <scameron@beardog.cca.cpqcorp.net>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This patch allows to use loop device with partitionned disk image.
Original behavior of loop is not modified.
A new parameter is introduced to define how many partition we want to be
able to manage per loop device. This parameter is "max_part".
For instance, to manage 63 partitions / loop device, we will do:
# modprobe loop max_part=63
# ls -l /dev/loop?*
brw-rw---- 1 root disk 7, 0 2008-03-05 14:55 /dev/loop0
brw-rw---- 1 root disk 7, 64 2008-03-05 14:55 /dev/loop1
brw-rw---- 1 root disk 7, 128 2008-03-05 14:55 /dev/loop2
brw-rw---- 1 root disk 7, 192 2008-03-05 14:55 /dev/loop3
brw-rw---- 1 root disk 7, 256 2008-03-05 14:55 /dev/loop4
brw-rw---- 1 root disk 7, 320 2008-03-05 14:55 /dev/loop5
brw-rw---- 1 root disk 7, 384 2008-03-05 14:55 /dev/loop6
brw-rw---- 1 root disk 7, 448 2008-03-05 14:55 /dev/loop7
And to attach a raw partitionned disk image, the original losetup is used:
# losetup -f etch.img
# ls -l /dev/loop?*
brw-rw---- 1 root disk 7, 0 2008-03-05 14:55 /dev/loop0
brw-rw---- 1 root disk 7, 1 2008-03-05 14:57 /dev/loop0p1
brw-rw---- 1 root disk 7, 2 2008-03-05 14:57 /dev/loop0p2
brw-rw---- 1 root disk 7, 5 2008-03-05 14:57 /dev/loop0p5
brw-rw---- 1 root disk 7, 64 2008-03-05 14:55 /dev/loop1
brw-rw---- 1 root disk 7, 128 2008-03-05 14:55 /dev/loop2
brw-rw---- 1 root disk 7, 192 2008-03-05 14:55 /dev/loop3
brw-rw---- 1 root disk 7, 256 2008-03-05 14:55 /dev/loop4
brw-rw---- 1 root disk 7, 320 2008-03-05 14:55 /dev/loop5
brw-rw---- 1 root disk 7, 384 2008-03-05 14:55 /dev/loop6
brw-rw---- 1 root disk 7, 448 2008-03-05 14:55 /dev/loop7
# mount /dev/loop0p1 /mnt
# ls /mnt
bench cdrom home lib mnt root srv usr
bin dev initrd lost+found opt sbin sys var
boot etc initrd.img media proc selinux tmp vmlinuz
# umount /mnt
# losetup -d /dev/loop0
Of course, the same behavior can be done using kpartx on a loop device,
but modifying loop avoids to stack several layers of block device (loop +
device mapper), this is a very light modification (40% of modifications
are to manage the new parameter).
Signed-off-by: Laurent Vivier <Laurent.Vivier@bull.net>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
None of these files use any of the functionality promised by
asm/semaphore.h. It's possible that they rely on it dragging in some
unrelated header file, but I can't build all these files, so we'll have
fix any build failures as they come up.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
This patch adds the missing include directive <linux/scatterlist.h> to the
cciss.c source file. This was discovered by our release team when building
the kernel for the Alpha architecture.
Errors were found as references to functions 'sg_init_table' and 'sg_page' do
not exist without the include for Alpha.
Signed-off-by: Mike Pagano <mpagano@gentoo.org>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: <mike.miller@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When __blk_end_request returns nonzero, it means that the request was
not completely processed and some BIOs are still attached. Since we
have dequeued it by that time, it means leaking requests and hanging
processes, which is why BUG() was in there. In ub this happens if
a packet request ends normally, but with residue (e.g. when scsi_id
issues INQUIRY).
The fix is to make sure that arguments passed to __blk_end_request
are correct: the full request length and not just transferred length.
The transferred length is indicated to applications by adjusting
rq->data_len with old, unchanged code outside of this patch.
Signed-off-by: Pete Zaitcev <zaitcev@redhat.com>
Cc: Kiyoshi Ueda <k-ueda@ct.jp.nec.com>
Cc: Greg KH <greg@kroah.com>
Cc: Boaz Harrosh <bharrosh@panasas.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
NBD does not protect the nbd_device's socket from becoming NULL during
receives.
This closes a race with the NBD_CLEAR_SOCK ioctl (nbd-client -d) setting
the nbd_device's socket to NULL right before NBD calls sock_xmit.
Signed-off-by: Mike Snitzer <snitzer@gmail.com>
Cc: Paul Clements <paul.clements@steeleye.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Robert P.J. Day proposed to use the macro FIELD_SIZEOF in replace of code
that matches its definition.
The modification was made using the following semantic patch
(http://www.emn.fr/x-info/coccinelle/)
// <smpl>
@haskernel@
@@
#include <linux/kernel.h>
@depends on haskernel@
type t;
identifier f;
@@
- (sizeof(((t*)0)->f))
+ FIELD_SIZEOF(t, f)
@depends on haskernel@
type t;
identifier f;
@@
- sizeof(((t*)0)->f)
+ FIELD_SIZEOF(t, f)
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Introduce per-net_device inlines: dev_net(), dev_net_set().
Without CONFIG_NET_NS, no namespace other than &init_net exists.
Let's explicitly define them to help compiler optimizations.
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Fix up so that the virtio_blk devices in sysfs link correctly to their
block device. This then allows them to be detected by hal, etc
Signed-off-by: Jeremy Katz <katzj@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
no longer working for some time.
A driver that had been marked as BROKEN for such a long time seems to be
unlikely to be revived in the forseeable future.
But if anyone wants to ever revive this driver, the code is still present in
the older kernel releases.
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Alan Cox <alan@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Floppy rmmod locks up when no such hardware was initialized, since there is
nobody to wake the remove code up. Remove the completion, because release is
called during platform_unregister anyway.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The iSeries viodasd drivers does some very strange things with
scatterlists, one of these causing a BUG_ON to trigger when
scatterlist debugging is enabled due to initializing the
scatterlist with memset instead of sg_init_table().
This fixes it by using sg_init_table(). The rest of the stuff
it does to that poor list is still pretty awful but it will work.
I may look into fixing things in a nicer way some other time.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
On my system, pkt_open() consumes 584 bytes because the compiler decides to
inline lots of functions that would not normally be part of long call chains.
The following patch fixes that problem on my system.
Signed-off-by: Peter Osterlund <petero2@telia.com>
Cc: Nix <nix@esperi.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch removes the #define READ_AHEAD 1024 from the driver and uses the
block layer defaults, instead. We have found that under certain workloads
the setting can cause a disk connected to the e200 controller to go offline.
If the disk hiccups the link may try to downshift but the controller is
never notified that the link successfully completed the renegotiation.
We've also found that performance using the block layer default of 32 pages
was on par with the 1024 setting. We tried setting it to zero at one time
based on info from our firmware guys but that killed performance. Turns out
we were talking about 2 different read ahead settings.
Please consider this for inclusion.
Signed-off-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
volumes
This patch allows us to display information about all of the logical volumes
configured on a particular controller without stepping on memory even when
there are many volumes (128 or more) configured.
Please consider this for inclusion.
Signed-off-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
NBD doesn't work well with CFQ (or AS) schedulers, so let's default to
something else.
The two problems I have experienced with nbd and cfq are:
1) nbd hangs with cfq on RHEL 5 (2.6.18) -- this may well have been
fixed
There's a similar debian bug that has been filed as well:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=447638
There have been posts to nbd-general mailing list about problems with
cfq and nbd also.
2) nbd performs about 10% better (the last time I tested) with deadline
vs. cfq (the overhead of cfq doesn't provide much advantage to nbd [not
being a real disk], and you end up going through the I/O scheduler on
the nbd server anyway, so it makes sense that deadline is better with
nbd)
Signed-off-by: Paul Clements <paul.clements@steeleye.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The below implements the getgeo hook for Xen block devices. Extracted
from the xen-unstable tree where it has been used for ages.
It is useful to have because it allows things like grub2 (used by the
Debian installer images) to work in a guest domain without having to
sprinkle Xen specific hacks around the place.
Signed-off-by: Ian Campbell <ijc@hellion.org.uk>
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The current pmac32_defconfig fails to build with the following error:
Building modules, stage 2.
ERROR: "check_media_bay" [drivers/block/swim3.ko] undefined!
WARNING: modpost: Found 23 section mismatch(es).
To see full details build your kernel with:
'make CONFIG_DEBUG_SECTION_MISMATCH=y'
make[2]: *** [__modpost] Error 1
This patch fixes that.
Signed-off-by: Tony Breeds <tony@bakeyournoodle.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Acked-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Cc: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Remove the arbitrary 128 device limit for NBD. nbds_max can now be set to
any number. In certain scenarios where devices are used sparsely we have
run into the 128 device limit.
Signed-off-by: Paul Clements <paul.clements@steeleye.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
I guess aoedev_init() can go away now.
Cc: Greg KH <greg@kroah.com>
Cc: "Ed L. Cashin" <ecashin@coraid.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Update the year in the copyright notices.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Andrew Morton pointed out that the "too many targets" message in patch 2 could
be printed for failing GFP_ATOMIC allocations. This patch makes the messages
more specific.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The aoedev aoeminor member doesn't need a long format.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
An AoE target provides an estimate of the number of outstanding commands that
the AoE initiator can send before getting a response. The aoe_maxout
parameter provides a way to set an even lower limit. It will not allow a user
to use more outstanding commands than the target permits. If a user discovers
a problem with a large setting, this parameter provides a way for us to work
with them to debug the problem. We expect to improve the dynamic window
sizing algorithm and drop this parameter. For the time being, it is a
debugging aid.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
An aoe driver user who had about 70 AoE targets found that he was hitting a
BUG in sysfs_create_file because the aoe driver was trying to tell the kernel
about an AoE device more than once. Each AoE device was reachable by several
local network interfaces, and multiple ATA device indentify responses were
returning from that single device.
This patch eliminates a race condition so that aoe always informs the block
layer of a new AoE device once in the presence of multiple incoming ATA device
identify responses.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
What this Patch Does
Even before this recent series of 12 patches to 2.6.22-rc4, the aoe
driver was reusing a small set of skbs that were allocated once and
were only used for outbound AoE commands.
The network layer cannot be allowed to put_page on the data that is
still associated with a bio we haven't returned to the block layer,
so the aoe driver (even before the patch under discussion) is still
the owner of skbs that have been handed to the network layer for
transmission. We need to keep track of these skbs so that we can
free them, but by tracking them, we can also easily re-use them.
The new patch was a response to the behavior of certain network
drivers. We cannot reuse an skb that the network driver still has
in its transmit ring. Network drivers can defer transmit ring
cleanup and then use the state in the skb to determine how many data
segments to clean up in its transmit ring. The tg3 driver is one
driver that behaves in this way.
When the network driver defers cleanup of its transmit ring, the aoe
driver can find itself in a situation where it would like to send an
AoE command, and the AoE target is ready for more work, but the
network driver still has all of the pre-allocated skbs. In that
case, the new patch just calls alloc_skb, as you'd expect.
We don't want to get carried away, though. We try not to do
excessive allocation in the write path, so we cap the number of skbs
we dynamically allocate.
Probably calling it a "dynamic pool" is misleading. We were already
trying to use a small fixed-size set of pre-allocated skbs before
this patch, and this patch just provides a little headroom (with a
ceiling, though) to accomodate network drivers that hang onto skbs,
by allocating when needed. The d->skbpool_hd list of allocated skbs
is necessary so that we can free them later.
We didn't notice the need for this headroom until AoE targets got
fast enough.
Alternatives
If the network layer never did a put_page on the pages in the bio's
we get from the block layer, then it would be possible for us to
hand skbs to the network layer and forget about them, allowing the
network layer to free skbs itself (and thereby calling our own
skb->destructor callback function if we needed that). In that case
we could get rid of the pre-allocated skbs and also the
d->skbpool_hd, instead just calling alloc_skb every time we wanted
to transmit a packet. The slab allocator would effectively maintain
the list of skbs.
Besides a loss of CPU cache locality, the main concern with that
approach the danger that it would increase the likelihood of
deadlock when VM is trying to free pages by writing dirty data from
the page cache through the aoe driver out to persistent storage on
an AoE device. Right now we have a situation where we have
pre-allocation that corresponds to how much we use, which seems
ideal.
Of course, there's still the separate issue of receiving the packets
that tell us that a write has successfully completed on the AoE
target. When memory is low and VM is using AoE to flush dirty data
to free up pages, it would be perfect if there were a way for us to
register a fast callback that could recognize write command
completion responses. But I don't think the current problems with
the receive side of the situation are a justification for
exacerbating the problem on the transmit side.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When an AoE device is detected, the kernel is informed, and a new block device
is created. If the device is unused, the block device corresponding to remote
device that is no longer available may be removed from the system by telling
the aoe driver to "flush" its list of devices.
Without this patch, software like GPFS and LVM may attempt to read from AoE
devices that were discovered earlier but are no longer present, blocking until
the I/O attempt times out.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Adam Richter suggested eliminating this goto.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
By returning unsigned long long, mac_addr does not generate compiler warnings
on 64-bit architectures.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
A remote AoE device is something can process ATA commands and is identified by
an AoE shelf number and an AoE slot number. Such a device might have more
than one network interface, and it might be reachable by more than one local
network interface. This patch tracks the available network paths available to
each AoE device, allowing them to be used more efficiently.
Andrew Morton asked about the call to msleep_interruptible in the revalidate
function. Yes, if a signal is pending, then msleep_interruptible will not
return 0. That means we will not loop but will call aoenet_xmit with a NULL
skb, which is a noop. If the system is too low on memory or the aoe driver is
too low on frames, then the user can hit control-C to interrupt the attempt to
do a revalidate. I have added a comment to the code summarizing that.
Andrew Morton asked whether the allocation performed inside addtgt could use a
more relaxed allocation like GFP_KERNEL, but addtgt is called when the aoedev
lock has been locked with spin_lock_irqsave. It would be nice to allocate the
memory under fewer restrictions, but targets are only added when the device is
being discovered, and if the target can't be added right now, we can try again
in a minute when then next AoE config query broadcast goes out.
Andrew Morton pointed out that the "too many targets" message could be printed
for failing GFP_ATOMIC allocations. The last patch in this series makes the
messages more specific.
Signed-off-by: Ed L. Cashin <ecashin@coraid.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Laurent Vivier