platform/chrome: Add ChromeOS EC ISHTP driver
Hello Rushikesh S Kadam,
The patch 4f346361cf42: "platform/chrome: Add ChromeOS EC ISHTP driver" from May 4, 2019, leads to the following static checker warning:
drivers/platform/chrome/cros_ec_ishtp.c:527 cros_ec_pkt_xfer_ish()
warn: inconsistent returns 'read_sem:&init_lock'.
Locked on: line 475
line 482
Unlocked on: line 467
line 527drivers/platform/chrome/cros_ec_ishtp.c
450 static int cros_ec_pkt_xfer_ish(struct cros_ec_device *ec_dev,
451 struct cros_ec_command *msg)
452 {
453 int rv;
454 struct ishtp_cl *cros_ish_cl = ec_dev->priv;
455 struct ishtp_cl_data *client_data = ishtp_get_client_data(cros_ish_cl);
456 struct device *dev = cl_data_to_dev(client_data);
457 struct cros_ish_in_msg *in_msg = (struct cros_ish_in_msg *)ec_dev->din;
458 struct cros_ish_out_msg *out_msg 459 (struct cros_ish_out_msg *)ec_dev->dout;
460 size_t in_size = sizeof(struct cros_ish_in_msg) + msg->insize;
461 size_t out_size = sizeof(struct cros_ish_out_msg) + msg->outsize;
462
463 /* Proceed only if reset-init is not in progress */
464 if (!down_read_trylock(&init_lock)) {
465 dev_warn(dev,
466 "Host is not ready to send messages to ISH. Try again\n");
467 return -EAGAIN;
468 }
469
470 /* Sanity checks */
471 if (in_size > ec_dev->din_size) {
472 dev_err(dev,
473 "Incoming payload size %zu is too large for ec_dev->din_size %d\n",
474 in_size, ec_dev->din_size);
475 return -EMSGSIZE;
^^^^^^^^^^^^^^^^^
476 }
477
478 if (out_size > ec_dev->dout_size) {
479 dev_err(dev,
480 "Outgoing payload size %zu is too large for ec_dev->dout_size %d\n",
481 out_size, ec_dev->dout_size);
482 return -EMSGSIZE;
^^^^^^^^^^^^^^^^^Could we move these sanity checks to before we take the lock? Otherwise we need to goto end_error;
483 } 484 485 /* Prepare the package to be sent over ISH TP */ 486 out_msg->hdr.channel = CROS_EC_COMMAND; 487 out_msg->hdr.status = 0; 488 489 ec_dev->dout += OUT_MSG_EC_REQUEST_PREAMBLE; 490 cros_ec_prepare_tx(ec_dev, msg); 491 ec_dev->dout -= OUT_MSG_EC_REQUEST_PREAMBLE; 492 493 dev_dbg(dev, 494 "out_msg: struct_ver=0x%x checksum=0x%x command=0x%x command_ver=0x%x data_len=0x%x\n", 495 out_msg->ec_request.struct_version, 496 out_msg->ec_request.checksum, 497 out_msg->ec_request.command, 498 out_msg->ec_request.command_version, 499 out_msg->ec_request.data_len); 500 501 /* Send command to ISH EC firmware and read response */ 502 rv = ish_send(client_data, 503 (u8 *)out_msg, out_size, 504 (u8 *)in_msg, in_size); 505 if (rv < 0) 506 goto end_error; 507 508 rv = prepare_cros_ec_rx(ec_dev, in_msg, msg); 509 if (rv) 510 goto end_error; 511 512 rv = in_msg->ec_response.data_len; 513 514 dev_dbg(dev, 515 "in_msg: struct_ver=0x%x checksum=0x%x result=0x%x data_len=0x%x\n", 516 in_msg->ec_response.struct_version, 517 in_msg->ec_response.checksum, 518 in_msg->ec_response.result, 519 in_msg->ec_response.data_len); 520 521 end_error: 522 if (msg->command = EC_CMD_REBOOT_EC) 523 msleep(EC_REBOOT_DELAY_MS); 524 525 up_read(&init_lock); 526 527 return rv; 528 }
regards,
dan carpenter
Hi Dan
Quoted 48 lines · Dan Carpenter
Hello Rushikesh S Kadam,
The patch
4f346361cf42: "platform/chrome: Add ChromeOS EC ISHTP driver" from May 4, 2019, leads to the following static checker warning:
drivers/platform/chrome/cros_ec_ishtp.c:527cros_ec_pkt_xfer_ish()
warn: inconsistent returns 'read_sem:&init_lock'.
Locked on: line 475line 482 Unlocked on: line 467 line 527
drivers/platform/chrome/cros_ec_ishtp.c
450 static int cros_ec_pkt_xfer_ish(structcros_ec_device*ec_dev,
451 structcros_ec_command*msg)452 { 453 int rv; 454 struct ishtp_cl *cros_ish_cl = ec_dev->priv; 455 struct ishtp_cl_data *client_data = ishtp_get_client_data(cros_ish_cl); 456 struct device *dev = cl_data_to_dev(client_data); 457 struct cros_ish_in_msg *in_msg = (struct cros_ish_in_msg *)ec_dev->din; 458 struct cros_ish_out_msg *out_msg > 459 (struct cros_ish_out_msg *)ec_dev->dout; 460 size_t in_size = sizeof(struct cros_ish_in_msg) + msg->insize; 461 size_t out_size = sizeof(struct cros_ish_out_msg) + msg->outsize; 462 463 /* Proceed only if reset-init is not in progress */ 464 if (!down_read_trylock(&init_lock)) { 465 dev_warn(dev, 466 "Host is not ready to send messages to ISH. Try again\n"); 467 return -EAGAIN; 468 } 469 470 /* Sanity checks */ 471 if (in_size > ec_dev->din_size) { 472 dev_err(dev, 473 "Incoming payload size %zu is too large for ec_dev->din_size %d\n", 474 in_size, ec_dev->din_size); 475 return -EMSGSIZE; ^^^^^^^^^^^^^^^^^ 476 } 477 478 if (out_size > ec_dev->dout_size) { 479 dev_err(dev, 480 "Outgoing payload size %zu is too large for ec_dev->dout_size %d\n", 481 out_size, ec_dev->dout_size); 482 return -EMSGSIZE; ^^^^^^^^^^^^^^^^^
Quoted 2 lines
Could we move these sanity checks to before we take the lock? Otherwise we need to goto
end_error;
Thanks for reporting. I submitted updated patch
with fix here https://lkml.org/lkml/2019/5/15/263
Regards
Rushikesh
Quoted 50 lines
483 } 484 485 /* Prepare the package to be sent over ISH TP */ 486 out_msg->hdr.channel = CROS_EC_COMMAND; 487 out_msg->hdr.status = 0; 488 489 ec_dev->dout += OUT_MSG_EC_REQUEST_PREAMBLE; 490 cros_ec_prepare_tx(ec_dev, msg); 491 ec_dev->dout -= OUT_MSG_EC_REQUEST_PREAMBLE; 492 493 dev_dbg(dev, 494 "out_msg: struct_ver=0x%x checksum=0x%x command=0x%x command_ver=0x%x data_len=0x%x\n", 495 out_msg->ec_request.struct_version, 496 out_msg->ec_request.checksum, 497 out_msg->ec_request.command, 498 out_msg->ec_request.command_version, 499 out_msg->ec_request.data_len); 500 501 /* Send command to ISH EC firmware and read response */ 502 rv = ish_send(client_data, 503 (u8 *)out_msg, out_size, 504 (u8 *)in_msg, in_size); 505 if (rv < 0) 506 goto end_error; 507 508 rv = prepare_cros_ec_rx(ec_dev, in_msg, msg); 509 if (rv) 510 goto end_error; 511 512 rv = in_msg->ec_response.data_len; 513 514 dev_dbg(dev, 515 "in_msg: struct_ver=0x%x checksum=0x%x result=0x%x data_len=0x%x\n", 516 in_msg->ec_response.struct_version, 517 in_msg->ec_response.checksum, 518 in_msg->ec_response.result, 519 in_msg->ec_response.data_len); 520 521 end_error: 522 if (msg->command = EC_CMD_REBOOT_EC) 523 msleep(EC_REBOOT_DELAY_MS); 524 525 up_read(&init_lock); 526 527 return rv; 528 }regards,
dan carpenter