gcc in the linux kernel was not unrolling the inner loops of the Huffman
decoder, which was destroying decoding performance. The compiler was
generating crazy code with all sorts of branches. I suspect because of
Spectre mitigations, but I'm not certain. Once the loops were manually
unrolled, performance was restored.
Additionally, when gcc couldn't prove that the variable left shift in
the 4X2 decode loop wasn't greater than 63, it inserted checks to verify
it. To fix this, mask `entry.nbBits & 0x3F`, which allows gcc to eliete
this check. This is a no op, because `entry.nbBits` is guaranteed to be
less than 64.
Lastly, introduce the `HUF_DISABLE_FAST_DECODE` macro to disable the
fast C loops for Issue #3762. So if even after this change, there is a
performance regression, users can opt-out at compile time.
detected by @terrelln,
these issue could be triggered in specific scenarios
namely decompression of certain invalid magic-less frames,
or requested properties from certain invalid skippable frames.
* add check for valid dest buffer and fuzz on random dest ptr when malloc 0
* add uptrval to linux-kernel
* remove bin files
* get rid of uptrval
* restrict max pointer value check to platforms where sizeof(size_t) == sizeof(void*)
decompression features automatic support of sparse files,
aka a form of "compression" where entire blocks consists only of zeroes.
This only works for some compatible file systems (like ext4),
others simply ignore it (like afs).
Triggering this feature relies of `fseek()`.
But `fseek()` is not compatible with non-seekable devices, such as pipes.
Therefore it's disabled for pipes.
However, there are other objects which are not compatible with `fseek()`, such as block devices.
Changed the logic, so that `fseek()` (and therefore sparse write) is only automatically enabled on regular files.
Note that this automatic behavior can always be overridden by explicit commands `--sparse` and `--no-sparse`.
fix#3583
This does the following:
1. Compress test data into multiple frames
2. Perform a series of small decompressions and seeks forward, checking
that compressed data wasn't reread unnecessarily.
3. Perform some seeks forward and backward to ensure correctness.
When decompressing a seekable file, if seeking forward within
a frame (by issuing multiple ZSTD_seekable_decompress calls
with a small gap between them), the frame will be unnecessarily
reread from the beginning. This patch makes it continue using
the current frame data and simply skip over the unneeded bytes.
Looking at the __builtin_expect in ZSTD_decodeSequence:
{ size_t offset;
#if defined(__clang__)
if (LIKELY(ofBits > 1)) {
#else
if (ofBits > 1) {
#endif
ZSTD_STATIC_ASSERT(ZSTD_lo_isLongOffset == 1);
From profile-annotated assembly, the probability of ofBits > 1 is about 75%
(101k counts out of 135k counts). This is much smaller than the recommended
likelihood to use __builtin_expect which is 99%. As a result, clang moved the
else block further away which hurts cache locality. Removing this
__built_expect along with two others in ZSTD_decodeSequence gave better
performance when PGO is enabled. I suggest to remove these branch hints and
rely on PGO which leverages runtime profiles from actual workload to calculate
branch probability instead.
Inlining `BIT_reloadDStream` provided >3% decompression speed improvement for
clang PGO-optimized zstd binary, measured using the Silesia corpus with
compression level 1. The win comes from improved register allocation which leads
to fewer spills and reloads. Take a look at this comparison of
profile-annotated hot assembly before and after this change:
https://www.diffchecker.com/UjDGIyLz/. The diff is a bit messy, but notice three
fewer moves after inlining.
In general LLVM's register allocator works better when it can see more code. For
example, when the register allocator sees a call instruction, it partitions the
registers into caller registers and callee registers, and it is not free to do
whatever it wants with all the registers for the current function. Inlining the
callee lets the register allocation access all registers and use them more
flexsibly.
Rather than remove the flag entirely, as proposed in #3499, this commit uses
the newest C++ standard the compiler supports. This retains the selection of
using only standardized features (excluding GNU extensions) and keeps the
recency requirements of the codebase explicit.
Tested with various versions of `g++` and `clang++`.
This fixes compilation with clang-cl on Windows. There
is a bug in cmake so that check_linker_flag() doesn't give
the correct result when using link.exe/lld-link.exe.
Details in CMake's gitlab: https://gitlab.kitware.com/cmake/cmake/-/issues/22023Fixes#3522
Nick Terrell