On Wed, 27 Feb 2002 07:16, David Brownell wrote:
<snip>
> The "fxload" utility (on linux-hotplug) works nicely to download
> FX2 firmware, including test firmware.  I just checked in some
> updates for second stage loader support ... needs more testing
> though.  (And note that "fxload" is now linux-hotplug/fxload, not
> linux-hotplug/admin/fxload, in CVS!)
I tested the following (from the 2.0B4 driver disk):
LEDCycle.hex - works (internal only, and using a3load.hex as a pre-loader)
dev_io.hex - works (claims to be internal only, but needs external load - I 
used a3load.hex). Not especially reliable (aka worked before, now stuffed)
mon-ext-sio1-c0.hex - appears to load, sets "monitor mode" LED(D7), untested. 
(this is big external load - around 4K, I used a3load.hex again)
mon-ext-sio1-e0.hex - fails to load - not compatible with FX2 device
mon-int-sio1.hex - appears to load, sets "monitor mode" LED (D7), untested. 
(internal only, I also tried it with an external load, and it is fine)
bulkloop.hex - appears to load, but doesn't re-enumerate. Need to reset or 
reconnect to get the device back. Under investigation. (external, used 
a3load.hex)
bulksrc.hex -as for bulkloop.hex
bulkext.hex - as for bulkloop.hex
imemtest.hex - appears to load OK, but nothing happens (no reappearing device, 
no 7-segment display activity. Under investigation (external, used 
a3load.hex)
memtest.hex - as for imemtest

In addition to the a3load.hex, I did some trials with Vend_ax.hex. Looks to be 
equivalent.

These might be problems with the developmental silicon, but some of it works, 
so I don't see why it all can't work.


> Turns out the Rev E FX2 silicon (current, the first "production"
> silicon) works a lot better than the Rev B stuff we first got from
> Cypress.  It enumerates reliably at full speed too, for starters!
My device is fine with both OHCI and EHCI. I no longer see enumeration 
problems.

> As for test apps, I've tweaked the "bulk src" example a bit
> so that it's a straight source/sink, and it's been invaluable
> for testing bulk and interrupt transfers.  I've got a driver I use
> with it via "dd bs=... count=... < ... > ..." to read and write
> whatever data I want.  And when time allows, a new version
> will be supporting ISO ... :)
With your driver and your test firmware (which also works reliably, internal 
only) on 2.4.18-rc2-gregkh-1, I got the following:
bs=1024, count=1024 -> around 5 Meg/sec (real = 0.2 sec)
bs=10240, count=10240 -> around 20 Meg/sec (real = 5 sec)

Brad

_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to