Re: valgrind issues on Fedora 28 - Mailing list pgsql-hackers

From Tomas Vondra
Subject Re: valgrind issues on Fedora 28
Date
Msg-id 21c39c42-061d-e31e-143b-ccda1a19f8e1@2ndquadrant.com
Whole thread Raw
In response to Re: valgrind issues on Fedora 28  (Andrew Dunstan <andrew.dunstan@2ndquadrant.com>)
List pgsql-hackers
On 12/15/18 12:32 AM, Andrew Dunstan wrote:
> 
> On 12/14/18 4:35 PM, Tomas Vondra wrote:
>> On 12/14/18 4:58 PM, Tom Lane wrote:
>>> ...
>>>
>>> In general, I'm not particularly on board with our valgrind.supp
>>> carrying suppressions for code outside our own code base: I think
>>> that's assuming WAY too much about which version of what is installed
>>> on a particular box.
>>>
>> Fair point.
>>
>>> Maybe we could do something to make it simpler to have custom
>>> suppressions?  Not sure what, though.
>>>
>> I was thinking that perhaps we could allows specifying path to extra
>> suppressions and pass that to valgrind.
>>
>> But we don't actually invoke valgrind, that's something people do on
>> their own anyway - so we don't have anywhere to pass the path to. And
>> whoever invokes valgrind can simply stick it directly into the command
>> they're using (as it allows specifying multiple --suppressions=<file>
>> options). Or perhaps just put it into ~/.valgrindrc.
>>
>> So perhaps we should simply revert that commit and be done with it.
>>
>> One place that will need to solve it is buildfarm client, but it could
>> pick either of the options I mentioned.
>>
>>
> 
> The buildfarm client has a parameter in the config file for valgrind
> options. All you would have to do is an an extra --suppressions setting
> in there.
> 

OK, makes sense. I'll revert the commit tomorrow.


regards

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services


pgsql-hackers by date:

Previous
From: Andres Freund
Date:
Subject: Re: reducing the footprint of ScanKeyword (was Re: Large writablevariables)
Next
From: Andres Freund
Date:
Subject: Displaying and dumping of table access methods