Fix compiler warnings #1609
Closed
Fix compiler warnings #1609
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
[ Upstream commit a69041a ] Update testing.md to properly reflect recent changes to the test suite execution. Fixes: d6c1c23 ('tests: Fix test locating process') Signed-off-by: Noa Osherovich <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 381de8c ] Avoid the following failure when running the tests from CWD of /usr/share/doc/rdma-core-28.0/tests [root@rdma-dev-25 tests]$ python3 run_tests.py -v Traceback (most recent call last): File "run_tests.py", line 11, in <module> tests = SourceFileLoader('tests', module_path).load_module() File "<frozen importlib._bootstrap_external>", line 407, in _check_name_wrapper File "<frozen importlib._bootstrap_external>", line 907, in load_module File "<frozen importlib._bootstrap_external>", line 732, in load_module File "<frozen importlib._bootstrap>", line 265, in _load_module_shim File "<frozen importlib._bootstrap>", line 696, in _load File "<frozen importlib._bootstrap>", line 677, in _load_unlocked File "<frozen importlib._bootstrap_external>", line 724, in exec_module File "<frozen importlib._bootstrap_external>", line 859, in get_code File "<frozen importlib._bootstrap_external>", line 916, in get_data FileNotFoundError: [Errno 2] No such file or directory: '/__init__.py Fixes: d6c1c23 ("tests: Fix test locating process") Signed-off-by: Kamal Heib <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 68c6238 ] After the change to distinguish QP numbers and QP handles, the cqe->qp field contains the QP handle so as to make lookup for QPs simple. Flushing CQEs upon destroy of QP needs to be done by comparing the cqe->qp field with the QP handle, not the QP number. This change fixes this issue. Fixes: 3dbaeba ("vmw_pvrdma: Use resource ids from physical device if available") Reviewed-by: Jorgen Hansen <[email protected]> Reviewed-by: Rajesh Jalisatgi <[email protected]> Signed-off-by: Bryan Tan <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 0203641 ] Avoid the following exception when no IB device found by modifying the ParentDomainTestCase to inherit from RDMATestCase that skip the test if there is no IB device. ERROR: test_default_allocators (tests.test_parent_domain.ParentDomainTestCase) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/kheib/git/upstream/rdma-core/tests/test_parent_domain.py", line 35, in setUp dev = d.get_device_list()[-1] IndexError: list index out of range Fixes: 4fec673 ("tests: Add a test for parent domain") Signed-off-by: Kamal Heib <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 55654e7 ] Avoid the following failure by call PyverbsRDMAErrno() with the exact number of arguments. Traceback (most recent call last): File "/home/rdma-core/tests/test_odp.py", line 56, in test_odp_xrc_traffic client, server = self.create_players('xrc') File "/home/rdma-core/tests/test_odp.py", line 35, in create_players client = self.qp_dict[qp_type](self.dev_name, self.ib_port, File "/home/rdma-core/tests/base.py", line 367, in __init__ super(XRCResources, self).__init__(dev_name, ib_port, gid_index) File "/home/rdma-core/tests/base.py", line 238, in __init__ self.init_resources() File "/home/rdma-core/tests/base.py", line 454, in init_resources self.create_xrcd() File "/home/rdma-core/tests/base.py", line 413, in create_xrcd self.xrcd = XRCD(self.ctx, init) File "xrcd.pyx", line 54, in pyverbs.xrcd.XRCD.__init__ TypeError: PyverbsRDMAErrno() takes exactly one argument (2 given) Fixes: b68ce6f ("pyverbs: Introducing XRCD class") Signed-off-by: Kamal Heib <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit ca2ecc7 ] CQ has its private error codes, there is no need to use other error codes in hns_roce_handle_recv_inl_wqe, which will affect the return value of the poll cq function. And if the err state is V2_CQ_POLL_ERR, the ci pointer needs to be updated as the cqe has already been generated. Fixes: b8bdc87 ("libhns: Optimize some codes for hns userspace") Fixes: a7d029e ("libhns: Introduce CQ operations referred to hip08 device") Signed-off-by: Yangyang Li <[email protected]> Signed-off-by: Weihang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 1260a28 ] This patch fixes two issues: 1- Avoid iterating out of the range of GIDs table length. 2- Make sure to iterate over the whole GID table as could be holes of ZERO_GID. Fixes: bb59586 ("tests: RDMATestCase") Signed-off-by: Kamal Heib <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit d04d466 ] This patch fixes the following Coverity complaint: CID 1490689 (#1 of 1): Resource leak (RESOURCE_LEAK) 15. leaked_storage: Variable dev going out of scope leaks the storage it points to. Signed-off-by: Bart Van Assche <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 2da066a ] Avoid code duplication by move the code of getting the available devices to a separate function. Fixes: 8c55be0 ("tests: Skip old tests when no IB devices are found") Signed-off-by: Kamal Heib <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit fdde742 ] Commit 26e05f8 ("ibacm: use ccan/list.h") introduced a bug in acm_get_ep() by not incrementing inx. Fixes: 26e05f8 ("ibacm: use ccan/list.h") Signed-off-by: Håkon Bugge <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit a58a5e9 ] For dual port HCA, which has an InfiniBand port and an Ethernet port, only open InfiniBand port will introduce segment fault issues. Because the Ethernet port did not open yet, segment fault when active the Ethernet port. The second segment fault issue happens when there is asyn event on the Ethernet port. We should skip pure iWARP or RoCE devices, but not device which has at least one InfiniBand port. This patch also reverts commits d04d466 and e9ffc0b. d04d466 ("ibacm: Fix a memory leak in an acm_open_dev() error path") e9ffc0b ("ibacm: only open InfiniBand port") Fixes: e9ffc0b ("ibacm: only open InfiniBand port") Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit b2b870c ] In some devices, the page_size_cap is equal to PAGE_SIZE. Fixes: 9d66a1a ("pyverbs: Add support for extended query_device") Signed-off-by: Kamal Heib <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 256f045 ] The multi-function devices have same system image GUID while presented with multiple IB devices. This leads to the failures for NAME_GUID option to rename second device. Instead of system image GUID use node GUID as a unique identifier. Cc: <[email protected]> Fixes: 6b4099d ("kernel-boot: Perform device rename to make stable names") Signed-off-by: Leon Romanovsky <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 323709b ] In traffic() helper method, do post_recv after a recv WQE was consumed by the hardware. Fixes: 6fb2b9b ('tests: Add traffic helper methods') Signed-off-by: Noa Osherovich <[email protected]> Reviewed-by: Edward Srouji <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit d8a052e ] qp->sq.lock is currently unlocked twice when the ring state is invalid. Fix this by removing the goto, and also set bad_wr correctly. Fixes: 36756a6 ("libpvrdma: Add queue pair functions") Cc: [email protected] Reviewed-by: Vishnu Dasa <[email protected]> Reviewed-by: Adit Ranadive <[email protected]> Reviewed-by: Rajesh Jalisatgi <[email protected]> Signed-off-by: Bryan Tan <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 366374e ] The state of ibvqp will be updated after calling modify_qp(), because the assignment of ibvqp->state is out of hns driver, it can't be guaranteed that the configuration of ibvqp->state will take effect immediately. This may cause hns_roce_u_v2_modify_qp() in post_send/post_recv process is skipped. Therefore, update value of ibvqp->state in modify_qp(). Fixes: f1a80cc ("libhns: Bugfix for flush cqe in case multi-process") Signed-off-by: Yangyang Li <[email protected]> Signed-off-by: Lijun Ou <[email protected]> Signed-off-by: Weihang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 9bb0fcf ] The user context alloc request structure introduced a new field to indicate doorbell recovery is supported. This field was added so that additional features could be added in the future by setting a capability flag. However, the field wasn't zeroed, and was initialized using "|=" instead of "=" leading to garbage in the other bits. For forward compatability, we need to make sure all other bits are zero. Cc: [email protected] # v27 v28 Fixes: d9b2ba4 ("libqedr: Add support for Doorbell Overflow Recovery") Signed-off-by: Ariel Elior <[email protected]> Signed-off-by: Michal Kalderon <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 7c3d2fa ] PyverbsRDMAErrno takes only a message as argument, and it automatically raises PyverbsRDMAError with an updated message that includes the errno. Fixes: 7572aa1 ("pyverbs: Introducing pyverbs and its context class") Signed-off-by: Edward Srouji <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit e3077b7 ] l[port_state] in the port_state_to_str function returns an str, this commit removes .name from l[port_state].name. Fixes: a316351 ("pyverbs: Changes to print-related functions") Signed-off-by: Amit Matityahu <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 9768181 ] CHECK_AND_SET_VAL((p->qos_class << 4), 16, -1, qos_class, PR, QOS_CLASS); This Macro is expanded as following lines: if ((int16_t) (p->qos_class << 4) != (int16_t) -1) { qos_class = htobe16((uint16_t) (p->qos_class << 4)); comp_mask |= IB_PR_COMPMASK_QOS_CLASS; }; Error: CONSTANT_EXPRESSION_RESULT (CWE-569): rdma-core-28.0/infiniband-diags/saquery.c:1062: result_independent_of_operands: "(int16_t)(p->qos_class << 4) != -1 /* (int16_t)-1 */" is always true regardless of the values of its operands. This occurs as the logical operand of "if". |# 1060| CHECK_AND_SET_VAL(p->pkey, 16, 0, pr.pkey, PR, PKEY); |# 1061| CHECK_AND_SET_VAL(p->sl, 16, -1, pr.qos_class_sl, PR, SL); |# 1062|-> CHECK_AND_SET_VAL((p->qos_class << 4), 16, -1, qos_class, PR, QOS_CLASS); |# 1063| pr.qos_class_sl = (pr.qos_class_sl & htobe16(IB_PATH_REC_SL_MASK)) | |# 1064| qos_class; Fixes: 0005d2a ("saquery.c: Fix changing qos_class flag value to network order twice") Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 45be555 ] The TX Low Latency Queue (LLQ) is stored on the device's BAR, hence mmio_flush_writes should be used instead of udma_to_device_barrier before ringing the doorbell. mmio_wc_spinlock is used when acquiring the lock in order to guarantee that prior mmio writes are ordered with regard to the WC memory writes inside the lock. However, mmio_wc_spinunlock isn't used as the doorbell write must be done while holding the spinlock. Fixes: f7f275f ("efa: Elastic Fabric Adapter (EFA) userspace RDMA provider") Fixes: f189a2a ("efa: Support send using extended QP API") Signed-off-by: Gal Pressman <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit f026ac0 ] Compare CA device names by using the maximum length between them in order to avoid matches of sub strings between CA device names that can match the wrong device. Example before the fix: Input CA name: abc_11 List CA names: abc_1, def_1 When running : ibstat abc_11, the output will be the details of abc_1 even though abc_11 does not exist. After the fix, the output of this case will be: 'abc_11' IB device can't be found Signed-off-by: Haim Boozaglo <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 482eb44 ] When a firmware reset occurred because of some errors, modify qp to error in post_send() will be failed. The bad_wr will be returned to the user, then the user won't wait for the corresponding wc. But as sq db has been updated, the driver will return a software wc to the user, which may cause unexpected behaviours. Because the result of modifying qp is meaningless in this scenario, codes to handle the return value are removed. Signed-off-by: Yangyang Li <[email protected]> Signed-off-by: Jiaran Zhang <[email protected]> Signed-off-by: Weihang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 85e410d ] Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 2b25ddd ] Error: UNUSED_VALUE (CWE-563): rdma-core-28.0/librdmacm/examples/rcopy.c:265: value_overwrite: Overwriting previous write to "ret" with value "0". rdma-core-28.0/librdmacm/examples/rcopy.c:262: assigned_value: Assigning value from "*__errno_location()" to "ret" here, but that stored value is overwritten before it can be used. |# 260| if (fd < 0) { |# 261| printf("unable to open destination file\n"); |# 262|-> ret = errno; |# 263| } |# 264| Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 4a0245e ] Error: CPPCHECK_WARNING: rdma-core-28.0/infiniband-diags/ibportstate.c:245: error[shiftTooManyBitsSigned]: Shifting signed 32-bit value by 31 bits is undefined behaviour |# 243| mod = portnum; |# 244| if (espeed_cap) |# 245|-> mod |= 1<<31; |# 246| if (!smp_set_via(data, dest, IB_ATTR_PORT_INFO, mod, 0, srcport)) |# 247| IBEXIT("smp set portinfo failed"); Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 00dba86 ] Error: CHECKED_RETURN (CWE-252): rdma-core-28.0/srp_daemon/srp_daemon.c:2164: check_return: Calling "umad_init" without checking return value (as is done elsewhere 8 out of 10 times). rdma-core-28.0/ibacm/src/acm.c:3331: example_checked: Example 1: "umad_init()" has its value checked in "umad_init() != 0". rdma-core-28.0/infiniband-diags/ibstat.c:305: example_checked: Example 2: "umad_init()" has its value checked in "umad_init() < 0". rdma-core-28.0/infiniband-diags/saquery.c:1855: example_checked: Example 3: "umad_init()" has its value checked in "umad_init()". rdma-core-28.0/infiniband-diags/smpdump.c:217: example_checked: Example 4: "umad_init()" has its value checked in "umad_init() < 0". rdma-core-28.0/libibmad/rpc.c:385: example_checked: Example 5: "umad_init()" has its value checked in "umad_init() < 0". |# 2162| } |# 2163| |# 2164|-> umad_init(); |# 2165| res = alloc_res(); |# 2166| if (!res) { Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 5f2099e ] Dump only supported port extended counters instead of dump all extedned port counters that some of them may be not supported and dumped with zero values. When IsExtendedWidthSupported detected, the following fields will be dumped: PortUnicastXmitPkts PortUnicastRcvPkts PortMulticastXmitPkts PortMulticastRcvPkts When IsAdditionalPortCountersExtendedSupported detected, the following fields will be dumped: CounterSelect2 SymbolErrorCounter LinkErrorRecoveryCounter LinkDownedCounter PortRcvErrors PortRcvRemotePhysicalErrors PortRcvSwitchRelayErrors PortXmitDiscards PortXmitConstraintErrors PortRcvConstraintErrors LocalLinkIntegrityErrors ExcessiveBufferOverrunErrors VL15Dropped PortXmitWait QP1Dropped Signed-off-by: Haim Boozaglo <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 5c80749 ] Error: CPPCHECK_WARNING (CWE-120): [#def33] rdma-core-28.0/infiniband-diags/ibtracert.c:935: error[invalidScanfFormatWidth]: Width 20 given in format string (no. 1) is larger than destination buffer 'srcbuf[20]', use %19s to prevent overflowing it. |# 933| continue; /* ignore comment lines */ |# 934| |# 935|-> if (sscanf(portsbuf, "%20s %20s", srcbuf, dstbuf) != 2) |# 936| IBEXIT("ports-file, %s, at line %i contains bad data", |# 937| ports_file, line_count); Error: CPPCHECK_WARNING (CWE-120): [#def34] rdma-core-28.0/infiniband-diags/ibtracert.c:935: error[invalidScanfFormatWidth]: Width 20 given in format string (no. 2) is larger than destination buffer 'dstbuf[20]', use %19s to prevent overflowing it. |# 933| continue; /* ignore comment lines */ |# 934| |# 935|-> if (sscanf(portsbuf, "%20s %20s", srcbuf, dstbuf) != 2) |# 936| IBEXIT("ports-file, %s, at line %i contains bad data", |# 937| ports_file, line_count); Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit d034e23 ] Error: RESOURCE_LEAK (CWE-772): [#def38] rdma-core-28.0/infiniband-diags/ibdiag_common.c:167: alloc_fn: Storage is returned from allocation function "strdup". rdma-core-28.0/infiniband-diags/ibdiag_common.c:167: var_assign: Assigning: "ibd_nd_format" = storage returned from "strdup(val_str)". rdma-core-28.0/infiniband-diags/ibdiag_common.c:167: overwrite_var: Overwriting "ibd_nd_format" in "ibd_nd_format = strdup(val_str)" leaks the storage that "ibd_nd_format" points to. |# 165| } else if (strncmp(name, "nd_format", |# 166| strlen("nd_format")) == 0) { |# 167|-> ibd_nd_format = strdup(val_str); |# 168| } |# 169| } Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 5072f22 ] Error: RESOURCE_LEAK (CWE-772): [#def40] rdma-core-28.0/infiniband-diags/ibroute.c:176: alloc_fn: Storage is returned from allocation function "remap_node_name". rdma-core-28.0/infiniband-diags/ibroute.c:176: var_assign: Assigning: "mapnd" = storage returned from "remap_node_name(node_name_map, nodeguid, nd)". rdma-core-28.0/infiniband-diags/ibroute.c:178: noescape: Resource "mapnd" is not freed or pointed-to in "printf". rdma-core-28.0/infiniband-diags/ibroute.c:220: leaked_storage: Variable "mapnd" going out of scope leaks the storage it points to. |# 218| "; MAD status 0x%x AM 0x%x\n", |# 219| status, mod); |# 220|-> return NULL; |# 221| } |# 222| } Error: RESOURCE_LEAK (CWE-772): [#def41] rdma-core-28.0/infiniband-diags/ibroute.c:348: alloc_fn: Storage is returned from allocation function "remap_node_name". rdma-core-28.0/infiniband-diags/ibroute.c:348: var_assign: Assigning: "mapnd" = storage returned from "remap_node_name(node_name_map, nodeguid, nd)". rdma-core-28.0/infiniband-diags/ibroute.c:350: noescape: Resource "mapnd" is not freed or pointed-to in "printf". rdma-core-28.0/infiniband-diags/ibroute.c:368: leaked_storage: Variable "mapnd" going out of scope leaks the storage it points to. |# 366| "; MAD status 0x%x AM 0x%x\n", |# 367| status, block); |# 368|-> return NULL; |# 369| } |# 370| i = block * IB_SMP_DATA_SIZE; Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit bb64369 ] Error: RESOURCE_LEAK (CWE-772): [#def48] rdma-core-28.0/infiniband-diags/mcm_rereg_test.c:322: alloc_fn: Storage is returned from allocation function "calloc". rdma-core-28.0/infiniband-diags/mcm_rereg_test.c:322: var_assign: Assigning: "list" = storage returned from "calloc(50UL, 32UL)". rdma-core-28.0/infiniband-diags/mcm_rereg_test.c:332: leaked_storage: Variable "list" going out of scope leaks the storage it points to. |# 330| if (!f) { |# 331| err("cannot open %s: %s\n", guid_file, strerror(errno)); |# 332|-> return -1; |# 333| } |# 334| Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit c2ae19d ] Error: DC.STREAM_BUFFER (CWE-120): [#def50] rdma-core-28.0/kernel-boot/rdma_rename.c:171: dont_call: "fscanf" assumes an arbitrarily long string, so callers must use correct precision specifiers or never use "fscanf". rdma-core-28.0/kernel-boot/rdma_rename.c:171: remediation: Use correct precision specifiers or implement your own parsing. |# 169| } |# 170| |# 171|-> ret = fscanf(fp, "%s", address); |# 172| fclose(fp); |# 173| Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 8f3ce6f ] Error: USE_AFTER_FREE (CWE-416): [#def59] rdma-core-28.0/providers/qedr/qelr_verbs.c:194: freed_arg: "free" frees "vmr". rdma-core-28.0/providers/qedr/qelr_verbs.c:196: pass_freed_arg: Passing freed pointer "vmr" as an argument to "fprintf". |# 194| free(vmr); |# 195| |# 196|-> DP_VERBOSE(cxt->dbg_fp, QELR_MSG_MR, |# 197| "MR DERegister %p completed successfully\n", vmr); |# 198| return 0; Error: USE_AFTER_FREE (CWE-416): [#def60] rdma-core-28.0/providers/qedr/qelr_verbs.c:324: freed_arg: "free" frees "cq". rdma-core-28.0/providers/qedr/qelr_verbs.c:326: pass_freed_arg: Passing freed pointer "cq" as an argument to "fprintf". |# 324| free(cq); |# 325| |# 326|-> DP_VERBOSE(cxt->dbg_fp, QELR_MSG_CQ, |# 327| "destroy cq: successfully destroyed %p\n", cq); |# 328| Error: USE_AFTER_FREE (CWE-416): [#def61] rdma-core-28.0/providers/qedr/qelr_verbs.c:1059: freed_arg: "free" frees "qp". rdma-core-28.0/providers/qedr/qelr_verbs.c:1061: pass_freed_arg: Passing freed pointer "qp" as an argument to "fprintf". |# 1059| free(qp); |# 1060| |# 1061|-> DP_VERBOSE(cxt->dbg_fp, QELR_MSG_QP, |# 1062| "destroy cq: successfully destroyed %p\n", qp); |# 1063| Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit a35be3d ] Adapt the send info write length to the dynamic number of actions. Before changing the actions number to be dynamic the length matched to the chunk capacity. Post moving to support the dynamic mode the real length should be used to prevent copying from smaller data buffer a larger chunk capacity which might lead to a segmentation fault. Fixes: 9d18743 ("mlx5: Extended modify header to support more than 8 modify actions") Signed-off-by: Hamdan Igbaria <[email protected]> Signed-off-by: Alex Vesker <[email protected]> Signed-off-by: Yishai Hadas <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 671b3e1 ] Error: RESOURCE_LEAK (CWE-772): rdma-core-29.0/libibverbs/examples/xsrq_pingpong.c:581: open_fn: Returning handle opened by "socket". [Note: The source code implementation of the function has been overridden by a builtin model.] rdma-core-29.0/libibverbs/examples/xsrq_pingpong.c:581: var_assign: Assigning: "sockfd" = handle returned from "socket(t->ai_family, t->ai_socktype, t->ai_protocol)". rdma-core-29.0/libibverbs/examples/xsrq_pingpong.c:583: noescape: Resource "sockfd" is not freed or pointed-to in "connect". [Note: The source code implementation of the function has been overridden by a builtin model.] rdma-core-29.0/libibverbs/examples/xsrq_pingpong.c:598: noescape: Resource "sockfd" is not freed or pointed-to in "send_local_dest". rdma-core-29.0/libibverbs/examples/xsrq_pingpong.c:599: leaked_handle: Handle variable "sockfd" going out of scope leaks the handle. |# 597| |# 598| if (send_local_dest(sockfd, 0)) |# 599|-> return 1; |# 600| |# 601| if (recv_remote_dest(sockfd, 0)) Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 5c9647a ] Error: RESOURCE_LEAK (CWE-772): rdma-core-29.0/librdmacm/examples/rstream.c:410: alloc_arg: "getaddrinfo" allocates memory that is stored into "ai". rdma-core-29.0/librdmacm/examples/rstream.c:428: leaked_storage: Variable "ai" going out of scope leaks the storage it points to. |# 426| if (ret) { |# 427| printf("getaddrinfo src_addr: %s\n", gai_strerror(ret)); |# 428|-> return ret; |# 429| } |# 430| } Error: RESOURCE_LEAK (CWE-772): rdma-core-29.0/librdmacm/examples/rstream.c:424: alloc_arg: "getaddrinfo" allocates memory that is stored into "ai_src". rdma-core-29.0/librdmacm/examples/rstream.c:494: leaked_storage: Variable "ai_src" going out of scope leaks the storage it points to. |# 492| else |# 493| freeaddrinfo(ai); |# 494|-> return ret; |# 495| } |# 496| Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 271b019 ] Error: RESOURCE_LEAK (CWE-772): rdma-core-29.0/iwpmd/iwarp_pm_helper.c:407: open_arg: "get_iwpm_tcp_port" opens handle stored into "new_sd". rdma-core-29.0/iwpmd/iwarp_pm_helper.c:416: leaked_handle: Handle variable "new_sd" going out of scope leaks the handle. |# 414| iwpm_debug(IWARP_PM_ALL_DBG, "reopen_iwpm_mapped_port: Could not make port mapping (%s).\n", |# 415| str_err); |# 416|-> return NULL; |# 417| } |# 418| Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
[ Upstream commit 271b019 ] Error: RESOURCE_LEAK (CWE-772): rdma-core-29.0/iwpmd/iwarp_pm_helper.c:407: open_arg: "get_iwpm_tcp_port" opens handle stored into "new_sd". rdma-core-29.0/iwpmd/iwarp_pm_helper.c:416: leaked_handle: Handle variable "new_sd" going out of scope leaks the handle. |# 414| iwpm_debug(IWARP_PM_ALL_DBG, "reopen_iwpm_mapped_port: Could not make port mapping (%s).\n", |# 415| str_err); |# 416|-> return NULL; |# 417| } |# 418| Signed-off-by: Honggang Li <[email protected]> Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
Signed-off-by: Nicolas Morey-Chaisemartin <[email protected]>
This change addresses several compiler warnings identified during a code review:
1. **Uninitialized Variables:**
* In `providers/cxgb4/cq.c` (c4iw_poll_cq_one): Initialized `union t4_cqe cqe` to {0}. This resolves a warning about potential use of an uninitialized variable.
* In `providers/mthca/qp.c` (mthca_tavor_post_send & mthca_arbel_post_send): Initialized `uint32_t f0` and `uint32_t op0` to 0. These variables are used to store properties of the first work request in a batch for doorbell ringing; while logic generally ensures they are set, explicit initialization prevents warnings and potential issues with obscure code paths.
2. **Missing Function Prototype:**
* In `librdmacm/preload.c`: Added a forward declaration for `int __fxstat(int ver, int fd, struct stat *buf);`. This resolves a `-Wmissing-prototypes` warning and ensures the compiler is aware of the function's signature.
**Other Review Findings (Not Addressed in this Commit):**
* **Array Subscript Out of Bounds (False Positives):** Warnings in `ibacm/src/acm.c` and `ibacm/prov/acmp/src/acmp.c` related to zero-length arrays were assessed as likely false positives, as the code uses the flexible array member idiom correctly.
* **Deprecated `pthread_yield`:** Usage in `librdmacm/rsocket.c` was noted.
* **Mismatched Array Parameter Bounds:** Warnings in `libibumad/umad.c` and `libibmad/register.c` were assessed as likely benign due to API design.
These changes enhance code safety and silence distracting compiler warnings.
|
What is this? It is badly formed |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.