Hi Alan, One of my colleagues reports a problem that the enumeration will fail if the storage has some problems. I can re-produce this problem if I add a big delay (like 500ms) at handle_exception, at condition FSG_STATE_DISCONNECT, please see below code:
3050 case FSG_STATE_DISCONNECT: 3051 for (i = 0; i < fsg->nluns; ++i) 3052 fsg_lun_fsync_sub(fsg->luns + i); 3053 mdelay(500); 3054 do_set_config(fsg, 0); // Unconfigured state 3055 break; When the problem occurs, the call sequence like below: reset interrupt-> fsg->disconnect-> handle_exception reset interrupt-> fsg->disconnect ... udc setup package -> USB_REQ_SET_CONFIGURATION -> raise_exception(fsg, FSG_STATE_CONFIG_CHANGE); The reason why this problem occurs is that the first fsg_lun_fsync_sub(fsg->luns + i) consumes too much times, the fsg->state is changed by raise_exception (2nd reset interrupt), but the handle_exception (2nd reset interrupt) is not called after raise_exception(fsg, FSG_STATE_CONFIG_CHANGE) is called. The fsg->state is still FSG_STATE_DISCONNECT when raise_exception(fsg, FSG_STATE_CONFIG_CHANGE) is called, Since FSG_STATE_DISCONNECT > FSG_STATE_CONFIG_CHANGE, the SET_CONFIGURATION will never be handled, the enumeration will be failed. A workaround for this problem is: 3050 case FSG_STATE_DISCONNECT: 3051 if (fsg->config != 0) 3052 for (i = 0; i < fsg->nluns; ++i) 3053 fsg_lun_fsync_sub(fsg->luns + i); 3054 do_set_config(fsg, 0); // Unconfigured state 3055 break; But it can't handle the problem that the reset occurs during the transfer due to some timeout. (bus reset -> bus reset -> SET_CONFIGURATION) Best regards, Peter Chen MAD Linux BSP Team Freescale Semiconductor Ltd. -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
