Bug 2130191

Summary: ipmitool crashes if a sdr reports a percentage not in table range
Product: Red Hat Enterprise Linux 8 Reporter: Renaud Métrich <rmetrich>
Component: ipmitoolAssignee: Pavel Cahyna <pcahyna>
Status: CLOSED ERRATA QA Contact: Laura Trivelloni <ltrivell>
Severity: low Docs Contact:
Priority: low    
Version: 8.6Keywords: Triaged
Target Milestone: rcFlags: pm-rhel: mirror+
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: ipmitool-1.8.18-19.el8 Doc Type: No Doc Update
Doc Text:
Story Points: ---
Clone Of:
: 2224578 (view as bug list) Environment:
Last Closed: 2023-11-14 15:37:21 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:
Bug Depends On: 2224578    
Bug Blocks:    

Description Renaud Métrich 2022-09-27 13:13:52 UTC
Description of problem:

We have a customer reporting a crash when issuing "ipmitool sdr list", on RHEL7, but the code is expected to die similarly on RHEL8 since code is similar.
Digging into this, I could find out the following sdr percentage being returned:
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------
(gdb) p sr.full->cmn.unit
$6 = {pct = 0 '\000', modifier = 0 '\000', rate = 0 '\000', analog = 0 '\000', type = {base = 112 'p', modifier = 0 '\000'}}
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------

Here above we have "type.base" == 112.

The code failing is:
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------
1397 struct sensor_reading *
1398 ipmi_sdr_read_sensor_value(struct ipmi_intf *intf,
1399                  struct sdr_record_common_sensor *sensor,
1400                  uint8_t sdr_record_type, int precision)
1401 {
 :
1496                 /* determine units string with possible modifiers */
1497                 sr.s_a_units = ipmi_sdr_get_unit_string(sr.full->cmn.unit.pct,
1498                                            sr.full->cmn.unit.modifier,
1499 >>> HERE                                   sr.full->cmn.unit.type.base,
1500                                            sr.full->cmn.unit.type.modifier);
 :

  77 /* ipmi_sdr_get_unit_string  -  return units for base/modifier
  78  *
  79  * @pct:        units are a percentage
  80  * @type:       unit type
  81  * @base:       base
  82  * @modifier:   modifier
  83  *
  84  * returns pointer to static string
  85  */
  86 const char *
  87 ipmi_sdr_get_unit_string(uint8_t pct, uint8_t type, uint8_t base, uint8_t modifier)
  88 {
  :
  94         char *pctstr = pct ? "% " : "";
  :
  96         switch (type) {
  :
 105         case 0:
 106         default:
  :
 114                         snprintf(unitstr, sizeof (unitstr), "%s%s",
 115 >>> HERE                         pctstr, unit_desc[base]);
 :
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------

Here above we see a "unit_desc[base]" is performed, hence here "unit_desc[112]" which is out of range:

-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------
777 /* unit description codes (IPMI v1.5 section 37.16) */
778 #define UNIT_MAX        0x90
779 static const char *unit_desc[] __attribute__ ((unused)) = {
780 "unspecified",
781             "degrees C", "degrees F", "degrees K",
782             "Volts", "Amps", "Watts", "Joules",
783             "Coulombs", "VA", "Nits",
784             "lumen", "lux", "Candela",
785             "kPa", "PSI", "Newton",
786             "CFM", "RPM", "Hz",
787             "microsecond", "millisecond", "second", "minute", "hour",
788             "day", "week", "mil", "inches", "feet", "cu in", "cu feet",
789             "mm", "cm", "m", "cu cm", "cu m", "liters", "fluid ounce",
790             "radians", "steradians", "revolutions", "cycles",
791             "gravities", "ounce", "pound", "ft-lb", "oz-in", "gauss",
792             "gilberts", "henry", "millihenry", "farad", "microfarad",
793             "ohms", "siemens", "mole", "becquerel", "PPM", "reserved",
794             "Decibels", "DbA", "DbC", "gray", "sievert",
795             "color temp deg K", "bit", "kilobit", "megabit", "gigabit",
796             "byte", "kilobyte", "megabyte", "gigabyte", "word", "dword",
797             "qword", "line", "hit", "miss", "retry", "reset",
798             "overflow", "underrun", "collision", "packets", "messages",
799             "characters", "error", "correctable error", "uncorrectable error",};
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------

See https://www.intel.com/content/dam/www/public/us/en/documents/product-briefs/second-gen-interface-spec-v1.5-rev1.1-2.pdf page 416 for the table in the specification.

There are 2 issues here:
1. "UNIT_MAX" is 90 in decimal, not 0x90 as in the code
2. There is no check made to see if "type.base" is greater or equal to "UNIT_MAX" ("UNIT_MAX" is just not used anywhere in the code)

Version-Release number of selected component (if applicable):

ipmitool RHEL7+

How reproducible:

Don't know, I don't have the hardware.

Comment 1 Pavel Cahyna 2023-07-21 14:04:25 UTC
Thank you for the report and analysis. The bug seems to occur on some atypical hardware, but I have found a way to reproduce it using a simulator.

Comment 2 Pavel Cahyna 2023-07-26 17:38:04 UTC
Fixed by upstream PR 120 (commit 12e2f5da63fcfdb544f87ec492e5b1bc4f89868c)

Comment 9 errata-xmlrpc 2023-11-14 15:37:21 UTC
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 (ipmitool 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-2023:7064