CVE-2026-89483

In the Linux kernel, the following vulnerability has been resolved:

nvme: zero the discard fallback page

nvme setup discard() always maps sizeof(struct nvme dsm range) * NVME DSM MAX RANGES = 4096 bytes as the DSM payload however many ranges the command declares, because some devices ignore the 'Number of Ranges' field - the Fixes: commit records two that read past the declared ranges. A single-range discard fills only the first 16 bytes.

Normally the buffer comes from kzalloc() and the other 4080 bytes are zero. When that allocation fails the code falls back to the per-controller ctrl->discard page, which nvme init ctrl() obtains with alloc page(GFP KERNEL) and nothing ever zeroes, so those 4080 bytes are whatever the page last held and are handed to the controller. Reaching it requires the kzalloc(GFP ATOMIC | GFP NOWARN) to fail, that is memory pressure; it is not remotely triggerable. Failing the allocation under KMSAN reproduces it, with the leaked tail full of vmemmap struct page pointers. The extent in the report is a partial transfer of the payload, not the whole 4096 bytes; the 16-byte boundary in it is the one declared range:

[ 11.991601] BUG: KMSAN: uninit-value in dma map phys+0x14c8/0x1900 [ 11.991969] dma map phys+0x14c8/0x1900 [ 11.992220] dma map page attrs+0xcf/0x130 [ 11.992485] e1000 xmit frame+0x4099/0x6d10 [ 11.992768] dev hard start xmit+0x22f/0xa80 [ 11.993068] sch direct xmit+0x35c/0xcb0 [ 11.993315] dev queue xmit+0x1ee5/0x5eb0 [ 11.993608] ip finish output2+0x1903/0x1c30 [ 11.993881] ip finish output+0x288/0x870 [ 11.994125] ip output+0x15e/0x400 [ 11.994365] ip queue xmit+0x1e85/0x1fb0 [ 11.994639] ip queue xmit+0x60/0x80 [ 11.994899] tcp transmit skb+0x4e71/0x5fa0 [ 11.995210] tcp write xmit+0x3a36/0x9160 [ 11.995533] tcp push pending frames+0xc5/0x3c0 [ 11.995854] tcp push+0x7dc/0x840 [ 11.996076] tcp sendmsg locked+0x766c/0x8400 [ 11.996371] tcp sendmsg+0x4b/0x90 [ 11.996572] inet sendmsg+0x134/0x2a0 [ 11.996823] sock sendmsg+0x265/0x360 [ 11.997076] sock sendmsg+0x100/0x1e0 [ 11.997293] nvme tcp try send+0x196f/0x6370 [ 11.997605] nvme tcp queue rq+0x1d54/0x20b0 [ 11.997882] blk mq dispatch rq list+0x5ee/0x2e50 [ 11.998175] blk mq sched dispatch requests+0x16dc/0x24a0 [ 11.998539] blk mq sched dispatch requests+0x11b/0x2c0 [ 11.998865] blk mq run work fn+0x13b/0x280 [ 11.999146] process scheduled works+0x966/0x1ad0 [ 11.999465] worker thread+0xe44/0x1480 [ 11.999709] kthread+0x53b/0x600 [ 11.999927] ret from fork+0x29f/0x7c0 [ 12.000191] ret from fork asm+0x1a/0x30 [ 12.000460] [ 12.000558] Uninit was created at: [ 12.000788] alloc frozen pages noprof+0x8bf/0xd30 [ 12.001096] alloc pages mpol+0x1d0/0x5f0 [ 12.001326] alloc pages noprof+0x102/0x290 [ 12.001627] nvme init ctrl+0x5a3/0x9f0 [ 12.001891] nvme tcp create ctrl+0xd75/0x19b0 [ 12.002170] nvmf dev write+0x4c68/0x4fd0 [ 12.002426] vfs write+0x587/0x1a10 [ 12.002636] x64 sys write+0x207/0x4f0 [ 12.002874] x64 sys call+0x2ff0/0x3ea0 [ 12.003123] do syscall 64+0x147/0x3b0 [ 12.003400] entry SYSCALL 64 after hwframe+0x77/0x7f [ 12.003680] [ 12.003777] Bytes 16-2843 of 2844 are uninitialized [ 12.004068] Memory access of size 2844 starts at ffff888109f82000 [ 12.004412] [ 12.004530] CPU: 0 UID: 0 PID: 101 Comm: kworker/0:1H Not tainted 7.2.0-rc5-NVMECTL-gf5098b6bae76 #1 PREEMPT(lazy) [ 12.005127] Hardware name: QEMU Ubuntu 24.04 PC v2 (i440FX + PIIX, arch caps fix, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 [ 12.005762] Workqueue: kblockd blk mq run work fn [ 12.006073] =====================================================

Allocate the page with GFP ZERO. The single allocation site covers every use of it: bytes no discard has written stay zero, and bytes one did write hold that controller's own range list, which it has already been sent.