CVE-2026-89581

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

bpf, x86: Fix per-CPU address resolution into an extended register

The destination of the per-CPU address MOV is encoded in ModRM.reg, which is extended by REX.R, but the REX prefix is built with add 1mod(), which sets REX.B. REX.B extends ModRM.rm and SIB.base, and this instruction addresses memory as disp32 with no base, so the bit has no effect at all and the high register bit is simply lost.

Every is ereg() destination therefore resolves to the wrong register, picking whichever one shares the low three bits:

R5 -> RAX R7 -> RBP R8 -> RSI R9 -> RDI

With BPF REG 5, whose reg2hex is 0, the emitted

65 49 03 04 25 add %gs:,%rax

adds the per-CPU offset to RAX rather than R8. The destination keeps the unadjusted address and RAX is clobbered, so the program goes on to dereference a pointer that was never made per-CPU:

BUG: unable to handle page fault for address: 0000607e386a8894 RIP: bpf prog 707837aafd2aa9ae update percpu data+0x93/0xc9 Call Trace: bpf prog test run raw tp+0x2dc/0x7d0 flush smp call function queue+0x1e9/0xc80 Kernel panic - not syncing: Fatal exception in interrupt

R5 is the mildest of the four, aliasing a scratch register and faulting at the store. R7 aliases RBP and would corrupt the frame pointer, R8 and R9 alias the argument registers.

Use add 2mod() so the register goes through REX.R, matching how add 2reg() places it in ModRM.reg and how emit priv frame ptr() hardcodes 0x4c for the same instruction with R9. Encodings for the non-extended registers are unchanged.

Problem showed up when trying to resurrect BPF GCC CI (selftests built with BPF GCC).

This has gone unnoticed because clang reloads the address into R1 before each per-CPU access, so the destination is never an extended register. GCC keeps several per-CPU addresses live at once, and test progs-bpf gcc panics the kernel in global percpu data/init, where the address of a .percpu variable ends up in R5.