Le 26/07/2018 à 19:58, Alex Bennée a écrit :
>
> Laurent Vivier <laur...@vivier.eu> writes:
>
>> Le 26/07/2018 à 15:29, Alex Bennée a écrit:
>>> I've slightly re-organised the check to more closely match the
>>> sequence that the kernel uses in do_mmap().
>>>
>>> Signed-off-by: Alex Bennée <alex.ben...@linaro.org>
>>> Cc: umarcor <1783...@bugs.launchpad.net>
>>> ---
>>> linux-user/mmap.c | 14 +++++++++++---
>>> 1 file changed, 11 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/linux-user/mmap.c b/linux-user/mmap.c
>>> index d0c50e4888..3ef69fa2d0 100644
>>> --- a/linux-user/mmap.c
>>> +++ b/linux-user/mmap.c
>>> @@ -391,14 +391,22 @@ abi_long target_mmap(abi_ulong start, abi_ulong len,
>>> int prot,
>>> }
>>> #endif
>>>
>>> - if (offset & ~TARGET_PAGE_MASK) {
>>> + if (!len) {
>>> errno = EINVAL;
>>> goto fail;
>>> }
>>>
>>> len = TARGET_PAGE_ALIGN(len);
>>> - if (len == 0)
>>> - goto the_end;
>>> + if (!len) {
>>> + errno = EINVAL;
>>> + goto fail;
>>> + }
>>
>> Why do you check twice len?
>> TARGET_PAGE_ALIGN() rounds up the value, so if it was not 0, it cannot
>> be now.
>
> I was trying to follow the kernel style but I realise TARGET_PAGE_ALIGN
> might be a different test compared to the kernel's PAGE_ALIGN(len) for
> overflows:
...
> /* Careful about overflows.. */
> len = PAGE_ALIGN(len);
> if (!len)
> return -ENOMEM;
>
OK, so keep it but you should use ENOMEM, not EINVAL (and add a comment :) )
Thanks,
Laurent
