Re: [PATCH] kasan: fix the missing underflow in memmove and memcpy with CONFIG_KASAN_GENERIC=y

From: Dmitry Vyukov <dvyukov@google.com>
To: Walter Wu <walter-zh.wu@mediatek.com>
Cc: Marc Gonzalez <marc.w.gonzalez@free.fr>,
	LKML <linux-kernel@vger.kernel.org>,
	kasan-dev <kasan-dev@googlegroups.com>,
	Alexander Potapenko <glider@google.com>,
	Matthias Brugger <matthias.bgg@gmail.com>,
	Andrey Ryabinin <aryabinin@virtuozzo.com>,
	Linux ARM <linux-arm-kernel@lists.infradead.org>
Subject: Re: [PATCH] kasan: fix the missing underflow in memmove and memcpy with CONFIG_KASAN_GENERIC=y
Date: Tue, 1 Oct 2019 05:01:58 +0200	[thread overview]
Message-ID: <CACT4Y+b3NPemYwJJsD_oC0vde5Ybz1qDNWb=cFu2HpOTMrGSnQ@mail.gmail.com> (raw)
In-Reply-To: <1569897400.17361.27.camel@mtksdccf07>

On Tue, Oct 1, 2019 at 4:36 AM Walter Wu <walter-zh.wu@mediatek.com> wrote:
>
> On Mon, 2019-09-30 at 10:57 +0200, Marc Gonzalez wrote:
> > On 30/09/2019 06:36, Walter Wu wrote:
> >
> > >  bool check_memory_region(unsigned long addr, size_t size, bool write,
> > >                                 unsigned long ret_ip)
> > >  {
> > > +       if (long(size) < 0) {
> > > +               kasan_report_invalid_size(src, dest, len, _RET_IP_);
> > > +               return false;
> > > +       }
> > > +
> > >         return check_memory_region_inline(addr, size, write, ret_ip);
> > >  }
> >
> > Is it expected that memcpy/memmove may sometimes (incorrectly) be passed
> > a negative value? (It would indeed turn up as a "large" size_t)
> >
> > IMO, casting to long is suspicious.
> >
> > There seem to be some two implicit assumptions.
> >
> > 1) size >= ULONG_MAX/2 is invalid input
> > 2) casting a size >= ULONG_MAX/2 to long yields a negative value
> >
> > 1) seems reasonable because we can't copy more than half of memory to
> > the other half of memory. I suppose the constraint could be even tighter,
> > but it's not clear where to draw the line, especially when considering
> > 32b vs 64b arches.
> >
> > 2) is implementation-defined, and gcc works "as expected" (clang too
> > probably) https://gcc.gnu.org/onlinedocs/gcc/Integers-implementation.html
> >
> > A comment might be warranted to explain the rationale.
> > Regards.
>
> Thanks for your suggestion.
> Yes, It is passed a negative value issue in memcpy/memmove/memset.
> Our current idea should be assumption 1 and only consider 64b arch,
> because KASAN only supports 64b. In fact, we really can't use so much
> memory in 64b arch. so assumption 1 make sense.

Note there are arm KASAN patches floating around, so we should not
make assumptions about 64-bit arch.

But there seems to be a number of such casts already:

$ find -name "*.c" -exec egrep "\(long\).* < 0" {} \; -print
    } else if ((long) delta < 0) {
./kernel/time/timer.c
    if ((long)state < 0)
./drivers/thermal/thermal_sysfs.c
    if ((long)delay < 0)
./drivers/infiniband/core/addr.c
    if ((long)tmo < 0)
./drivers/net/wireless/st/cw1200/pm.c
    if (pos < 0 || (long) pos != pos || (ssize_t) count < 0)
./sound/core/info.c
        if ((long)hwrpb->sys_type < 0) {
./arch/alpha/kernel/setup.c
    if ((long)m->driver_data < 0)
./arch/x86/kernel/apic/apic.c
            if ((long) size < 0L)
    if ((long)addr < 0L) {
./arch/sparc/mm/init_64.c
    if ((long)lpid < 0)
./arch/powerpc/kvm/book3s_hv.c
            if ((long)regs->regs[insn.mm_i_format.rs] < 0)
            if ((long)regs->regs[insn.i_format.rs] < 0) {
            if ((long)regs->regs[insn.i_format.rs] < 0) {
./arch/mips/kernel/branch.c
            if ((long)arch->gprs[insn.i_format.rs] < 0)
            if ((long)arch->gprs[insn.i_format.rs] < 0)
./arch/mips/kvm/emulate.c
            if ((long)regs->regs[insn.i_format.rs] < 0)
./arch/mips/math-emu/cp1emu.c
        if ((int32_t)(long)prom_vec < 0) {
./arch/mips/sibyte/common/cfe.c
    if (msgsz > ns->msg_ctlmax || (long) msgsz < 0 || msqid < 0)
    if (msqid < 0 || (long) bufsz < 0)
./ipc/msg.c
    if ((long)x < 0)
./mm/page-writeback.c
    if ((long)(next - val) < 0) {
./mm/memcontrol.c