| United States-English |
|
|
|
![]() |
Jazz home > Software > Beta Test patches > C.65.00 > MPEMXH9Apatch details: MPEMXH9A |
|
Patch Description
Problems with spoolfiles created with posix apps (FSERR 105, SA514)
SR Number:
8606-230812 8606-310669 8606-310956 8606-318248
This patch is dependent upon: None
General Release Text
Known Conflicts
===============
Not applicable
Patch-to-Patch Dependencies
===========================
Not applicable
External Symptoms, Cause, Fix Description
=========================================
8606230812 - Spoolfiles can become corrupted using posix libraries
A mixture of "posix" and "non-posix" applications writing to the
same spoolfile can result in corruption in one of the spoolfile
blocks and results in FSERROR 105 (Bad variable block structure)
when the spoolfile is read. This has typically been seen in the
$STDLIST spoolfile for jinetd.
8606310669 - CCTL wrong on some spoolfile recs created by posix apps
If a posix application performs writes to a spoolfile which are not
terminated by \n (new line), then no new line will be generated, even
when a subsequent write does include a new line (\n) character.
For example, when writing to a spoolfile (opened with CCTL), the code:
printf("one ");
printf("two\n");
gets printed as if the \n was omitted from the second write.
8606310956 - SA514 when reading active spoofle of job running POSIX app
System Abort 514 can occur if a POSIX application running in a job
writes to its job $stdlist spoolfile at the same time that another
user is reading the same spoolfile (for example, via the CI PRINT
command). The POSIX application writes to the spoolfile as if it is
a byte stream file, using byte stream access. The spoolfile also
needs to be very large, 2 GB or larger, for the failure to occur.
8606318248 - Spoolfiles can become corrupted using posix libraries
BAD VARIABLE BLOCK STRUCTURE (FSERR 105) can also happen if a posix
application does a write to a spoolfile and size of that write is
greater than the block size of the spoolfile.
Customer Environment Changes
============================
Not applicable
Hardware Dependencies
=====================
Not applicable
Software Product/Application Dependencies
=========================================
Not applicable
Product Version Information
===========================
See filename information for specific version changes for each module.
Dump Analysis
=============
The stack trace of the system abort 514 looks like this. This is from
the process writing to the spoolfile:
$148 ($34e) nmdat > tr,d,i
PC=a.0017edec system_abort
NM* 0) SP=418607a0 RP=a.00a4518c disc_sm_start_read+$4b4
NM 1) SP=418607a0 RP=a.01073ac8 tm_msb_validate_meof_common+$b4
NM 2) SP=41860660 RP=a.010749c4 tm_msb_write_prime_common+$13c
NM 3) SP=418605e0 RP=a.01075690 tm_msb_write_common+$1ac
NM 4) SP=41860420 RP=a.00e9ac3c tm_msb_log_disc+$dc
NM 5) SP=41860320 RP=a.0143dcc4 FWRITE+$9a4
The footprint table shows a trap in protect_vp_range, which is called
by prefetch_:
173 ($34e) nmdat > event_footprint
Pin $34e ($c3c4db40, $b9357894 4) -- No entry in message catalog for
warning # 32765, subsys #109. (TRAP_HANDLER Escape)
ISM: $41861120 from NL.PUB.SYS/protect_vp_range+$110
OS Release VUF: C.65.00
Criticality:
(1) The system could experience a System failure without the patch.
Update Required: Yes
This page last updated on Mon, Mar 10, 2008 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||