github ColinIanKing/stress-ng V0.18.12
on GitHub

5 days ago

stress-ng V0.18.12 "manic memory mangler"

New stressors

  • easy-opcode - simple easy opcode stressor for stressing front end decoder and instruction cache
  • physmmap - physical memory mapping (/dev/mem) stressor
  • umask - umask system call exercising stressor

New options

  • --vmstat-units - vmstat memory scaling option
  • --aggressive - aggressive mode now works on more stressors
  • --minimize - size minimization mode now works on more stressors
  • --maximize - size maximization mode now works on more stressors

Detailed Changelog

[Dmitry Antipov]physmmap: add physical memory mapping (/dev/mem) stressor

  • stress-link/memfd/mmaphuge/mmaptorture: fix global buffer overflow

[Dmitry Grand]

  • fix error: assignment to ‘void (* const*)(stress_fma_t *)’

[Nicolas Bouton]

  • stress-prefetch: add aarch64 load prefetch instructions (prfm)

[Sergey Fedorov]

  • Enable PPC32 builds, use PPC32 opcodes

[Zong Li]

  • core-asm-ret: add lpad instruction in return code on RISC-V

[Colin Ian King]

  • debian/control: update standards version to 4.7.2
  • stress-easycopy: ensure direction flag is cleared
  • core-asm-ret: add PPC64 big endian and PPC little and big endian support
  • stress-flushcache: call dcbst_page for PPC and PPC64
  • stress: improve checking for ppc and ppc64
  • README.md: update contributor list
  • stress-nop: make policy non-clobberable for ppc32 gcc 4.9.2 builds
  • stress-cyclic: make policy non-clobberable for ppc32 gcc 4.9.2 builds
  • README.md: add another research paper link
  • stress-hyperbolic: long double trig should be at least as accurate as double
  • stress-trig: long double trig should be at least as accurate as double
  • stress-opcode: Fix incorrect opcode mmap size, 2 pages short
  • stress-lockbus: add timer to break out stuck split locks on ppc64
  • stress-regex: Fix digits regex name, should be digits and not alphas
  • stress-physmmap: Fix spelling mistake "mappping" -> "mapping"
  • core-shim: fix spelling mistake "obsolute" -> "obsolete"
  • stress-tlb-shootdown: page align two more offsets
  • stress-tlb-shootdown: msync memfd mapping using the correct size
  • stress-malloc: only declare tmp_align for aligned allocations
  • stress-tlb-shootdown: page align offsets
  • stress-opcode: fix void * pointer arithmetic, use uint8_t *
  • stress-easy-opcode: fix void * pointer arithmetic, use uint8_t *
  • stress-misaligned: perform __uint128_t access using uint64_t pointers
  • stress-open: always unlink file on O_TRUNC open
  • stress-stream: add hex string compare of checksums
  • stress-stream: fix checksum string termination, should be using index j
  • Manual: Fix spelling mistake "execises" -> "exercises"
  • stress-physmmap: add upper region size bounds check
  • stress-easy-opcode: ensure enough end spaces is available for return ops(s)
  • core-arch.h: move endian macros to core-arch.h
  • stress-easy-opcode: fix ppc64 build, replace . with ,
  • stress-opcode: clean up object code size handling
  • stress-easy-opcode: add a simple easy opcode stressor
  • stress-tlb-shootdown: make TLB shootdowns more aggressive, add metrics
  • core-cpu-cache: only use clflushopt if it is supported by the processor
  • stress-cpu: disable FP denormalization to improve performance
  • core-cpu: add helpers to disable/enable fp denormalization
  • kernel-coverage: add --physmmap-read test
  • stress-physmmap: use VERIFY_NONE since there are no verifications
  • core-cpu-cache: fix build issue, use renamed variable
  • stress-physmmap: add --physmmap-read option to force read data in every page
  • stress-misaligned: use atomic_fetch_add if 2,4,8 byte variants unavailable
  • core-cpu-cache: use uint8_t for pointers to avoid void * pointer arithmetic
  • stress-cache: prefetch data closer to the CPU
  • stress-physmmap: attempt to mmap entire region in one mmap
  • stress-physmmap: make mappable_pages a register size_t
  • stress-physmmap: report if no pages were mappable
  • stress-physmmap: add physical memory mapping (/dev/mem) stressor
  • README.md: Add Nicolas Bouton to contributors list
  • Makefile: add sanitization build flags
  • README.md: Add Zong Li to contributors list
  • Remove stress-mpfr.
  • core-parse-opts: rename scales to type of scaling
  • stress-uprobe: cast size_t to ssize_t for sane type comparison
  • stress-uprobe: prevent buffer overflows when parsing data
  • stress-race-sched: fix off-by-one comparison on method_all_index
  • stress-malloc: ensured aligned_alloc is a multiple of alignment
  • README.md: Add Dmitry Antipov and Dmitry Grand to contributors list
  • README.md add another research citation
  • README.md: add some more kernel bug references
  • stress-shellsort: use mmap data, collapse pages to try to sort on hugepage mapped data
  • stress-insertionsort: use mmap data, collapse pages to try to sort on hugepage mapped data
  • stress-heapsort: used mmap data, collapse pages to try to sort on hugepage mapped data
  • stress-bubblesort: used mmap data, collapse pages to try to sort on hugepage mapped data
  • stress-bitononicsort: fix mmap'd anon mapped name
  • stress-bitonicsort: used mmap data, collapse pages to try to sort on hugepage mapped data
  • stress-qsort: collapse pages to try to sort on hugepage mapped data
  • stress-mergesort: collapse pages to try to sort on hugepage mapped data
  • stress-workload: force no hugepages on mapping for more PTEs
  • stress-numa: force no hugepages on mapping for more PTEs
  • stress-memthrash: force no hugepages on mapping for more PTEs
  • core-madvise: add stress_madvise_nohugepage, clean up madvise helpers
  • core-madvise: move stress_madvise_collapse from core-helper to core-madvise
  • stress-vm: add collapse to vm-madvise options for MADV_COLLAPSE
  • stress-matrix-3d: improve performance by using MADV_COLLAPSE
  • stress-matrix: improve performance by using MADV_COLLAPSE
  • stress-memrate: improve performance by using MADV_COLLAPSE
  • core-heler: add stress_madvise_collapse to collapse pages to hugepages
  • README.md: update number of stressors and stressor types
  • stress-umask: add umask system call exercising stressor
  • stress-mmaptorture: add MADV_RANDOM on zero sized allocation
  • README.md: add a couple more research paper links
  • core-shim: add build-time check for close_range, use it in core-shim
  • stress-prefetch: perform uint64_t pointer addition correctly
  • core-helper: check for pipe on stdin and sort out tty width accordingly
  • README.md: add another research paper link
  • matrix-methods.job: remove hot from hot-cpu
  • core-cpu-cache: add #elif for cacheflush, helps static analyzer
  • stress-bigheap: use bit-wise & for logical anding
  • stress-numa: use bit-wise & for logical anding
  • stress-zombie: voidify return from shim_usleep()
  • stress-zlib: maximize/minimize --zlib-level, --zlib-mem-level
  • stress-yield: maximize/minimize --yield-procs
  • stress-swap: enable --swap-self on --aggressive
  • stress-stack: enable --stack-{fill|mlock|pageout|swap} on --aggressive
  • stress-sock: maximize/minimize --sock-msgs
  • stress-shm-sysv: enable --shm-sysv-mlock on --aggressive
  • stress-shm: enable --shm-mlock on --aggressive
  • Manual: fix missing sysv shared memory heading
  • stress-sem: enable --sem-shared on --aggressive
  • stress-resources: enable --resources-mlock on --aggressive
  • stress-remap: minimize --remap-pages
  • stress-readahead: limit --readahead-bytes maximized setting to MAX_32
  • stress-randlist: maximize/minimize --randlist-items, --randlist-size
  • stress-ramfs: fix minimize option, bump max ramfs size to 2GB
  • stress-ramfs: make --ramfs-fill fill pages with random data
  • stress-ng: check for stressors that should be ignore
  • stress-ramfs: maximize/minimize --ramfs-size
  • stress-pty: maximize/minimize --pty-max
  • stress-pseek: maximize/minimize --pseek-io-size
  • stress-pseek: enable --pseek-rand on --aggressive
  • stress-poll: maximize/minimize --poll-fds
  • stress-pipeherd: enable --pipeherd-yield on --aggressive
  • stress-pipe: maximize/minimize --pipe-data-size, --pipe-size
  • Manual: fix indentation on --pci-ops-rate option
  • stress-pagemove: enable --pagemove-mlock and --pagemove-numa on --aggressive
  • stress-open: minimize --open-max
  • stress-numa: enable numa-shuffle-addr and numa-shuffle-node on --aggressive
  • stress-numa: maximize/minimize --numa-bytes
  • Manual: fix missing netlink task title
  • stress-nanosleep: maximize/minimize --nanosleep-threads
  • stress-msg: maximize/minimize --msg-bytes
  • stress-mpfr: maximize/minimize --mpfr-precision
  • stress-monte-carlo: maximize/minimize --monte-carlo-samples
  • stress-mmaptorture: maximize/minimize --mmaptorture-bytes, --mmaptorture-msync
  • stress-mmaphuge: maximize/minimize --mmaphuge-mmaps
  • stress-mmapfork: maximize --mmapfork-bytes
  • stress-mlockmany: maximize/minimize --mlockmany-procs
  • Manual: fix missing title for mmapmany stressor
  • stress-malloc: maximize/minimize --malloc-pthreads
  • stress-jpeg: maximize/minimize --jpeg-height, --jpeg-width, --jpeg-quality
  • stress-ipsec-mb: maximize/minimize --ipsec-mb-jobs
  • stress-io-uring: maximize/minimize --io-uring-entries
  • Manual: remove idle-page help
  • stress-icmp-flood: maximize/minimize --icmp-flood-max-size
  • stress-fractal: maximize/minimize --fractal-iterations, --fractal-sizex, --fractal-sizey
  • stress-fpunch: maximize/minimize --fpunch-bytes
  • stress-forkheavy: maximize/minimize --forkheavy-allocs, --forkheavy-procs
  • stress-flipflop: maximize/minimize --flipflop-bits
  • stress-fifo: maximize/minimize --fifo-data-size
  • stress-fd-fork: maximize/minimize --fd-fork-fds
  • stress-far-branch: maximize/minimize --far-branch-pages
  • stress-factor: maximize/minimize --factor-digits
  • stress-exec: maximize/minimize --exec-max
  • stress-ng: fix help, iostate -> iostat
  • stress-epoll: maximize/minimize --epoll-sockets
  • stress-dirmany: maximize/minimize --dirmany-bytes
  • stress-dir: maximize/minimize --dir-dirs
  • stress-dccp: maximize/minimize dccp-msgs
  • stress-chdir: maximize/minimize chdir-dirs
  • stress-bigheap: maximize/minimize bigheap-bytes
  • stress-malloc: flush pages when using --aggressive
  • core-resources: ensure all mapped pages are touched
  • stress-bigheap: use memory clearing calloc when using --aggressive
  • stress-vm: flush pages when using --aggressive
  • stress-vm-addr: flush pages when using --aggressive
  • stress-vma: flush pages when using --aggressive
  • stress-stack: flush touched stack pages when using --aggressive
  • stress-stack: make --stack-fill option fill stack with random data
  • stress-workload: use stress_cpu_data_cache_flush to flush data
  • stress-tlb-shoodown: use stress_cpu_data_cache_flush to flush data
  • stress-sysbadaddr: use stress_cpu_data_cache_flush to flush page
  • stress-sigsegv: use stress_cpu_data_cache_flush to flush data
  • stress-mcontend: use stress_cpu_data_cache_flush to flush cache line
  • stress-prefetch: use stress_cpu_data_cache_flush for cache flushing
  • core-cpu-cache: increment addr by cache line size
  • core-cpu-cache: add more optimized data cache flush helper
  • stress-pageswap: madvise populate read when using --aggressive option
  • stress-munmap: for pages to page out with --aggressive option
  • stress-madvise: force more swapouts/swapins with --aggressive options
  • core-out-of-memory: Fix off-by-one array overflow error
  • stess-bigheap: increase reallocs and memory checks with --aggressive
  • stress-affinity: add more sched_setaffinity call with --aggressive option
  • stress-access: add more access calls with --aggressive option enabled
  • stress-lockbus: add 1 byte and 2 byte misaligned 32 bit lock bus operations
  • stress-ng: enable specifying number of instances as % of cpus
  • core-, stress-: rename num_instances to instances
  • core-vmstat: add vmstat memory scaling option --vmstat-units U
Check out latest releases or
releases around ColinIanKing/stress-ng V0.18.12

Don't miss a new stress-ng release

NewReleases is sending notifications on new releases.

Get notifications