340.46 on i386: REDZONE: Buffer overflow detected

David Wolfskill of FreeBSD had reported getting the following messages from kernel while running 340.46 version of the driver on his i386 (32-bit) laptop:

Nov 18 09:59:20 localhost kernel: REDZONE: Buffer overflow detected. 16 bytes corrupted after 0xd2375200 (256 bytes allocated).
Nov 18 09:59:20 localhost kernel: Allocation backtrace:
Nov 18 09:59:20 localhost kernel: #0 0xc0d8b3ba at redzone_setup+0xda
Nov 18 09:59:20 localhost kernel: #1 0xc0ab9bd4 at malloc+0x244
Nov 18 09:59:20 localhost kernel: #2 0xc1d2f9d1 at os_alloc_mem+0x21
Nov 18 09:59:20 localhost kernel: #3 0xc1d1cc1e at _nv013362rm+0x12
Nov 18 09:59:20 localhost kernel: Free backtrace:
Nov 18 09:59:20 localhost kernel: #0 0xc0d8b790 at redzone_check+0x310
Nov 18 09:59:20 localhost kernel: #1 0xc0ab9c18 at free+0x38
Nov 18 09:59:20 localhost kernel: #2 0xc1d2f9f9 at os_free_mem+0x19
Nov 18 09:59:20 localhost kernel: #3 0xc195bcc7 at _nv004407rm+0x19d7

Which looks a fair bit as if nvidia.ko is doing something… inadvisable.

Is this a known issue? Was it fixed in later versions (if yes, which ones, and who can confirm that)?