QEMU Meson TCG tests, stable 10.0.13, and PostgreSQL backup fixes

QEMU development led the day with large patch series on test infrastructure, a stable queue, and architecture support, while PostgreSQL saw focused work on backup verification and WAL preallocation. Maintainers and contributors advanced Meson migration, confidential computing, migration performance, and recovery tooling across both projects.

QEMU migrates TCG tests to Meson

A 103-patch v6 series on qemu-devel proposes running QEMU TCG tests with Meson instead of makefiles. The series drew extensive review and Reviewed-by tags from maintainers. The change aligns the TCG test suite with the project's Meson-based build and matters for anyone maintaining or extending QEMU testing.

QEMU 10.0.13 stable patch queue posted

A 102-patch round-up for stable 10.0.13 was posted to qemu-devel, with freeze set for 2026-08-24. The queue includes a fix for CVE-2026-18054 in virtio-gpu plus NVMe and virtio-gpu reset work. Stable-branch users and downstream packagers should track the freeze ahead of the point release.

PostgreSQL fixes truncated compressed backup detection

A patchset on pgsql-hackers fixes pg_verifybackup so it detects truncated zstd, gzip, and lz4 compressed backups. Review and test discussion accompanied the proposal among five participants. Operators who verify compressed backups gain more reliable integrity checks when archives are incomplete.

QEMU RFC for RISC-V CoVE confidential VMs

An RFC series of 17 patches proposes KVM CoVE confidential VM support for RISC-V in QEMU, modeled on SEV and TDX. Two participants discussed the design on qemu-devel. The work would extend confidential computing options for RISC-V guests under KVM.

QEMU RISC-V disassembler reorganization

Richard Henderson posted v2 of a 52-patch series that reorganizes the RISC-V disassembler to use pointers. The series targets the disas/riscv code path. A clearer structure should ease maintenance of RISC-V instruction decoding.

QEMU fast snapshot load with userfaultfd

A v4 series of 11 patches proposes faster VM snapshot resume using userfaultfd for lazy RAM load plus an eager background thread. Three participants reviewed the migration changes on qemu-devel. The approach can cut restore latency for large-memory guests.

QEMU ARM splits max CPU into max-v8 and max-v9

A v2 series of 10 patches introduces cpu types max-v8 and max-v9 on the ARM target to handle Armv9 deprecation rules. Two participants discussed the change. The split helps QEMU track architecture version constraints more accurately.

PostgreSQL proposes pg_wal_preallocate()

A patch on pgsql-hackers adds a superuser SQL function pg_wal_preallocate() to eagerly create future WAL segments. Discussion covered the interface, caps via max_wal_size, and automatic alternatives among six participants. The feature would let administrators prepare WAL capacity ahead of write-heavy workloads.