LATX, fix: Release SMC backing after the last mapping disappears

[LaurenIsACoder]

Scope

Based on upstream bfe17ae7c1. Replace the append-only, process-global SMC memfd allocator with one backing file per converted host page. Close the descriptor after creating the guest and shadow mappings; the kernel retains the backing while any mapping, including a fork descendant's mapping, still owns it.

This prevents unmapped pages from retaining their contents in an ever-growing open memfd. It does not punch holes or truncate backing that a child may still use. Preparatory memfd/ftruncate/shadow-mmap failures clean up and use the existing fallback; guest MAP_FIXED failure remains fatal because the original mapping may already have been removed. Handle fd 0 and do not retry Linux close(EINTR).

The default LATX_SMC=0 is unchanged. Per-page memfd/ftruncate/close operations have an unmeasured throughput cost when this optional shared-backing mode is enabled; this is a resource-lifetime fix, not a speedup claim.

Author self-review and combined validation

The actual published diffs were reviewed individually and checked against the local candidates. Follow-up corrections were folded into their existing commits; author/sign-off metadata updates preserved the per-PR source trees. The series remains six commits across #472, #473, #474, #475 and #476.

All five branches merge without conflicts on bfe17ae7c1. The combined source tree 80aaaaf13688387640f528d1c0bccb6a800f8d6e builds and passes 33/33 lat-pr-fast tests. On that combined binary, the 500-cycle KZT unload and retained-handle checks, 520-cycle SMC/fork checks, and 512 successful TSYNC operations alongside thread creation also complete successfully. A tests-disabled configuration registers no tests.

The integration suite still reports 27 skips, including after selecting the available guest sysroot with LATX_X86_64_SYSROOT: 25 require clang and two require the x86 cross compiler. These are not passes. This author self-review does not substitute for the upstream CI matrix, maintainer review or real-application acceptance.

Validation

LoongArch64, GCC 15.3, x86_64-linux-user, LAT O1, KZT compiled in, release:

./configure --target-list=x86_64-linux-user --enable-latx --optimize-O1 \
  --extra-ldflags=-ldl --enable-kzt --enable-tests --disable-docs \
  --disable-werror --meson=meson
ninja -C build -j2 all meson-test-prereq
env -u LAT_LD_PREFIX meson test -C build --no-rebuild \
  --suite lat-pr-fast --print-errorlogs --num-processes 2
env -u LAT_LD_PREFIX meson test -C build --no-rebuild \
  --suite latx-integration --print-errorlogs --num-processes 2

• Isolated PR branch: 25/25 lat-pr-fast passed.
• Maintained production-body regression uses real memfd/mmap/fork for 1,024 lifecycles, alias permissions, parent-unmap/child-use, fd 0, close(EINTR) and allocation/fatal-map failure cleanup. The original allocator fails the no-retained-fd check. ASan/UBSan runs pass in debug and NDEBUG configurations.
• Actual x86 guest code performed 520 map/execute/self-modify/unmap cycles, checking results 1 then 2 with LATX_AOT=0, LATX_KZT=0, LATX_SMC=2. The final live sample has two SMC mappings and no retained SMC descriptor; after unmap it has zero mappings. After fork and parent unmap, the child still executes result 2, writes the code and executes result 3, then exits successfully. The LATX_SMC=0 control also succeeds.
• Integration suite: 27 skipped (exit 77) because the configured guest-building environment is unavailable (missing clang and the suite's default guest sysroot). Skips are not passes; broader guest compatibility and enabled-mode performance remain gates.

Independent of the other memory-lifetime PRs. No installed runtime replacement or application restart.