Summary:
This change implements support for the custom event logging sleds and
intrinsics at runtime. For now it only supports handling the sleds in
x86_64, with the implementations for other architectures stubbed out to
do nothing.
NOTE: Work in progress, uploaded for exposition/exploration purposes.
Depends on D27503, D30018, and D33032.
Reviewers: echristo, javed.absar, timshen
Subscribers: mehdi_amini, nemanjai, llvm-commits
Differential Revision: https://reviews.llvm.org/D30630
llvm-svn: 302857
A problem arises if a machine supports the rdtscp instruction, but the processor
frequency cannot be determined by the function getTSCFrequency(). In this case,
we want to use the emulated TSC instead. This patch implements that by adding a
call to getTSCFrequency() from probeRequiredCPUFeatures(), and the function only
returns true if both the processor supports rdtscp and the CPU frequency can be
determined.
This should fix PR32620.
Reviewers: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D32067
llvm-svn: 300525
This patch applies a work-around to the XRay FDR tests when TSC emulation is
needed because the processor frequency cannot be determined.
This fixes PR32620 using the suggestion given by Dean in comment 1.
Reviewers: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31967
llvm-svn: 300017
Previously in r297800, a work-around was created to use TSC emulation on x86_64 when RDTSCP was not available on the host. A similar change was needed in the file xray_fdr_logging.cc which this patch ports over to that file.
Eventually the code should be refactored as there will be 3 locations with the same code, but that can be done as a separate step. This patch is just to keep the test from failing on my machine due to an illegal instruction since RDTSCP is not available on my x86_64 linux VM.
Reviewers: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31909
llvm-svn: 299922
Summary:
Currently the FDR log writer, upon flushing, dumps a sequence of buffers from
its freelist to disk. A reader can read the first buffer up to an EOB record,
but then it is unclear how far ahead to scan to find the next threads traces.
There are a few ways to handle this problem.
1. The reader has externalized knowledge of the buffer size.
2. The size of buffers is in the file header or otherwise encoded in the log.
3. Only write out the portion of the buffer with records. When released, the
buffers are marked with a size.
4. The reader looks for memory that matches a pattern and synchronizes on it.
2 and 3 seem the most flexible and 2 does not rule 3 out.
This is an implementation of 2.
In addition, the function handler for fdr more aggressively checks for
finalization and makes an attempt to release its buffer.
Reviewers: pelikan, dberris
Reviewed By: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31384
llvm-svn: 298982
Summary:
This change exercises the end-to-end functionality defined in the FDR
logging implementation. We also prepare for being able to run traces
generated by the FDR logging implementation from being analysed with the
llvm-xray command that comes with the LLVM distribution.
This also unblocks D31385, D31384, and D31345.
Reviewers: kpw, pelikan
Subscribers: llvm-commits, mgorny
Differential Revision: https://reviews.llvm.org/D31452
llvm-svn: 298977
Instead of std::atomic APIs for atomic operations, we instead use APIs
include with sanitizer_common. This allows us to, at runtime, not have
to depend on potentially dynamically provided implementations of these
atomic operations.
Fixes http://llvm.org/PR32274.
llvm-svn: 298833
Summary:
Separated the IO and the thread local storage state machine of logging
from the writing of log records once the contents are deterministic.
Finer granularity functions are provided as inline functions in the same
header such that stack does not grow due to the functions being separated.
An executable utility xray_fdr_log_printer is also implemented to use the
finest granularity functions to produce binary test data in the FDR format
with a relatively convenient text input.
For example, one can take a file with textual contents layed out in rows
and feed it to the binary to generate data that llvm-xray convert can then
read. This is a convenient way to build a test suite for llvm-xray convert
to ensure it's robust to the fdr format.
Example:
$cat myFile.txt
NewBuffer : { time = 2 , Tid=5}
NewCPU : { CPU =1 , TSC = 123}
Function : { FuncId = 5, TSCDelta = 3, EntryType = Entry }
Function : { FuncId = 5, TSCDelta = 5, EntryType = Exit}
TSCWrap : { TSC = 678 }
Function : { FuncId = 6, TSCDelta = 0, EntryType = Entry }
Function : { FuncId = 6, TSCDelta = 50, EntryType = Exit }
EOB : { }
$cat myFile.txt | ./bin/xray_fdr_log_printer > /tmp/binarydata.bin
$./bin/llvm-xray convert -output-format=yaml -output=- /tmp/binarydata.bin
yaml format comes out as expected.
Reviewers: dberris, pelikan
Reviewed By: dberris
Subscribers: llvm-commits, mgorny
Differential Revision: https://reviews.llvm.org/D30850
llvm-svn: 297801
Summary:
Functions with the LOG_ARGS_ENTRY sled kind at their beginning will be handled
in a way to (optionally) pass their first call argument to your logging handler.
For practical and performance reasons, only the first argument is supported, and
only up to 64 bits.
Reviewers: javed.absar, dberris
Reviewed By: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D29703
llvm-svn: 297000
Summary:
Use a common definition of a "this variable is unused" annotation for useless
variables only present for their lambda global initializers, to silence gcc's
warning.
Reviewers: dberris
Reviewed By: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D29860
llvm-svn: 296449
Revert "Fix -Wsign-compare - this might not be quite right, but preserves behavior"
Revert "[XRay] Implement powerpc64le xray."
This reverts commit r294826.
This reverts commit r294781.
llvm-svn: 294842
Summary:
The implementation, however, is in different arch-specific files, unless it's emulated.
Reviewers: dberris, pelikan, javed.absar
Subscribers: aemerson, llvm-commits
Differential Revision: https://reviews.llvm.org/D29796
llvm-svn: 294777
Summary:
Fixing a bug I found when testing a reader for the FDR format. Function ID is
now correctly packed into the 28 bits which are documented for it instead of being
masked to all ones.
Reviewers: dberris, pelikan, eugenis
Reviewed By: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D29698
llvm-svn: 294563
Summary:
As pointed out in casual reading of the XRay codebase, that we had
some interesting named functions that didn't quite follow the LLVM coding
conventions.
Reviewers: chandlerc, dblaikie
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D29625
llvm-svn: 294373
Summary:
This was pointed out that FDR mode didn't quite put the thread ID in the
buffers, but instead would write down the parent process ID.
Reviewers: pelikan, rSerge
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D29484
llvm-svn: 294166
Summary:
In this change we introduce the notion of a "flight data recorder" mode
for XRay logging, where XRay logs in-memory first, and write out data
on-demand as required (as opposed to the naive implementation that keeps
logging while tracing is "on"). This depends on D26232 where we
implement the core data structure for holding the buffers that threads
will be using to write out records of operation.
This implementation only currently works on x86_64 and depends heavily
on the TSC math to write out smaller records to the inmemory buffers.
Also, this implementation defines two different kinds of records with
different sizes (compared to the current naive implementation): a
MetadataRecord (16 bytes) and a FunctionRecord (8 bytes). MetadataRecord
entries are meant to write out information like the thread ID for which
the metadata record is defined for, whether the execution of a thread
moved to a different CPU, etc. while a FunctionRecord represents the
different kinds of function call entry/exit records we might encounter
in the course of a thread's execution along with a delta from the last
time the logging handler was called.
While this implementation is not exactly what is described in the
original XRay whitepaper, this one gives us an initial implementation
that we can iterate and build upon.
Reviewers: echristo, rSerge, majnemer
Subscribers: mehdi_amini, llvm-commits, mgorny
Differential Revision: https://reviews.llvm.org/D27038
llvm-svn: 293015
Summary:
In this change we introduce the notion of a "flight data recorder" mode
for XRay logging, where XRay logs in-memory first, and write out data
on-demand as required (as opposed to the naive implementation that keeps
logging while tracing is "on"). This depends on D26232 where we
implement the core data structure for holding the buffers that threads
will be using to write out records of operation.
This implementation only currently works on x86_64 and depends heavily
on the TSC math to write out smaller records to the inmemory buffers.
Also, this implementation defines two different kinds of records with
different sizes (compared to the current naive implementation): a
MetadataRecord (16 bytes) and a FunctionRecord (8 bytes). MetadataRecord
entries are meant to write out information like the thread ID for which
the metadata record is defined for, whether the execution of a thread
moved to a different CPU, etc. while a FunctionRecord represents the
different kinds of function call entry/exit records we might encounter
in the course of a thread's execution along with a delta from the last
time the logging handler was called.
While this implementation is not exactly what is described in the
original XRay whitepaper, this one gives us an initial implementation
that we can iterate and build upon.
Reviewers: echristo, rSerge, majnemer
Subscribers: mehdi_amini, llvm-commits, mgorny
Differential Revision: https://reviews.llvm.org/D27038
llvm-svn: 290852
Dean Michael Berris