When shutting down with redzone enabled, I get the following: REDZONE: Buffer overflow detected. 16 bytes corrupted after 0xfffffe0091e67688 (6710920 bytes allocated). Allocation backtrace: #0 0xffffffff80d66c09 at redzone_setup+0xe9 #1 0xffffffff80a35323 at malloc+0x273 #2 0xffffffff80a8c967 at blist_create+0x97 #3 0xffffffff80d6bcd7 at swaponsomething+0xa7 #4 0xffffffff80d6a3d3 at sys_swapon+0x413 #5 0xffffffff80f06ceb at amd64_syscall+0x79b #6 0xffffffff80ee58eb at Xfast_syscall+0xfb Free backtrace: #0 0xffffffff80d66f74 at redzone_check+0x304 #1 0xffffffff80a353f4 at free+0x54 #2 0xffffffff80d6ac2d at swapoff_one+0x61d #3 0xffffffff80d6adb7 at swapoff_all+0xd7 #4 0xffffffff80b05c18 at bufshutdown+0x398 #5 0xffffffff80a5a259 at kern_reboot+0x1d9 #6 0xffffffff80a5a024 at sys_reboot+0x3c4 #7 0xffffffff80f06ceb at amd64_syscall+0x79b #8 0xffffffff80ee58eb at Xfast_syscall+0xfb Uptime: 1h29m58s This is with -CURRENT as of today, zfsroot with a separate swap partition that's considerably larger than current RAM (approx 6x as large). 16GB of RAM, 96GB of swap.
Possibly related to bug 224485, also filed today.
This is the same bug as PR 231116: for an allocation of 6710920 bytes we get 6710896 / 16 = 419431 nodes, which is the number of nodes needed to map 33554432 = 2^24 + 2^23 blocks.
*** This bug has been marked as a duplicate of bug 231116 ***