Re: [libusb] Timeouts then Access Denied Errors on Win 7

[Richard B. <ra...@lu...>]

Actually, mea culpa.  I'm pretty sure its the same problem as described in
this open issue https://github.com/libusb/libusb/issues/16



On Wed, Dec 10, 2014 at 1:34 PM, Richard Bryan <ra...@lu...> wrote:

>
>
>>
>> No, you have received no "consensus" of any kind.  You're going to need
>> to figure out what combination of things triggers this.  Are the calls
>> to libusb_get_device_list in a separate thread?  Does this happen only
>> when you call libusb_get_device_list while one of your reads is
>> outstanding?  If this is caused by some interference between the two
>> processes, that's a problem that can be tackled, but it's going to take
>> some analysis.
>>
>> As it is, you have nothing but vague suspicions.
>>
>>
> I know I haven't received a "consensus", that's why I was asking...
>
> The program interacts with libusb within a single thread.  Calls to
> libusb_get_device_list never occur when a bulk transfer read or write is
> occurring.  As it was I had no suspicions at all as to the cause of the
> problem, I was just repeating what Xiaofan had said the cause of the
> problem might be.
>
> Some less-vagues supsicions though... Based on the absence of certain
> error/warning messages in the debug log, I believe the error is triggered
> at calls to one of the following places:
>
>
>
> https://github.com/libusb/libusb/blob/ada59db186aa95215158129db8cd42e079daff07/libusb/os/windows_usb.c#L118
>
>
> https://github.com/libusb/libusb/blob/ada59db186aa95215158129db8cd42e079daff07/libusb/os/windows_usb.c#L123
>
>
> https://github.com/libusb/libusb/blob/ada59db186aa95215158129db8cd42e079daff07/libusb/os/windows_usb.c#L1412
>
> The debug message associated with the last place (line 1412) happens
> regularly -- before any errors with devices occur.  Additionally, the
> devices with which the problem occurs are not HID devices.  So its probably
> coming from CHECK_WINUSBX_AVAILABLE?  But I wouldn't expect the sub_api
> to somehow become uninitialized after being initialized correctly at the
> start... And there are no errors in the log that would indicate this.
>
> I'll see if I can make a simple test program that reproduces this
> consistently.
>
>
>
>