www/forgejo16: Deprecate
This is no longer supported. Users who want bleeding-edge Forgejo
should switch to www/forgejo17 if they haven't already.
MFH: 2026Q4
(cherry picked from commit 9ec40f3cd30755a589827ad9eea2a391f77c9308)
www/forgejo16: Deprecate
This is no longer supported. Users who want bleeding-edge Forgejo
should switch to www/forgejo17 if they haven't already.
MFH: 2026Q4
Revert "vm_object: Fix synchronization with teardown in the object list loop"
This was pushed by accident.
This reverts commit 4e6e4fa8be9940e711b791073eee5bcfa77ec5d4.
vm_object: Fix synchronization with teardown in the object list loop
Checking for OBJT_DEAD is racy since pagers update the object type
without the object lock held. Check for the flag as well, as that is
set beforehand. Otherwise pager dealloc can race with, e.g.,
cdev_pager_get_path().
Try to minimize accesses of the object after dropping the object lock in
vm_object_list_handler().
gve: Fix return value in gve_tx_has_desc_room_dqo()
When the cached tx->dqo.desc_head does not show enough available
descriptor ring slots, gve_tx_has_desc_room_dqo() reloads desc_head from
tx->dqo.hw_tx_head and re-checks availability. If the refreshed
desc_head provides sufficient space, the function mistakenly returns 0
(false) instead of true, causing callers to treat a successful
descriptor head refresh as ENOBUFS, stop the TX queue, and retry or
abort packet transmission (including unloading already-mapped DMA
segments) unnecessarily.
Return true when sufficient descriptor space is available after
refreshing desc_head.
In 60s iperf benchmarks between two FreeBSD 14.4 c4-standard-192
instances, this eliminates spurious tx->stopped transitions across all
workloads:
- TCP 1-stream: 54,859 -> 13 (-99.98%)
- TCP 16-stream: 9,623,851 -> 6,324,299 (-34.3%, -3.30M spurious stops)
[9 lines not shown]
gve: Fix DQO TX free-list memory ordering and pending_pkt state
The DQO TX path uses lockless producer/consumer free lists
(free_pending_pkts_prd/csm and free_qpl_bufs_prd/csm) shared between
the completion and transmit taskqueues. On weakly ordered architectures
such as arm64, the current synchronization has three issues:
1. gve_free_pending_packet() updates pending_pkt state and links
pending_pkt->next before pushing the entry onto
free_pending_pkts_prd using a relaxed atomic_cmpset_32(), while
gve_alloc_pending_packet() steals free_pending_pkts_prd using a
relaxed atomic_swap_32(). Without release/acquire ordering, the
transmit thread can observe a stale pending_pkt->next pointer after
stealing the producer list (either orphaning freed entries or
re-allocating entries whose completion tags are still in flight), or
prior completion stores (such as setting state to FREE or clearing
mbuf) can become visible after the slot is re-allocated.
2. Similarly, in QPL mode, gve_tx_alloc_qpl_buf() steals
free_qpl_bufs_prd with a relaxed atomic_swap_32() without an acquire
[53 lines not shown]