File: CHANGES

package info (click to toggle)
mcelog 104-1
  • links: PTS, VCS
  • area: main
  • in suites: jessie, jessie-kfreebsd
  • size: 996 kB
  • ctags: 1,508
  • sloc: ansic: 7,739; sh: 481; makefile: 87
file content (106 lines) | stat: -rw-r--r-- 4,859 bytes parent folder | download | duplicates (2)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
<newer changes first>

Add Linux Kongress 2010 paper
Add Sandy Bridge Support
Write pid file by default in daemon mode
Reopen log files on SIGUSR1 in daemon mode
Default --daemon mode to logging to /var/log/mcelog
Add Core i3/5 and more Westmere support
Add Xeon75xx support
Add Intel architectural mca fallback for new family 6 Intel CPUs
Add --pidfile option to write daemon pid into file.
Enable DIMM, socket, cache tracking and per page tracking by default
Support changing mcelog daemon to other user
	Disabled by default because this breaks the advanced triggers
	which need root rights.
Allow to disable memory error logging
Add per socket memory accounting/trigger support
Add per page memory error accounting and predictive page offlining
	Support for accounting errors per memory page and offlining
	pages and executing shell script triggers when error thresholds
	are exceeded.
	Offlining requires a new kernel that supports page offlining.
	Works with specific Intel systems with integrated memory controller
Add cache error ``yellow bit trigger'' support
	An cache error threshold indication on Intel CPUs can
	execute a special trigger program to offline affected CPUs.
Add in memory dimm database in daemon mode
	Errors are accounted per DIMM and can execute shell scripts when
	error thresholds are exceeded.
	This can be all configured using the configuration file.
	Add mcelog --client to query a running mcelog daemon for errors
	Works only on specific Intel systems for now with integrated 
	memory controllers that report DIMM errors.
Keep stdout open in daemon mode when no syslog
Add --foreground option
Avoid %a scanf extension that broke old glibc version
Fixes to Tulsa and Dunngton decoding (Youkang Song)
Fixes to Nehalem decoding and (Ted Barragy, Hidetoshi Seto)
Fix parallel logging in daemon mode
Fixes to the manpage (Thomas Renninger)
More consistent number printing (Huang Ying)
mcelog builds now with more gcc warning options enabled
Improved man page; now actually describing what a machine check is
Fix and finally document TSC decoding (still doesn't work everywhere and is obsolete now)
Add support for --ascii --file option to specify input file
Add support for /etc/mcelog.conf config file. This can set all options on the command line
	and some more.
Add new --cpu=cputype option and deprecate old k8/core2/generic/intel-cpu options
Use official Nehalem names
Support re-parsing mcelog output with --ascii
Add Intel Tulsa (Xeon 71xx) decoding support
Add Dunnington (Xeon 7400) decoding support
Add Nehalem decoding support
Add --no-syslog option
Various bug fixes
Add --logfile option
Implement daemon mode
Various bug fixes to the decoder
Add --intel-family=... argument
Better decoding of core2 events and explicit decoding of pre Core P6 cores
Add decoding of broadcast timeouts
Add --version argument
Fix parsing of multiple fields on a line in --ascii 
Add --raw mode for easier machine decoding (Ying Huang)
Skip printk timestamp headers in --ascii (Ying Huang)
Support more than 255 CPUs (requires updated kernel)
Update for upcoming kernel interface "struct mce" extensions
Support AMD Fam10h/11h CPUs (Joachim Deguara)
Add switch to use LOG_ERR for syslog messages
Various misc. cleanups
mcelog now logs summaries of some serious events into syslog by default
Fix syslog multiline logging
Separate decoding output and error messages
Automatic dependency generation in Makefile
Many cleanups in DMI decoding
Add DIMM database and error triggers
Automatic sanity check for DMI information and enable by default
Add support for decoding Intel Core2 machine checks
Simple decoding of the TSC value into uptime for Intel CPUs
Add the MCE design paper.
Decode Intel thermal events properly
Add some "RAMs" to K8 ECC strings in futile hope that users will get 
	the hint
Allow modifier command line options after --ascii
Don't print decoded address twice for --ascii
Fix SMBIOS anchor scan to work on more machines and don't crash when 
	no anchor found.
Fix --ascii reparsing of mcelog output.
Add --filter and filter out known broken K8 GART errors
Add --ignorenodev argument and use in cron script (avoids cron errors
in Xen guest kernels) 
Add new --dmi argument to look up machine check addresses in SMBIOS
(warning unreliable due to wide spread bios bugs) 
Fix argument decoding (support --, allow arguments in any order)
Clarify --ascii in the manpage
Support for AMD K8 Revision F machine check DRAM error thresholding 
	from Jacob Shin <jacob.shin@amd.com>
Add P4 decoder contributed by "Guo, Racing" <racing.guo@intel.com> for
	Intel P4 and Xeon.
Add K8 decoder from 2.4 kernel code to decode Opteron/Athlon64 logs.
	(code mostly from Eric Morton and Andi Kleen)  
Add --ascii function to decode fatal kernel output.
Improve manpage
Fix 32bit bugs
Fix uninitialized variable in check_cpu
Minor cleanups