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

forcedeth: fix off-by-one when saving/restoring non-PCI config space

nvsuspend() and nvresume() walk the non-PCI configuration space with

for (i = 0; i <= np->registersize/sizeof(u32); i++)

which runs one iteration too many. savedconfigspace is declared as

u32 savedconfigspace[NVPCIREGSZMAX/4];

and NVPCIREGSZVER3 is equal to NVPCIREGSZMAX (0x604), so on a VER3 device registersize/sizeof(u32) is exactly the array length and the last iteration addresses one element past the end.

The element it lands on is np->namerx[0..3]: savedconfigspace[] is followed immediately by char namerx[IFNAMSIZ + 3], and char needs no padding. Nothing observable is corrupted by that, because nvrequestirq() rewrites namerx with sprintf() before it is ever passed to requestirq(). The bug is the out-of-bounds access itself, which UBSAN reports and which CONFIGUBSANTRAP=y turns into a trap that aborts the running kernel code, plus an MMIO read and, on resume, an MMIO writel() to base + 0x604, one dword past the range the driver mapped:

np->base = ioremap(addr, np->registersize);

VER1 and VER2 devices stay inside the array, but they too get the stray read and the stray write one dword past their own window.

Caught by UBSAN on an Apple Macmini3,1 (MCP79) during a deep S3 cycle. The splat below is trimmed: the build path in the file name, the CPU and taint lines, the Workqueue line, the "?" hint frames, and the frames below devicesuspend are all cut. The kernel was tainted, with an out-of-tree nouveau and CPUOUTOFSPEC; forcedeth itself was the stock module.

UBSAN: array-index-out-of-bounds in drivers/net/ethernet/nvidia/forcedeth.c:6225:25 index 385 is out of range for type 'u32 [385]' Call Trace: dumpstacklvl+0x5d/0x80 ubsanepilogue+0x5/0x2b ubsanhandleoutofbounds.cold+0x54/0x59 thismodule+0xe398c/0xe9010 [forcedeth] pcipmsuspend+0x80/0x170 dpmruncallback+0x51/0x160 devicesuspend+0x1a2/0x4a0 ...

Both loops are hit. UBSAN reports each source location only once per module load (ubsanhandleoutofbounds() calls suppressreport(), which does testandsetbit(REPORTEDBIT, ...) on the struct sourcelocation), so the two splats land in the first S3 cycle after the module is loaded and later cycles are silent even though the access still runs off the end every time. In that first cycle line 6225 is reported from pcipmsuspend and line 6240 from pcipmresume.

The same off-by-one was fixed in nvgetregs() by commit ba9aa134287f ("forcedeth: fix buffer overflow") in 2012; these two loops were missed. The suspend and resume side was reported on LKML in September 2013 by Marc Weber, with the same analysis and the same one-character fix, but the patch was attached rather than sent inline and the thread ended there.

Use < instead of <=, which saves and restores exactly registersize bytes.