| Age | Commit message (Collapse) | Author |
|---|
|
virtio spec requires that all drivers set DRIVER_OK
before using devices. While rpmsg isn't yet
included in the virtio 1 spec, previous spec versions
also required this.
virtio rpmsg violates this rule: is calls kick
before setting DRIVER_OK.
The fix isn't trivial since simply calling virtio_device_ready earlier
would mean we might get an interrupt in parallel with adding buffers.
Instead, split kick out to prepare+notify calls. prepare before
virtio_device_ready - when we know we won't get interrupts. notify right
afterwards.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Adjust the number of rpmsg buffers to rely on the size of the
vring, instead of using the hard coded value of 512 (256 per
direction).
This is needed when small vrings are being used, where 256
buffers are too much to fit in a vring.
While considering the vring size, keep using the 512 hard coded
value as an upper limit to avoid wacky resource tables consuming
unreasonable amount of memory.
NOTE: The number of buffers is already assumed to be symmetrical
in each direction, and that logic is unchanged.
Signed-off-by: Suman Anna <s-anna@ti.com>
[edit commit message, small code and comment simplification]
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/ohad/rpmsg
Pull rpmsg changes from Ohad Ben-Cohen:
"A small pull request consisting of:
- Make rpmsg process all pending messages instead of just one, from
Robert Tivy
- Fix Kconfig dependency on VIRTUALIZATION, from Suman.
Note: this was submitted late during the 3.9 rc cycle and it seemed
appropriate to wait with it for the merge window.
- Belated addition of an rpmsg entry to the MAINTAINERS file. People
seem to look for this"
* tag 'rpmsg-3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/rpmsg:
rpmsg: fix kconfig dependencies for VIRTIO
MAINTAINERS: add rpmsg entry
rpmsg: process _all_ pending messages in rpmsg_recv_done
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux
Pull virtio & lguest updates from Rusty Russell:
"Lots of virtio work which wasn't quite ready for last merge window.
Plus I dived into lguest again, reworking the pagetable code so we can
move the switcher page: our fixmaps sometimes take more than 2MB now..."
Ugh. Annoying conflicts with the tcm_vhost -> vhost_scsi rename.
Hopefully correctly resolved.
* tag 'virtio-next-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux: (57 commits)
caif_virtio: Remove bouncing email addresses
lguest: improve code readability in lg_cpu_start.
virtio-net: fill only rx queues which are being used
lguest: map Switcher below fixmap.
lguest: cache last cpu we ran on.
lguest: map Switcher text whenever we allocate a new pagetable.
lguest: don't share Switcher PTE pages between guests.
lguest: expost switcher_pages array (as lg_switcher_pages).
lguest: extract shadow PTE walking / allocating.
lguest: make check_gpte et. al return bool.
lguest: assume Switcher text is a single page.
lguest: rename switcher_page to switcher_pages.
lguest: remove RESERVE_MEM constant.
lguest: check vaddr not pgd for Switcher protection.
lguest: prepare to make SWITCHER_ADDR a variable.
virtio: console: replace EMFILE with EBUSY for already-open port
virtio-scsi: reset virtqueue affinity when doing cpu hotplug
virtio-scsi: introduce multiqueue support
virtio-scsi: push vq lock/unlock into virtscsi_vq_done
virtio-scsi: pass struct virtio_scsi to virtqueue completion function
...
|
|
Return a negative error code from the error handling case instead of 0, as
returned elsewhere in this function.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Cc: Ohad Ben-Cohen <ohad@wizery.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Change virtqueue callback function rpmsg_recv_done() to process all
available messages instead of just one message.
Signed-off-by: Robert Tivy <rtivy@ti.com>
[split _recv function instead of adding indentation]
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
|
|
We never add buffers with input and output parts, so use the new accessors.
Cc: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
Convert to the much saner new idr interface.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
idr_destroy() can destroy idr by itself and idr_remove_all() is being
deprecated. Drop its usage.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Conflicts:
drivers/devfreq/exynos4_bus.c
Sync with Linus' tree to be able to apply patches that are
against newer code (mvneta).
|
|
Correct spelling typo in printk within various drivers.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
|
|
CONFIG_HOTPLUG is going away as an option. As a result, the __dev*
markings need to be removed.
This change removes the use of __devinit, __devexit_p, __devinitdata,
__devinitconst, and __devexit from these drivers.
Based on patches originally written by Bill Pemberton, but redone by me
in order to handle some of the coding style issues better, by hand.
Cc: Bill Pemberton <wfp5p@virginia.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
We simplified virtqueue_add_buf(), make it clear in the callers.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
|
dma_alloc/free_coherent APIs requires the platform specific remoteproc
device as the device parameter. We are passing vdev->dev.parent to the
dma_free_coherent function which is the generic rproc device and it is
wrong, it has to be vdev->dev.parent->parent instead, same as when we
call dma_alloc_coherent function.
Signed-off-by: Fernando Guzman Lugo <fernando.lugo@ti.com>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/ohad/remoteproc
Pull remoteproc update from Ohad Ben-Cohen:
- custom binary format support from Sjur Brændeland
- groundwork for recovery and runtime pm support
- some cleanups and API simplifications
Fix up conflicts in drivers/remoteproc/remoteproc_core.c due to clashes
with earlier cleanups by Sjur Brændeland (with part of the cleanups
moved into the new remoteproc_elf_loader.c file).
* tag 'remoteproc-for-3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/remoteproc:
MAINTAINERS: add remoteproc's git
remoteproc: Support custom firmware handlers
remoteproc: Move Elf related functions to separate file
remoteproc: Add function rproc_get_boot_addr
remoteproc: Pass struct fw to load_segments and find_rsc_table.
remoteproc: adopt the driver core's alloc/add/del/put naming
remoteproc: remove the get_by_name/put API
remoteproc: support non-iommu carveout assignment
remoteproc: simplify unregister/free interfaces
remoteproc: remove the now-redundant kref
remoteproc: maintain a generic child device for each rproc
remoteproc: allocate vrings on demand, free when not needed
|
|
When rpmsg drivers are built into the kernel, they must not initialize
before the rpmsg bus does, otherwise they'd trigger a BUG() in
drivers/base/driver.c line 169 (driver_register()).
To fix that, and to stop depending on arbitrary linkage ordering of
those built-in rpmsg drivers, we make the rpmsg bus initialize at
subsys_initcall.
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Federico Fuga <fuga@studiofuga.com>
[ohad: rewrite the commit log]
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
|
|
For each registered rproc, maintain a generic remoteproc device whose
parent is the low level platform-specific device (commonly a pdev, but
it may certainly be any other type of device too).
With this in hand, the resulting device hierarchy might then look like:
omap-rproc.0
|
- remoteproc0 <---- new !
|
- virtio0
|
- virtio1
|
- rpmsg0
|
- rpmsg1
|
- rpmsg2
Where:
- omap-rproc.0 is the low level device that's bound to the
driver which invokes rproc_register()
- remoteproc0 is the result of this patch, and will be added by the
remoteproc framework when rproc_register() is invoked
- virtio0 and virtio1 are vdevs that are registered by remoteproc
when it realizes that they are supported by the firmware
of the physical remote processor represented by omap-rproc.0
- rpmsg0, rpmsg1 and rpmsg2 are rpmsg devices that represent rpmsg
channels, and are registerd by the rpmsg bus when it gets notified
about their existence
Technically, this patch:
- changes 'struct rproc' to contain this generic remoteproc.x device
- creates a new "remoteproc" type, to which this new generic remoteproc.x
device belong to.
- adds a super simple enumeration method for the indices of the
remoteproc.x devices
- updates all dev_* messaging to use the generic remoteproc.x device
instead of the low level platform-specific device
- updates all dma_* allocations to use the parent of remoteproc.x (where
the platform-specific memory pools, most commonly CMA, are to be found)
Adding this generic device has several merits:
- we can now add remoteproc runtime PM support simply by hooking onto the
new "remoteproc" type
- all remoteproc log messages will now carry a common name prefix
instead of having a platform-specific one
- having a device as part of the rproc struct makes it possible to simplify
refcounting (see subsequent patch)
Thanks to Stephen Boyd <sboyd@codeaurora.org> for suggesting and
discussing these ideas in one of the remoteproc review threads and
to Fernando Guzman Lugo <fernando.lugo@ti.com> for trying them out
with the (upcoming) runtime PM support for remoteproc.
Cc: Fernando Guzman Lugo <fernando.lugo@ti.com>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
|
|
When inbound messages arrive, rpmsg core looks up their associated
endpoint (by destination address) and then invokes their callback.
We've made sure that endpoints will never be de-allocated after they
were found by rpmsg core, but we also need to protect against the
(rare) scenario where the rpmsg driver was just removed, and its
callback function isn't available anymore.
This is achieved by introducing a callback mutex, which must be taken
before the callback is invoked, and, obviously, before it is removed.
Cc: stable <stable@vger.kernel.org>
Reported-by: Fernando Guzman Lugo <fernando.lugo@ti.com>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
|
|
When an inbound message arrives, the rpmsg core looks up its
associated endpoint and invokes the registered callback.
If a message arrives while its endpoint is being removed (because
the rpmsg driver was removed, or a recovery of a remote processor
has kicked in) we must ensure atomicity, i.e.:
- Either the ept is removed before it is found
or
- The ept is found but will not be freed until the callback returns
This is achieved by maintaining a per-ept reference count, which,
when drops to zero, will trigger deallocation of the ept.
With this in hand, it is now forbidden to directly deallocate
epts once they have been added to the endpoints idr.
Cc: stable <stable@vger.kernel.org>
Reported-by: Fernando Guzman Lugo <fernando.lugo@ti.com>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
|
|
dev_dbg() in rpmsg_probe() made use of the %x formatting that
expects an 'unsigned int' which dma_addr_t is not in cases where
dma_addr_t is 64-bit (CONFIG_ARCH_DMA_ADDR_T_64BIT). Casting to
a 'unsigned long long' and using %llx will avoid this.
Signed-off-by: Mark Asselstine <mark.asselstine@windriver.com>
CC: Ohad Ben-Cohen <ohad@wizery.com>
CC: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
|
|
After processing an incoming message, always publish the real size
of its containing buffer when putting it back on the available rx ring.
Using any different value might erroneously limit the remote processor
(leading it to think the buffer is smaller than it really is).
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Mark Grosen <mgrosen@ti.com>
Cc: Suman Anna <s-anna@ti.com>
Cc: Fernando Guzman Lugo <fernando.lugo@ti.com>
Cc: Rob Clark <rob@ti.com>
Cc: Ludovic BARRE <ludovic.barre@stericsson.com>
Cc: Loic PALLARDY <loic.pallardy@stericsson.com>
Cc: Omar Ramirez Luna <omar.luna@linaro.org>
|
|
When an inbound message arrives, validate its reported length before
propagating it, otherwise buggy (or malicious) remote processors might
trick us into accessing memory which we really shouldn't.
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Mark Grosen <mgrosen@ti.com>
Cc: Suman Anna <s-anna@ti.com>
Cc: Fernando Guzman Lugo <fernando.lugo@ti.com>
Cc: Rob Clark <rob@ti.com>
Cc: Ludovic BARRE <ludovic.barre@stericsson.com>
Cc: Loic PALLARDY <loic.pallardy@stericsson.com>
Cc: Omar Ramirez Luna <omar.luna@linaro.org>
|
|
The name service endpoint wasn't destroyed, so fix it.
This is achieved by introducing an internal __rpmsg_destroy_ept
function which doesn't assume the given ept is bound to an rpmsg
channel (much like the existing __rpmsg_create_ept).
This is needed because the name service ept belongs to the rpmsg bus,
and is never bound with a specific rpdev.
Reported-by: Omar Ramirez Luna <omar.ramirez@ti.com>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Mark Grosen <mgrosen@ti.com>
Cc: Suman Anna <s-anna@ti.com>
Cc: Fernando Guzman Lugo <fernando.lugo@ti.com>
Cc: Rob Clark <rob@ti.com>
Cc: Ludovic BARRE <ludovic.barre@stericsson.com>
Cc: Loic PALLARDY <loic.pallardy@stericsson.com>
Cc: Omar Ramirez Luna <omar.ramirez@ti.com>
|
|
Since commit 7bb7aef2 "virtio: rename virtqueue_add_buf_gfp to virtqueue_add_buf",
virtqueue_add_buf_gfp is already rename to virtqueue_add_buf now.
This patch fixes below build error:
CC [M] drivers/rpmsg/virtio_rpmsg_bus.o
drivers/rpmsg/virtio_rpmsg_bus.c: In function 'rpmsg_send_offchannel_raw':
drivers/rpmsg/virtio_rpmsg_bus.c:723: error: implicit declaration of function 'virtqueue_add_buf_gfp'
make[2]: *** [drivers/rpmsg/virtio_rpmsg_bus.o] Error 1
make[1]: *** [drivers/rpmsg] Error 2
make: *** [drivers] Error 2
Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
|
|
Add a virtio-based inter-processor communication bus, which enables
kernel drivers to communicate with entities, running on remote
processors, over shared memory using a simple messaging protocol.
Every pair of AMP processors share two vrings, which are used to send
and receive the messages over shared memory.
The header of every message sent on the rpmsg bus contains src and dst
addresses, which make it possible to multiplex several rpmsg channels on
the same vring.
Every rpmsg channel is a device on this bus. When a channel is added,
and an appropriate rpmsg driver is found and probed, it is also assigned
a local rpmsg address, which is then bound to the driver's callback.
When inbound messages carry the local address of a bound driver,
its callback is invoked by the bus.
This patch provides a kernel interface only; user space interfaces
will be later exposed by kernel users of this rpmsg bus.
Designed with Brian Swetland <swetland@google.com>.
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au> (virtio_ids.h)
Cc: Brian Swetland <swetland@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Greg KH <greg@kroah.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
|
