Age | Commit message (Collapse) | Author | Files | Lines |
|
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
|
|
If our buffer gets completely drained, wait for it to refill before
submitting more packets. If our buffers overflows, drop not just the package
causing the overflow, but drop half the buffer size packets, so that it
should be at its target size of being half-filled again.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
|
|
If a control packet is a clear stall, we need to do an actual clear stall,
rather then just forward the control packet, so that the usbhost usbstack
knows the stall is cleared.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
|
|
@libdir@ is expanded to ${exec_prefix}/lib on my machine so we
need exec_prefix to be defined in the .pc file.
Thanks to Marc-André Lureau for reporting this.
|
|
Oops.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
|
|
Having separate set_locking_funcs functions causes the locking functions
to be available potentially too late. And it just is not very pretty.
This commit changes the API (in a way which is backward compat with 0.3.1), so
that locking functions can be specified at creation time, just like all the
other callbacks.
This commit also bumps the reported version to 0.3.2, so that apps using
libusbredirhost can check for this version to check the presence of the new
usbredirhost_open_full function.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
|
|
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
|
|
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
|
|
Replace the current build system with an autotools based one.
HdG, 3 small changes:
1) LIBUSBREDIRPARSER_SO_VERSION 1:0:0 should be 0:0:0, so that the soname
does not change by going from Makefiles to autofoo
2) Drop LIBGPOD_SO_VERSION from configure.ac
3) Make libusb pkgconfig check require version >= 1.0.9
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
|
|
usbredirhost_cancel_iso_stream should call
usbredirhost_cancel_iso_stream_unlocked instead of calling itself. This fixes
usbredirhost_cancel_iso_stream going into an endless recursion loop.
Thanks to coolper chen <lixin.chen@saicocch.com> for reporting this!
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Document that libusbredirhost_open closes the past in libusb device handle
on failure, and make sure this is always true.
|
|
|
|
Add a usbredirparser_free_packet_data function for symetry with
usbredirparser_free_write_buffer.
|
|
|
|
The return statement which made this a nop was added as a quick test to
work around an issue with a morphing device, it should have never ended up
in git ...
|
|
|
|
This:
-fixes a memleak where we would not free the transfer + transfer buffer
for iso / int transfers
-stops libusb from complaining there are still transfers pending on device
close
|
|
And add a new usb_redir_interface_info packet type which contains the
interface info.
|
|
And document how a single usbredirproto connection may be reused for
a newly plugged in / re-plugged device.
|
|
|
|
|
|
Now that there no longer is a usb_reset_status packet type / callback having
an id argument makes no sense.
|
|
Drop usb_redir_reset_status, instead if reconnecting to the device fails
after reset the usb-host will send a usb_redir_device_disconnected packet
|
|
Drop usb_redir_disconnected status, instead the usb-host should always
send a usb_redir_device_disconnected packet on device disconnection. The
reason behind this is that having to handle disconnection from data packet
handlers make things unnecessarily hard for the usb-guest
|
|
|
|
|
|
|
|
When cancelling an active iso stream some iso packets will likely complete
while we are cancelling, but we will see them as uncomplete as long as we've
not asked libusb to check for complete transfers, this is inherently racy and
there is nothing we can do. Thus we will sometimes try to cancel already
completed transfers which will fail.
This is not a problem, but libusb spews ugly error messages to stderr when
this happens. This patch silences these error messages.
|
|
|
|
|
|
|
|
|
|
|