develooper Front page | perl.perl5.porters | Postings from April 2019

Re: [perl #134061] panic: pad_findmy_pvn illegal flag bits0x20000000

Thread Previous | Thread Next
From:
Karl Williamson
Date:
April 26, 2019 22:25
Subject:
Re: [perl #134061] panic: pad_findmy_pvn illegal flag bits0x20000000
Message ID:
4e9245e3-d277-cb3c-0206-d97d504b6228@khwilliamson.com
On 4/26/19 7:54 AM, James E Keenan via RT wrote:
> On Fri, 26 Apr 2019 13:49:01 GMT, jkeenan wrote:
>> On Fri, 26 Apr 2019 13:06:15 GMT, arc wrote:
>>> Sergey Aleynikov (via RT) <perlbug-followup@perl.org> wrote:
>>>> use utf8; eval "sort \x{100}%"; warn $@;
>>>> panic: pad_findmy_pvn illegal flag bits 0x20000000 at (eval 1) line
>>>> 1.
>>>
>>> Patch attached.
>>>
>>> Sawyer, is this OK to merge before 5.30?
>>>
>>
>> My two cents:  At this point in the dev cycle we should not be making
>> changes in *.c or *.h files.  The potential ramifications of changes
>> to the guts are vast.  Our ability to thoroughly assess the impact in
>> the time remaining to production release is very limited.
>>
>> That's why we have code freezes.
> 
> I would like to add my thanks to Sergey Aleynikov for the tremendous work he has done during this development cycle at identifying weaknesses in our source code with his fuzzing work.  See, for example, all the items added to our "fuzzer-detected issues" META ticket (https://rt.perl.org/Ticket/Display.html?id=126546) which come from his work.
> 
> Nonetheless, we have generally not been treating these bug reports as blockers for perl-5.30, whether or not we have been able to write patches for them.  If we're not treating this particular RT as a blocker, then we should not be applying a patch for it during code freeze.

It is my experience that these reports from Sergey, once investigated, 
show real underlying problems that should be fixed; they don't tend to 
be just edge cases that won't ever come up in real code, but symptomatic 
of real code potential issues once reduced to simpler test cases

The ones that bisect to this development cycle might actually ought to 
be blockers, depending on the outcome of an investigation.

But my point is that they really should be investigated to be sure. 
I've looked so far at all but two of the ones for the areas I have 
expertise in.

> 
> Thank you very much.
> 

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About