It's OK if we cannot map every register LLVM knows to a Fail register
ID, but we need to explicitly skip these cases in the
RegisterImporter.
Change-Id: I2152f819fb94aa4de5720c5798b229b66988d382
This change became necessary as we observed weird fail-client SIGSEGV
crashes with both Bochs and Gem5 backends and different experiments.
Some Fail* components are instantiated statically: the
SimulatorController instance "simulator", containing the
ListenerManager and the CoroutineManager, and the active
ExperimentFlow subclass(es)
(experiments/instantiate-experiment*.ah.in). The experiment(s) is
registered as an active flow in the CoroutineManager at startup.
As plugins (which are ExperimentFlows themselves) are often created on
an experiment's stack, ExperimentFlows deregister themselves on
destruction (e.g., when leaving the plugin variable's scope). The
core problem is, that the creation and destruction order of statically
instantiated objects depends on the link order; if the experiment is
destroyed after the CoroutineManager, its automatic self-deregistering
feature talks to the smoking ruins of the latter.
This change removes all static instantiations of ExperimentFlow and
replaces them with constructions on the heap. Additionally it makes
sure that the CoroutineManager recognizes that a shutdown is in
progress, and refrains from touching potentially already destroyed
data structures when a (mistakenly globally instantiated)
ExperimentFlow deregisters in this case.
Change-Id: I8a7d42fb141222cd2cce6040ab1a01f9de61be24
Contemporary AspectC++ versions can deal with the LLVM headers very
well, and #ifdef __puma stuff in Fail* headers results in
unmaintainable #ifdef __puma blocks in other parts of Fail* (e.g., the
trace importer).
Make sure you're using a 64-bit ac++ when living in a 64-bit userland
(the 32-bit version doesn't know about __int128), and be aware that
AspectC++ r325 introduced a regression that has not been fixed yet.
Change-Id: I5bb759b08995a74b020d44a2b40e9d7a6e18111c
For def/use pruning, the linear address mapping of the x86
sub-registers (e.g., AX represents the lower 16 bits of EAX) must
overlap. If it doesn't, e.g., AX and EAX are considered separate
registers by def/use pruning, resulting in a failure to correlate an
EAX def with a subsequent AX use.
The only user of this mapping up to now, RegisterImporter, forced all
register widths to 8 bits, thereby fortunately canceling out this
problem. Nevertheless it makes no sense to continue encoding a
constant width in these virtual addresses.
Existing trace, fspgroup, fsppilot and result tables may be converted
to the new encoding by using this query:
UPDATE <tablename> SET data_address = ((data_address >> 4) & ~0xf) | data_address & 0xf;
Change-Id: I7a942b78c34f6140803a86af639eeedef3550f34
This change adapts the gem5 backend to the Register class interface change
in commit 52723a8. The necessary modifications suggested adding the "misc"
registers from gem5, too.
Change-Id: I32561c3fc905b9cd396e32ce80c791c01d5682fb
This change introduces a CMake-style FindMySQL.cmake properly looking for
libmysqlclient_r with mysql_config. This also fixes linking on some
machines.
Change-Id: Ifdbfdc3c7440dead37a8b63aaa86732d636aa0e2
The extended trace register list is needed in multiple locations; the CPU
class is the logical module to contain this information.
Increased number of x86 registers to be traced; we can remove those
that prove unusable for fault-space pruning later on.
Change-Id: Ic46ecdbc55167a6d92872c190317fc0d1a3ad92d
This allows a register to be a member of multiple UniformRegisterSets.
Needed to architecture-specifically store, e.g., a list of registers to
record in traces.
Change-Id: Ia5d20768540efe252a84df967d43e569f107da3a
Several database implementations, including, e.g., MySQL and PostgreSQL,
allow multiple rows to be INSERTed at once. At least for MySQL this
improves INSERT speed significantly.
Change-Id: Ic223f7915053f30b0cb00c567310f3e780c9fa59
Database::insert_id() retrieves the auto-generated ID from the last INSERT
or UPDATE on a table containing an AUTO_INCREMENT column.
Change-Id: I6ec94d1820bb7e77bd37df33145caa93016fd328
This was necessary because LLVM 3.1's
ELFObjectFile<ELFT>::sectionContainsSymbolsectionContainsSymbol() (needed
for proper functioning of our LLVM disassembler, unless you're using
-ffunction-sections) is "unimplemented".
Change-Id: I81112627ebd1c92b718ac6f4ed58d7f188aedf0c
- Don't enforce the join order, MariaDB usually gets this right.
- Update DB statistics before terminating.
Change-Id: If7bbbe146321430d199811062d05b3c179c5732f
data_address is definitely part of the unique key to trace entries, but
instr2 is arbitrary (could be instr1, time1 or time2 as well). Moving
data_address up the hierarchy to speed up certain FSP experiments.
Change-Id: I37a1f6c1e5b3957ba2f5bf46e0cd1a9c4aa7bfef
When recording a "full" trace (register values, plus dereferenced
values), we use a fixed dereference size of 4 bytes.
Change-Id: Ifbf6730377e7df2e7f1fc1e6b91bc114ccf10fbc
The new parameter -s/--stats shows trace statistics such as number of
instructions, number of memory reads/writes, and duration in simulator
time.
Change-Id: I39a48ff62d9c308c420fe52e7ed17ed57ae9c139
Quoting connect(3posix): "If connect() fails, the state of the socket is
unspecified. Conforming applications should close the file descriptor and
create a new socket before attempting to reconnect."
Change-Id: Ibcdcc0f546560a41009832894659a37947243f2f
We need to cover the complete fault space when a memory map is
provided. This means fake trace entries for completely unaccessed
addresses (e.g., unused parts of the stack), and makes sure we include
all of our fault space in the statistics later.
Without a memory map, we just don't know the extents of our fault
space; just guessing by using the minimum and maximum addresses is not
a good idea, we might have large holes in the fault space.
Change-Id: Ieb4b30a2e41a9921b969bf88513bfbacfc290c93
This prevents client and server from being sent a SIGPIPE (and
terminating) when the other side unexpectedly closes the connection.
It's way easier to handle this condition when checking the write()
return value, than to do anything smart in a SIGPIPE handler. More
details:
<http://stackoverflow.com/questions/108183/how-to-prevent-sigpipes-or-handle-them-properly>
Change-Id: I1da5bf5ef79c8b7b00ede976e96ed4f1c560049d
We're not yet using the common DatabaseCampaign as it doesn't allow
for additional experiment parameters (such as "variant" and
"benchmark") yet. TODO: Integrate changes in DatabaseCampaign in a
generic way and use it.
Change-Id: I45480003be433654aea8d3a417fbfa66be31155b
- Variants/benchmarks can now be selected with wildcards
(--variant/--benchmark), and can be excluded from pruning
(--variant-exclude/--benchmark-exclude).
- The database clearing step can be skipped with --no-delete to
avoid deadlocks with concurrent DB accesses.
- Internals:
* injection_instr / injection_instr_absolute moves from
fspgroup to fsppilot. fsppilot now contains all information we
need for running FI experiments.
TODO: generic campaign needs to be modified, too.
* Force MySQL to use an efficient join order (STRAIGHT_JOIN).
Change-Id: I6241ea2de9da1a1e709fae6374df4fc06ef262a0
- Import timing information from traces that were recorded with timing.
- Allow restricting import to a memory map ("vertical" restriction).
- Proper fault-space right-margin handling.
- Cleanups, data-type usage, etc.
Change-Id: I7a49e8e9e49894c458e884bfc234f36b9ba8b130
TODO: Showing timing or extended trace information should be configurable
with a commandline switch later.
Change-Id: I72ac95ddd1d54dfef87f212ec5afa30b2ed9a6ad
This adds an interface for a backend-specific notion of time, e.g. CPU
cycles since simulator start, and a concrete implementation for the
Bochs backend. This is needed to record CPU idle times (e.g., HLT
instruction), and for target backends capable of more timing-accurate
execution.
This change also modifies the tracing plugin to add the time to all
trace events.
Change-Id: I93ac1d54c07f32b0b8f84f333417741d8e9c8288
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
This allows the commandline parameter parser to modify argc, as it finds
arguments for the Fail* client. Additionally argv is correctly null
terminated when removing arguments.
This fixes a bug introduced in eb17e9ef82.
Change-Id: Iabe84530790ecb7c587b0af139127015aad868d5
This allows us to generate prerequisites (traces, state snapshots, memory
maps) in parallel, and without the previous shell script hacks.
Change-Id: I05a0321a794b4033d05eed20f5bffbd1e910cf1b
Without --with-all-libs (as documented in doc/how-to-build.txt), at least
the "nogui" GUI does not work anymore -- the default for headless
experiments.
(Note the UNDEFINED->UNDEF enum rename; this was necessary due to a vga.h
include side-effect in bochs/gui/svga.cc.)
Change-Id: I1bc3208e905783505a35bbc48ff29f00eef599d6
It doesn't make sense to override database, hostname or username parameters
coming from ~/.my.cnf with arbitrary own defaults. Instead allow
libmysqlclient to use the builtin default parameter mechanism.
See <https://dev.mysql.com/doc/refman/5.1/en/option-files.html> for what a
~/.my.cnf may look like; example:
[client]
user=hsc
password=secret42
database=fail_demo
Change-Id: I90aadc04375c624d5e906557c8bd4049e5caa17c
Imported and adapted FindProtobuf.cmake from cmake 2.8.9 into the project to
make it build on Debian 6 again.
Change-Id: I4eda087b1757073224defc643dab86d0bd162409
An experiment talking to a campaign server via the JobClient/JobServer
interface needs the FailControlMessage.proto compiler to run before the
experiment is compiled. A dependency on fail-comm ensures this.
If CommandLine.hpp and (indirectly) optionparser.h is #included in
FailBochsInit.ah, bochs compilation fails (for, e.g., gui/x.cc, at least
on Debian 6).
When we want to use a bounded job queue, job producer and job consumer must
run in different threads in order not to deadlock the campaign. Ideally
this functionality moves to the CampaignManager later (including the
retrieval of existing experiment results and storing new results).
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1946 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
Synchronize re-sending jobs in sendPendingExperimentData() and modifying
(or indirectly, via getDone() and the campaign, deleting) jobs in the
m_runningJobs queue.
a) sendPendingExperimentData needs an intact job to serialize and send it.
b) After moving the job to m_doneJobs, it may be retrieved and deleted
by the campaign at any time.
Additionally, receiving a result overwrites the job's contents. This
already may cause breakage in sendPendingExperimentData (a).
git-svn-id: https://www4.informatik.uni-erlangen.de/i4svn/danceos/trunk/devel/fail@1943 8c4709b5-6ec9-48aa-a5cd-a96041d1645a
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
Horst Schirmeier