LLVM/project e24f712lldb/source/Plugins/Process/Linux NativeRegisterContextLinux_arm64.h NativeRegisterContextLinux_arm64.cpp

[lldb][AArch64][Linux] Add function to get regset dependencies (#212476)

This function encodes relations between sets so that you 
don't have to remember all the dependencies.

There are some places where we do: 
if some condition:
  Invalidate(some set);

I've chosen not to put that logic in this new function so
that it can be a simple switch case.

The few places we need extra state to make the decision
are also all SVE/SME related. This code will likely remain
bespoke anyway.

Whereas the simple paths will all get collapsed into
a generic handler later.


    [2 lines not shown]
DeltaFile
+38-22lldb/source/Plugins/Process/Linux/NativeRegisterContextLinux_arm64.cpp
+11-1lldb/source/Plugins/Process/Linux/NativeRegisterContextLinux_arm64.h
+49-232 files

LLVM/project a424bbfllvm/lib/Target/ARM ARMAsmPrinter.cpp ARMTargetMachine.cpp, llvm/test/CodeGen/ARM float-abi-module-flag.ll

ARM: Read float ABI from the "float-abi" module flag (#212981)

Use the value from the module flag if present, otherwise
fall back on the legacy TargetOptions field until that is
removed.

Co-authored-by: Claude (Claude-Opus-4.8) <noreply at anthropic.com>
DeltaFile
+43-0llvm/test/CodeGen/ARM/float-abi-module-flag.ll
+10-1llvm/lib/Target/ARM/ARMTargetMachine.cpp
+6-1llvm/lib/Target/ARM/ARMAsmPrinter.cpp
+59-23 files

NetBSD/pkgsrc RBXAs5Ldoc CHANGES-2026

   pdated lang/oracle-jdk2[15]
VersionDeltaFile
1.5217+3-1doc/CHANGES-2026
+3-11 files

NetBSD/pkgsrc 8whOCMGlang/oracle-jdk25 Makefile distinfo

   oracle-jdk25: updated to 25.0.4

   25.0.4
   https://www.oracle.com/java/technologies/javase/25-0-4-relnotes.html
VersionDeltaFile
1.2+13-13lang/oracle-jdk25/distinfo
1.2+2-2lang/oracle-jdk25/Makefile
+15-152 files

NetBSD/pkgsrc UAHLxVAlang/oracle-jdk21 Makefile distinfo

   oracle-jdk21: updated to 21.0.12

   21.0.12
   https://www.oracle.com/java/technologies/javase/21-0-12-relnotes.html
VersionDeltaFile
1.7+13-13lang/oracle-jdk21/distinfo
1.8+2-2lang/oracle-jdk21/Makefile
+15-152 files

LLVM/project 832b5acclang/test/SemaCXX GH106182.cpp concepts-subsumption.cpp

[Clang][test] Relocate concepts regression test to concepts-subsumption.cpp NFC (#215762)
DeltaFile
+0-16clang/test/SemaCXX/GH106182.cpp
+16-0clang/test/SemaCXX/concepts-subsumption.cpp
+16-162 files

NetBSD/pkgsrc-wip 7492e61trivy Makefile go-modules.mk

trivy: updated to 0.73.0
DeltaFile
+753-1,071trivy/distinfo
+250-356trivy/go-modules.mk
+1-1trivy/Makefile
+1,004-1,4283 files

NetBSD/src DAUNAJRsys/dev/sdmmc sdmmc.c sdmmc_mem.c

   sdmmc: dmamap_load bounce buffer on demand

   When using a bounce buffer, cmd->c_dmamap->dm_segs[0]->ds_len did not agree with cmd->c_datalen because the bounce buffer is loaded once at initialization, so cmd->c_dmamap->dm_segs[0]->ds_len never gets updated before executing a sdmmc command.

   This is problematic when the dmamap is passed on to an API like fdt_dma that treats the dmamap as the source of truth.

   This problem probably hasn't surfaced yet because it needs three preconditions:
   - using bounce buffers is already an edge case for when we can't allocate block size aligned dma memory
   - the driver needs CAPS_DMA, but not CAPS_MULTI_SEG_DMA
   - the driver needs to use cmd->c_dmamap->dm_segs[0]->ds_len and not cmd->c_datalen

   The fix is to bus_dmamap_load the bounce buffer just when we need it. This allows us to load it with the correct size.
VersionDeltaFile
1.81+18-2sys/dev/sdmmc/sdmmc_mem.c
1.46+2-10sys/dev/sdmmc/sdmmc.c
+20-122 files

LLVM/project f9d1cdallvm/lib/CodeGen/SelectionDAG LegalizeFloatTypes.cpp, llvm/test/CodeGen/NVPTX fp128-conv-no-libcall-error.ll

DAG: Gracefully diagnose missing FP conversion libcalls when softening

Diagnose the missing libcall and return poison instead.

Co-authored-by: Claude (Claude-Opus-4.8) <noreply at anthropic.com>
DeltaFile
+70-0llvm/test/CodeGen/NVPTX/fp128-conv-no-libcall-error.ll
+46-14llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
+116-142 files

LLVM/project 742031cclang/lib/CIR/Dialect/Transforms LoweringPrepare.cpp, clang/test/CIR/CodeGenCUDA device-stub.cu

[CIR][HIP] Adjust HIP fatbin alignment to 4096 (#215314)
DeltaFile
+7-1clang/lib/CIR/Dialect/Transforms/LoweringPrepare.cpp
+3-3clang/test/CIR/CodeGenCUDA/device-stub.cu
+10-42 files

LLVM/project c003bfbllvm/lib/Target/AMDGPU SIRegisterInfo.cpp, llvm/test/CodeGen/AMDGPU eliminate-frame-index-flat-scratch-frame-reg-no-sgpr.ll

[AMDGPU] Fix eliminateFrameIndex assertion when reusing the frame register. (#215182)

This is a follow-up patch that solves the crashing test in the #215180 

On a flat-scratch target, a frame index used by a VALU instruction with
a live frame register but no scavengeable SGPR entered the branch that
assumes no frame register exists, tripping `assert(!FrameReg && "there
is a frame register!")`.

Restore the `!FrameReg` term in the guard so this case falls through to
the existing path that reuses the frame register as the temporary
(offset folded in, then restored). The `!FrameReg` assertion is moved
into the SVS
fallback, where it actually holds.
DeltaFile
+9-9llvm/test/CodeGen/AMDGPU/eliminate-frame-index-flat-scratch-frame-reg-no-sgpr.ll
+10-2llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
+19-112 files

LLVM/project 848e563llvm/test/CodeGen/ARM float-abi-module-flag.ll

Fix test comment
DeltaFile
+1-1llvm/test/CodeGen/ARM/float-abi-module-flag.ll
+1-11 files

NetBSD/src K6lDL1dsys/arch/arm/ti ti_edma.h ti_sdhc.c

   evbarm/ti_edma: start using fdt_dma

   In preparation for the am18xx sdmmc driver.
VersionDeltaFile
1.7+341-113sys/arch/arm/ti/ti_edma.c
1.13+59-150sys/arch/arm/ti/ti_sdhc.c
1.2+1-36sys/arch/arm/ti/ti_edma.h
+401-2993 files

LLVM/project fbaa112llvm/lib/Target/ARM ARMAsmPrinter.cpp ARMTargetMachine.cpp, llvm/test/CodeGen/ARM float-abi-module-flag.ll

ARM: Read float ABI from the "float-abi" module flag

Use the value from the module flag if present, otherwise
fall back on the legacy TargetOptions field until that is
removed.

Co-authored-by: Claude (Claude-Opus-4.8) <noreply at anthropic.com>
DeltaFile
+43-0llvm/test/CodeGen/ARM/float-abi-module-flag.ll
+10-1llvm/lib/Target/ARM/ARMTargetMachine.cpp
+6-1llvm/lib/Target/ARM/ARMAsmPrinter.cpp
+59-23 files

OPNSense/core 3fa8a4csrc/opnsense/mvc/app/views/OPNsense/Diagnostics fw_log.volt

Firewall: Log Files: Live View - htmlSafe() action search value

PR: GHSA-9h93-hhcx-957c
(cherry picked from commit 14de8b0235246b83a1824ca9a5f052817fbd035b)
DeltaFile
+1-1src/opnsense/mvc/app/views/OPNsense/Diagnostics/fw_log.volt
+1-11 files

OPNSense/core af24a9fsrc/opnsense/mvc/app/models/OPNsense/Interfaces NetworkInterface.xml

Interfaces: Assignments - strict description fields as before in interfaces_assign.php

Earlier code dropped all unallowed fields in https://github.com/opnsense/core/blob/8cc69b21e0f4c2622fc8a62df2a15ba7cb1e731f/src/www/interfaces_assign.php#L121

PR: GHSA-9h93-hhcx-957c
(cherry picked from commit 7f6811fea620c94d6bd13aaa13ecf5df90d939a0)
DeltaFile
+3-1src/opnsense/mvc/app/models/OPNsense/Interfaces/NetworkInterface.xml
+3-11 files

OPNSense/core 8356822src/www system_authservers.php interfaces.php

ui: add some more legacy_html_escape_form_data() safeguards

PR: GHSA-6vv6-fpw9-8r2g
(cherry picked from commit 096071af4be64380a34f396e1718008247c91dc0)
DeltaFile
+6-2src/www/interfaces.php
+4-2src/www/system_authservers.php
+10-42 files

OPNSense/core 4a87f70src/opnsense/mvc/app/controllers/OPNsense/OpenVPN/Api ServiceController.php

VPN: OpenVPN: Connection Status - add some input validation for control characters

PR: GHSA-fqmm-g6xw-qhm4
(cherry picked from commit 024d986d43538b5261413f6cf5b815175e9253b1)
DeltaFile
+4-0src/opnsense/mvc/app/controllers/OPNsense/OpenVPN/Api/ServiceController.php
+4-01 files

OPNSense/core a2708c7src/opnsense/mvc/app/views/OPNsense/Firewall filter_rule.volt alias.volt, src/opnsense/mvc/app/views/OPNsense/IDS policy.volt

mvc: add some missint htmlSafe() calls for generated HTML

PR: GHSA-75fw-j788-j4hj
(cherry picked from commit 5786beb80c5410a170b44ef6f1532ac071e135bd)
DeltaFile
+3-3src/opnsense/mvc/app/views/OPNsense/IDS/policy.volt
+1-1src/opnsense/mvc/app/views/OPNsense/IPsec/sad.volt
+1-1src/opnsense/mvc/app/views/OPNsense/Firewall/filter_rule.volt
+1-1src/opnsense/mvc/app/views/OPNsense/Firewall/alias.volt
+6-64 files

NetBSD/src 9FRb9Cksys/arch/arm/ti files.ti ti_edma.c, sys/arch/evbarm/conf GENERIC_V5

   evbarm/ti_edma: enable on am18xx SoCs

   The linux dt-bindings allow two different ways to specify power/clock domains for the edma3. This commit adds support for the way used on the am18xx (powerdomains).
VersionDeltaFile
1.3+16-7sys/arch/arm/ti/ti_tptc.c
1.6+14-5sys/arch/arm/ti/ti_edma.c
1.33+3-3sys/arch/arm/ti/files.ti
1.11+3-1sys/arch/evbarm/conf/GENERIC_V5
+36-164 files

LLVM/project 11eba5cllvm/test/Transforms/LoopVectorize/AArch64 sve2-histcnt-epilogue.ll interleaving-load-store.ll

[LV] Prefer fixed VFs for epilogue vectorization (#211814)

Unless forced (via `-scalable-vectorization=always`), or if the fixed
vector cost is much higher than scalable vectors (which occurs for some
loops, such as histograms), prefer fixed vectors for the epilogue.

There are currently some hidden costs for scalable epilogues that are
not modeled (such as reduced post-vectorization unrolling).

This restriction likely can be lifted/reworked once we support
predicated vector epilogue.
DeltaFile
+32-42llvm/test/Transforms/LoopVectorize/AArch64/reduction-recurrence-costs-sve.ll
+24-32llvm/test/Transforms/LoopVectorize/AArch64/interleave-with-gaps.ll
+15-22llvm/test/Transforms/LoopVectorize/AArch64/vector-reverse.ll
+15-20llvm/test/Transforms/LoopVectorize/AArch64/induction-costs-sve.ll
+13-19llvm/test/Transforms/LoopVectorize/AArch64/interleaving-load-store.ll
+14-14llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-epilogue.ll
+113-1499 files not shown
+166-19815 files

OPNSense/core 14de8b0src/opnsense/mvc/app/views/OPNsense/Diagnostics fw_log.volt

Firewall: Log Files: Live View - htmlSafe() action search value

PR: GHSA-9h93-hhcx-957c
DeltaFile
+1-1src/opnsense/mvc/app/views/OPNsense/Diagnostics/fw_log.volt
+1-11 files

OPNSense/core 7f6811fsrc/opnsense/mvc/app/models/OPNsense/Interfaces NetworkInterface.xml

Interfaces: Assignments - strict description fields as before in interfaces_assign.php

Earlier code dropped all unallowed fields in https://github.com/opnsense/core/blob/8cc69b21e0f4c2622fc8a62df2a15ba7cb1e731f/src/www/interfaces_assign.php#L121

PR: GHSA-9h93-hhcx-957c
DeltaFile
+3-1src/opnsense/mvc/app/models/OPNsense/Interfaces/NetworkInterface.xml
+3-11 files

NetBSD/src LkH7rp9sys/arch/arm/ti am18xx_psc.c

   evbarm/am18xx: expose the PSC as powerdomain

   Some peripherals in the DTB expect to manage clocks as power domains.
VersionDeltaFile
1.6+27-3sys/arch/arm/ti/am18xx_psc.c
+27-31 files

OPNSense/core 096071asrc/www system_authservers.php interfaces.php

ui: add some more legacy_html_escape_form_data() safeguards

PR: GHSA-6vv6-fpw9-8r2g
DeltaFile
+6-2src/www/interfaces.php
+4-2src/www/system_authservers.php
+10-42 files

OpenBSD/ports hUmiNv1databases/py-sqlalchemy distinfo Makefile

   update to py3-sqlalchemy-2.0.52
VersionDeltaFile
1.90+2-3databases/py-sqlalchemy/Makefile
1.56+2-2databases/py-sqlalchemy/distinfo
+4-52 files

OpenBSD/ports BoLkhLLmail/postfix/stable Makefile distinfo

   MFC update to postfix-3.11.6
VersionDeltaFile
1.164.2.4+2-2mail/postfix/stable/distinfo
1.281.2.5+1-1mail/postfix/stable/Makefile
+3-32 files

OpenBSD/ports WI0jz7xmail/postfix/stable35 Makefile distinfo

   MFC update to postfix-3.5.27
VersionDeltaFile
1.4.10.3+2-2mail/postfix/stable35/distinfo
1.11.2.3+1-1mail/postfix/stable35/Makefile
+3-32 files

FreeBSD/ports 23a9390net-mgmt/netbox pkg-plist distinfo

net-mgmt/netbox: Update to 4.6.8

Changelog:

Performance Improvements:
* Avoid N+1 queries when resolving generic relations (e.g. assigned
  objects) via the GraphQL API
* Improve performance when provisioning new custom fields
* Omit implicit pagination when prefetching to-one relations via
  the GraphQL API
* Improve caching logic when retrieving custom fields via
  get_for_model()

Bug Fixes:
* Clear a device's stale rack assignment when changing its site
* Enforce object permissions on custom script write operations via
  the REST API
* Avoid re-executing the LDAP configuration file on every
  permission check

    [24 lines not shown]
DeltaFile
+9-9net-mgmt/netbox/Makefile
+3-3net-mgmt/netbox/distinfo
+4-0net-mgmt/netbox/pkg-plist
+16-123 files

FreeBSD/ports e77a522www/py-dj60-drf-spectacular-sidecar Makefile distinfo

www/py-dj60-drf-spectacular-sidecar: Update to 2026.8.1

Changelog:

https://github.com/tfranzel/drf-spectacular-sidecar/compare/2026.7.1...2026.8.1

MFH:            2026Q3
(cherry picked from commit be2c64b6f7affbd51c6e4cceabc2b9bc8e378fc9)
DeltaFile
+3-3www/py-dj60-drf-spectacular-sidecar/distinfo
+1-1www/py-dj60-drf-spectacular-sidecar/Makefile
+4-42 files