Sync

amdvlk-opt:
  > Add 'glslang' and 'ocaml-stdlib-shims' to makedepends
  > Merge pull request #3 from rialpa84/master

community-patches:
  > wine-tkg-git: Update wine_wayland_driver patchset - https://gitlab.collabora.com/alf/wine/-/tree/wayland
  > wine-tkg-git: move "MR" LFH patchset to its own and add back proton versions for staging and staging+winesync
  > linux515-tkg: Fixup for AMD_CPPC on 5.15.88
  > linux61-tkg: Add BBRv2 TCP IPv4 algorithm patch - https://github.com/google/bbr/tree/v2alpha
  > wine-tkg-git: Update LFH patchset for 8.0 RCs - https://gitlab.winehq.org/wine/wine/-/merge_requests/1628
  > linux61-tkg: Add x86_split_lock-Add_sysctl_to_control_the_misery_mode.mypatch

gamescope-git:
  > Fix build in clean chroots by adding deps (#8)
  > Don't attempt reloading the main .cfg when moving away
  > arch-meson frogs us over so use meson directly, and add vkroots to the fallbacks
  > Fixups

linux-tkg:
  > linux 6.1.y: misc_additions: Remove `Revert "drm/display/dp_mst: Move all payload info into the atomic state"`
  > cfg: Enable _futex_waitv by default and move all the legacy or deprecated options to a "legacy options" section.
  > Revert "linux 6.1.y: Update Project C patchset to v6.1-r3 - https://gitlab.com/alfredchen/linux-prjc/-/commits/linux-6.1.y-prjc"
  > linux 6.1.y: NR_CPUS is buggy on 6.1, let's ignore the setting with a warning
  > linux 6.2 rc: Import and allow Project C CPU schedulers
  > linux 6.2 rc: misc_additions: Get rid of merged patch "drm/i915: improve the catch-all evict to handle lock contention"
  > linux 6.1.y: Update Project C patchset to v6.1-r3 - https://gitlab.com/alfredchen/linux-prjc/-/commits/linux-6.1.y-prjc
  > linux 6.1.y: misc_additions: Remove merged patch `netfilter: nft_payload: incorrect arithmetics when fetching VLAN header bits`
  > Typo
  > Don't rewrite _kernel_work_folder_abs and _kernel_source_folder_abs values when they are already available
  > cfg: Add tt and bore to the cpu schedulers options comment
  > readme: Add TT to the cpu schedulers list
  > linux 6.1.y: misc_additions: Add `revert "drm/display/dp_mst: Move all payload info into the atomic state"` and `netfilter: nft_payload: incorrect arithmetics when fetching VLAN header bits`
  > CI: Make sure we're using strings for _kernel_work_folder and _kernel_source_folder
  > Always nuke remnant customization.cfg
  > Write _kernel_work_folder_abs and _kernel_source_folder_abs to BIG_UGLY_FROGMINER
  > Get rid of _srcpath
  > Drop broken Void Linux explicit support. Void Linux users should use the Generic install as a replacement.
  > readme: Warn about the disk space requirements to build the kernel
  > README: Add BORE to alternative schedulers (#686)
  > linux 6.1.y: Update bcachefs patch for 6.1.4
  > Apply glitched-cfs patch to all the CFS schedulers
  > CI: Use "/tmp" as work and source dir
  > Always create a subfolder for source and work directories with hardcoded names (linux-src-git and linux-kernel.git for source and work dirs respectively). If an absolute path is used, create a linux-tkg subdir to contain those. This allows using the same root path for both as well as defaulting to an empty value that mimics our previous behavior. The main motivation for this is to offer more safety as using an existing non-empty dir might result in unwanted behavior.
  > makepkg: Fix not dynamically setting our source symlink, which is needed as of a38ef1643cee7a4041494bbcc2474a9e1824f1ca
  > linux 6.1.y: misc_additions: Remove patch `drm/i915: improve the catch-all evict to handle lock contention`
  > Various fixes for the RPM target (#684)
  > linux 6.1.y: Update prjc to v6.1-r1 (official release) with the pending fix for boot failure - https://gitlab.com/alfredchen/linux-prjc/-/issues/68#note_1226803051
  > Improve customization of source and work folders (#672)
  > linux 5.15.y: Update fsync backport for new futex dir hierarchy
  > linux 5.15.y: Update compatibility fix for old Fsync ABI (#680)
  > Remove futex patch, it got merged into upstream 6.1.2 (#677)
  > linux 5.15.y: futex_waitv fixup (#676)
  > Add initial 6.2-rc support (#673)
  > Add bcachefs for 6.1 (#670)
  > Only set custom command line options if the variable is not empty
  > Remove NR_CPUS prompt and use defconfig's default if the value is empty instead of an enforced 128.
  > linux 6.1.y: Add `futex: Resend potentially swallowed owner death notification` to misc additions.
  > Misc fixes  - Enforce sane defaults for ticks selector  - Make sure we have a default governor set early on (schedutil)  - Build BLK_DEV_LOOP as module instead of builtin
  > linux 6.1.y: Update our unprivileged CLONE_NEWUSER patch to current
  > Remove useless dir deletion. We're now checking that it doesn't exist, making it redundant.
  > Handle existing source dir in tmpfs - That's actually safe to do thanks to @AdelKS 's fine work.
  > Add an option to clone the source in tmpfs everytime instead of using the root PKGBUILD dir. This effectively brings back a similar approach to makepkg in the way of handling git sources. It'll require ~20GB of space by itself, so if your tmpfs is in RAM, make sure you have enough free, and it's definitely not recommended to use this on a <32GB RAM machine.
  > Refresh RT patches mapping
  > linux 5.15.y: prjc fixup following https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/kernel/sched/core.c?h=v5.15.83&id=f435c66d2336fff104e2c3cbfff60905733e7d89
  > linux 6.1.y: misc_additions: Get rid of `Bluetooth: fix deadlock for RFCOMM sk state change` and `x86/split_lock: Add sysctl to control the misery mode`. The latter will be added to community patches instead and we'll now pass `split_lock_detect=off` by default to the command line. We'll see how the situation pans out and adapt as needed.
  > linux 6.1.y: base: Remove our `watermark_scale_factor` patch, which doesn't feel needed anymore, and add a selection of patches from Zen: `Input: evdev - use call_rcu when detaching client`, `mm: Lower the non-hugetlbpage pageblock size to reduce scheduling delays` and `mm: Don't hog the CPU and zone lock in rmqueue_bulk()`.
  > linux 6.1: Import 6.0 based prjc patchset from ptr1337 until we can find why current upstream is frogged. Enable back pds/bmq and set as unofficial.
  > Import 'file' invocation fix from Arch

mesa-git:
  > Update customization.cfg (#29)
  > android-libbacktrace was enabled with 6291d4d but only affects android. Let's disable.
  > cfg: Add a note regarding dri drivers deprecation following 911f4b457755133c7c10e8e57132aed08f6a53f8
  > Fix PKGBUILD for dri-drivers removal (#28)
  > Move back to using gitlab as default source. The github mirror has been out of date for a couple weeks with no resolution in sight yet.

nvidia-all:
  > 525.47.06
  > Disable kernel 6.2 patch on 525.8*+
  > 525.85.05 + bit (#144)
  > Remove explicit examples for version number and use expected formatting instead.
  > Remove some legacy series having valid checks against newer kernels than actually patchable or supported.
  > 525.78.01
  > Update md5sum for patches/kernel-6.2.patch (#140) (#141)
  > Fix 6.2 patchset for nvidia, nvidia-dkms will compile also now for 6.1 or lower version correctly (#139)
  > Add 6.2-rc support (#138)
  > 525.47.04

wine-tkg-git:
  > Skip proton-win10-default patches on 69154f0+ as wine now default's to win10 prefixes
  > Hotfixes: Get rid of our patch to fix staging's patchinstall.py python3 compat
  > Move to using staging's patchinstall.py script instead of patchinstall.sh which is considered deprecated now.
  > fsync: Rename toggle options (#941)
  > Fixup for 69e1ed9fba963a45fe7a7a847ad5394bfbee929e
  > Workaround for Fedora freetype2 libs not being detected now that it's been moved to a subdir
  > Hotfixes: fixup MR 1148 for 4827ae13928259b22e48765be3f14aef5bc26146
  > Rebase shared gpu resource for wine git (#950)
  > hotfixer: early: Fix wrong staging patches application order on 89aa62e ("ntdll-WRITECOPY" needs to be applied before "ntdll-Placeholders").
  > Reverse logic for some _use_vkd3dlib checks, to properly handle empty values.
  > Move futex2 patches to legacy, and mark it as deprecated in comments
  > proton: disable _proton_nvapi_disable by default.
  > cfg: Move deprecated/legacy options to a legacy config file.
  > Get rid of FAudio hard depencency
  > makepkg: Only get rid of src early if it's a symlink
  > proton: uninstaller: On fresh builds, don't include the new build in the array of removable builds.
  > Switch our _NOLIB32 option that used to simply disable 32-bit support altogether (making worthless builds) to using Wine 8.0+ WoW64 (running 32-bit apps through 64-bit). This is still WIP, so disable by default.
  > deps: Split dependency resolution arch for fedora and archlinux to prepare for WoW64
  > proton: gstreamer: Set orc-source to auto (due to https://gitlab.freedesktop.org/gstreamer/gstreamer/-/commit/6caa8524dd9a5fd1763b78f1745f08644267d889) and disable lib32 stuff when _NOLIB32 is enabled
  > CI: Disable Proton nopackage Ubuntu autobuilds
  > deps: debuntu: CI doesn't prompt. Let's run update-alternatives as root.
  > deps: debuntu: Attempt to use the --set argument for update-alternatives, which should be available on Ubuntu at least. This allows for automation, where --config is interactive.
  > deps: debuntu: Run update-alternatives on mingw while we're at it
  > Hotfix for shared gpu resources (#945)
  > CI: proton-ubuntu: Use buntu 22.04
  > proton: makepkg: Tweak the chroot workaround so it's *a bit* less hacky, and hopefully a bit more robust
  > profiles: Set _NOINITIALPROMPT="true" in Valve profiles
  > proton: makepkg: The pkgver limitation is due to our -git pkgname. Let's get rid of that as we did with wine-tkg, and let's use full "as desired" pkgver.
  > makepkg: Nuke the "$_where"/src dir/symlink before makepkg creates it.
  > Rebase sharedgpuresources patchset  for Wine 8.0 (#861)
  > proton: makepkg: Use a hacky loop to properly populate pkgver through wine-tkg without having the user prompted twice.
  > proton: makepkg: Mention the path if not supported nor maintained - anymore
  > proton: readme: Remove mentions of the makepkg way
  > proton: makepkg: Let's be more annoying
  > proton-tkg: makepkg: HACK: Get rid of the "overcomplicated BS" along with pkgver() and set pkgver's value later on through the proton-tkg.sh script instead.
  > non-makepkg: Output the mismatching remotes details and restart _nomakepkgsrcinit on failure instead of exiting
  > profiles: chaotic-staging: Use `0002-proton_LFH-staging` community patch following renaming
  > non-makepkg: Nuke wine and staging source dirs on remote mismatch
  > cfg: chaotic: Add 0002-proton_LFH.mypatch back to the _community_patches array.. Until it breaks again.
  > fastsync: Add some "QoL things" for testers (#935)
  > Disable _clock_monotonic by default, and enforce it to disabled on 8.0+ trees
  > pefixup: Ignore *.conf files
  > Hotfixes: autoconf-opencl-hotfix: Fixup for legacy trees older than 34289e7 (<6.23)
  > Use `patchinstall.sh --upstream-commit` on older staging trees which are missing the `"staging/upstream-commit` file.
  > Separate staging dirs from different sources
  > Hotfixes: Add NosTale mouse fix - PSzczepanski1996/proton-ge-nostale@a4078f0
  > Always define mainline and staging sources. Custom source can overwrite set values as needed later on.
  > Proton: Fixup for custom d3d9 support
  > Fix wrong remote comparison and always pass a default _stgsrctarget
  > Rename _plain_mirrorsrc to _github_mirrorsrc and make it affect the staging repo used along mainline. Set it to false by default, effectively switching to gitlab repos by default.
  > Valve: Disable wined3d-zero-inf-shaders staging patch
  > Valve: Move the noswap patch handling to valve hotfixes and add a specific patch for experimental+
  > Hotfixes: rdr2: Update revert list