* [email protected] <[email protected]> wrote:
> From: Kan Liang <[email protected]> > > The perf fuzzer triggers a warning which map to: > > if (WARN_ON_ONCE(idx >= ARRAY_SIZE(pt_regs_offset))) > return 0; > > The bits between XMM registers and generic registers are reserved. > But perf_reg_validate() doesn't check these bits. > > Add REG_RESERVED for reserved bits. > Check the reserved bits in perf_reg_validate(). > > Fixes: 878068ea270e ("perf/x86: Support outputting XMM registers") > Reported-by: Vince Weaver <[email protected]> > Signed-off-by: Kan Liang <[email protected]> > --- > arch/x86/kernel/perf_regs.c | 7 +++++-- > 1 file changed, 5 insertions(+), 2 deletions(-) > > diff --git a/arch/x86/kernel/perf_regs.c b/arch/x86/kernel/perf_regs.c > index 86ffe5a..3f8c1fc 100644 > --- a/arch/x86/kernel/perf_regs.c > +++ b/arch/x86/kernel/perf_regs.c > @@ -79,6 +79,9 @@ u64 perf_reg_value(struct pt_regs *regs, int idx) > return regs_get_register(regs, pt_regs_offset[idx]); > } > > +#define REG_RESERVED (((1ULL << PERF_REG_X86_XMM0) - 1) & \ > + ~((1ULL << PERF_REG_X86_MAX) - 1)) This is just randomly polluting the macro namespace with a new variant. We have PERF_REG_X86_ pattern - why not name the new one within that pattern? Thanks, Ingo
