On 12/02/2013 12:53 AM, Magnus Fromreide wrote:
> On Fri, 2013-11-29 at 11:24 +0800, yzhu1 wrote:
>> On 11/26/2013 02:58 PM, Magnus Fromreide wrote:
>>> On Tue, 2013-11-26 at 11:05 +0800, yzhu1 wrote:
>>>> Hi, Magnus
>>>>
>>>> Follow your advice, I made the 3 new patches. Now the patches are in the
>>>> attachment. Please check them.
>>>> Thanks a lot.
>>> Are that really the new patches?
>>> It seems all the issues I spoke about are still present.
>>>
>>> /MF
>>>
>>>
>> Hi, Magnus
>>
>> To now, there are 7 patches in total.
>>
>> 0001-snmpTargetParams-supports-zero-OID.patch <----support zero OID
>> in snmpTargetParams
>> 0002-fix-param-length.patch <----fix a defect which result from params
>> length
>> 0003-conf-support.patch <----support save/restore zero OID in conf file
>> 0004-use-params-hex-string-as-index.patch <----use params
>> hex string as index in tlstmparams
>> 0005-use-nameData-hex-string-as-index.patch <----use namedata
>> hex string as index in tlstmaddr
>> 0006-replace-nameData-with-nameDate-hex-string-in-log.patch <----use hex
>> string of namedata in log
>> 0007-Use-paramsLen-to-replace-params.patch <----fix a bug since
>> params == NULL can not exactly suggest that params has no value
>>
>> After these 7 patches are applied, net-snmp can work well to now.
>> Please comment these patches and merge them.
> Thanks for the patch collection - that made it easier to review them.
>
> #6 and #5:
> * An SnmpAdminString is no more than 255 octets, so the tmp buffer
> is unnecessarily big, maxlen * 2 + 2 (0x) + 1 (\0) = 513 bytes
> is enough.
> * I still fail to understand why you prefer to write 1024 NUL's to
> the tmp buffer before reading the value instead of writing a
> single NUL to it after having read the value, and in the case of
> read_config_save_octet_value the output is a NUL-terminated
> string in any case so there is no reason for it at all there.
> #4:
> * A variant of #6 and #5. The buffer to
> read_config_save_octet_string is not big enough at only 512
> bytes.
> * Tricky and Important: Earlier the value FOO matched a string
> with value FOO but with your patch it matches the string "FOO".
> This changes the interface of the agent.
Hi, Magnus
Do you mean this
"
+ /**To get len and param**/
*line = read_config_read_octet_string(*line, (u_char **)&tmp, &len);
tmp[len] = 0;
/** xxx-rks: validate snmpadminstring? */
- if (len)
- stp->name = strdup(buf);
+ if (len){
+ char temp[SNMP_MAXBUF_SMALL];
+ read_config_save_octet_string(temp, buf, len);
+ stp->name = strdup(temp);/**use param hex value as index**/
+ }
"
will change name from ascii string to hex string ? Yes. I confirmed
this. But I checked the source code.
Maybe there is no place except this to use this string. So it will not
introduce any risk.
Thanks a lot.
Zhu Yanjun
>
> Thanks for doing this work.
>
> /MF
>
>
------------------------------------------------------------------------------
Sponsored by Intel(R) XDK
Develop, test and display web and hybrid apps with a single code base.
Download it for free now!
http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
_______________________________________________
Net-snmp-coders mailing list
Net-snmp-coders@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
