- 25 Mar, 2013 40 commits
-
-
Johan Hovold authored
Add generic TIOCMIWAIT implementation which correctly handles hangup, USB-device disconnect, does not rely on the deprecated sleep_on functions and hence does not suffer from the races currently affecting several usb-serial drivers. This makes it much easier to add TIOCMIWAIT support to subdrivers as the tricky details related to hangup and disconnect (e.g. atomicity, that the private port data may have been freed when woken up, and waking up processes at disconnect) have been handled once and for all. To add support to a subdriver, simply set the tiocmiwait-port-operation field, update the port icount fields and wake up any process sleeping on the tty-port modem-status-change wait queue on changes. Note that the tty-port initialised flag can be used to detect disconnected as the port will be hung up as part of disconnect (and cannot be reactivated due to the disconnected flag). However, as the tty-port implementation currently wakes up processes before calling port shutdown, the tty-hupping flag must also be checked to detect hangup for now. Signed-off-by:
Johan Hovold <jhovold@gmail.com> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Johan Hovold authored
Add tiocmiwait operation to struct usb_serial_driver. Signed-off-by:
-
Johan Hovold authored
Make sure to unregister the tty-device before calling subdriver port_remove. This way remove will reverse probe, and specifically any port data released in port_remove will be available throughout tty unregister. Note that the order currently does not matter as the tty-layer can make callbacks also after the device has been unregistered. This is handled in usb-serial core using the disconnected flag, which is already set when usb-serial bus device remove is called. Cc: Peter Hurley <peter@hurleysoftware.com> Reported-by:
Peter Hurley <peter@hurleysoftware.com> Signed-off-by:
-
Johan Hovold authored
Use usb_poison_urb to reliably kill all urbs on disconnect and suspend. This way there will be no question that the urbs cannot be resubmitted by buggy subdrivers. Signed-off-by:
-
Johan Hovold authored
The calls to usb_poison_urb and usb_unpoison_urb are expected to be balanced. However, if an urb that has not yet been submitted is poisoned, its reject counter will not be increased as its ep-field is NULL. A consecutive call to unpoison will thus in fact poison the urb as its reject counter will be decremented to a negative value, effectively preventing the urb from being submitted. Note that there are currently no in-kernel drivers affected by this. Cc: Alan Stern <stern@rowland.harvard.edu> Acked-by:
Alan Stern <stern@rowland.harvard.edu> Signed-off-by:
-
Johan Hovold authored
Rename port_release so that all usb_serial_port functions have a common prefix. Signed-off-by:
-
Johan Hovold authored
We should not call kill_traffic (and usb_kill_urb) once disconnect returns. Any pending urbs are killed at disconnect and new submissions are prevented by usb_unbind_interface (and usb_disable_interface). Signed-off-by:
-
Johan Hovold authored
Make sure the interrupt urb submitted in port_probe is killed in port_remove. The interrupt-urb completion handler references the port and may get called after port_remove has returned and the port has been unregistered (although this is currently prevented by usb-serial core as we are using a non-private urb). Signed-off-by:
-
Johan Hovold authored
Most USB serial drivers are, and should be, using as much of the generic implementation as possible. Rename the fixup_generic function to a more descriptive name. Reword the related debug message in a more neutral tone (and remember to add the missing newline). Finally, move the operations initialisation to after the initial sanity checks. Signed-off-by:
-
Johan Hovold authored
Remove empty generic release implementation and make the release callback non-mandatory (like attach, probe and disconnect). Signed-off-by:
-
Johan Hovold authored
Remove the now empty generic disconnect callback and make the disconnect callback non-mandatory. Signed-off-by:
-
Johan Hovold authored
There is no need for the generic disconnect callback to stop the read and write urbs a second time as this has already been taken care of by close (which is called from hangup as part of disconnect). Signed-off-by:
-
Johan Hovold authored
The generic close operation will be used if the close field is left uninitialised. Signed-off-by:
-
Johan Hovold authored
The disconnect callback is set to the generic implementation by usb-serial core if NULL. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in cleanup. The device is never cleared, and cleanup is never called after a completed disconnect anyway. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in close. The device is never cleared, and close is never called after a completed disconnect anyway. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in close. The device is never cleared, and close is never called after a completed disconnect anyway. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in close. The device is never cleared, and close is never called after a completed disconnect anyway. Simplify urb killing, and remove some related debug and dead code while at it. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in close. The device is never cleared, and close is never called after a completed disconnect anyway. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in dtr_rts. The device is never cleared, and disconnect is handled for dtr_rts in usb-serial core anyway. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in close. The device is never cleared, and close is never called after a completed disconnect anyway. Remove some out-commented bogus code while at it. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in close. The device is never cleared, and close is never called after a completed disconnect anyway. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in close. The device is never cleared, and close is never called after a completed disconnect anyway. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in close. The device is never cleared, and close is never called after a completed disconnect anyway. Signed-off-by:
-
Johan Hovold authored
Remove bogus (and unnecessary) test for serial->dev being NULL in close. The device is never cleared, and close is never called after a completed disconnect anyway. Signed-off-by:
-
Johan Hovold authored
Remove redundant port number from debug output (already printed as part of device name). Signed-off-by:
-
Johan Hovold authored
Failed allocations already get an OOM message and a stack dump. Signed-off-by:
-
Johan Hovold authored
Remove redundant port number from debug output (already printed as part of device name). Signed-off-by:
-
Johan Hovold authored
Remove redundant comments and fix some minor coding style issues. Signed-off-by:
-
Johan Hovold authored
Rename the tty-port callbacks using a common prefix to more clearly separate them from the tty and usb driver callbacks. Rename serial_down to serial_port_shutdown to match the callback name. Signed-off-by:
-
Johan Hovold authored
Pass usb-serial port rather than usb device to set_control_lines, and make sure port device is used for all port related debugging. Signed-off-by:
-
Johan Hovold authored
Use interface rather than usb-serial device for debugging interface related operations. This gives more descriptive messages, such as [ 905.669436] pl2303 1-4.1:1.0: 0x40:0x1:0x8:0x0 0 rather than [ 341.943535] usb 1-4.1: 0x40:0x1:0x8:0x0 0 Signed-off-by:
-
Johan Hovold authored
Remove kill of interrupt-in urb on close as it has never been submitted. Signed-off-by:
-
Johan Hovold authored
Remove kill of interrupt-in urb on close as it has never been submitted. Signed-off-by:
-
Johan Hovold authored
Remove the port data refcounting and release the private data explicitly at port remove. The port data refcounting was used to make sure the port data was not freed until the last tty reference was closed. Since moving over to tty ports, the underlying assumptions are no longer valid as close is now called as part of tty port shutdown, which can occur before the final tty reference is dropped on device disconnect. This means that the private port data refcounting is now completely useless, as the last reference will always be dropped on port_remove. Signed-off-by:
-
Alan Stern authored
This patch (as1669) removes the check_unlinks_later flag in ehci-hcd's unlink_empty_async(). It wasn't being used for anything and should have been removed in an earlier patch, but I forgot about it. Signed-off-by:
-
Alan Stern authored
This patch (as1665) changes the way ehci-hcd's end_unlink_async() routine works in order to avoid recursive execution and to be more efficient: Now when an IAA cycle ends, a new one gets started up right away (if it is needed) instead of waiting until the just-unlinked QH has been processed. The async_iaa list is renamed to async_idle, which better expresses its new purpose: It is now the list of QHs which are now completely idle and are waiting to be processed by end_unlink_async(). A new flag is added to track whether an IAA cycle is in progress, because the list formerly known as async_iaa no longer stores the QHs waiting for the IAA to finish. The decision about how many QHs to process when an IAA cycle ends is now made at the end of the cycle, when we know the current state of the hardware, rather than at the beginning. This means a bunch of logic got moved from start_iaa_cycle() to end_unlink_async(). Signed-off-by:
-
Alan Stern authored
This patch (as1664) converts ehci-hcd's async_unlink, async_iaa, and intr_unlink from singly-linked lists to standard doubly-linked list_heads. Originally it didn't seem necessary to use list_heads, because items are always added to and removed from these lists in FIFO order. But now with more list processing going on, it's easier to use the standard routines than continue with a roll-your-own approach. I don't know if the code ends up being notably shorter, but the patterns will be more familiar to any kernel hacker. Signed-off-by:
-
Alan Stern authored
This patch (as1668) consolidates two nearly identical code paths in ehci_urb_dequeue(). The test for !qh can be removed because it will never succeed; the fact that usb_hcd_check_unlink_urb() returned 0 means that urb must be queued and therefore urb->hcpriv must point to a QH. Signed-off-by:
-
Alan Stern authored
This patch (as1662) does some more QH-related cleanup in ehci-hcd. The qh->needs_rescan flag is currently used for two different purposes; the patch replaces it with two separate flags for greater clarity: qh->dequeue_during_giveback indicates that a completion handler dequeued an URB (implying that a rescan is needed), and qh->exception indicates that the QH is in an exceptional state requiring an unlink (either it encountered an I/O error or an unlink was requested). The new flags get set where the dequeue, exception, or unlink request occurred, rather than where the unlink is started. This is so that in the future, if we need to, we will be able to tell apart unlinks that truly were required from those that were carried out merely because the QH wasn't being used. Signed-off-by:
-
