Re: [Qemu-devel] [PATCH] sd: Fix out-of-bounds assertions

[Lidong Chen <lidong.chen@oracle.com>]

Hi,

Thank you all for the reviews and comments! Since the fixes in sd.c have 
gone through the review, I can fix the issue in util/main-loop.c 
(mentioned in the reviews of Peter and Liam) in a separate patch.

Thanks,

Lidong

On 4/9/2019 3:39 AM, Liam Merwick wrote:
> On 09/04/2019 06:51, Markus Armbruster wrote:
>> Lidong Chen <lidong.chen@oracle.com> writes:
>>
>>> Due to an off-by-one error, the assert statements allow an
>>> out-of-bounds array access.
>>>
>>> Signed-off-by: Lidong Chen <lidong.chen@oracle.com>
>
>
> Reviewed-by: Liam Merwick <liam.merwick@oracle.com>
>
>
>>> ---
>>>   hw/sd/sd.c | 4 ++--
>>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/hw/sd/sd.c b/hw/sd/sd.c
>>> index aaab15f..818f86c 100644
>>> --- a/hw/sd/sd.c
>>> +++ b/hw/sd/sd.c
>>> @@ -144,7 +144,7 @@ static const char *sd_state_name(enum 
>>> SDCardStates state)
>>>       if (state == sd_inactive_state) {
>>>           return "inactive";
>>>       }
>>> -    assert(state <= ARRAY_SIZE(state_name));
>>> +    assert(state < ARRAY_SIZE(state_name));
>>>       return state_name[state];
>>>   }
>>>   @@ -165,7 +165,7 @@ static const char 
>>> *sd_response_name(sd_rsp_type_t rsp)
>>>       if (rsp == sd_r1b) {
>>>           rsp = sd_r1;
>>>       }
>>> -    assert(rsp <= ARRAY_SIZE(response_name));
>>> +    assert(rsp < ARRAY_SIZE(response_name));
>>>       return response_name[rsp];
>>>   }
>> This is the second fix for this bug pattern in a fortnight. Where's
>> one, there are more:
>
>
> As Lidong mentioned, an internal static analysis tool (Parfait) was 
> used to catch these. Not every arch/board is compiled but I had 
> eyeballed the code of most interest to me and they seemed fine (e.g. 
> for array accesses, the subsequent loops used a less-than check)
>
> However, this WIN32 code in util/main-loop.c seems wrong.
>
> 425     g_assert(n_poll_fds <= ARRAY_SIZE(poll_fds));
> 426
> 427     for (i = 0; i < w->num; i++) {
> 428         poll_fds[n_poll_fds + i].fd = (DWORD_PTR)w->events[i];
> 429         poll_fds[n_poll_fds + i].events = G_IO_IN;
> 430     }
>
> Seems like this should be:
>
> g_assert(n_poll_fds + w->num <= ARRAY_SIZE(poll_fds));
>
> Otherwise, the rest seem fine.
>
> Regards,
>
> Liam
>
>
>>
>> $ git-grep '<= ARRAY_SIZE'
>> hw/intc/arm_gicv3_cpuif.c:    assert(aprmax <= 
>> ARRAY_SIZE(cs->ich_apr[0]));
>> hw/intc/arm_gicv3_cpuif.c:    assert(aprmax <= 
>> ARRAY_SIZE(cs->ich_apr[0]));
>> hw/net/stellaris_enet.c:        if (s->tx_fifo_len + 4 <= 
>> ARRAY_SIZE(s->tx_fifo)) {
>> hw/sd/pxa2xx_mmci.c:        && s->tx_len <= ARRAY_SIZE(s->tx_fifo)
>> hw/sd/pxa2xx_mmci.c:        && s->rx_len <= ARRAY_SIZE(s->rx_fifo)
>> hw/sd/pxa2xx_mmci.c:        && s->resp_len <= ARRAY_SIZE(s->resp_fifo);
>> hw/sd/sd.c:    assert(state <= ARRAY_SIZE(state_name));
>> hw/sd/sd.c:    assert(rsp <= ARRAY_SIZE(response_name));
>> hw/usb/hcd-xhci.c:    assert(n <= ARRAY_SIZE(tmp));
>> target/mips/op_helper.c:    if (base_reglist > 0 && base_reglist <= 
>> ARRAY_SIZE (multiple_regs)) {
>> target/mips/op_helper.c:    if (base_reglist > 0 && base_reglist <= 
>> ARRAY_SIZE (multiple_regs)) {
>> target/mips/op_helper.c:    if (base_reglist > 0 && base_reglist <= 
>> ARRAY_SIZE (multiple_regs)) {
>> target/mips/op_helper.c:    if (base_reglist > 0 && base_reglist <= 
>> ARRAY_SIZE (multiple_regs)) {
>> target/ppc/kvm.c:           <= ARRAY_SIZE(hw_debug_points));
>> target/ppc/kvm.c:           <= ARRAY_SIZE(hw_debug_points));
>> target/ppc/kvm.c:    assert((nb_hw_breakpoint + nb_hw_watchpoint) <= 
>> ARRAY_SIZE(dbg->arch.bp));
>> tcg/tcg.c:    tcg_debug_assert(pi <= ARRAY_SIZE(op->args));
>> util/main-loop.c:    g_assert(n_poll_fds <= ARRAY_SIZE(poll_fds));
>> util/module.c:    assert(n_dirs <= ARRAY_SIZE(dirs));
>>
>> Lidong Chen, would you like to have a look at these?
>>
>> Cc'ing maintainers to help with further investigation.
>>
>