Just reviewed it, looks good, thanks.
Ethan
On Fri, Aug 12, 2011 at 15:02, Ethan Jackson <et...@nicira.com> wrote:
> That fixes it, thanks.
>
> Ethan
>
> On Fri, Aug 12, 2011 at 15:00, Ben Pfaff <b...@nicira.com> wrote:
>> Reported-by: Ethan Jackson <et...@nicira.com>
>> ---
>> Ethan, does this fix the problem that you reported?
>>
>> diff --git a/lib/ofp-parse.c b/lib/ofp-parse.c
>> index e6a6af1..5c7feb2 100644
>> --- a/lib/ofp-parse.c
>> +++ b/lib/ofp-parse.c
>> @@ -812,14 +812,20 @@ parse_field_value(struct cls_rule *rule, enum
>> field_index index,
>> static void
>> parse_reg_value(struct cls_rule *rule, int reg_idx, const char *value)
>> {
>> - uint32_t reg_value, reg_mask;
>> + /* This uses an oversized destination field (64 bits when 32 bits would
>> do)
>> + * because some sscanf() implementations truncate the range of %i
>> + * directives, so that e.g. "%"SCNi16 interprets input of "0xfedc" as a
>> + * value of 0x7fff. The other alternatives are to allow only a single
>> + * radix (e.g. decimal or hexadecimal) or to write more sophisticated
>> + * parsers. */
>> + unsigned long long int reg_value, reg_mask;
>>
>> if (!strcmp(value, "ANY") || !strcmp(value, "*")) {
>> cls_rule_set_reg_masked(rule, reg_idx, 0, 0);
>> - } else if (sscanf(value, "%"SCNi32"/%"SCNi32,
>> + } else if (sscanf(value, "%lli/%lli",
>> ®_value, ®_mask) == 2) {
>> cls_rule_set_reg_masked(rule, reg_idx, reg_value, reg_mask);
>> - } else if (sscanf(value, "%"SCNi32, ®_value)) {
>> + } else if (sscanf(value, "%lli", ®_value)) {
>> cls_rule_set_reg(rule, reg_idx, reg_value);
>> } else {
>> ovs_fatal(0, "register fields must take the form <value> "
>> --
>> 1.7.2.5
>>
>>
>
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
