Commit Graph

61 Commits

Author SHA1 Message Date
07968377b3 ecos: fix golden-run runtime conversion
BochsController::getTimerTicksPerSecond() only works reliably when the
simulation is already running (e.g., after a restore()).  This broke
timeout conditions for the very first experiment in a FailBochs instance.

Change-Id: Ice5f0aa0c6759f2d9341ad4f21d5c346307b4c12
2014-04-27 19:04:05 +02:00
940fa6035d ecos: record simulated runtime
As the eCos kernel tests do not sufficiently check for failures, a
(compared to the golden run) measurably *reduced* simulation time can be
used as an indication for SDC.

Change-Id: I303f6c3f7ef56cb691c1f226eff74d113dd16629
2014-04-03 14:45:46 +02:00
6e03753b6e ecos: modifications for mibench benchmarks
Change-Id: Ifecbf24912dbffa814b189aed9336a5420ec6392
2014-02-25 13:32:56 +01:00
cbf9daea97 ecos: rewrite for DatabaseCampaign
+ modified resulttypes

Change-Id: I463759e66e7497c80eeee9a065fc95e058ec3dc1
2014-02-25 13:32:56 +01:00
4cb97a7fa5 formatting, typos, comments, details
Change-Id: Iae5f1acb653a694622e9ac2bad93efcfca588f3a
2014-01-22 13:08:13 +01:00
0a5e54e9aa ecos_kernel_test experiment bugix: don't resume if 'experiment reached finish() before FI'
Change-Id: Id0bb9400b8aa28307ed385a8c32b91b17254ba1c
2014-01-15 12:52:44 +01:00
eccbc61b1d ecos kernel test: memory access listeners' ranges (high/low) may be equal
Change-Id: I02f53d9d698a56c606ef354a37d7a3c467ec8127
2013-07-22 14:03:20 +02:00
955f89b3eb ecos: make LOCAL builds compile again
Change-Id: Icd992aa20443426bbcaa507c39453d6ecb9174c0
2013-07-03 13:46:55 +02:00
01e7f8c8a1 ecos: bugfix: cyg_test_output may be called before FI
This is a pretty old bug that unfortunately affects both DSN 2013 and
SOBRES 2013 results.

Change-Id: I64a2790a4d55515a23a34d108be99646d5dd345d
2013-07-03 13:46:55 +02:00
66ecedd864 ecos: removed configuration checks
Change-Id: I5df0b5f3435f3f695ba1ab9ca624dfa238509e46
2013-07-03 13:46:55 +02:00
f3c36e70ef ecos: stack-protection eval
Change-Id: I576c2ef3834f61bb9017af37541afc7639672782
2013-05-29 16:43:46 +02:00
091e8dcae0 ecos: baseline assessment integrated into main experiment
Change-Id: Iaf2a31c917b6ddd50568e5fb784ab8457193ee7d
2013-04-29 14:15:53 +02:00
403886e541 ecos: minor changes, cleanup
- Count experiments, not jobs
 - Debug output

Change-Id: Ide5e1219cdcc8112d1a0d4e7367beca2dd5821ef
2013-04-29 14:15:51 +02:00
c0b36f6236 ecos: use new timer iface to record benchmark runtime
... instead of the previous TimerListener workaround.

Change-Id: I3e712540e93b668301f50ecf4f5a5760e0a8fdb3
2013-04-29 14:15:49 +02:00
0f16f18d75 cosmetics
Change-Id: Ifae805ae1e2dac95324e054af09a7b70f5d5b60c
2013-04-22 14:24:02 +02:00
2d45a2c52c ecos: use MemoryMap materialization instead of own code
Change-Id: I8615a066c53e1d6a02c78bce3199fa1f73edfda9
2013-04-10 13:01:04 +02:00
d7a3a28431 ecos: split valid mem access range
When eCos is built as a multiboot binary, some of its data structures
are still at very low (<<1M) addresses, but the rest moves to
addresses >1M.  This change makes sure our invalid mem access
detection is not overly generous.

Change-Id: If8265a407b3706a4ff71562b316e05aa22255f62
2013-04-09 13:16:13 +02:00
001d036613 ecos: ignore VGA mem accesses
Ignore VGA mem accesses for valid address range detection.

Change-Id: I4a85b6d4a2de52ecbd977d2dba474df818710600
2013-04-06 18:59:12 +02:00
a0293b9d18 ecos: use commandline parameters for local test runs, too
Change-Id: Iac43732408420a9e3687d5756c942c2f2256ad86
2013-04-03 15:18:08 +02:00
373c19dd68 ecos: store traceinfo.txt in correct location
This commit completes the change introduced in commit
59e5fd3169.

Change-Id: I0a6ab7b35fbb69cbb8ef91e187b0d0bc32d01071
2013-04-03 15:18:08 +02:00
59e5fd3169 ecos: fail-client accepts commandline parameters for prerequisites phase
This allows us to generate prerequisites (traces, state snapshots, memory
maps) in parallel, and without the previous shell script hacks.

Change-Id: I05a0321a794b4033d05eed20f5bffbd1e910cf1b
2013-03-28 18:09:39 +01:00
4686c27d3d ElfReader: Support for Section and Symbol size.
- getSection/getSymbol now returns an ElfSymbol reference.

Searching by address now searches if address is within
symbol address and symbol address + size.
So we can test, if we are *within* a function, object or
section and not only at the start address.
2013-03-04 15:18:52 +01:00
b996617a97 ecos_kernel_test: improved dependency check
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@2074 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2013-02-07 10:43:03 +00:00
5f2364e1a2 ecos_kernel_test: updates due to architecture changes (ElfReader still NOT working properly)
At the moment, the experiment only works with a hard-coded address for the _stext symbol because the ElfReader cannot extract the symbol's address from the elf-binary. The experiment has been tested locally with PREREQUISITES = 1 and PREREQUISITES = 0. In the latter case, the test only considered the client/server communication, i.e., no FI has been performed at all (NOT tested yet).

git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@2066 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2013-02-06 15:39:57 +00:00
b8df5d6a11 ecos_kernel_test experiment updated (NOT TESTED)
The loop is running now until undone already fetched jobs are available.

git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@2044 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2013-01-31 16:33:50 +00:00
125914a305 BochsRegister.hpp and BochsRegisterIDs.hpp not needed anymore
The includes of these headers have already been removed from the experiments. In the current code, the content of the header BochsRegister.hpp is rather simply copied to x86/Architecture.hpp. It is therefore necessary to revisit the code soon (especially the FIXME related to register IDs).

Another problem is that there is no generalization of register IDs. Thus, all experiments are currently specific to a concrete architecture (which is not desired).

git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@2010 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2013-01-17 13:41:23 +00:00
hsc
87ee9df37b ecos: additional burst fault model
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1961 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-27 17:06:32 +00:00
hsc
5fe61e0f3f ecos: cosmetics
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1960 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-27 17:06:29 +00:00
hsc
5135c79c05 TimerListener: microsecond granularity (ms is too coarse)
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1952 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-23 15:35:08 +00:00
hsc
c65c4936ab ecos: compress traces
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1950 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-23 15:33:27 +00:00
hsc
35026de2d8 ecos: use multiple intermediate states to speed up experiments (disabled)
This modification creates and uses multiple intermediate snapshot states
(one every 1,000,000 instructions) to fast-forward to the FI site.
Unfortunately this doesn't work yet; the trace seems to change in many (not
all!) cases we do this.  One possible cause could be an incorrect
(off-by-one or alike?) restoration of the serial device timers, and
therefore an earlier/later transition to "output buffer empty", resulting
in eCos' serial putc function needing a different amount of polling loop
iterations.  Needs more investigation.

git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1940 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-20 15:01:43 +00:00
hsc
041746741d ecos: include variant and benchmark in job info
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1939 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-20 15:01:40 +00:00
hsc
4f48eb3232 ecos: send mandatory ecos_test_result when sanity check fails
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1929 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-13 16:26:58 +00:00
hsc
7b3e5986d1 ecos: save experiment runtime
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1927 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-13 00:17:31 +00:00
hsc
9decf95a09 ecos: specific traps are OK for two benchmarks
The "except1" and "clockcnv" benchmarks explicitly cause a trap and handle
it in their own trap handler.  This, of course, isn't a bad experiment
outcome.

git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1923 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-13 00:17:18 +00:00
hsc
464bc9390b ecos: campaign rewritten
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1920 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-12 10:59:35 +00:00
hsc
4957cfb71b ecos: remove hard-coded trace filename
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1913 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-10 16:13:59 +00:00
hsc
0ab4f12b7c ecos: no need for clearListeners anymore
Since r1700, listeners remove themselves when they leave their scope.
A restore() call also implicitly clears all listeners.

git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1912 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-09 15:37:33 +00:00
hsc
661563125e ecos: cosmetics / compiler warning
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1910 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-09 15:37:27 +00:00
hsc
c2242f2bb4 ecos: guest-image independent fail client
TODO: campaign needs to be updated, too

git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1908 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-09 15:37:21 +00:00
hsc
a26aeb52ee ecos: preparations for guest-image independent fail client
Additional change: prerequisite data files are not opened in append mode
anymore.  These files can be easily reproduced, and append mode does more harm
than good here.

git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1906 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-09 15:37:14 +00:00
chb
126fb1a1ff ecos_kernel_test: remove 2nd reboot (which was just before a restore)
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1889 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-07 16:01:35 +00:00
chb
b7bbcd389f ecos_kernel_test: record error detected/corrected
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1882 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-06 10:51:41 +00:00
hsc
9e5de7b14a ecos_kernel_test bugfix: timer granularity of 10s is way too coarse
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1877 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-11-02 16:59:07 +00:00
34d89000cf ecos_kernel_test: simplification (no need to call clearListeners() here).
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1843 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-10-25 15:52:55 +00:00
chb
8eb7bd129a ecos_kernel_test: take width of memory accesses into account for determining bounds
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1591 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-09-05 09:05:11 +00:00
chb
b33609ef1d ecos_kernel_test: removed crude workarounds for bugs that have been fixed lately
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1588 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-09-03 13:14:47 +00:00
chb
e6546c1365 limit number of sequential FAIL* experiments to 50, in order to prevent swapping
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1494 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-08-15 11:05:27 +00:00
chb
2b72988203 ecos_kernel_test: wait more than once for test_output breakpoint
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1493 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-08-14 12:47:41 +00:00
chb
b139c71053 ecos_kernel_test: experiment calculates timeout in STEP 2
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1492 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
2012-08-14 09:32:07 +00:00