|
Message
From: rtl2gds@S...<rtl2gds@S...>
Date: Sun Sep 5 10:14:30 CEST 2004
Subject: [usb] timeout
Hi Pranav,
Big thanks for your response, now I understand.
Michal
----- Original Message ----- From: pranav kumar<pranav_c_java@y...> To: Date: Fri Aug 27 12:25:21 CEST 2004 Subject: [usb] timeout
> hi michal, > > I think this might help you. > > For full-/low-speed transactions, the timer starts > counting on the SE0-to-J transition of the EOP > strobe and > stops counting when the Idle-to-K SOP transition is > detected. For high-speed transactions, the timer > starts > counting when the data lines return to the squelch > level and stops counting when the data lines leave the > squelch level. > If a response is not received within this worst case > timeout(18 bit timres for FS), then the transmitter > considers that the packet transmission has failed. > Then the concerned register bit is changed and the > next Token packet is sent. > pranav > > > Hi, > > I mean resetting/starting timeout timer not USB > > reset. > > I understand spec like that: > > For eg. when device receives setup token in control > > transfer it expects > > data0 packet to arrive before timeout after setup > > token(18 bit times for > > Full Speed). So, it starts timeout timer after > > succesfull reception of > > SETUP (ie after EOP = SE0 to J transition) and check > > if timout occurs. > > > > I am not sure, but timeout timer maybe should be set > > to value 18 bit > > times minus SETUP packet length duration time in > > this example, because > > timer starts after SETUP packet is received. > > Am i right? > > > > Michal > > > > > > > > > > I would like to ask about timeout values in usb > > 2.0 core, why > > > these > > > > are > > > > 622 ns FS(7.5b times) and 400 ns HS (192b > > times)? > > > > Shouldn't these be 18 bit times in FS and 816 > > bit times in HS > > > as > > > > stated > > > > in 7.1.18 of usb spec (end to end delay)? > > > > And how timeout is checked? Is timeout counter > > started when > > > > linestate > > > > enters squelch level (or SE0 to J transition for > > FS) and reset > > > when > > > > linestate leaves squelch (or J to K transition > > for FS), when > > > device > > > > waits > > > > for data or response from usb host)? > > > > > > > > > > > > > > > > _______________________________________________ > > http://www.opencores.org/mailman/listinfo/usb > > > __________________________________________________ > Do You Yahoo!? > Tired of spam? Yahoo! Mail has the best spam protection around > http://mail.yahoo.com > >
|
 |