This is a piece of code from psr_init.c that bombs on my system:
uns32 ps_cnt = 0;
...
ps_cnt = scandir(NCS_PSS_DEF_PSSV_ROOT_PATH, &ps_list, filter, NULL);
m_NCS_CONS_PRINTF("ps_cnt: %d\n", ps_cnt);
if (ps_cnt < 0)
perror("scandir");
if(ps_cnt > 2)
{
/* Log that there are multiple formats of persistent store */
m_LOG_PSS_STR(NCSFL_SEV_CRITICAL, "pss_check_n_reformat(): Too many
formats of persistent store found");
for(;ps_cnt > 0; ps_cnt--)
free(ps_list[ps_cnt-1]); <============= segmentation fault
scandir is a system call that returns -1 on errors!
Basically the same code fragment exist in psr_amf.c but in that code ps_cnt is
an int. Should be a function used from the two places!
The OpenSAF code is not compiled with -Wall but the code fragment above
generates no warning with -Wall. Perhaps lint would have. A more powerful
static
analysis tool would find things like this in no time.
(Somehow this bug is triggered when I lower the default stack size for all
processes. I can't understand why.)
Regards,
Hans
Murthy E-G19462 wrote:
> Hans,
>
> We tested with your stack patch. In PC environment under Debian/Redhat
> we are seeing crash in
> Pssv process. So the patch needs further investigation.
>
> Based on testing we will fine tune the stack size and include it in the
> subsequnet release or
> provide it as a patch.
>
> Regards
> Murthy
>
>> -----Original Message-----
>> From: Hans Feldt [mailto:[EMAIL PROTECTED]
>> Sent: Thursday, November 15, 2007 3:29 PM
>> To: Betham R-G19443
>> Cc: Murthy E-G19462; [email protected]
>> Subject: Re: [Users] Opensaf1.0-4 Release
>>
>> Not solved (thread stack size):
>>
>> http://list.opensaf.org/archives/users/2007-August/000179.html
>>
>> I think this one is really important. Core dumps gets too big
>> and truncated, memory is wasted & the system looks less
>> attractive to applications (no memory left).
>>
>> I spent a lot of time and effort creating this patch, why is
>> it not included?
>>
>> Why bring in new features without fixing issues in the existing ones?
>>
>> Regards,
>> Hans
>>
>>
>> Betham R-G19443 wrote:
>>> Yes, it is correct.
>>>
>>> Thanks,
>>> Ramesh.
>>>
>>>> -----Original Message-----
>>>> From: [EMAIL PROTECTED]
>>>> [mailto:[EMAIL PROTECTED] On Behalf Of Hans Feldt
>>>> Sent: Wednesday, November 14, 2007 5:47 PM
>>>> To: Murthy E-G19462
>>>> Cc: [email protected]
>>>> Subject: Re: [Users] Opensaf1.0-4 Release
>>>>
>>>> Thanks! Just need some clarification what's in and what's not.
>>>>
>>>> Not solved:
>>>> http://list.opensaf.org/archives/users/2007-August/000252.html
>>>>
>>>> Solved:
>>>> http://list.opensaf.org/archives/users/2007-September/000392.html
>>>>
>>>> Is that correct?
>>>>
>>>> Regards,
>>>> Hans
>>>>
>>>> Murthy E-G19462 wrote:
>>>>> All,
>>>>>
>>>>> You can download the opensaf1.0-4 release from below link.
>>>>> http://www.opensaf.org/download
>>>>>
>>>>> Key features of the drop.
>>>>>
>>>>> - HPI B spec support
>>>>> - Tested openSAF1.0-4 release on Debian4.0R1 (going
>> forward Debian
>>>>> major releases are going to be
>>>>> the reference platform for testing opensaf).
>>>>> - Cleanup of unwanted shared libraries, SPSv code and
>> inclusion of
>>>>> public header files in the rpm.
>>>>> - UML patch to use net-snmp source code.
>>>>>
>>>>> Detailed description of the features are available in Changelog.
>>>>> Also read the updated INSTALL file.
>>>>>
>>>>> Note: Dynamic loading of the SAF agent libraries is not
>> included in
>>>>> this drop.
>>>>> We experienced some timing issues and inconsistent behavior while
>>>>> loading these Libraries. So these changes are taken out of the
>>>>> release.
>>>>>
>>>>> Regards
>>>>> Murthy
>>>>> _______________________________________________
>>>>> Users mailing list
>>>>> [email protected]
>>>>> http://list.opensaf.org/maillist/listinfo/users
>>>>>
>>>> _______________________________________________
>>>> Users mailing list
>>>> [email protected]
>>>> http://list.opensaf.org/maillist/listinfo/users
>>>>
>>> _______________________________________________
>>> Users mailing list
>>> [email protected]
>>> http://list.opensaf.org/maillist/listinfo/users
>>>
>>
> _______________________________________________
> Users mailing list
> [email protected]
> http://list.opensaf.org/maillist/listinfo/users
>
_______________________________________________
Users mailing list
[email protected]
http://list.opensaf.org/maillist/listinfo/users
