Re: [Qemu-devel] [PATCH] qemu: json: Fix parsing of integers >= 0x8000000000000000

[Anthony Liguori]

On 05/23/2011 08:04 AM, Daniel P. Berrange wrote:
> On Fri, May 20, 2011 at 01:11:05PM -0500, Anthony Liguori wrote:
> > On 05/20/2011 01:03 PM, Richard W.M. Jones wrote:
> > > There seem to be a few unsafe uses of strto* functions.  This patch
> > > just fixes the one that affects me :-)
> >
> > Sending an integer of this size is not valid JSON.
> >
> > Your patch won't accept negative numbers, correct?
> >
> > JSON only supports int64_t.
>
> That's not really true. JSON supports arbitrarily large numbers
> &  integers.

This really blows my mind:

alert(9223372036854775807 == 9223372036854775808);

Regards,

Anthony Liguori

  It is merely the QEMU parser / object model which
> is artifically limiting them to int64_t. The core of the problem
> is with the QInt implementation in QEMU, which uses an 'int64_t'
> as its canonical form, rather than just holding a string representation
> of the number. The JSON parser should only validate that the
> data is a valid JSON number, and then pass the number as a string
> to QInt. The conversion to int_64 or other integer sizes / formats
> should be done at time of use, according to the type of data the
> command actually wants, whether int64t, int32t, int16t etc. eg the
> QInt API should look more like:
>
>    QInt *qint_from_string(const char *number);
>    QInt *qint_from_int64(int64_t val);
>    QInt *qint_from_int32(int64_t val);
>    QInt *qint_from_int16(int64_t val);
>    QInt *qint_from_uint64(uint64_t val);
>    QInt *qint_from_uint32(uint32_t val);
>    QInt *qint_from_uint16(uint16_t val);
>
>    int qint_get_int64(QInt *qi, int64t *val);
>    int qint_get_int32(QInt *qi, int32t *val);
>    int qint_get_int16(QInt *qi, int16t *val);
>    int qint_get_uint64(QInt *qi, uint64t *val);
>    int qint_get_uint32(QInt *qi, uint32t *val);
>    int qint_get_uint16(QInt *qi, uint16t *val);
>
>
> Regards,
> Daniel