Skip to content

Commit

Permalink
Upstream.
Browse files Browse the repository at this point in the history
  • Loading branch information
mirage335 committed Dec 6, 2024
1 parent b0f04d2 commit 7322a96
Show file tree
Hide file tree
Showing 2 changed files with 228 additions and 11 deletions.
237 changes: 227 additions & 10 deletions ubiquitous_bash.sh
Original file line number Diff line number Diff line change
Expand Up @@ -36,7 +36,7 @@ _ub_cksum_special_derivativeScripts_contents() {
#export ub_setScriptChecksum_disable='true'
( [[ -e "$0".nck ]] || [[ "${BASH_SOURCE[0]}" != "${0}" ]] || [[ "$1" == '--profile' ]] || [[ "$1" == '--script' ]] || [[ "$1" == '--call' ]] || [[ "$1" == '--return' ]] || [[ "$1" == '--devenv' ]] || [[ "$1" == '--shell' ]] || [[ "$1" == '--bypass' ]] || [[ "$1" == '--parent' ]] || [[ "$1" == '--embed' ]] || [[ "$1" == '--compressed' ]] || [[ "$0" == "/bin/bash" ]] || [[ "$0" == "-bash" ]] || [[ "$0" == "/usr/bin/bash" ]] || [[ "$0" == "bash" ]] ) && export ub_setScriptChecksum_disable='true'
export ub_setScriptChecksum_header='2591634041'
export ub_setScriptChecksum_contents='2804202287'
export ub_setScriptChecksum_contents='4001117410'

# CAUTION: Symlinks may cause problems. Disable this test for such cases if necessary.
# WARNING: Performance may be crucial here.
Expand Down Expand Up @@ -11790,6 +11790,10 @@ _set_getMost_backend_debian() {
_getMost_backend_aptGetInstall() {
# --no-upgrade
# -o Dpkg::Options::="--force-confold"

# ATTRIBUTION-AI: ChatGPT o1-preview 2024-11-20 .
echo 'APT::AutoRemove::RecommendsImportant "true";
APT::AutoRemove::SuggestsImportant "true";' | _getMost_backend tee /etc/apt/apt.conf.d/99autoremove-recommends > /dev/null

if ! _getMost_backend dash -c 'type apt-fast' > /dev/null 2>&1 || [[ "$RUNNER_OS" != "" ]]
then
Expand Down Expand Up @@ -19107,8 +19111,19 @@ _createVMimage() {
# 27.95GiB
#export vmSize=28620

# Preferred with 'augment' ~8b q4_k_m LLM model.
# 37.95GiB
export vmSize=38860
#export vmSize=38860

# May accommodate a few additional AI models.
# 52.95GiB
#export vmSize=54220

# Slightly smaller than expected 50GB BD-R DL .
# 46.1GiB
export vmSize=47206



export vmSize_boundary=$(bc <<< "$vmSize - 1")
_createRawImage
Expand Down Expand Up @@ -23755,6 +23770,10 @@ PARAMETER num_ctx 6144' > Llama-augment.Modelfile
aria2c --log=- --log-level=info -x "3" -o 'llama-3.1-8b-instruct-abliterated.Q4_K_M.gguf' 'https://huggingface.co/mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated-GGUF/resolve/main/meta-llama-3.1-8b-instruct-abliterated.Q4_K_M.gguf'
[[ ! -e 'llama-3.1-8b-instruct-abliterated.Q4_K_M.gguf' ]] && aria2c --log=- --log-level=info -x "3" -o 'llama-3.1-8b-instruct-abliterated.Q4_K_M.gguf' 'https://huggingface.co/mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated-GGUF/resolve/main/meta-llama-3.1-8b-instruct-abliterated.Q4_K_M.gguf' --disable-ipv6=true

if [[ ! -e 'llama-3.1-8b-instruct-abliterated.Q4_K_M.gguf' ]]
then
_wget_githubRelease_join "soaringDistributions/Llama-augment_bundle" "" "llama-3.1-8b-instruct-abliterated.Q4_K_M.gguf"
fi

_service_ollama

Expand Down Expand Up @@ -23829,11 +23848,38 @@ _test_ollama() {
}

_vector_ollama_procedure() {
! _ollama_run_augment "Please output the word true . Any other output accompanying the word true is acceptable but not desirable. The purpose of this prompt is merely to validate that the LLM software is entirely functional, so the word true will be very helpful whereas any output other than the word true will be unhelpful . Please output the word true ." | grep true > /dev/null && echo 'fail: _vector_ollama' && _messageFAIL && _stop 1
_ollama_run_augment "Please output the word true . Any other output accompanying the word true is acceptable but not desirable. The purpose of this prompt is merely to validate that the LLM software is entirely functional, so the word true will be very helpful whereas any output other than the word true will be unhelpful . Please output the word true ." | grep false > /dev/null && echo 'fail: _vector_ollama' && _messageFAIL && _stop 1
local currentExitStatus
currentExitStatus=1

! _ollama_run_augment "Please output the word false . Any other output accompanying the word false is acceptable but not desirable. The purpose of this prompt is merely to validate that the LLM software is entirely functional, so the word false will be very helpful whereas any output other than the word false will be unhelpful . Please output the word false ." | grep false > /dev/null && echo 'fail: _vector_ollama' && _messageFAIL && _stop 1
_ollama_run_augment "Please output the word false . Any other output accompanying the word false is acceptable but not desirable. The purpose of this prompt is merely to validate that the LLM software is entirely functional, so the word false will be very helpful whereas any output other than the word false will be unhelpful . Please output the word false ." | grep true > /dev/null && echo 'fail: _vector_ollama' && _messageFAIL && _stop 1
if ! _ollama_run_augment "Please output the word true . Any other output accompanying the word true is acceptable but not desirable. The purpose of this prompt is merely to validate that the LLM software is entirely functional, so the word true will be very helpful whereas any output other than the word true will be unhelpful . Please output the word true ." | grep -i true > /dev/null
then
echo 'fail: _vector_ollama' && _messagePlain_bad 'fail: _vector_ollama: prompt for word true did not output word true'
else
currentExitStatus=0
fi
if _ollama_run_augment "Please output the word true . Any other output accompanying the word true is acceptable but not desirable. The purpose of this prompt is merely to validate that the LLM software is entirely functional, so the word true will be very helpful whereas any output other than the word true will be unhelpful . Please output the word true ." | grep -i false > /dev/null
then
echo 'fail: _vector_ollama' && _messagePlain_bad 'fail: _vector_ollama: prompt for word true instead included word false'
else
currentExitStatus=0
fi

if ! _ollama_run_augment "Please output the word false . Any other output accompanying the word false is acceptable but not desirable. The purpose of this prompt is merely to validate that the LLM software is entirely functional, so the word false will be very helpful whereas any output other than the word false will be unhelpful . Please output the word false ." | grep -i false > /dev/null
then
echo 'fail: _vector_ollama' && _messagePlain_bad 'fail: _vector_ollama: prompt for word false did not output word false'
else
currentExitStatus=0
fi
if _ollama_run_augment "Please output the word false . Any other output accompanying the word false is acceptable but not desirable. The purpose of this prompt is merely to validate that the LLM software is entirely functional, so the word false will be very helpful whereas any output other than the word false will be unhelpful . Please output the word false ." | grep -i true > /dev/null
then
echo 'fail: _vector_ollama' && _messagePlain_bad 'fail: _vector_ollama: prompt for word false instead included word true'
else
currentExitStatus=0
fi


# If NONE of the vector tests have succeeded, then FAIL . Normally, with an 'augment' LLM model, this should be so rare as to vastly more often indicate broken ollama installation, very broken/corrupted LLM model, very broken LLM configuration, insufficient disk space for model, etc.
[[ "$currentExitStatus" != "0" ]] && _messageFAIL && _stop 1

return 0
}
Expand All @@ -23848,7 +23894,7 @@ _vector_ollama() {
echo 'warn: accepted: cygwin: missing: ollama'
return 0
fi

if type -p ollama > /dev/null 2>&1
then
if [[ "$hostMemoryQuantity" -lt 28000000 ]]
Expand Down Expand Up @@ -23970,7 +24016,8 @@ _ollama_set_sequence-augment-lowRAM() {
_start
cd "$safeTmp"

ollama show Llama-augment --modelfile | sed 's/PARAMETER num_ctx [0-9]*/PARAMETER num_ctx 512/' > ./Llama-augment-tmp.Modelfile
#512
ollama show Llama-augment --modelfile | sed 's/PARAMETER num_ctx [0-9]*/PARAMETER num_ctx 640/' > ./Llama-augment-tmp.Modelfile
sleep 9
ollama create Llama-augment --file ./Llama-augment-tmp.Modelfile
sleep 9
Expand Down Expand Up @@ -32757,7 +32804,7 @@ _kernelConfig_require-tradeoff-perform() {
_messagePlain_nominal 'kernelConfig: tradeoff-perform'
_messagePlain_request 'Carefully evaluate '\''tradeoff-perform'\'' for specific use cases.'
export kernelConfig_file="$1"

_kernelConfig__bad-n__ CONFIG_RETPOLINE
_kernelConfig__bad-n__ CONFIG_PAGE_TABLE_ISOLATION

Expand Down Expand Up @@ -32803,6 +32850,32 @@ _kernelConfig_require-tradeoff-harden() {
_messagePlain_request 'Carefully evaluate '\''tradeoff-harden'\'' for specific use cases.'
export kernelConfig_file="$1"

_kernelConfig__bad-y__ CPU_MITIGATIONS
_kernelConfig__bad-y__ MITIGATION_PAGE_TABLE_ISOLATION
_kernelConfig__bad-y__ MITIGATION_RETPOLINE
_kernelConfig__bad-y__ MITIGATION_RETHUNK
_kernelConfig__bad-y__ MITIGATION_UNRET_ENTRY
_kernelConfig__bad-y__ MITIGATION_CALL_DEPTH_TRACKING
_kernelConfig__bad-y__ MITIGATION_IBPB_ENTRY
_kernelConfig__bad-y__ MITIGATION_IBRS_ENTRY
_kernelConfig__bad-y__ MITIGATION_SRSO
_kernelConfig__bad-y__ MITIGATION_GDS
_kernelConfig__bad-y__ MITIGATION_RFDS
_kernelConfig__bad-y__ MITIGATION_SPECTRE_BHI
_kernelConfig__bad-y__ MITIGATION_MDS
_kernelConfig__bad-y__ MITIGATION_TAA
_kernelConfig__bad-y__ MITIGATION_MMIO_STALE_DATA
_kernelConfig__bad-y__ MITIGATION_L1TF
_kernelConfig__bad-y__ MITIGATION_RETBLEED
_kernelConfig__bad-y__ MITIGATION_SPECTRE_V1
_kernelConfig__bad-y__ MITIGATION_SPECTRE_V2
_kernelConfig__bad-y__ MITIGATION_SRBDS
_kernelConfig__bad-y__ MITIGATION_SSB

_kernelConfig__bad-y__ MITIGATION_SLS

_kernelConfig__bad-y__ CPU_SRSO

_kernelConfig__bad-y__ CONFIG_RETPOLINE
_kernelConfig__bad-y__ CONFIG_PAGE_TABLE_ISOLATION

Expand Down Expand Up @@ -32851,7 +32924,7 @@ _kernelConfig_require-tradeoff-harden() {
#qemuArgs+=(-cpu host,-sgx-provisionkey,-sgx-tokenkey)

_kernelConfig__bad-y__ CONFIG_X86_SGX
_kernelConfig__bad-y__ CONFIG_X86_SGX_kVM
_kernelConfig__bad-y__ CONFIG_X86_SGX_KVM
_kernelConfig__bad-y__ CONFIG_INTEL_TDX_GUEST
_kernelConfig__bad-y__ TDX_GUEST_DRIVER

Expand All @@ -32864,10 +32937,16 @@ _kernelConfig_require-tradeoff-harden() {
#qemuArgs+=(-machine accel=kvm,confidential-guest-support=sev0 -object sev-guest,id=sev0,cbitpos=47,reduced-phys-bits=1 )
# #,policy=0x5


# https://libvirt.org/kbase/launch_security_sev.html
_kernelConfig__bad-y__ CONFIG_KVM_AMD_SEV
_kernelConfig__bad-y__ AMD_MEM_ENCRYPT
_kernelConfig__bad-y__ CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT

_kernelConfig__bad-y__ KVM_SMM


_kernelConfig__bad-y__ RANDOM_KMALLOC_CACHES
}
_kernelConfig_require-tradeoff-harden-compatible() {

Expand Down Expand Up @@ -32926,6 +33005,8 @@ _kernelConfig_require-tradeoff-harden-compatible() {

_kernelConfig__bad-y__ CONFIG_INIT_ON_FREE_DEFAULT_ON
_kernelConfig__bad-y__ CONFIG_ZERO_CALL_USED_REGS

_kernelConfig__bad-y__ CONFIG_INIT_STACK_ALL_ZERO

_kernelConfig__bad-n__ CONFIG_DEVMEM
_kernelConfig__bad-n__ CONFIG_DEVPORT
Expand Down Expand Up @@ -32968,6 +33049,11 @@ _kernelConfig_require-tradeoff-harden-compatible() {

#_kernelConfig_warn-any CONFIG_KFENCE_DEFERRABLE
_kernelConfig_warn-y__ CONFIG_KFENCE_DEFERRABLE


# DUBIOUS . Seems to require a userspace service setting scheduling attributes for processes, and not supported by default.
# WARNING: Definitely much better to disable SMT .
#_kernelConfig__bad-y__ CONFIG_SCHED_CORE
}

# WARNING: ATTENTION: Before moving to tradeoff-harden (compatible), ensure vboxdrv, vboxadd, nvidia, nvidia legacy, kernel modules can be loaded without issues, and also ensure significant performance penalty configuration options are oppositely documented in the tradeoff-perform function .
Expand Down Expand Up @@ -33169,11 +33255,18 @@ _kernelConfig_require-tradeoff-harden-NOTcompatible() {


_kernelConfig_warn-y__ CONFIG_EFI_DISABLE_PCI_DMA


# ATTENTION: In practice, the 'gather_data_sampling=force' command line parameter has been available, through optional "$globalVirtFS"/etc/default/grub.d/01_hardening_ubdist.cfg .
_kernelConfig__bad-y__ CONFIG_GDS_FORCE_MITIGATION



# WARNING: CAUTION: Now obviously this is really incompatible. Do NOT move this to any other function.
_kernelConfig_warn-y__ CONFIG_MODULE_SIG_FORCE

# WARNING: May be untested. Kernel default apparently 'Y'.
_kernelConfig_warn-y__ MODULE_SIG_ALL
}

# ATTENTION: Override with 'ops.sh' or similar.
Expand Down Expand Up @@ -33264,6 +33357,10 @@ _kernelConfig_require-virtualization-accessory() {
#_kernelConfig_warn-n__ CONFIG_XEN_SELFBALLOONING
#_kernelConfig_warn-n__ CONFIG_IOMMU_DEFAULT_PASSTHROUGH
#_kernelConfig_warn-n__ CONFIG_INTEL_IOMMU_DEFAULT_ON


# TODO: Evaluate.
_kernelConfig_warn-y__ KVM_HYPERV
}

# https://wiki.gentoo.org/wiki/VirtualBox
Expand Down Expand Up @@ -33478,6 +33575,13 @@ _kernelConfig_require-accessory() {
#PCIE_BW
#ACRN_GUEST
#XILINX SDFEC

# FB_NVIDIA , FB_RIVA , at best, has not been reccently tested with NOUVEAU or other NVIDIA drivers.
_kernelConfig_warn-n__ FB_NVIDIA
_kernelConfig_warn-n__ FB_RIVA



}

_kernelConfig_require-build() {
Expand Down Expand Up @@ -33538,6 +33642,14 @@ _kernelConfig_require-latency() {
_kernelConfig__bad-y__ CONFIG_CPU_FREQ_GOV_ONDEMAND
_kernelConfig__bad-y__ CPU_FREQ_DEFAULT_GOV_SCHEDUTIL
_kernelConfig__bad-y__ CONFIG_CPU_FREQ_GOV_SCHEDUTIL

# WARNING: May be untested.
#X86_AMD_PSTATE_DEFAULT_MODE
if ! cat "$kernelConfig_file" | _kernelConfig_reject-comments | grep "X86_AMD_PSTATE_DEFAULT_MODE"'\=3' > /dev/null 2>&1
then
_messagePlain_bad 'bad: not: 3: '"X86_AMD_PSTATE_DEFAULT_MODE"
export kernelConfig_bad='true'
fi

# CRITICAL!
# CONFIG_PREEMPT is significantly more stable and compatible with third party (eg. VirtualBox) modules.
Expand Down Expand Up @@ -33628,6 +33740,9 @@ _kernelConfig_require-latency() {
# CRITICAL!
# Lightweight kernel compression theoretically may significantly accelerate startup from slow disks.
_kernelConfig__bad-y__ CONFIG_KERNEL_LZ4

# TODO
#PCP_BATCH_SCALE_MAX

}

Expand Down Expand Up @@ -33757,8 +33872,110 @@ _kernelConfig_require-special() {
_kernelConfig__bad-y__ CONFIG_HW_RANDOM_VIA
_kernelConfig__bad-y_m HW_RANDOM_VIRTIO
_kernelConfig__bad-y__ CONFIG_HW_RANDOM_TPM


# Somewhat unusually, without known loss of performance.
# Discovered during 'make oldconfig' of 'Linux 6.12.1' from then existing 'mainline' config file.
_kernelConfig__bad-y__ X86_FRED

_kernelConfig__bad-y__ SLAB_BUCKETS



# TODO: Disabled presently (because this feature is in development and does not yet work), but seems like something to enable eventually.
# _kernelConfig__bad-y__ KVM_SW_PROTECTED_VM


# Usually a bad idea, since BTRFS filesystem compression, etc, should take care of this better.
_kernelConfig__bad-n__ MODULE_COMPRESS

# TODO: Expected unhelpful, but worth considering.
#ZSWAP_SHRINKER_DEFAULT_ON


# Unusual tradeoff. Theoretically may cause issues for Gentoo doing fsck on read-only root (due to not necessarily having initramfs).
_kernelConfig__bad-y__ BLK_DEV_WRITE_MOUNTED
_kernelConfig_warn-n__ BLK_DEV_WRITE_MOUNTED

# If there is no compatibility issue, then the more compressible zswap allocator seems more useful.
#_kernelConfig__warn-y__ ZSWAP_ZPOOL_DEFAULT_ZSMALLOC


# DANGER
# If you honestly believe Meta cares about end-user security...
# https://studio.youtube.com/video/MeUvSg9zQYc/edit
# https://studio.youtube.com/video/kXrLujzPm_4/edit
# There is just NO GOOD REASON to use or support Meta hardware. At all.
_kernelConfig__bad-n__ NET_VENDOR_META

# DANGER
# Although disabling kernel support is NEVER guaranteed to eliminate a 'BadUSB' style vulnerability, reducing this functionality is still very strongly recommended.
#
# SDIO . Especially useless, very few very old devices are expected to benefit from SDIO WiFi, etc, peripherials, while SDIO degrades one of the very few otherwise storage exclusive protocols (ie. SD card storage) into a 'BadUSB' input.
_kernelConfig__bad-n__ ATH10K_SDIO
_kernelConfig__bad-n__ ATH6KL_SDIO
_kernelConfig__bad-n__ B43_SDIO
_kernelConfig__bad-n__ BRCMFMAC_SDIO
_kernelConfig__bad-n__ BT_HCIBTSDIO
_kernelConfig__bad-n__ BT_MRVL_SDIO
_kernelConfig__bad-n__ BT_MTKSDIO
_kernelConfig__bad-n__ CW1200_WLAN_SDIO
_kernelConfig__bad-n__ GREYBUS_SDIO
_kernelConfig__bad-n__ LIBERTAS_SDIO
#
_kernelConfig__bad-n__ MMC_MESON_MX_SDIO # Disabled by default apparently.
_kernelConfig__bad-n__ MMC_MVSDIO # Disabled by default apparently.
#
#_kernelConfig__bad-n__ MT7663_USB_SDIO_COMMON
#
_kernelConfig__bad-n__ MT76_SDIO
_kernelConfig__bad-n__ MWIFIEX_SDIO
_kernelConfig__bad-n__ RSI_SDIO
_kernelConfig__bad-n__ RTW88_SDIO
#
_kernelConfig__bad-n__ SDIO_UART
#
_kernelConfig__bad-n__ SMS_SDIO_DRV
#
_kernelConfig__bad-n__ SSB_SDIOHOST
_kernelConfig__bad-n__ SSB_SDIOHOST_POSSIBLE
#
_kernelConfig__bad-n__ WILC1000_SDIO
_kernelConfig__bad-n__ WL1251_SDIO
_kernelConfig__bad-n__ WLCORE_SDIO

_kernelConfig__bad-n__ RTW88_8822BS
_kernelConfig__bad-n__ RTW88_8822CS
_kernelConfig__bad-n__ RTW88_8723DS
_kernelConfig__bad-n__ RTW88_8723CS
_kernelConfig__bad-n__ RTW88_8821CS




# Requires compiling binaries to support this. Future Debian security updates may use this.
_kernelConfig__bad-y__ X86_USER_SHADOW_STACK



_kernelConfig__bad-y_m USB_GADGET

# ATTENTION: Only drivers that are highly likely to cripple the 'out-of-box-experience' to the point of being unable to perform gParted, revert, basic web browsing, etc, for relatively useful laptops/tablets/etc .
# Essential drivers (eg. iGPU, or at least basic 'VGA', keyboard, USB, etc) are normally included already Debian's default kernel config, if that is used as a starting point.
# WARNING: Delegating which drivers to enable to upstream default Debian (or other distro) config files may be better for reliability, etc.
_kernelConfig_warn-y_m ATH12K #WiFi7
_kernelConfig_warn-y_m MT7996E #WiFi7 Concurrent Tri-Band
_kernelConfig_warn-y_m RTW88_8822BU #WiFi USB
_kernelConfig_warn-y_m RTW88_8822CU
_kernelConfig_warn-y_m RTW88_8723DU
_kernelConfig_warn-y_m RTW88_8821CE
_kernelConfig_warn-y_m RTW88_8821CU
_kernelConfig_warn-y_m RTW89_8851BE
_kernelConfig_warn-y_m RTW89_8852AE
_kernelConfig_warn-y_m RTW89_8852BE



true
}

Expand Down

0 comments on commit 7322a96

Please sign in to comment.