On 14.09.22 18:21, Bill Wendling wrote: > The ZERO_CALL_USED_REGS feature may zero out caller-saved registers > before returning. However, alternate code may call this function without > first saving %[re]di, because the proper clobbers aren't taken into > account. > > This shows up in spurious_kernel_fault() where the "pte_offset_kernel()" > call results in this assembly code: > > .Ltmp151: > #APP > # ALT: oldnstr > .Ltmp152: > .Ltmp153: > .Ltmp154: > .section .discard.retpoline_safe,"",@progbits > .quad .Ltmp154 > .text > > callq *pv_ops+536(%rip) > > .Ltmp155: > .section .parainstructions,"a",@progbits > .p2align 3, 0x0 > .quad .Ltmp153 > .byte 67 > .byte .Ltmp155-.Ltmp153 > .short 1 > .text > .Ltmp156: > # ALT: padding > .zero (-(((.Ltmp157-.Ltmp158)-(.Ltmp156-.Ltmp152))>0))*((.Ltmp157-.Ltmp158)-(.Ltmp156-.Ltmp152)),144 > .Ltmp159: > .section .altinstructions,"a",@progbits > .Ltmp160: > .long .Ltmp152-.Ltmp160 > .Ltmp161: > .long .Ltmp158-.Ltmp161 > .short 33040 > .byte .Ltmp159-.Ltmp152 > .byte .Ltmp157-.Ltmp158 > .text > > .section .altinstr_replacement,"ax",@progbits > # ALT: replacement 1 > .Ltmp158: > movq %rdi, %rax > .Ltmp157: > .text > #NO_APP > .Ltmp162: > testb $-128, %dil > > The %dil register was zeroed out by the call to "*pv_ops+536(%rip)". > > In general, the _paravirt_ident_64() function appears like it shouldn't > have any instrumentation or other modifications applied to it. Thus just > write it in assembly to avoid having to continually modify it whenever a > new feature comes along. > > Link: https://github.com/KSPP/linux/issues/192 > Cc: Kees Cook > Cc: Nick Desaulniers > Cc: Juergen Gross > Cc: "Srivatsa S. Bhat (VMware)" > Cc: Alexey Makhalov > Cc: Thomas Gleixner > Cc: Ingo Molnar > Cc: Borislav Petkov > Cc: Dave Hansen > Cc: "H. Peter Anvin" > Cc: VMware PV-Drivers Reviewers > Cc: x86@kernel.org > Cc: virtualization@lists.linux-foundation.org > Cc: linux-kernel@vger.kernel.org > Cc: llvm@lists.linux.dev > Signed-off-by: Bill Wendling > Suggested-by: Peter Zijlstra > Reported-and-tested-by: Nathan Chancellor Reviewed-by: Juergen Gross Juergen