Merge branch 'epson2/better_timeouts' into 'master'

epson2: better timeouts

See merge request sane-project/backends!605

backend/epson2-ops.c

@@ -588,7 +588,7 @@ e2_discover_capabilities(Epson_Scanner *s)
 	/* ESC F, request status */
 	status = esci_request_status(s, &scanner_status);
 	if (status != SANE_STATUS_GOOD)
-		return status;;
+		return status;
 
 	/* set capabilities */
 	if (scanner_status & STATUS_OPTION)
@@ -1593,7 +1593,7 @@ e2_check_adf(Epson_Scanner * s)
 
 		status = esci_request_extended_status(s, &buf, NULL);
 		if (status != SANE_STATUS_GOOD)
-			return status;;
+			return status;
 
 		t = buf[1];
 
@@ -1646,11 +1646,23 @@ e2_start_ext_scan(Epson_Scanner * s)
 	if (buf[0] != STX)
 		return SANE_STATUS_INVAL;
 
-	if (buf[1] & 0x80) {
+	if (buf[1] & STATUS_FER) {
 		DBG(1, "%s: fatal error\n", __func__);
 		return SANE_STATUS_IO_ERROR;
 	}
 
+	/*
+	 * The 12000XL signals busy only with FS+G, all other status queries
+	 * say non-busy. Probably because you can in deed communicate with the
+	 * device, just scanning is not yet possible. I tried polling with FS+G
+	 * every 5 seconds, but that made scary noises. So, bail out and let
+	 * the user retry manually.
+	 */
+	if (buf[1] & STATUS_NOT_READY) {
+		DBG(1, "%s: device not ready\n", __func__);
+		return SANE_STATUS_DEVICE_BUSY;
+	}
+
 	s->ext_block_len = le32atoh(&buf[2]);
 	s->ext_blocks = le32atoh(&buf[6]);
 	s->ext_last_len = le32atoh(&buf[10]);

backend/epson2.c

@@ -910,6 +910,7 @@ sane_init(SANE_Int *version_code, SANE_Auth_Callback __sane_unused__ authorize)
 					  EPSON2_BUILD);
 
 	sanei_usb_init();
+	sanei_usb_set_timeout(60 * 1000);
 
 	return SANE_STATUS_GOOD;
 }