Bug 2157658
| Summary: | Please backport upstream fix 1cc2c66d7d5234ed0aa70155a0755c2f59ca20f1 | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 8 | Reporter: | Renaud Métrich <rmetrich> |
| Component: | rsyslog | Assignee: | Attila Lakatos <alakatos> |
| Status: | CLOSED ERRATA | QA Contact: | Dalibor Pospíšil <dapospis> |
| Severity: | medium | Docs Contact: | Jan Fiala <jafiala> |
| Priority: | medium | ||
| Version: | 8.7 | CC: | alakatos, dapospis, peter.vreman, rsroka, tmeader |
| Target Milestone: | rc | Keywords: | Patch, Triaged |
| Target Release: | --- | Flags: | pm-rhel:
mirror+
|
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | rsyslog-8.2102.0-13.el8 | Doc Type: | Bug Fix |
| Doc Text: |
.Rsyslog no longer crashes due to a bug in `imklog`
Previously, Rsyslog could encounter a segmentation fault if the `imklog` module was enabled and a `free()` call using an invalid object was freed during use. With this update, the freed object is correctly deallocated at the correct place. As a result, the segmentation fault no longer occurs.
|
Story Points: | --- |
| Clone Of: | Environment: | ||
| Last Closed: | 2023-05-16 08:36:49 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
Attaching GitHub PR, as the one attached to the bug was just an initiative (a new one was created to solve the issue). Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (rsyslog bug fix and enhancement update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2023:2831 |
Description of problem: A customer sees regularly crashes of *rsyslogd* when the process exits. This happens because of a bug in *imklog* (which is disabled by default now, but not on customer's system): -------- 8< ---------------- 8< ---------------- 8< ---------------- 8< -------- (gdb) bt #0 __GI___libc_free (mem=0x203a226c6e5f7361) at malloc.c:3110 #1 0x00007f692f03633f in modExit () at imklog.c:478 #2 0x000055be3605ff97 in modPrepareUnload (pThis=0x55be3707b4f0) at modules.c:555 #3 modUnlinkAndDestroy (ppThis=ppThis@entry=0x7ffeec6cd1f0) at modules.c:1008 #4 0x000055be36060198 in modUnloadAndDestructAll (modLinkTypesToUnload=eMOD_LINK_ALL) at modules.c:1033 #5 0x000055be36023f94 in deinitAll () at rsyslogd.c:2074 #6 main (argc=<optimized out>, argv=0x7ffeec6cd418) at rsyslogd.c:2155 (gdb) f 3 #3 modUnlinkAndDestroy (ppThis=ppThis@entry=0x7ffeec6cd1f0) at modules.c:1008 1008 CHKiRet(modPrepareUnload(pThis)); (gdb) p *pThis $1 = {pPrev = 0x55be370611e0, pNext = 0x55be3707c7d0, iIFVers = 6, eType = eMOD_IN, eLinkType = eMOD_LINK_DYNAMIC_LOADED, eKeepType = eMOD_NOKEEP, pszName = 0x55be3707c020 "imklog", cnfName = 0x55be3707c000 "imklog", uRefCnt = 0, bSetModCnfCalled = 1 '\001', modInit = 0x0, modQueryEtryPt = 0x7f692f035bc0 <queryEtryPt>, isCompatibleWithFeature = 0x55be3605ee40 <dummyIsCompatibleWithFeature>, freeInstance = 0x0, dbgPrintInstInfo = 0x0, tryResume = 0x0, modExit = 0x7f692f036310 <modExit>, modGetID = 0x7f692f035b10 <modGetID>, doHUP = 0x0, doHUPWrkr = 0x0, beginCnfLoad = 0x7f692f036430 <beginCnfLoad>, setModCnf = 0x7f692f035f40 <setModCnf>, endCnfLoad = 0x7f692f036260 <endCnfLoad>, checkCnf = 0x7f692f035eb0 <checkCnf>, activateCnfPrePrivDrop = 0x7f692f035ea0 <activateCnfPrePrivDrop>, activateCnf = 0x7f692f0365a0 <activateCnf>, freeCnf = 0x7f692f036230 <freeCnf>, mod = {im = { willRun = 0x7f692f036530 <willRun>, runInput = 0x7f692f036560 <runInput>, afterRun = 0x7f692f036500 <afterRun>, newInpInst = 0x0, bCanRun = 0}, om = { beginTransaction = 0x7f692f036530 <willRun>, commitTransaction = 0x7f692f036560 <runInput>, doAction = 0x7f692f036500 <afterRun>, endTransaction = 0x0, parseSelectorAct = 0x0, newActInst = 0x0, SetShutdownImmdtPtr = 0x0, createWrkrInstance = 0x0, freeWrkrInstance = 0x0, supportsTX = 0 '\000'}, lm = { dummy = 48 '0'}, pm = {newParserInst = 0x7f692f036530 <willRun>, freeParserInst = 0x7f692f036560 <runInput>, parse2 = 0x7f692f036500 <afterRun>, parse = 0x0}, sm = {strgen = 0x7f692f036530 <willRun>}, fm = { getFunctArray = 0x7f692f036530 <willRun>}}, pModHdlr = 0x55be3707af30} -------- 8< ---------------- 8< ---------------- 8< ---------------- 8< -------- Through rebuilding with ASAN support, ASAN reported the issue: -------- 8< ---------------- 8< ---------------- 8< ---------------- 8< -------- ==53456==ERROR: AddressSanitizer: heap-use-after-free on address 0x606000032cf8 at pc 0x7f0e62c03939 bp 0x7fff175ccf10 sp 0x7fff175ccf00 READ of size 8 at 0x606000032cf8 thread T0 #0 0x7f0e62c03938 (/usr/lib64/rsyslog/imklog.so+0xc938) #1 0x55f8081aa510 in modPrepareUnload modules.c:555 #2 0x55f8081aa510 in modUnlinkAndDestroy modules.c:1008 #3 0x55f8081ab1a8 in modUnloadAndDestructAll modules.c:1033 #4 0x55f8080b0eca in deinitAll rsyslogd.c:2074 #5 0x7f0e68ca7d84 in __libc_start_main (/lib64/libc.so.6+0x3ad84) #6 0x55f8080a6dbd in _start (/usr/sbin/rsyslogd+0x2d0dbd) 0x606000032cf8 is located 56 bytes inside of 64-byte region [0x606000032cc0,0x606000032d00) freed by thread T0 here: #0 0x7f0e6b2037e0 in __interceptor_free (/lib64/libasan.so.5+0xef7e0) #1 0x7f0e62c0349c (/usr/lib64/rsyslog/imklog.so+0xc49c) previously allocated by thread T0 here: #0 0x7f0e6b203db0 in calloc (/lib64/libasan.so.5+0xefdb0) #1 0x7f0e62c03b7d (/usr/lib64/rsyslog/imklog.so+0xcb7d) SUMMARY: AddressSanitizer: heap-use-after-free (/usr/lib64/rsyslog/imklog.so+0xc938) -------- 8< ---------------- 8< ---------------- 8< ---------------- 8< -------- The root cause is reading deallocated memory on line 478: -------- 8< ---------------- 8< ---------------- 8< ---------------- 8< -------- 454 BEGINfreeCnf 455 CODESTARTfreeCnf 456 ENDfreeCnf : 473 BEGINmodExit 474 CODESTARTmodExit 475 if(pInputName != NULL) 476 prop.Destruct(&pInputName); 477 478 free(runModConf->pszBindRuleset); 479 /* release objects we used */ 480 objRelease(glbl, CORE_COMPONENT); 481 objRelease(net, CORE_COMPONENT); 482 objRelease(datetime, CORE_COMPONENT); 483 objRelease(ruleset, CORE_COMPONENT); 484 objRelease(prop, CORE_COMPONENT); 485 ENDmodExit -------- 8< ---------------- 8< ---------------- 8< ---------------- 8< -------- At the time of exit, the "runModConf" memory has already been freed. The memory has to be freed in freeCnf() instead (line 455). This was fixed by Upstream commit 1cc2c66d7d5234ed0aa70155a0755c2f59ca20f1. Version-Release number of selected component (if applicable): rsyslogd RHEL8 and RHEL9 How reproducible: Sometimes Steps to Reproduce: 1. Enable imklog 2. Restart rsyslog service repeatedly Actual results: Some crashes Expected results: No crash