Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
[StaticScan] Changes due to minor issues found by a static scan
Modules changes:
- sr_port/bm_getfree.c - Issues: free_bit (line 216), map_size (line 216), cs1 (line 223),
and depth (line 221) are all uninitialized. Additionally,
blkhist.cr and blkhist.cycle undefined at line 238.
- Loop changed so always do one iteration. Will do a new scan and check to see which of
these issues are fixed by the loop change and which remain.
- sr_port/dse_m_rest.c - Issue: The variable r_top was flagged as uninitialized in line 87.
- The variable r_top was re-initialized a few lines further down so line 87 was not
necessary and was removed.
- Some reformats due to non-standard formatting pre-existing.
- sr_port/dse_shift.c - Issue: lbp undefined in line 99
- Initialize 'lbp' to NULL earlier so no paths to it are left for it to be undefined.
- sr_port/gtmsource_ctl_init.c - Issue: tmp_ctl resource leak at line 182.
- Free 'tmp_ctl' when returning or erroring out.
- sr_port/is_canonic_name.c - Issue: Issue was noted via an issue in a different module.
In the parse_gv_name_and_subscripts() routine, near the top,
seq is set to *subscripts though is never referenced again.
- Remove variable 'seq' which is not used and may have issues dereferencing 'subscripts'
when not provided.
- sr_port/mu_int_blk.c - Issue: free_blk_base resource leak going out of scope line 323.
- Free malloc'd storage (in 'free_blk_base') before return.
- sr_port/mupip_set_journal.c - Issue: Resource leak rewriting gds_info, gds_info not
cleaned up.
- Remove unused gds_info allocation (not plugged into anything) and changed entire clause
to FILE_CNTL_INIT_IF_NULL() macro to do initialization correctly.
- sr_port/mur_output_show.c - Issue: first_time uninitialized in line 328.
- Initialize 'first_time' to TRUE so no path to it being uninitialized.
- sr_port/op_fnzsocket.c - Issue: tls_options_mask uninitialized at 598.
- Move initialization of 'tls_options_mask' up earlier so no paths to uninitialized use.
- sr_port/op_indincr.c - Issue: s is uninitialized at line 101
- Only do put_tref() if something to put. While this leaves v undefined if there was an
error, this does not matter because comp_fini() ignores v in that case.
- sr_port/stp_gcol_src.h - Issue: uninitialized value of cstr at line 1063.
- Make sure cstr is initialized whether expansion succeeds or fails by moving point
where cstr is set.
- sr_port/tcp_open.c - Issues: 1. resource leak - ai_ptr going out of scope at line 190.
2. remote_ai_head uninitialized at line 302.
3. calling close (gtm_close) without checking return value at
line 355.
4. resource leak - remote_ai_head out of scope at line 340.
- Remove some old Tru64 stuff.
- Remove unnecessary initialization of remote_ai_ptr.
- Remove some unused variables.
- Move hostname validation to client section and remove the conditional parsing of the
host (required for client) which solves YottaDB#2.
- Change close() to CLOSEFILE() macros which solves YottaDB#3.
- Make sure to release linked list at ai_ptr which solves YottaDB#1.
- Remove assert when erroring with timeout (now tested by online_bkup/online6). This was
required since the tests in online6 now timeout instead getting GETADDRINFO for invalid
passive hosts being specified.
- Make sure release linked list at remote_ai_head (which solves YottaDB#4).
- sr_port/ydberrors.msg
- Added TCPCONNTIMEOUT message as the other timeout messages didn't quite do what was
needed. There were a couple util_out_print() timeout message in tcp_open() but it
really needed to be a real error - especially since it was now being tested in a test
(online_bkup/online6).
- sr_unix/anticipatory_freeze.c - Issue: Resource leak - handle goes out of scope at line 358.
- Add an FCLOSE() and error check for same prior to module returns.
- sr_unix/bin_load.c - Issue: Several allocated blocks and buffers need cleanup when leave
- Implement mechanism to track all of the allocated buffers and free any that are
allocated if an error occurs or on normal routine return.
- sr_unix/cli_lex.c - Issue: retptr uninitialized at line 488 (return from module).
- Initialize return value to null so is initialized when nothing is read.
- sr_unix/gtm_getpwuid.c - Originally had changes for an issue but backed those out but left
these changes intact anyway.
- Removed an unneeded return value check from malloc (gtm_malloc does not return if
no storage).
- Reformatted an #ifdef block.
- sr_unix/gtmcrypt_entry.c - Issue: Resource leak - handle goes out of scope at line 135.
- release/close the handle before return
- sr_unix/gtmrecv_end.c - Issue: jnlpool_strm_seqno[idx] uninitialized at line 194.
- Only do the loop to dump the streams if the journal pool exists.
- sr_unix/gtmsecshr.c - Issue: Resource leak - procstrm goes out of scope at line 1026.
- Close 'procstrm' before return on error.
- sr_unix/gtmsource_shutdown.c - Issue: maxindex is uninitialized at line 317.
- Add '!auto_shutdown' to clause as 'maxindex' is only set when !auto_shutdown.
- sr_unix/iosocket_tls.c - Issue: 1. errlen2 is uninitialized at line 416.
2. errlen is unintialized at line 466.
- Errors aren't using the correct method of getting error out there. The 'errlen'
var is only seldomly set. Change to use LEN_AND_STR() macro to provide length.
- sr_unix/mu_all_version_standalone.c - Issue: save_errno is unintialized at line 172.
- Save 'errno' to 'save_errno' to initialize it for the error message.
- sr_unix/mucblkini.c - Issue: Vars bp1, bp2, and bmp were not released before errors.
- Free these vars appropriately before leaving routine.
- sr_unix/op_zlink.c - Issue: srcnamelen uninitialized line 227.
- Move initialization of 'srcstr' closer to where used and only for those options
where srcnamelen is set.
- sr_unix/relinkctl.c - Issue: shm_hdr uninitialized in line 1069
- Initialize 'shm_hdr' to prevent usage when is still uninitialized.
- Add error checks and messages for SHMDT invocations.
- sr_unix/ss_anal_shdw_file.c - Issue: bitmap_buffer and bp allocations not freed prior
to error returns (lines 118 and 123).
- Add frees for bitmap_buffer and bp before error returns.
- sr_unix/trigger_source_read_andor_verify.c - Issue: rttabent uninitialized on line 356.
- Initialize 'rttabent' to NULL
- If not initialized, set 'rttabent' to our rtn_names list to begin search.
- sr_unix/util_output.c - Issue: Origin of this change is lost. It was from an issue
raised for another module and digging down through the calls,
we ended up here where a var as uninitialized after this loop
because of the lack of a default clause such that 'length'
didn't get set IIRC (the chose switch value in the scan's
simulation did not match a case).
- Add default clause to switch with an assert in it. Also added code to keep static
scan from complaining about 'length'.
- sr_unix/wait_for_disk_space.c - Issue: freeze_cleared uninitialized at line 161
- Add a default for 'freeze_cleared' so is initialized.
- sr_unix_cm/gtcm_bgn_net.c - Issue: Resource leak - ai_ptr goes out of scope at 135, 148.
- Release 'ai_ptr' linked list before normal and returns.
- sr_unix_cm/omi_prc_conn.c - Issue: Neither agname or ag_pass allocated memory is
released before normal or error return.
- Define OMI_FREE() macro for better cleanup. Add to any place before error or normal
return.
- Initialize 'agname' and 'ag_pass' so know if they are allocated.
- Remove clauses that test memory allocation result. Note gtm_malloc() does not return
if allocation fails.
- sr_unix_cm/rc_srvc_xct.c - Issue: Resource leak - elst var goes out of scope at line 198.
- Added a macro to do the cleanups and added it before all return points.
- sr_unix_gnp/cmi_init.c - Issue: Resource leak - local_ai_ptr out of scope lines
67, 74, 85, 95,122.
- Some minor formatting changes for code standards.
- Make sure local_ai_ptr linked list is freed before leave routine.
- sr_unix_gnp/cmi_open.c - Issue: Resource leak - ai_head out of scope line 123.
- Some minor formatting changes for code standards.
- Make sure ai_head linked list is freed before leave routine.
- sr_unix_gnp/cmj_get_port.c - Issue: Resource leak - ai_ptr out of scope line 123.
- Removed - no users of this routine were found.
- sr_unix_gnp/cmj_getsockaddr.c - Issue: Resource leak - ai_ptr out of scope line 152.
- Make sure ai_ptr linked list is freed before leave routine with this error return.
- sr_unix_gnp/cmu_getclb.c - Issue: Resource leak - ai_ptr out of scope lines 52, 57.
- Minor changes for coding standards.
- Make sure ai_ptr linked list is freed before leave routine.
- sr_unix_gnp/gtcm_gnp_server_main.c - Issue: status uninitialized line 287.
- Change static routines to STATICFN{DCL,DEF} (coding standards)
- The 'status' variable was not set until after it was referenced. Removed the test
of the uninitialized status after the fetch.
- sr_unix_gnp/gtcm_open_cmerrlog.c - Issue: Potential to overrun end of lfn_path at
line 80.
- Change max usable length before truncation to make room for a null terminator.- Loading branch information
