Age | Commit message (Collapse) | Author |
|
The main mutex in a comedi device can get held for quite a while when
processing comedi instructions, so for performance reasons, the "read"
and "write" file operations do not use it; they use use the
`attach_lock` rwsemaphore to protect against the comedi device becoming
detached at an inopportune moment. Do the same for the "poll" file
operation.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Currently, the "poll" file operation checks if an asynchronous "read"
(or "write" command is active on the "read" (or "write" subdevice, but
does not consider whether the command was started from the file object
being polled. Since that is the only file object able to read (or
write) data, take it into consideration.
With this change, if no read (or write) command is running on the
subdevice, or it is started by a different file object, the file object
is marked as readable (or writeable) regardless, but the read (or write)
file operation will return an error.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
When handling the "poll" file operation and checking for `POLLOUT`,
don't allocate space from the buffer for writing, just check that space
is available for writing. That check is done after checking that an
asynchronous "write" command is running on the subdevice. Allocating
the buffer space before checking a "write" command is running can cause
problems if the subdevice supports commands in either direction and
currently has an active "read" command.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Comedi subdevices that support asynchronous acquisition commands have a
wait queue head used for blocking reads or writes and for the poll file
operation. The comedi device may have several subdevices that support
"read" and/or "write" commands, but each open file object has at most
one "read" subdevice and one "write" subdevice. It's possible (though
rare) for those to be the same subdevice if the subdevice supports
commands in either direction. In that case, the "poll" file operation
doesn't really need to do a `poll_wait()` on the same subdevice twice.
Although harmless, it wastes a poll table entry. Check for that, and
avoid it.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
n this file, function names are otherwise used as pointers without &.
A simplified version of the Coccinelle semantic patch that makes this
change is as follows:
// <smpl>
@r@
identifier f;
@@
f(...) { ... }
@@
identifier r.f;
@@
- &f
+ f
// </smpl>
Signed-off-by: Shraddha Barke <shraddha.6596@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Adjust the kernel-doc in this file for consistency of capitalization and
punctuation. Make more use of the special kernel-doc markers for
parameter names, constant names, etc. Use the correct kernel-doc tag
for the return values sections.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
The comment before the `COMEDI_SRF_...` macros starts with a kernel-doc
mark but isn't valid kernel-doc. There isn't a kernel-doc template for
a group of macro constants. These macros are all private anyway, so
don't really need to be in kernel-doc. Just change it to a normal
comment.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
With two exceptions (drm/qxl and drm/radeon) all vm_operations_struct
structs should be constant.
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Oleg Nesterov <oleg@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
This function is only called by comedi_cleanup_board_minors() and the
'minor' parameter will always be < COMEDI_NUM_BOARD_MINORS.
For aesthetics, absorb the function and remove the unnecessary BUG_ON().
Split the comedi_clear_board_minor() out to clarify that the return value
is a comedi_device pointer.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
comedi_dev_get_from_board_minor()
This function is only called by comedi_dev_get_from_minor() and the 'minor'
value will always be < COMEDI_NUM_BOARD_MINORS. Remove the unnecessary
BUG_ON().
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Drivers should not crash the kernel.
This function is only called by comedi_device_detach_cleanup() and the
s->minor will always be valid or the device wouldn't have attached in
the first place.
Leave the checks for safety in accessing the comedi_subdevice_minor_table
array but remove the BUG_ON() calls.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
The BUG_ON() checks in this function are not necessary.
comedi_cleanup_board_minors() clears all the entries in the
comedi_board_minor_table array and will call comedi_device_cleanup()
for all attached devices. comedi_device_cleanup() will then
clear the entries in the comedi_subdevice_minor_table array with
comedi_free_subdevice_minor().
Remove the BUG_ON(), drivers should not crash the kernel.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
The BUG_ON() checks in comedi_subdevice_from_minor() and
comedi_dev_get_from_subdevice_minor() are not necessary.
The 'minor' numbers for a given comedi driver are setup by
comedi_dev_get_from_subdevice_minor() and will always be in
the correct range.
Drivers should not crash the kernel, remove the BUG_ON() checks.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
This patch fixes the checkpatch.pl warning:
WARNING: ENOSYS means 'invalid syscall nr' and nothing else
try_module_get fails when the reference count of the module is not
allowed to be incremented ,and hence -ENXIO is returned indicating
no device or address.
[IA - combined two of santhosh's changes to the error return value!]
Signed-off-by: santhosh pai <santhosh.pai88@yahoo.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
If the "comedi" module has been loaded with the
"comedi_num_legacy_minors" module parameter set to a non-zero value,
some reserved comedi devices get created. These can be attached to a
low-level comedi driver using the `COMEDI_DEVCONFIG` ioctl command,
which checks for the `CAP_SYS_ADMIN` capability. Of course, the comedi
device needs to be opened before the ioctl command can be sent. If the
comedi device is unattached, `comedi_open()` currently requires the
`CAP_NET_ADMIN` capability. It makes more sense to just require the
`CAP_SYS_ADMIN` capability here, so change it.
For the curious, commit a8f80e8ff94e ("Networking: use CAP_NET_ADMIN
when deciding to call request_module") changed this capability from
`CAP_SYS_MODULE` to `CAP_NET_ADMIN`, even though it doesn't seem
relevant here. The original `CAP_SYS_MODULE` capability was due to the
function having some code to request a module using a "char-major-%i-%i"
alias, but that was never compiled in and was removed by commit
f30f2c2d417b ("staging: comedi: remove check for CONFIG_KMOD").
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
This patch removes commented code.This was a checkpatch warning.
Signed-off-by: Hari Prasath Gujulan Elango <hgujulan@visteon.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Fixed multiple spelling errors.
Signed-off-by: Carlos E. Garcia <carlos@cgarcia.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
The `COMEDI_SRF_...` macros define flag combinations in the `runflags`
member of `struct comedi_subdevice`. They are only used directly in
"comedi_fops.c", so move them to there.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
The `COMEDI_SRF_FREE_SPRIV` flag in the `runflags` member of `struct
comedi_subdevice` indicates that the memory pointed to by the `private`
member can be automatically freed by the comedi core on subdevice
clean-up (when the low-level comedi device is being "detached"). the
flag doesn't really belong in `runflags`, but it was somewhere
convenient to keep it without having to add a new member to the
structure.
Rather than access the `COMEDI_SRF_FREE_SPRIV` flag directly, use some
new wrapper functions:
* comedi_can_auto_free_spriv(s) - checks whether the subdevice's
`s->private` points to memory that can be freed automatically.
* comedi_set_spriv_auto_free(s) - marks the subdevice as having a
`s->private` that points to memory that can be freed automatically.
Export `comedi_set_spriv_auto_free()` for use by the low-level comedi
driver modules, in particular the "amplc_dio200_common" module.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`comedi_event()` is called from low-level drivers to handle comedi
asynchronous command event flags. As a safety check, it checks the
subdevice's "run" flags to make sure an asynchronous command is running.
It can also change the run flags to mark the command as no longer
running (possibly also marking it as terminated with an error).
Checking the runflags and modifying them involves two uses of the
subdevice's spin-lock. It seems better to do it with a single use of
the spin-lock. This also avoids possible interactions with
`do_become_nonbusy()`.
Acquire the subdevice's spin-lock at the start of `comedi_event()` and
release it near the end, before a possible call to `kill_fasync()` (but
after it's parameter values have been determined).
Add and make use of few new inline helper functions:
* `__comedi_clear_subdevice_runflags()` -- clears some run flags without
using the spin-lock
* `__comedi_set_subdevice_runflags()` -- sets some run flags without
using the spin-lock
* `__comedi_get_subdevice_runflags()` -- a spin-lockless version of
`comedi_get_subdevice_runflags()
* `__comedi_is_subdevice_running()` -- a spin-lockless version of
* `comedi_is_subdevice_running()`
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`comedi_event()` is called from low-level drivers to handle comedi
asynchronous command event flags. Some events cause waiting tasks to be
woken up, and a `SIGIO` signal to be sent via `kill_fasync()`. The
signal code is `POLL_OUT` if the subdevice supports commands in the
"write" direction, or `POLL_IN` for the "read" direction. If the
subdevice supports commands in either direction, it sends two `SIGIO`
signals, one with each code. Change that latter case to only send one
`SIGIO` signal, using the direction of the current command to determine
the signal code. If the `CMDF_WRITE` flag is set in the current
command, it's in the "write" direction, otherwise it's in the "read"
direction.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`comedi_event()` is called from low-level drivers to handle asynchronous
command event flags that are stored in `s->async->events` for subdevice
`s`. It normally clears the event flags as well. As a safety check, it
does nothing if no asynchronous command is running, but it leaves
`s->async->events` unchanged in this case. For additional safety,
change it to always clear the event flags to avoid leaving stale event
flags set when another asynchronous command is set up.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
In places where the `s->async` value has been stored in a local
variable, use the variable instead of repeating `s->async`.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Don't use a conditional operator when a simple "not" will do.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`comedi_is_subdevice_in_error()` is only used by `comedi_read()` and
`comedi_write()` and is only called (soon) after
`comedi_is_subdevice_running()` returns `false` (with extra conditions
in the case of `comedi_write()`). `comedi_is_subdevice_running()` and
`comedi_get_subdevice_runflags()` both call
`comedi_get_subdevice_runflags()` which uses the subdevice's spin-lock.
Eliminate one use of the subdevice's spin-lock in `comedi_read()` and
`comedi_write()` by calling `comedi_get_subdevice_runflags()` and
checking the runflags directly. Add a couple of inline functions to
check the runflags: `comedi_is_runflags_running()` and
`comedi_is_runflags_in_error()`. These do the same test on runflags as
`comedi_is_subdevice_running()` and `comedi_is_subdevice_in_error()` but
get passed the runflags value directly.
`comedi_is_subdevice_in_error()` is no longer used, so remove it.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`comedi_set_subdevice_runflags()` changes the comedi subdevice's
`runflags` member according to a bit-mask and new bit values. It's name
might suggest that it only "sets", not "clears". Rename it to
`comedi_update_subdevice_runflags()` to avoid confusion.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
This patch replace pr_err with dev_err, when appropriate device
structure is found.
Issue found using the following Coccinelle script:
@r exists@
identifier f, s, i;
position p;
@@
f(...,struct s *i,...) {
<+...
when != i == NULL
pr_err@p(...);
...+>
}
@rr@
identifier r.s, fld;
@@
struct s {
...
struct device *fld;
...
};
@@
identifier r.i, rr.fld;
position r.p;
@@
-pr_err@p
+dev_err
(
+ i->fld,
...)
Signed-off-by: Haneen Mohammed <hamohammed.sa@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Reformat remaining block comments to use the usual block comment style.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
The unlocked_ioctl handler calls a different function to handle each
supported ioctl command code. Most of these have a block comment
indicating which command code it handles, a brief description, and an
informal description of the inputs and outputs. These block comments
are formatted in various styles. Reformat them to the usual block
comment style and do a bit of rewording for consistency. The comment
block for the `COMEDI_CMD` ioctl is missing, so add one.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Add kerneldoc for exported functions `comedi_dev_put()`,
`comedi_dev_get_from_minor()`, `comedi_is_subdevice_running()`, and
`comedi_event()`.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Use the usual block comment style.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Tidy up and document the subdevice "runflags". Rename them so they have
comedi namespace.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Now that Comedi has the structures in place to support setting the
current "read" and/or "write" subdevice on a per-file object basis, add
new ioctls to set them. The newly chosen "read" ("write") subdevice
needs to support "read" ("write") commands, and the file cannot be busy
handling a "read" ("write") command on the previous subdevice (if any).
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Comedi devices may have several subdevices that support "read" and/or
"write" asynchronous commands that use the "read" or "write" file
operations for data transfer. The low-level Comedi drivers may nominate
a default "read" subdevice and/or a default "write" subdevice, but it
may have other subdevices that support asynchronous commands.
The Comedi core provides a somewhat clunky mechanism to provide access
to the asynchronous command support of the non-default subdevices. When
a low-level device is "attached" to a core Comedi device, it dynamically
allocates a minor device number for each of the subdevices that support
asynchrounous commands and associates them with files created in SysFS
named "comediX_subdY", where "X" is the minor device number of the main
comedi device, and "Y" is the subdevice number. An application can open
these subdevice-specific files and they behave like the regular
"comediX" files except that the "read" and/or "write" subdevice may be
different to the default chosen by the low-level driver.
This patch adds a layer of indirection between the file object and the
comedi device object to allow the current "read" and/or "write"
subdevice to be altered after opening the Comedi device, on a per-file
object basis. The advantage is that an application only needs to open
the main Comedi device file and can then choose which subdevice it wants
to "read" or "write". The main Comedi device file can be opened more
than once, and each file object can choose the "read" and "write"
subdevices independently.
The new `struct comedi_file` is created on "open" and freed on
"release". It includes pointers to the main Comedi device structure,
and to the current "read" and "write" subdevice structures (which may be
NULL). It also has information to keep track of when a low-level device
has been attached or detached since the previous time the file object
was used. In that case, the current "read" and "write" subdevices in
the `struct comedi_file` will be changed to the new defaults (or set to
NULL). (The change to new defaults is done by `comedi_file_reset()`.
The checking for attach/detach is done by `comedi_file_check()` which
will call `comedi_file_reset()` if there have been any attach/detach
operations since the previous call.)
A subsequent patch will add the ioctls to change the current "read" and
"write" subdevices.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
This inline helper function has been replaced with comedi_bytes_per_sample().
The same commit (bf33eb4b4f57) introduced a couple other related helper
functions a manipulate the sample size.
Use the new helper functions to remove the use of 'bytes_per_sample()' and
remove it.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`do_bufinfo_ioctl()` handled the `COMEDI_BUFINFO` ioctl. It is supposed
to update the read or write positions in the buffer depending on the
direction of data transfer set up by the asynchronous command.
Currently it checks the `SDF_CMD_READ` and `SDF_CMD_WRITE` subdevice
flags. That's fine for most subdevices - the ones that only support one
direction, but is incorrect for those subdevices that allow the command
to be set up in either direction. Since we now set the `CMDF_WRITE`
flag according to the data transfer direction of the current command
running on the subdevice, check that flag instead.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`comedi_poll()` handles the poll() file operation for comedi devices.
If no asynchronous command has been set up on the current "read"
subdevice, it sets the `POLLIN` and `POLLRDNORM` bits in the return
value to indicate that the read() file operation would not block as it
would return an error. Add a check so it also does that if the
asynchronous command has been set up in the "write" direction as that
also causes the read() file operation to return an error.
Similarly, if no asynchronous command has need set up on the current
"write" subdevice, it sets the `POLLOUT` and `POLLWRNORM` bits in the
return value to indicate that the write() file operation would not block
as it would return an error. Add a check so it also does that if the
asynchronous command has been set up in the "read" direction as that
also causes the write() file operation to return an error.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
If a Comedi asynchronous command has been set up for data transfer in
the "read" direction on the current "write" subdevice (for those
subdevices that support both directions), don't allow the "write" file
operation as that would mess with the data in the comedi data buffer
that is written by the low-level comedi hardware driver.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
If a Comedi asynchronous command has been set up for data transfer in
the "write" direction on the current "read" subdevice (for those
subdevices that support both directions), don't allow the "read" file
operation as that would mess with the data in the comedi data buffer
that is read by the low-level comedi hardware driver.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Most comedi subdevices that support asynchronous commands only support
data transfer in either the "read" or "write" direction, as indicated by
the `SDF_CMD_READ` and `SDF_CMD_WRITE` subdevice flags, although a few
support both directions on the same subdevice (though not
simultaneously). The `struct comedi_cmd` structure passed via ioctl
call to set up the command contains a `CMDF_WRITE` flag that can be used
to choose the direction if the subdevice supports both directions, but
the flag is optional if the subdevice only supports data transfer in one
direction.
If the subdevice only supports asynchronous data transfer in a sing
direction, set the `CMDF_WRITE` flag to the correct state so that Comedi
can make use of it later. In the case of the `COMEDI_CMDTEST` ioctl,
the updated flag will be written back to the `struct comedi_cmd` in
user-space. In the case of the `COMEDI_CMD` ioctl, the flag only gets
written back if an error is detected while testing the command, or if
the `CMDF_BOGUS` command flag is set. Since `__comedi_get_user_cmd()`
is called for both ioctls, that's a good place to set the flag.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
We want the upstream fixes in here as well.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
As a follow-up to commit 6cab7a37f5c04 ("staging: comedi: (regression)
channel list must be set for COMEDI_CMD ioctl"), Hartley Sweeten pointed
out another couple of bugs stemming from commit 6cab7a37f5c04 ("staging:
comedi: comedi_fops: introduce __comedi_get_user_chanlist()").
Firstly, `do_cmdtest_ioctl()` never frees the kernel copy of the user
chanlist allocated by `__comedi_get_user_chanlist()`, so that memory is
leaked. Fix it by freeing the allocated kernel memory pointed to by
`cmd.chanlist` before that pointer is overwritten with its original
pointer to user memory before `cmd` is copied back to user-space.
Secondly, if `__comedi_get_user_chanlist()` returns an error,
`cmd->chanlist` is left unchanged and in fact will be a pointer to user
memory. This causes `do_cmd_ioctl()` to `goto cleanup` and call
`do_become_nonbusy()` which would attempt to free the memory pointed to
by the user-space pointer. Fix it by setting `cmd->chanlist` to NULL at
the start of `__comedi_get_user_chanlist()`.
Fixes: c6cd0eefb27b ("staging: comedi: comedi_fops: introduce __comedi_get_user_chanlist()")
Reported-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: <stable@vger.kernel.org> # 3.15.y 3.16.y 3.17.y: 6cab7a37f5c04
Cc: <stable@vger.kernel.org> # 3.15.y 3.16.y 3.17.y
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
For the `COMEDI_LOCK`, `COMEDI_UNLOCK`, `COMEDI_CANCEL`, and
`COMEDI_POLL` ioctls the third argument is a comedi subdevice number.
This is passed as an `unsigned long`, but when it is passed down to the
ioctl command-specific handler functions `do_lock_ioctl()`,
`do_unlock_ioctl()`, `do_cancel_ioctl()`, and `do_poll_ioctl()`, the
value has been narrowed to an `unsigned int`. Pass through the argument
as an `unsigned long` to avoid truncating the value on 64-bit
architectures.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
Introduce COMEDI_CB_ERROR_MASK and COMEDI_CB_CANCEL_MASK to clarify the
async->events that indicate errors and cancel an async command.
Use the new defines to tidy up and clarify the code.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`do_cmd_ioctl()`, the handler for the `COMEDI_CMD` ioctl can incorrectly
call the Comedi subdevice's `do_cmd()` handler with a NULL channel list
pointer. This is a regression as the `do_cmd()` handler has never been
expected to deal with that, leading to a kernel OOPS when it tries to
dereference it.
A NULL channel list pointer is allowed for the `COMEDI_CMDTEST` ioctl,
handled by `do_cmdtest_ioctl()` and the subdevice's `do_cmdtest()`
handler, but not for the `COMEDI_CMD` ioctl and its handlers.
Both `do_cmd_ioctl()` and `do_cmdtest_ioctl()` call
`__comedi_get_user_chanlist()` to copy the channel list from user memory
into dynamically allocated kernel memory and check it for consistency.
That function currently returns 0 if the `user_chanlist` parameter
(pointing to the channel list in user memory) is NULL. That's fine for
`do_cmdtest_ioctl()`, but `do_cmd_ioctl()` incorrectly assumes the
kernel copy of the channel list has been set-up correctly.
Fix it by not allowing the `user_chanlist` parameter to be NULL in
`__comedi_get_user_chanlist()`, and only calling it from
`do_cmdtest_ioctl()` if the parameter is non-NULL.
Thanks to Bernd Porr for reporting the bug via an initial patch sent
privately.
Fixes: c6cd0eefb27b ("staging: comedi: comedi_fops: introduce __comedi_get_user_chanlist()")
Reported-by: Bernd Porr <mail@berndporr.me.uk>
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Bernd Porr <mail@berndporr.me.uk>
Cc: <stable@vger.kernel.org> # 3.15.y 3.16.y 3.17.y
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
The following patch compressed two lines of code into one using
coccinelle and removed an unused variable.
The semantic patch used is as follows:
@@
expression ret;
identifier f;
@@
-ret =
+return
f(...);
-return ret;
Signed-off-by: Heena Sirwani <heenasirwani@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`TRIG_BOGUS` is now just a synonym for `CMDF_BOGUS`. Change
`do_cmd_ioctl()` to use the new name.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
`TRIG_WAKE_EOS` is now just a synonym for `CMDF_WAKE_EOS`. Change
`do_cmd_ioctl()` to use the new name.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
(*buf_change)
This parameter is never used by any of the comedi drivers that provide a
(*buf_change) callback. If the 'new_size' is needed in the callback it can
be found from the 's->async->prealloc_bufsz' as done in the ni_pcidio driver.
Remove the unused parameter.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|