Hi,
I discovered this little correctness bug when messing with trying to
work out why my flyswatter was unreliable.
in ft2232.c
ftdi_set_interface needs to be called to set the active interface BEFORE
ftdi_usb_open_desc, because ftdi_usb_open_desc needs to know which
interface to open.
There is no bug exhibited in openocd, because the interface defaults to
Interface A. But the code as wrote is still incorrect, and if libftdi
ever changes the default, then it will break. Also, this will cause
problems if anyone makes a jtag device using Interface B of the new
2232H chip from FTDI.
Patch attached.
Note: I haven't tested this yet, because SVN Head is segfaulting for
me. More on that later. Yet, I am confident that it will work OK.
Strontium
Index: src/jtag/ft2232.c
===================================================================
--- src/jtag/ft2232.c (revision 1743)
+++ src/jtag/ft2232.c (working copy)
@@ -1766,6 +1766,12 @@
if (ftdi_init(&ftdic) < 0)
return ERROR_JTAG_INIT_FAILED;
+ if (ftdi_set_interface(&ftdic, INTERFACE_A) < 0)
+ {
+ LOG_ERROR("unable to select FT2232 channel A: %s", ftdic.error_str);
+ return ERROR_JTAG_INIT_FAILED;
+ }
+
/* context, vendor id, product id */
if (ftdi_usb_open_desc(&ftdic, vid, pid, ft2232_device_desc,
ft2232_serial) < 0)
@@ -1779,12 +1785,6 @@
return ERROR_JTAG_INIT_FAILED;
}
- if (ftdi_set_interface(&ftdic, INTERFACE_A) < 0)
- {
- LOG_ERROR("unable to select FT2232 channel A: %s", ftdic.error_str);
- return ERROR_JTAG_INIT_FAILED;
- }
-
if (ftdi_usb_reset(&ftdic) < 0)
{
LOG_ERROR("unable to reset ftdi device");
_______________________________________________
Openocd-development mailing list
Openocd-development@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/openocd-development
