On Wed, 29 Oct 2003, Julian Back wrote:

> Here is the relevant section from the device manual, it may explain the 
> situation more clearly:
> 
> "18.6.2 Forcible Stall by Application
> 
> The application uses the EPSTL register to issue a stall request for the 
> USB function module. When the application wishes to stall a specific 
> endpoint, it sets the corresponding bit in EPSTL (1- 1 in figure 18.13). 
> The internal status bits are not changed at this time. When a 
> transaction is sent from the host for the endpoint for which the EPSTL 
> bit was set, the USB function module references the internal status bit, 
> and if this is not set, references the corresponding bit in EPSTL (1-2 
> in figure 18.13). If the corresponding bit in EPSTL is set, the USB 
> function module sets the internal status bit and returns a stall 
> handshake to the host (1-3 in figure 18.13). If the corresponding bit in 
> EPSTL is not set, the internal status bit is not changed and the 
> transaction is accepted.
> 
> Once an internal status bit is set, it remains set until cleared by a 
> Clear Feature command from the host, without regard to the EPSTL 
> register. Even after a bit is cleared by the Clear Feature command (3-1 
> in figure 18.13), the USB function module continues to return a stall 
> handshake while the bit in EPSTL is set, since the internal status bit 
> is set each time a transaction is executed for the corresponding 
> endpoint (1-2 in figure 18.13). To clear a stall, therefore, it is 
> necessary for the corresponding bit in EPSTL to be cleared by the 
> application, and also for the internal status bit to be cleared with a 
> Clear Feature command (2-1, 2-2, and 2-3 in figure 18.13)."

Judging from that, it looks like clearing a halt requires letting the 
device software know by sending some vendor-specific message, in addition 
to simply sending a Clear Feature command.  You have to wonder why they 
designed their controller that way.

This doesn't bode well for the SuperH's ability to cope with "flash" BBB.

Alan Stern



-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?   SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to