[kbuild-all] [kbuild] [vireshk-pm:opp/linux-next 4/11] drivers/opp/of.c:662:7-14: ERROR: reference preceded by free on line 625

tree: https://git.kernel.org/pub/scm/linux/kernel/git/vireshk/pm.git opp/linux-next head: f4111e2e1ae16a32002ba348a18a59691fcedec3 commit: 0f438abc246a03c0c7a212738329f87000a9e527 [4/11] OPP: Add support for parsing interconnect bandwidth If you fix the issue, kindly add following tag as appropriate Reported-by: kbuild test robot <lkp(a)intel.com&gt; Reported-by: Julia Lawall <julia.lawall(a)lip6.fr&gt; coccinelle warnings: (new ones prefixed by >>) # https://git.kernel.org/pub/scm/linux/kernel/git/vireshk/pm.git/commit/?id... git remote add vireshk-pm https://git.kernel.org/pub/scm/linux/kernel/git/vireshk/pm.git git remote update vireshk-pm git checkout 0f438abc246a03c0c7a212738329f87000a9e527 vim +662 drivers/opp/of.c 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 580 static int _read_opp_key(struct dev_pm_opp *new_opp, struct device_node *np, 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 581 bool *rate_not_available) 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 582 { 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 583 struct property *peak, *avg; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 584 u32 *peak_bw, *avg_bw; 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 585 u64 rate; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 586 int ret, i, count; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 587 bool found = false; 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 588 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 589 ret = of_property_read_u64(np, "opp-hz", &rate); 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 590 if (!ret) { 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 591 /* 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 592 * Rate is defined as an unsigned long in clk API, and so 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 593 * casting explicitly to its type. Must be fixed once rate is 64 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 594 * bit guaranteed in clk API. 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 595 */ 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 596 new_opp->rate = (unsigned long)rate; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 597 found = true; 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 598 } 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 599 *rate_not_available = !!ret; 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 600 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 601 peak = of_find_property(np, "opp-peak-kBps", NULL); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 602 if (peak) { 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 603 /* 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 604 * Bandwidth consists of peak and average (optional) values: 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 605 * opp-peak-kBps = <path1_value path2_value>; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 606 * opp-avg-kBps = <path1_value path2_value>; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 607 */ 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 608 count = peak->length / sizeof(u32); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 609 peak_bw = kmalloc_array(count, sizeof(*peak_bw), GFP_KERNEL); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 610 if (!peak_bw) 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 611 return -ENOMEM; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 612 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 613 ret = of_property_read_u32_array(np, "opp-peak-kBps", peak_bw, 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 614 count); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 615 if (ret) { 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 616 pr_err("%s: Error parsing opp-peak-kBps: %d\n", 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 617 __func__, ret); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 618 goto free_peak_bw; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 619 } 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 620 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 621 for (i = 0; i < count; i++) 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 622 new_opp->bandwidth[i].peak = kBps_to_icc(peak_bw[i]); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 623 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 624 found = true; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 @625 kfree(peak_bw); ^^^^^^^^^^^^^ 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 626 } 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 627 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 628 avg = of_find_property(np, "opp-avg-kBps", NULL); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 629 if (peak && avg) { 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 630 count = avg->length / sizeof(u32); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 631 avg_bw = kmalloc_array(count, sizeof(*avg_bw), GFP_KERNEL); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 632 if (!avg_bw) { 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 633 ret = -ENOMEM; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 634 goto free_peak_bw; ^^^^^^^^^^^^^^^^^^ Obviously this goto leads to a double free etc... 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 635 } 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 636 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 637 ret = of_property_read_u32_array(np, "opp-avg-kBps", avg_bw, 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 638 count); This code is really confusing. This assignment over writes the "ret" variable. I guess that's intended instead of using "found = true;" on this path??? It really feels like this function should be split into multiple functions. If I read it correctly, then if it can find any property then that's counted as a success? I guess we'd like to find them all but if we can find one then that's enough? 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 639 if (ret) { 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 640 pr_err("%s: Error parsing opp-avg-kBps: %d\n", 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 641 __func__, ret); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 642 goto free_avg_bw; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 643 } 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 644 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 645 for (i = 0; i < count; i++) 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 646 new_opp->bandwidth[i].avg = kBps_to_icc(avg_bw[i]); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 647 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 648 kfree(avg_bw); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 649 } 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 650 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 651 if (!of_property_read_u32(np, "opp-level", &new_opp->level)) 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 652 found = true; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 653 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 654 if (found) 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 655 return 0; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 656 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 657 return ret; 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 658 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 659 free_avg_bw: 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 660 kfree(avg_bw); 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 661 free_peak_bw: 0f438abc246a03 drivers/opp/of.c Georgi Djakov 2020-05-12 @662 kfree(peak_bw); 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 663 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 664 return ret; 6c591eec67cbb4 drivers/opp/of.c Saravana Kannan 2020-05-12 665 } --- 0-DAY CI Kernel Test Service, Intel Corporation https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org _______________________________________________ kbuild mailing list -- kbuild(a)lists.01.org To unsubscribe send an email to kbuild-leave(a)lists.01.org